Project Description | Parent Description | Capabilities
This is the executive skills toy temporary system.
Note that our
toy systems tend to do most of the work.
- Emacs-Client-7302, eval (find-file "/var/lib/myfrdcsa/codebases/minor/normal-form/to.do")
- Fix packager, to be able to do `packager -y ./cff.tar.gz`
- How could one ever be bored when there is AI to do.
- We can do NLP on our Lisp and Perl function names that read like sentences.
- Add a search function to elog, then check whether kmax/to.do ever existed, or if it was deleted from git somehow.
- Have a filesystem listening process that listens for new .do and .notes files and automatically adds them to the list of files for our todo system.
- Write something for when.do that does textual entailment to determine when when conditions could be being satisfied and recommends the follow up.
Also develop a menu, for free life planner, that allows you to select the when conditions from a menu, and includes the RTE system?
- Create an index of which systems can do which functions.
- Periodically review to.do files and try to determine if we have accomplished the items mentioned.
- do better tracking of where things are asserted.
- Write system to process honey-do notes from emails.
- Have to-do mode highlight questions and items that are wrapped by a question mark.
- Add to C-cC-eC-ek and x to only do the FCMS stuff if it's installed.
- Investigate what added the has-formalization assertions to Org::FRDCSA::verber::PSEx2::do.
- When the AI is working, we will want to do regression testing of it.
- Just thought of a system.
Knowing all the different things that we use, we can surjectively search the cso for everything that has to do with important topics.
For instance, we could search elpa and melpa for Prolog, etc. And make a ranking over all packages.
- Write a function that reminds us to get back to certain topics, so we don't forget the new language we have learned.
i.e. like here: /var/lib/myfrdcsa/codebases/minor/what-to-do/what-to-do.flr
- Log our dribble file and emacs recording sessions through memax, and then apply deep learning to them to antipate what we should do at each given point in an emacs session.
See if we can record emacs sessions that are invokes with 'emacs' instead of 'emacs -nw'.
- Write a function to search all to.do files.
- do something about my propensity to forget that C-ccsc which runs cmh-comment-for-referent is a thing-at-point using instead of the top of the stack.
Maybe make some mode that you can shift between.
- write a function that uses architect to do a choose from emacs between systems related to the system in the cwd directory
- index solution answers into to.do for quick answers to questions
- Add some more detail to do-mode's 'in progress'
- The free-life-planner should take into account facts about other people's lives, so as to understand how to work with them, and plan with them.
Can also use a multiagent capability to do so
- Should probably learn how to do things with DBIx::Class before I write my freekbs2 source, so I can take advantage of that infrastructure.
- do 1 hour + existing commitments on Encryption/Decryption and Prime Toggle Flags Bit Encodings implementation.
Make perl model, make available through picform.org.
- Add a possible or necessary modal operator to normal-form do-todo-mode
- Get IAEC VM set up, control IAEC from within Emacs, have it have the ability to open files of interest and annotate them.
Kind of like do::ListProcessor4, only all from within Emacs.
Allow the user to control it as well.
Monitor the state of Emacs, as input to the planning process.
- Process our old to.do files to get an idea for concepts that we have to be sensistive to.
build ontology of affairs that way.
- Go through all of our do-todo mode entries and automatically recognize questions.
- Resolve the issue of files being not on the machine they need to be.
Probably do auto git commit and pull and such.
- do a study of all the files which are not showing up in 'boss updatedb', i.e. things like frdcsa/emacs/ and maybe things like cyc-mode/hybrid/..
- Fix clear up to when you click on a word pronounce for someone.
Make font resizeable - and integrate TTS and literacy efforts.
Install on several computers that we own, and use to teach children literacy.
Have it correlate and do flashcard like training of words and sounds.
Bob's our uncle.
- Set up searching of dated/daily todos as part of do-todo search capability
- Syncronize the color schemes of SPSE2 and Normal-Form/do-Todo-Mode etc.
- Create a new capability to say that a given goal is clear in the sense that nothing is obstructing it and it can be worked on immediately - do analysis through Prolog or Cyc or PrologCYC
- Develop a system for, in addition to the kmax-object-manipulation-system, classical planning for it's use for people who do not yet know the functions.
- Fix the projects radar dir dir completion, or figure out how to do it.
- Analyze 'do tell!' with NLU.
- Add a font highlighting for 'skipped' to do-todo-mode
- The ideal way to do contexts in KBS2::Reasoner::Prolog is with Prolog modules.
- Set up a system with kbfs using Prolog which understands whether a directory has been updated or not.
And reminds us, when doing the kmax-update-frdcsa-repositories to do all the different ones that need updating.
- Be sure to pace my learning of new languages with repeated work on existing languages.
For instance, I should certainly attempt to keep up my abilities with different programming languages, like my abilities with Perl, by repeatedly refreshing myself on different topics.
To do this involves actively refreshing my memory - the task should ideally be guided by a program, either study, etc. Empirically quantify the decay of knowledge by computing a difficulty ordering over the set of concepts (relative to the user), and then testing questions for each concept as needed.
- do learning of what additions were added over each revision.
- We should definitely look into whether "(5) Otherwise, is this formula something you'd prove by induction?
If you can't simplify it, it may be because you ``need this fact to prove this fact,'' in which case, induction is the right thing to do." means that circular dependencies are related to induction.
- In order to do ACL2-Doc, we really need a better "I have seen/skimmed/read this page before" capability.
We should add a seen capability, keybindings for have read, and a modeline thing which tells us if it is read or not.
- Can write notes for to.do in lisp, especially now that we know about `, and ,@.
- I wonder if we could use unilang as the record for IAEC, and have it fill out things and observations, just as I do, with provenance information provided.
Maybe can even train on this.
- One possible thing to do to get the auto-programmer and the IAEC and system-implementor working, would be to have it try out every different possibility of mereological type such as when KNext is saying that something may be a certain way, but it is unclear which way it is. Could also order them by probability, such as by using WSD results.
This way, the program can direct it's evolution.
Then we can simulate everything from the structure describe by the text, as input to the program for trying to evolve programs to solve problems.
- Something we can do for packages that we might to make for people of software that isn't released under open source license is have it download from internet archive, so we are not hosting it.
- Look into object databases for common lisp and emacs lisp as a way to do sayer2.
- I really do not understand programming languages, what they're capable of.
- Develop a context mechanism for assessing meaning, add to NLU.
For instance, the expression "Chris might know someone" had a lot of context, that we were looking at the time for someone to do tech support, things like that.
all these assertions could be made in a KB and used as contextual knowledge for the statement.
Build this manually at first, followed by automatically.
- do some kind of evolutionary search and rating scheme for AIE.
- Use kbfs, have it learn when to automatically attribute facts to files based on certain correlations.
Also use Sayer and Sayer-learner, and Thinker to learn when these facts apply to the files.
Then, for instance, do automatic classification of text files into subject headings.
Ultimately organize all of the research papers and documents I have into a coherent, cohesive whole.
- Fix FRDCSA Cyc access to not reconnect all the time, to do proper error handling, and to , hrm forget
- do like a paperless office type frontend for academician which tells you which pages have been read, of which texts.
- do query expansion searches for other materials on AI and stuff, and also start mapping out the internet
- do social network analysis.
- Make a new mode for do todo mode that allows you to talk about the epistemic status of something, such as whether it is currently under review, or decided upon.
Motivating example: (daily (go on a walk)), wanted to say, well wait a minute, maybe not.
- For js-rapid-response, have it do regex-based keyword based analysis of which ones to respond to including also using the classifier I wrote.
I this way get better performance out of selecting items to write with.
Get the classification probability if possible and only send if very high.
- I wonder if we should take unilang entries like 'Also, do this' and if there is no agent for them, log them anyway as commands by the user - after trying first to autovivify.
- You can do knowledge acquisition from corpora by asking questions of the data for which the answer is known, using a system like Nutcracker, LogAnswer, or Groundhog, except that you find situations where the answer is present, and you look to see which literals in the logic form were unable to be proven, and add those literals to the knownledge-base - or something close to this.
- Possibly have FRDCSA index all .do and .notes files and index the answers to all (solution (X) (Y)) statements, and make that searchable.
- Write a function for tab-completing inserting arbitrary text into nest to.do style things, like (yes ())
- Make it so you can sync a buffer with the phone software, so you can edit in a to.do file stuff that will be executed with the interactive execution manager.
- Use kbfs to tag which files are known to be software archives, versus which are known to be data sets, versus which are not known to be either, versus which we know nothing about at all.
This database could be bootstrapped by iterating over /var/lib/myfrdcsa/codebases/external versus /var/lib/myfrdcsa/codebases/datasets, and then tagging several of neither kind.
Then extract features of those data points, using Sayer, etc, to do machine learning to automatically classify items as being software collections or not.
- do an analysis of what is wrong with the current kbfs system, as to why we need to reengineer it.
- For kmax-object-manipulation-system, the thing at point, and like for instance, in w3m, a URL under point, should automatically hint at certain actions.
For instance, the URL will have different actions possible to undertake on it.
Perhaps what we need to do is use EIEIO and duck typing or something.
- Figure out why I am fatigued, and what I can do to get more done each day.
- do regression testing to see which domains that are handled by LPG-td work with SGPLAN-6.1
- Get used to the idea that although we want to make a flexible system, I am still going to have to do a ton of programming.
What exactly to program is difficult though, because we might get bogged down in a particular system, like verber - and not have a correct picture of the relation of that particular project to the overall goal.
What is needed is to locate and package existing useful systems, write some code, and rerelease a tarball and package.
- do taint analysis with Sayer information.
- Come up with a system of saying things like something is a research paper, how do we do that
- Add to has-OCR result in paperless office the engine name that did it if it doesn't already do that.
- Make a flora2 flr representation of the 2.do agent notes and others on logics, i.e. have metadata about the various predicates stored.
use that to compute the semantics.
- Add something to the rayon-load-windows, and verber similar thing to choose which file to load from among several options (such as what-to-do.flr and frdcsa.flr).
- rewrite the crontab todo stuff into flora2 and add additional facets such as accumulating warnings, etc. come up with specialist logic, especially, what happens if you don't do it, in your particular situation.
- New years resolutions.
Aggregate all resources of the FRDCSA.
to.do, .notes, daily todos, unilang logs, etc, into a unified information access system across the FRDCSA, add search abilities
- Develop a logic representing the meaning of our .do and .notes files.
- Develop an ontology of the kinds of entries in our .do and .notes files, and have operations that operate on the different types.
For instance, some tasks should be able to be deferred until later, like a later day, etc
- Set it up so that we do searches for different combinations of technologies that could be integrated, for instance ATEL + Flora2, etc. trying to dig up documents that relate them.
- kbfs should have the capability to do filesystem archaeology, basically, work out movements and large scale actions that happened to different files and so forth, to explain the present location, etc.
- Add an emacs function for starting a new ~/.config/frdcsa/todo/.do file.
- Add stuff to .do mode, (actually create .do, .notes modes, etc) to be able to highlight important items in red or something.
- Apparently unilang or freekbs2 do not support \f character.
- Read up on Emacs version control integration, or ask Jess how to do it.
- Write a function for making the Emacs region become a .do entry
- Make what-to-do prioritize and take into consideration situations.
- do project management that we learned at the last job.
- Schedule major system work and do it then, such as the backup.
- Create the ability to do suppositional reasoning like stuff in SPSE2 by fixing some constraints and then sending to the planner or inferencer to fill out the remaining details, or decide a question about the feasibility of something.
Make a gametree like view.
- Add to the pse-assigned to the ability to offer to someone that they work on a task, and if they accept, the ability to pay them the agreed upon price, and state the time duration on the offer.
Have a pool of workers who offer to do the work.
- have C-csK do a syntax check first before clearing the existing contents.
- Have spse2 automatically sync with to.do files, with unilang entries.
- Setup a system for referring to the entry in a to.do or .notes files, but that you can assert things about it.
so basically like C-csl.
- do housekeeping, clean my directories and sort things appropriately.
- do this later.
- We can do all of this and more with this new system.
- Add to do.el the ability to decompose a task
- Add failed to do.el
- Develop a planner to determine what is optimal to do.
- For everything I'd like someone else to do, see what we could do of equal value.
- Some things I could do: practice programming with Java environments that have performatives already put in.
- Develop a question which ask what I can do with the nugget at point.
- Develop a system based on affordances which asks what we might want to do with a given document.
For instance, the information extraction tool can be used to process html pages.
Need a logic to specify this situation.
- Develop a language for rapid importation of goals into SPSE2, so that for complex structures we can do it rapidly.
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Sep 22 06:00:00 CDT 2010
|("due-date-for-entry" "150200" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Sep 15 06:00:00 CDT 2010
|("due-date-for-entry" "149574" "3 hours")|
- KBS, MySQL:freekbs:default assert ("due-date-for-entry" "select * from messages where Contents like '%Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Sep 1 06:00:00 CDT 2010%'
148685" "3 hours")
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Sep 1 06:00:00 CDT 2010
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Aug 25 06:00:00 CDT 2010
|("due-date-for-entry" "148241" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Aug 18 06:00:00 CDT 2010
|("due-date-for-entry" "147935" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Aug 11 06:00:00 CDT 2010
|("due-date-for-entry" "147372" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Aug 4 06:00:00 CDT 2010
|("due-date-for-entry" "146073" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Fri Jul 30 19:22:21 CDT 2010
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Jul 21 06:00:00 CDT 2010
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Jul 14 06:00:00 CDT 2010
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Jul 7 06:00:00 CDT 2010
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Jun 30 06:00:00 CDT 2010
|("due-date-for-entry" "142916" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Jun 23 06:00:00 CDT 2010
|("due-date-for-entry" "141254" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Jun 16 06:00:02 CDT 2010
|("due-date-for-entry" "140651" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Jun 9 06:00:00 CDT 2010
|("due-date-for-entry" "140105" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Jun 2 06:00:00 CDT 2010
|("due-date-for-entry" "138465" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed May 26 06:00:00 CDT 2010
|("due-date-for-entry" "137975" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed May 12 06:00:00 CDT 2010
|("due-date-for-entry" "136464" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed May 5 06:00:00 CDT 2010
|("due-date-for-entry" "135377" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Apr 28 06:00:00 CDT 2010
|("due-date-for-entry" "133925" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Apr 21 06:00:00 CDT 2010
|("due-date-for-entry" "132391" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Apr 14 06:00:00 CDT 2010
|("due-date-for-entry" "131135" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Apr 7 06:00:00 CDT 2010
|("due-date-for-entry" "129524" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Mar 10 06:00:00 CST 2010
|("due-date-for-entry" "127301" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Mar 3 06:00:00 CST 2010
|("due-date-for-entry" "127264" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Jan 13 06:00:00 CST 2010
|("due-date-for-entry" "125600" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Dec 23 06:00:00 CST 2009
|("due-date-for-entry" "125173" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Nov 25 06:00:00 CST 2009
|("due-date-for-entry" "122100" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Oct 28 06:00:01 CDT 2009
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Oct 21 06:00:00 CDT 2009
|("due-date-for-entry" "115401" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Sep 30 06:00:00 CDT 2009
|("due-date-for-entry" "114455" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Aug 19 06:00:00 CDT 2009
|("due-date-for-entry" "113535" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Jul 8 06:00:00 CDT 2009
|("due-date-for-entry" "113214" "3 hours")|
- Setup the ability to do per channel keyword search (e.g. Chicago on gnucash)
- Every now and then do a long ball-in-court thing, looking back very deeply.
also have it reorder by date...
- Need the ability to do things like search for the phone numbers of posi members in Chicagoland with one search.
- do a conference call using teamspeak for posi.
- do a tutorial on how to avoid scams for older library patrons.
- do something similar to Joomla.
- Figure out what to do for radar-web-search if touch fails, probably just add a timeout
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Mar 4 06:00:00 CST 2009
|("due-date-for-entry" "107857" "3 hours")|
- do an analysis of why Tarawatch and CST are fighting
- Make sure that the business we do business with all have good models, like perhaps diamondcard.us, etc.
- do a video for diamondcard.us
- Write a scheduler to remind me to do things on certain dates - include taxes.
|("due-date-for-entry" "106714" "3 days")|
- do something about all the files in my home dir.
- do a semantic map of goals and then process writings based on goal priority
|("comment" "106641" "for priority classification")|
- do get my resume done.
As for importance of a given topic, compute it by NLP over the unilang entries.
For instance, if resume is appearing more frequently than it should statistically.
- do a survey to figure out what skills everyone has - like Emacs... and then send those people stuff.
Implement a decision tree to determine skills...
- Make sure bbdb can do european numbers.
- Fix all the strange data that's just sitting there in /s1/temp/frdcsa.org and getting all confused on other systems.
do more data handling.
- do an analysis of the people on sourceforge.
- Chris would like to do electronics manufacturing
- Need to develop a list of places that we can do consulting work, as well as products that we can sell, etc.
|("assigned-to" "105907" "Justin Coslor")|
|("assigned-to" "105907" "Sara Masters")|
|("assigned-to" "105907" "Chris Lampkin")|
- Generate an enormous resume based on all my AI tools - set a day aside to do this.
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Feb 11 06:00:00 CST 2009
|("due-date-for-entry" "105776" "3 hours")|
- Need to make a glossary of all my demos and tests (demonstrations), so that I can quickly find how to do certain things.
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Feb 4 06:00:00 CST 2009
|("due-date-for-entry" "105394" "3 hours")|
- Get more people using Ekiga so we can do the ekiga conference call.
- Email him a list of capabilities that my text based AI tools can do (free open source)
|("assigned-by" "104978" "Justin Coslor")|
- Write a system that helps me to think about things that I can do.
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Jan 21 06:00:00 CST 2009
|("due-date-for-entry" "104701" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Jan 14 06:00:00 CST 2009
|("due-date-for-entry" "104473" "3 hours")|
- Are you test you are testing this is a test sleep modeThis is anFigure out what is wrong with the speech recognition.
I figured it out as nothing to do with the speech recognition
- Figure out a way to have the speech recognition bind to a particular window so that it can continue to do all the work in that window will open elsewhere.
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Jan 7 06:00:00 CST 2009
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Dec 31 06:00:00 CST 2008
- Have clear parse paragraphs too, and do other kinds of analyses...
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Dec 24 06:00:00 CST 2008
|("due-date-for-entry" "103811" "3 hours")|
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Dec 17 06:00:00 CST 2008
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Dec 10 06:00:00 CST 2008
|("due-date-for-entry" "102069" "3 hours")|
- do benchmarks on freekbs.
- Agenda: <do all-day-task> <do an all day programming task (list is in manager Todo doc)> Wed Dec 3 06:00:00 CST 2008
|("due-date-for-entry" "99534" "3 hours")|
- code-monkey should do more code generation.
Right now boss framework is a very simple instance of that.
- do not depend on things like the previous entry.
- How do we detect inconsistencies in the requirements?
probably by mapping things to ACE>
- Tom Murphy said he was interested in a place where people go to do whatever they want.
- Get akahige or whatever to do real social therapy (like get rid of seeking approval, saying certain).
- Lessons learned: Step back when debugging nasty systems.
If it doesn't look pretyt, try to avoid it.
Chances are it's not going to work.
Don't do little calculations unless absolutely necessary.
This advice should be computable based on measureable proeprties of the programming situation.
In other words, if we were to actaully analyze w/ some system, the code being written, we should be able to see whether or not the solution is elegant and warn the user.
- Now all we have to do is fix that data and worry about complexity/efficiency issues with corpus doing autoclassification.
We also have to add more elegant classification than bayes, as well as talking amongst agents to support better categorization and command execution from our notes.
- fweb in dynamic content generation should do coloring dynamic to user.
- Eventually should get meal planner working.
I think for starters, only get pgourmet going strong.
I could do that as soon as pverber was operating nominally.
There is still a lot of work to be done on manager's end to get it working.
- Ensure that we look at all logs and files eventually.
Model which files we have seen when!
This ties in perfectly with this log system.
That is what it is doing.
So from that we may update timing information on all files.
Maybe we do want to do an SQL db.
- Develop terminology management system that creates giant definition of all terms we are using (Justin's thesauri).
I do not need to wait for him to work on that.
- Design a folding chair (I have no money, but when I do, I better have priorities mapped out).
We will always have storage for whatever I build.
- I find that writing a first generation simplistic system gets the basic functionality in place without all the difficulties.
It would then be fitting to do an entire rewrite.
boss ought to keep track of that kind of information.
- Should use WSD on corpus entries and also run ExtractAbbrev.java on them to get abbreviations like Emacs unilang Agent, run this through termios, and use that to boot strap translation.
Also do anaphora resolution by running it on chunks of aligned texts.
- Here is what I resolved.
Best thing to do right now is work on system for determining where user is.
- Justin is having a telepathic test.
That's what a subtree is and you can make a super tree by looking at the current configuration state and you can form ontological possibility branches off of each of those decision point possiblilities and that's a and you can make a sub tree by back tracking.
and you can do that recursively, super sub super sub and possibility thinking decision tree point network and as long as you use formal system methodology it can be compatible with the map of the current state or with the recontextualization of the map of the current state but since its virtual there's no way to prove consistency and completeness because there's no way to prove those states exist but it can be useful if its compatible with the ..
of the current state, cause that at least can give a person possibilities for prioritizing goal structures.
- Track activities, it is possible to be doing many things at once.
Use the event-log to do this.
- Justin has already written on how to do this.
- Have Emacs do intelligent programming.
Use the editor itself as the system for intelligence.
- I am thinking that by eschewing simplistic yet exceedingly formal models we can make progress towards having a human like AI.
Again the basic question is how do we work on this.
We need to calibrate some volunteer base.
- do collab filtering or something.
Also, we should really steal online collab filtering and summarization data by scraping to save effort.
Need to create a free digital library.
Maybe work with internetarchive.org or some other such thing.
When I get in with Stallman ask him if I can use my weight to do those kinds of things.
- unilang should do basic RPC.
- The scheduling system has to learn to deal with backtracking.
For instance, if there is a goal to get gate working, and we can't do it without internet access, it has to realize this.
- Relate it if possible to goals.
do not work on the goal system tomorrow, but the day after that.
- do something about the computer getting shaked whenever I bump the table.
We need essentially to move the computer to its own table or something.
- do something about those files.
- do topic mapping, based on current readings and zipper things together.
clear should detect useful leads in documents and store those for later use and analysis.
- There are other things going on: interpretation of Stallman's statement.
There are hundreds of programs (things you have to do and run) in a Unix-"LIKE" operating system.
- do some online ordering.
- Write code to detect what other users are trying to do, and guide them to the solutions.
This should be part of kmax
- The code mapping system si going to be complex in that there are going to be refactorings.
Maybe it would be useful to integrate this into a version tracking system and do autoamtic analysis of code changes.
- Justin says do something about the positive ions from the computer.
- Okay, eventually get an assignments system going, so that I can check in with people.
Drop them if they do not complete their assignments.
- Some program should obsess over these entries, and making sure they are satisfied.
I should probably start with a particular class.
Namely, time oriented material.
do this by then, etc.
- do phone soliciting of money using audience when it is working better.
Hello, I am the personal A.I. secretary of the humanitarian inventory Andrew Dougherty.
do you have a moment to chat?
Good, well, Andrew Dougherty has been working for 6 years to develop a free artificial intelligence to aid the public good and to solve problems affecting people.
He is not motivated by greed but by his values which instruct him to help other people.
He is wondering whether you might be interested in reading more about his work to solve problems affecting people....
(this is terrible but you get the idea.)
- Keep some database of solutions.
I guess this is what architect does?
But we need something more general.
Example, if I can't program, and need help getting started, a good thing to do is to rebuild the fweb stuff.
- do a locate after the next updatedb to find whatever voy files were not correctly converted.
- When we want specific functionality, the best thing to do would be to do analogical reasoning on existing systems.
I.e., find or remember some system that looks/behaves similarly, and just steal that code.
- We can actually model who is saying to do what, so that we may learn from what groups of people are saying, and based on calculations about many things - about deceptions they might be interested (using inference models), i.e. motives, and other things, determine what are areas that are good to study, so that we can learn by social osmosis.
- In order to do better testing for the study project, Dean should draft notes regarding failure scenarios.
- Rather than just marching out and trying to program in a language you don't really know you should do a full study of it.
- study should ask people what they want to study, and then have them do so.
- setanta - one way to do this would be to look at packages and the way in which they were installed on machines at security conscious places
- * Allow graphic display of requirements or subsets wherever it make sense to do so.
- What you propose would make perfect sense if I had the social sense to do it.
- Any fool can criticize, condemn and complain and most fools do
- (in response to: the human mind is smarter than a computer) How do you know you haven't been taught to think that by a very intelligent computer?
- fieldgoal should determine whether the people like video games and if they do, get rid of them.
- do homework online.
- Can do dependency analysis on parts of the text.
- Could even write a program to do that.
- Suppose we have some concept: R-tree, we want to look up the definition, etc. Should be able to do this, no reason we can't.
- Suppose should do the same with complements.
- I've made quite some progress on gourmet finally - we can create an isa hierarchy, and do planning over that.
- Need to have some system that answers questions like: How do I delete redundant emails from my inbox?
- Or have the dictionary do that.
- There is really nothing that I can do about it - no more than I could kill off the English empire - no more than I could cause the sun to burn out (though I suppose that when that happens, twill be my fault!).
- When seeing him next, do not get excited, simply do not talk much.
- Twas everything my father's fault, there was nothing he could do about it.
- Don't do typing tutorial or really anything else when you haven't eaten.
- I wonder how does one express constraints on packages like python2.2, python2.4, say you want to say python (>= 2.2), how do you do that/
- Perhaps job search can use theorem proving or classical planning to do certain things, like, if the job is not from CMU, then it is okay to put that I am president of CMU chess club., if the job is related significantly to chess, it is also okay, etc.
- probably want to do laundry
- In order to do that, well, we'll have to have a way to verify the spider results.
- Some of the things I would like to be able to do, is to have everything planned out in advance, including parking, when to carry my backpack, etc.
- Dou, Ghe, Rty do Ug Her Ty
- However, do not lay off entirely as he will notice.
- The tough part is figuring out what to do.
- For this stuff like: when I see Mom, do this: have to figure out what sort of preconditions there are for the actions I want to do.
- do not install anything that conflicts with RSR.
- Need to find someway to do this shit.
- This seems to be the case - I'm not stupid enough to be your average American, but I'm not smart enough to do much about it.
- Our systems will wind up being burlesque if we do not get other people involved immediately.
- How do we do the translator (you just speak in picform, uh-duh!)
- Here is a bias - that people from older times know less than we do.
- do not sing these stupid songs.
- Difficult to do transaction system in Perl.
- On the one hand you admit you can't cut it in these fields, and on the other you say that these fields can't do anything.
- Functionality I need now: the ability to do complex searches.
- Predator could conceivably do something where it installs to a chroot and then does change detection to generate debian/dirs.
- J- If you do get upset - realize that that is your body telling you that you are not conceiving of things in the proper context.
- please do not waste so much time tomorrow.
- Functionality I need now: the ability to do complex searches that utilize information like previous searches.
- If you are in danger of dying, play a puzzle (for it in map cube pegs rect solo dominosa samegame guess loopy mines slant untangle lightup sixteen fifteen blackboxgame netgame twiddle netslide pattern flipgame inertia; do $it ; done)
- Need to intermix the different things we need to do
- do not get monster energy drink anymore, it gives me heart trouble.
- Now I realize that the basic calculation everyone is making is whether it is fun and therefore fun to do.
- Can't do any real changes to the system until we have some backups.
- Note that each time I do a move I don't have enough for what I need, so anticipate next move and what I will need.
- Get a extension hand to do things like turn off the light.
- One of the first things I can do is this: generate a meal plan temporal planning system to handle recipes, with a trial translation system.
- How do you know which one is which instance - pgourmet.
- fweb in dynamic content generation should do coloring dynamic to user.
- Ultimately, verber does not have the technology to handle everything, so what do we do then?
- Write a pervasive letter to UofI and if they do not accept Award myself an honorary degree from SFAI.
- If it's getting bearable, why do you keep inviting it back.
- Ultimately verber does not have the technology to handle everything, so what do we do then?
- If you do make such a movie, make sure it ends in a surging rush.
- Codemonkey can look at the AST for given expressions and do pattern matching to automatically write and update functions.
- It's funny how I don't do much with theorem proving except to make my life easier.
- do not depend on things like the previous entry.
- unilang should do basic RPC.
- Next I need to improve manager to the point where it can direct me to do certain things at certain times.
- do something about those files.
- Use the interaction logs to compile an enormous amount of statistics on what we do.
- Already we can do analysis of how efficient our Emacs sessions are by simply compressing that file.
- But it is prioritization and computing a function from the task, through inference, to what they need to do.
- ONly do typing with a real keyboard.
- Should focus on the development of systems that use probabilistic information, and can do viterbi type calculations on the distributions.
- Talk to other application develops and see how they do things.
- do a locate after the next updatedb to find whatever voy files were not correctly converted.
- Model futures probabilistically - i.e. the probabilities of future states holding and what to do in those cases.
- obtain .tty files from other developers to learn how to do things
- Can do many exercises lying down.
- Getting in to heaven has something to do with calculus and saddle points or whatever you call it when the second derivative moves through zero.
- Automatically determine how frequent to do certain tasks based on estimated wear tear/importance.
- We can do this to some degree by maintaining a joint activity calendar.
- do some online ordering.
- If there ever is another time where I don't have something in particular to do, read manuals.
- Keep up to date on what everyone else is up to, as that's how they must do things.
- Tomorrow morning, do shopping trip.
- The new goal can be added to the end of the list, etc. We can do it like that.
- To do this, we use spark...
- I do have a pretty amazing ability to code nonstop.
- Find functions that do the same thing and name them the same.
- Using the system architecture information we construct, along with intersystem relations, we can do code generation.
- I can do this manually.
- Maybe that incident with Natalie had something to do with Mom being upset today.
- One thing to do would be to automatically compute this from their file information.
- Before doing things we should come up with checklists for what to do.
- Then again, the names are unrelated to what the functions actually do.
- Lookup how to do Aikido stretches for my hands.
- For some time now I've wanted a system that will do machine learning on derivative theorems.
- Since fravia was good, do linguistic analysis to find related items.
- With any SQL we do from now on, we will want to use inserts using hashes to be flexible wrt upgrades.
- Although, why do so?
- I just realized that I could do both mental programming/networking w/ normal programming/networking.
- do a stochastic planner.
- Obviously we need our script index to do a find instead of an ls, it missed lookup, I think.
- do something for my legs to stretch out without hurting my knees
- Start having to say prayers after we do a sin.
- Add to daily routine to do puzzles.
- Should definitely fix computer in the morning, do laundry, set alarm for morning.
- (write system to browse fravia (seeker itself should do that))
- Should tackle problems very methodically in everything that we do, so that we don't explode the solution.
- Another thing we can do is determine how often errors occur with a particular goal or method, for instance Eclipse development versus Emacs development.
- do semantic spreading activation.
- Should do a dist upgrade pretty soon, or a complete reinstall or something.
- We can have kbfs do transparent compression of non-active, inefficient files.
- Figure out how best to do multiple sound things at the same time.
- It's very easy to do a mind to machine, the key is sound, sight, but I haven't figured out the other direction, other than video
- Another thing is we have learning similar games, so we can't just learn to do ksokoban, we must first learn about directions, etc...
- Just do roving downloads of sourceforge, compile and release packages.
- Nothing we do except through god can be licensed according to the fail.
- do our documentation axiomatically, so that when things change, we just adjust the KB and regenerate the documentation.
- Another thing that is kind of confusing is this: how do we get the project activity information
- Need to do lots of reading.
- Where do we edit the wiki from?
- One thing to do is to segment .bashrc logs, etc.
- do web log analysis
- RB::Wikize - do not clobber previous information
- do a backup of this system over to the frdcsa.onshore.net
- Maybe do it to frdcsa.dyndns.org
- Setup my archive and add an entry to the wiki on how to do this.
- do some kind of summary of most troublesome networks and determine why they are troublesome and what we can do about it.
- For setanta/ rt-analysis, do a full text index of the database for searching
- do text clustering to identify words.
- Use setanta to do various network tasks like rsyncing data to a different machine.
- Need to correlate tasks to code so we know what is already done and do not repeat that.
- do something to decrease the loading time.
- KBS could use vampire or something to do its inferencing.
- We could make code-monkey interactive and dialog with the user about what to do.
- clear should do code detection, etc.
- clear could use language detection software to do this.
- do accessory tasks to work, like reading w/ clear, requires less expense...
- If we want to have clear do this, we can use pitch to emphasize certain words.
- Need to use corpus to classify email/aim logs for what to do with them.
- boss should do the software risk management
- audience needs to interface with verber, and specifically, do conformant planning over communication acts.
- Secondly - audience should do the classification of what events are interesting, I believe anyway at least for now.
- Some more ideas for audience - of course it should regulate interaction in many ways - another things it should do is that if you are talking with people on IM (this being permitted by diplomat and RSR in the first place) Yeah - that's great - RSR can act as the guardian - approving various things.
- architect should contain information on how to do things freely.
- Obviously apartment-finder has something to do with housingmaps.com
- architect can review unimplemented capabilities from internal codebases and locate coresponding external codebases to do the work.
- antispam-console has to help with ensuring that certain mail gets through, like "URGENT - don't do this" when it is legitimate.
- antispam-console make sure do not drop encrypted zip as per Thu Aug 3 17:09:50 CDT 2006 incident.
- all you have to do is scrape out the members of certain interests or whatever, like "cslounge"
- all I really need to do is stock up the truck and drive out west!
- Now all we have to do is have it give the results of the autoclassification after we enter it.
- Why do they get mad at us for being sexual when that's all the world shoves down our throats?
- do entropic minimization of all the programs I have written to develop a shortcut language that implements them.
- Could eventually do all speech through a speech recognition interface.
- Make list of all things to do, etc.
- Set up a central debian repository so that all anyone has to do is add that one sources.list entry
- Create apartment finding software (helps do all the work)
- bard should do all the anagrams that make sense.
- Add a general program to do all tasks, so that if I am unsure how to do a task, I use that program, and it helps me to find the right functionality.
- Fix all the contact information in fweb and do a rebuild.
- need a program to list what all the other ones do
- all should do ontology extraction to build its phrase dictionary.
- Get akahige or whatever to do real social therapy (like get rid of seeking approval, saying certain).
- Based on estimated email message importance, if people do not acknowledge, audience should escalate to phone calls and such.
- Create a system for saying, let's do this now, and it execute code that helps.
E.g. pause for tara starts TTS that says a remembrence for Tara, or plays music during which we pause.
- Add a system that uses timex3 and sends a message to unilang-client that says something like "Authorize: 3025823 ("pse-due-date" "235523" "1 month")" for something that says "do this next month".
- Many programs make the assumption that the only items worth analyzing are those from unilang-Client.
So do we refactor that?
- Day by day do a bit of cleaning of the FRDCSA code.
|("pse-has-property" "84164" "habitual")|
- Develop a calculus of work substitution, that is, a formalization of when you have reason to do something instead of your "regularly scheduled program"
- do an inner-annotator agreement test on the FRDCSA Cleanse project, make sure all interested parties are signed up to the FRDCSA mailinglist, and email them with a request to join the release party.
Put the date on a calendar, maybe the posi calendar.
|("pse-has-property" "83499" "very important")|
- do a bash, shopt sometime in the future and inspect the stuff there
- do project archeology on the FRDCSA.
for instance, manager/scripts/misc-scripts is full of historic scripts
- Develop a software (part of manager) that displays todo list items after every shell, kind of like the new mail feature.
Figure out how to do this.
- do various image processing techniques on the resultant images, like object recognition, facial recognition, etc.
- do Sara's resume...
- pse: Have a system that, while not necessarily forcing you to do things, at least has you look at it by a certain time...
- Have a system that, while not necessarily forcing you to do things, at least has you look at it by a certain time...
- do over the google-set expanded capability search using the search data from al.
- do more Web as corpus searches.
- Develop a system to search a site for a concept.
So for instance, I wondered whether they reinstated the service to a particular website in an article.
So, in otherwords, do query expansion.
- email firstname.lastname@example.org, saying about: what you should do if there is something you want to use but you can't find license information for it
- Develop a system for spotting which goals are related.
So if I say, I want to do something related to inventory, suggest to them that they work on the inventory control software.
- do more remedial mathematics.
- Find out how to do randomize and playlists on rockbox
- Record which services people use, like whether they do RSS or not
- do something about posi being "open source"...
- do automated analysis of news channels to see which topics are being discussed.
- do not hold back - get in touch with the right people soon
- Zlatan said that a bad leadership style is to listen to what everyone has to say and then just do what you wanted to do anyway.
- Learn how to do exception programming.
- Finish the due-date calculator and have it do reminders and all kinds of neat things.
- Add the ability to mark certain goals as habitual.
For instance, do daily commits is required to be done once per day.
|("same-as" "67968" "66881")|
- do daily commits.
- for it in `seq 2 52 | tac`; do sudo zgrep -i justin access.log.$it.gz >> /tmp/justin-results.txt ; done
- do more to stand up against US imperialism.
|("critic-unilang-classification" "67301" "political-action-item")|
- Setup an IRC bot for posi that allows you to do things like they do in Debian, i.e. 'tell soandso about rt'
- Setup an IRC bot that allows you to do things like they do in Debian, i.e. 'tell soandso about rt'
- Expand the freekbs system to do more than just unilang entries.
- do not let anyone go in Mom's bedroom while she is gone.
- KBS, MySQL:freekbs:default assert ("feeling-for-entry" "41642" "I plan to do this later")
- do an rsync to frdcsa.onshore.net
- Get radar-web-search to do multi-threaded downloading.
- do tests by randomly generating lots of freekbs entries and testing performance.
- do a thorough review of all the people at Barcamp Chicago and see who wants to collaborate.
- The audience system for tracking whehter people have gotten back should do so by a given date, i.e. if so and so has not responded by 2 days, alert me, and/or resend.
- Sayer: The notion of what kind of features to extract is highly variable, for instance, if you were looking at integers, well there are thousands and thousands of number-theoretic features that could be used.
So the idea is, for the part of sayer that looks at features to decide what else to do, it should make use of a general purpose feature extractor based on this concept that there are so many potential features.
In fact the space of features is roughly synonmyous with that of the predicates.
- Setup code to do sentiment analysis using: file:///var/lib/myfrdcsa/sandbox/lingpipe-3.5.1/lingpipe-3.5.1/demos/tutorial/sentiment/read-me.html
- How do you list emacs modes?
- After all this time I still do not have a good web page caching agent.
|("comment" "51167" "Actually there is HTTP::Cache::Transparent")|
- How do I unjar a file?
|("comment" "51067" "jar -xvf jarfile")|
- Get started using TAGS for emacs...
Also do code search.
- What gases do biological plants give off at night?
- do something immediately about the Ethiopian children.
|("unilang-message-classify" "44537" "political-action-item")|
- Modify do/coach to handle freekbs and then you have a system for computing waiting-on stuff.
- Figure out a way to do twitter from unilang so it is dual logged.
- do an analysis of keywords in my writings and use that to generate a search for systems.
- Convert FRDCSA Emacs modules to autoload, and document how to do it.
- Have a "van week" where I do everything I can for the van.
- Figure out how to incrementally do mini-dinstall
- classify should maintain a database of files (and their content) which has been approved for release.
Maybe do this as part of subversion or something.
- Start developing graphical user interfaces to do some tasks, as they are inevitably necessary.
- Should run the build and everything else in a chroot to prevent damage, then do file integrity checks before and afterwards.
- packager: What to do when there already is a makefile?
- Make a password manager that assists with changing all passwords, remembers when to do it, and assists with executijng it.
- Figure out what to do about Enju server.
- do a test to see who is getting my mails.
- Actually do a rough shod system for now, that only approximates the mapping but does the collaborative filtering.
- do more searching for food ontologies.
- do laundry
- The problem of choosing what to do is similar to the problem of choosing what to eat.
- A good way to find new channels is to do a whois on the channels someone is on.
- Should have a code mapping database with translations between different programming languages to do the same thing.
- Need to write documentation for several projects.
do that one day.
- Record all schemas.
do a regular backup to DVD on systems.
- So, I looked into how to court report for fast typing, but I don't know that they type in the normal sense.
I did do a speed drill and I was at 94 WPM.
That's pretty good.
- Should get into the groove, where I can do such things instantly.
I am out of shape to say the least.
Must note, that getting isolated should be a piece of cake.
- What does pse do about circular dependencies?
- Find out how to standardize the database back end.
Does DBI already do this?
- job-search is one of the "rewards" that people who do good work for open source software get.
We set it up on a secure server, and people connect to it to obtain their profiles, etc.
- svrs: do not dig unless conditions (sound, light, etc), correct.
do not leave unless also correct.
Build test shelter hatch.
Use rock crib blast valve for toilet valve.
- do computing workshop reports.
- Would be neat to be able to tell bard to create poem to do certain thing, like: bard, satirize Ian's letter to me.
- RSR should note that going outside to do anything exposes me to women.
- When internet access is detected, plan for an immediate task as to what to do with the access.
- do not think about small animals.
- Our basic model is that I control the development of the project, but that other people who have interest in doing things do various large tasks for the project, get all the credit, etc.
- Things to do once connected to the internet: send url of plans to Justin, find out when Mary is going to take me shopping, email Mike asking for a time that we can move the server in and maybe swing by the store, look into writing a spider for craigs list shopping items for broker.
- Note that the reason that I become unproductive after _that_ is because my hands can't type as well.
So I do other things than programming, and the cycle continues.
Luckily, my mind can recognize such basic things.
So should RSR.
- Instead of typnign all the student reports up at once, do a couple at a time.
- corpus should be sure to do things linearly.
That way, when manually classifying things, we can assert a "continuation" between entries if they represent the same topic.
- When verber can't figure out how to contact someone, to tell them what to do, should contact me and ask for help.
- Have fieldgoal print out a schedule of when everyone is going to get here, etc. Another thing to do is start migrating the POSIThon functionaity to programming-marathon, maybe rename this POSITHON
- Threat analysis is important.
Consider what to do against Bears, Snakes.
Software should tell you what to do.
Of course, you should already have training.
- Only do planned internet connections.
- The FRDCSA is separate from posi, so I do not face a total threat of difficulty, I can always choose to secede.
- How long it takes to do certain tasks should be recorded, like, for instance, how long it takes to read the Emacs manual.
- do busroute reload.
- In general I just realized that we can perlify all of our software, just like we did with ORA.
We could do this for MinorThird, etc.
- Code monkey should be able to realize functions that do the same thing and generalize them into one function.
- Also, I need to finish the clear package.
I do not need to create a live CD, unlike my previous thought just now.
I have only to release a package of it.
- Need a system to answer questions like, why do I have to put new socks on each day.
- Also need stuff to remind me when to do certain exercises.
- Certain tasks need to be done for next years computing workshop.
We should brainstorm about what we can do to help improve certain things.
I didn't do any real lesson planning due to time and effort constraints and this was my main difficulty.
I think clear has applications of course, but to be an ITS its going to have to know more about learning styles than it presently seems to.
- Could display reminders of opportunities to do certain things, for instance, catch a bus.
- Must make sure that I do not get lazy with reading.
all skills - typing listening and reading should be enhanced.
- You can do this partially with export, where COLUMNS="179".
- Here are two things.
If you engage in a bad habit, you must do some penalty.
- Thought of a possible method for saving space: search for archives, and expand the archives contents, the do subgraph matching.
- However, see way out.
Also realize that time saved now will be of great advantage later.
So how do we go abou t that?
- When releasing AI, could do it in a mad dash fashion - before countermeasures can be deployed.
- I am start enough to train myself, the question is how can I et there in these circumstances.
The answer has to do with RSR?
- Should make a little package for cleaning up when the time comes - to do an interview - enough money for a haircut, and clean clothes, etc.
- I can't wait until verber is operational to the point where I do get a good schedule out.
- Here is a great example.
For a long time I did not know about w3m-el's advanced features for opening new items in new windows.
The software should help to make me aware of such things.
It can do this using collaborative filtering, and/or conceptual understanding.
Remember Javelin's capabilities.
- Here is a great way to gain influence.
Join a bunch of OTHER people's projects, and just do what we need.
- The system can automatically determine which functions it needs to optimize based on which ones are taking too much time.
For instance, if in query planning results in generating busroute tables, and busroute takes a long time to do this, verber's learning component would then determine that this is taking a long time and automatically add a goal to optimize that.
- Perhaps when I first wake up we should do a new typing lesson so that my muscles get ready.
- Need to read pod manual and then do documentation.
- busroute can do this: it can allow bounding of the search in specific dimensions instead of just quality, i.e. distance, safety, etc.
- I need to do a backup, like now.
- If we ever want to do an impressive demo, use ttyplay -speed 10000 on one of our emacs log files.
- Using ttyrec and emacs -nw we can do this.
Perhaps can set a flag for killing excessive dead space.
- Event log should be forked off again to a separate project since there is no direct responsibility for emotional training to do the actual event-logging.
- One thing that Machiavelli could do is spot things like the previous one.
They are logical possibilities.
It could do this by taking a common scam format and then pattern matching to events.
- do not the whites rule the blacks?
And do not the Jews rule the whites?
- For any text which coauthor uses to generate a book that is not by a particular author - coauthor can do the citation.
- I noticed that I can't get real work done inmy room because all I really need to do is sit down and flesh out these resumes and in so many days I was unable tol
- Great - Tabari and other coding software provides the data for a great taxonomy of action types (may do hierarchial classification already) for use with audience/diplomat
- Here are some different kinds of pse input.
do this when this happens.
- I just realized that being a hacker has very high appeal - thus - by doing AI and hacking (well appearing to do hacking, at any rate) we could actually draw a lot more people.
- I think we might be able to start some stupid thing like "Extreme programming" which is where people face off to do ginormous programming tasks over grueling days, etc...
- If I do get a nice paycheck from a grant, be sure to purchase necessities first, if at all.
The money should be stored, as much as possible.
College loans complicate this process.
- That's one thing is I haven't been studying formal models enough lately.
Like this PI calculus stuff looks interesting.
We want to map it out and use clear to ensure familiarity.
Another thing we could do is spider citeseer again.
Perhaps slowly over time.
- One possible thing to do is write something for unilang that interfaces with the corpus classifier.
- Need to do - no talk training.
- When you say "it can't be done", you are effectively saying that no one in the world can do this - that your knowledge of what is and isn't possible with computers is such hot stuff.
When you actually learn what is going on you will be like the frog from the well who sees the ocean.
- That is the attitude of this project.
We have a tremendous opportunity here, to collect and make a giant ocean of software.
There are programs out there that do things that most people don't even know are possible, or feasible, yet alone already done.
- http://www.43things.com/ has already set up the modeling of people want to do.
- I am intent that they do not see me at CMU - I can pick up my education elsewhere.
- do not deny Justin access to my computer since he'll think he's being lied to.
- Justin - "But I'm tired..."
Andy - "At least sit up while I do the real work".
- Also supply basic political reasoning for our project - by detecting sentences like: Unlike software restricted by the GNU licence, you do not have to distribute any source code with your application.
- I can do all the stuff for the gnu on their servers.
- verber should do contingency planning - for instance if the person misses their bus or the bus doesn't come.
- Questions are how do we master complex class libaries in general?
- Note that we perhaps want to keep the funding separate from my systems since people will try to guide it and force me to do things against my better judgement.
- With audience - on IM it should model normal users behaviour - for instance it should not simply respond to everyone.
I don't do that.
- do I have a choice over whether or not you will stop saying I have a choice?
- I would like to focus on some software stuff - I believe the other thing I was supposed to do is generate busroute data for various stops.
- Maybe make a catalog of companies that do work for the NSA - like Vivisimo
- I told him I would get him military stuff - If I don't do that today he will not trust me.
- busroute should do something: should learn which stops work but not official, for instance the 13C stopping at LaRoche.
- I knew what I had to do.
- Clickish in open source that stifle innovation.
We shall try not to do that.
Get rid of these people out of the open source community.
Turns off people - its hard its wierd its different - modelling.
Get put off.
- Areas for improvement with corpus.
It's quite clear.
The procedure I came up with isn't efficient but it is easily adaptable to an efficient procedure.
Two things - don't do n^2, rather only look at reasonably related documents - this will cut down dramatically on run time.
Second, make more efficient the sharedterm code, and lastly profile to see where spending all time.
- I know the social networking on how to do this.
- Tell Brian to check out http://www.advokit.net/community/?q=book/view/15 (advokit), and that we should meet regularly.
One thing we want to do is map out existing groups and see if we can't improve their connections.
- Anticipating User Tasks Using Commonsense Reasoning, Alex Faaborg, Chris Tsai, Henry Lieberman, The goal of this project is to create an agent that predicts tasks users may be interested in adding to their to-do list, based on the context of appointments they are adding to their calendar.
The agent uses Commonsense Reasoning to predict tasks that users are likely to add.
For instance, if the user makes the calendar appointment \u201cgo fishing\u201d, the agent may recommend the task \u201cpack fishing equipment.\u201d By automatically generating potential tasks, the agent will save user\u2019s time, and will help them remember actions they may commonly forget.
- One thing we could consider is simply to identify all the things we don't know how to do.
- That's really what all these RDF tools do, etc.
- Need to create a backup pretty soon.
Need to check out all of our current backups.
Need to create secure system division.
Have much work to do.
- Better would be for it to do face recognition on them.
- Is sourceforge all we need?
Or do we need more?
- Another is more important - has to do with students and teaching - I could probably show them all of these systems.
I think it was something about the kind of power they could get using this software.
- Fortunately I do not believe we are eaten, we are simply beasts of burden to my knowledge.
- Try to get window's machine working so that we can do the OCR stuff soon.
- Should set verber up to do that stuff like "verber -t laundry" to launch a task like doing launch.
We need a task hierarchy from pse or something.
There are certain tasks that are repeatable and we should usually have those in verber.
- I should also get verber working to the point that it can do this stuff where I wake up, etc. That would be nice.
- Also get gourmet item scaled down and do resume.
all this should be done first thing in the morning - it will be like a race.
- However now I'm too tired to do this.
So I need to do that and other related things, like build a better critic::Classifier type system for corpus, tomorrow.
But I must also eat tomorrow.
- The only way to do this ultimately is to get shelter.
- In order to get verber operational we should begin executing basic actions, not entire plans.
For instance it should schedule the laundry, by launching it with verber -w laundry, etc. Also, since we don't have the interactive system, it should have the detached operation mode working - i.e. where it informs you of the steps to do and then return.
- In order to determine what tasks people can do, we have to determine willingness (due to interest or mindset) - social capital - and capability (whether they have the requisite skills).
We can also ask certain people whether they can do certain things.
- Should incorporate resource consumption reasoning within verber i.e., how many wearable pairs of clothes do I have and therefore when do I need to do laundry.
- We need a tool to do definition detection.
- Could do the gourmet data extraction just by asking - is this a subtype of this?
- do not use hotmail or any other service that shows you pretty women on the side.
all mail should be done with fetchmail and a text reader.
- I am breaking down and perhaps I need to go take a break - say, to Boston - perhaps I should go do that tomorrow.
- Dad's statement at the end about how we were going to get rolling on this is perfectly indicative of his plan - all along.
He wants to make me feel as though I've made a commitment.
In other words I opened a new front.
I do best when I reduce complexity so this was an error - now I have anxiety about it.
And that is all fine and well.
That may be enough to do me in. Mike is right I have to write that software and yet I am pressured by all these factors.
The truth is isolation but even that is complex.
I should none-the-less ask Mary to pick up the drums and perhaps drive them to Justin's.
But not after we pick up a dolly.
- while not individually, collectively it would be much easier on me to simply read books, rather than spend shit loads of time solving particular problems.
In other words - the planning system should almost always ahve me study the work before I do it.
Kind of like the educational system.
- do clothes in the morning.
- Need to create new software that models Ryan's interaction, in addition to an RSR module.
In other words, he might develop coping strategies if I am not adaptive to his needs.
For instance, he said he is not interested in reading.
I should show him clear and see what he thinks.
I should record his reaction, and base decisions off this.
In fact, software to do this would be quite simple.
So that is what I should do.
Maybe I should write it by tomorrow.
- Mary said she can write grant proposals and Mike said Hunt might do it.
- One scheduling rule we could incorporate is get done the things which take the littlest time first, then maybe we won't have to do the big things, i.e. other people might do them
- I do feel slightly sick.
It is strange, I think its from too much sleep.
It feels as though I am crying, but I'm not.
At least though the situation is good.
- Sean Boros expressed an interest in "obtaining time to do what he finds interesting".
I tried to impress upon him that that is what I am working on.
He seemed mildly amused, especially since he was quitting in a day.
- We were waiting around, having no money, and we were trying to get somewhere.
Hence we were waiting by the highway.
We were discussing, for lack of better things, how Arnold Schwarzenaegger was half black or half asian or half whatever you happened to be.
Then we moved to the other side of the road and someone stopped and let two of us in. The other person went back and watched the motorcycle and bikss.
We went to a restaurant and got some food.
The person in the car was advocating the idea that I was always tired and exhausted as a child because of my mental excerstions.
I said it was rather social because even on days I did not do any thinking I was still exhausted.
- Why do you keep foolishly trying to hold me back from doing what I am supposed to do to help us?
- Could use VerbOcean to determine which things to do first.
- Also, try to find things that Fred can do and to have these prepared before interaction so that he does not lose interest.
- The answer to these difficulties is AI.
I am sitting here wondering how to do this stuff.
The answer is AI.
- There are a lot of good programmers out there and if we seek to start a correspondance with them and lead them to the right path on their work, for instance someone who has a background with theorem provers and chess, and that isn't part of the system, we can do well.
- Ask Mike how to do that.
- do not check your email if you wish to avoid being dragged into stuff,.
- I think I was able to defeat this silly mindframe as soon as I remembered AI.
Perhaps I believed the bullshit arguments I was making.
My purpose is to gain my isolation, and to do this I have to appear to kathleen as competent to do what she needs.
I have a meeting with her but each time I prepare I get shunned.
However, this is a meeting with Kathleen.
I will prepare extensively, including doing C/C++ learning, as well as intelligence work and even grid programming.
I will write a globus program tonight so that I may say that I have done this.
- Here are some more items for RSR - do not take off your pants before removing the stuff from your pockets and putting it in the appropriate place.
- I can do an all night study/sleep thing, actually.
Best to go home.
- Here is some thinking on the part of the job.
The major problem of course is convincing Jeff that I know what to do.
Now, I can study some here to improve my understanding of C/C++, but that may or may not be sufficient to convince him.
Jeff is skeptical of me.
- Tell parents to note that Note that it is useless to ask someone to do something they cannot.
- Note that it is useless to ask someone to do something they cannot.
- Note that customer data mining technqiues do apply as well to our collaborations ystem.
- note the conflict between the notion of do it yourself, and the notion of automate it
- Here is where verber must do what ifs- what happens if I do not do this?
|("belongs-to" "2721" "verber")|
- didn't look right when getting off bus, got bad look from homeless person, left security cable in an unknown state, allowed laptop to discharge fully in bookbag, didn't finish a task in the correct time, allowed Jesse to see a political sigsite I was watching, was made fun of by Marcus, was not perfectly polite to il-chun, don't remember the substance of conversations, left my cubile, failed to do a task I said I would, made an unplanned movement, did not plan out day in advance, lost compsure and awareness multiple times, offended Mike Stevens subtly, did not tend to Justin's emotional state, shifted mentality away from AI, tempted to make unplanned movemnts, made an offensive possibly racist comment, not using my logic enough, did not relecutant to pursue certain important goals, procrastination, made noise early in the morning, talked too much, given too much credence to a person like seth, not acted soon enough to tell Justin how to handle the case with Aaron,
- Maybe this complex visualization that I do where I wakl thorough the whole day is what I should do every day, selecting whether I did everything in sequence, verber can also plan for everything and many things may be executed from memorization.
- Model what people need - what they're willing to do.
- * Allow graphic display of requirements or subsets wherever it make sense to do so.
- Requirements are much more than just a "to-do list" with a hierarchical number scheme, although many projects rely on exactly that, in a spreadsheet or a word processing document, which may or may not be kept up-to-date and version-controlled.
- 2. Requirements must be validated.
(do they pertain to fulfilling the need?)
- 1. Requirements must be classified.
(How do they fit into a dependence hierarchy?)
- Talk to CMU people who are experts in theory of C++, etc. do some research there, and get really top notch sources, and then talk my way past the programmers.
- Dynamically draw AST in other window.
(mstevens will talk to people he has already written stuff in this area).
If you can add in type checking systems, and highlight the color, if the line of code looks OKAY, you can do static analyzers while the guy is running the code, look this line of code.
- Any time I can do my own work for work, I should
- They will never help me. The vultures are weighting for me to do die.
Turn off their "fight" programs and help.
- RSR should have a feature where you warn it you are tempted to do something, and it tells you things that you need to here.
Like, Pleasure is a place, there are better places to go, etc.
- This is much nicer since clear doesn't do a bunch of BS here.
- With RSR, every habit should have a dual imperative form.
For instance, "didn't clean up after eating" would be "clean up when done eating".
This form will then be fed to the user at plan execution time, so that they will have a strong reminder not to do the bad habit.
- The important thing is knowing the truth.
do not believe in all the constructs of the movies - power is beyond the jedi, etc, etc.
- Presentation of released items is important.
Therefore, do not connect to my project directly, and secondly, use language models on the important project descriptions to generate mine.
- Releasing stuff on the internet is a far better way to get attention than to talk to people!
Duh, let the internet do your advertising.
- do not attempt to persuade them.
- do I really need to work on these people communication systems?
Or am I better off simply making packages?
That's a tough call.
- Here is an example of a lie - Bush thought it was the right thing to do.
- That resulted in me being totally useless the next day, but at least I bought my shelf.
The basic thing I need to do is work out one mission whereby I go around and purchase everything I need.
This must be on payday next week.
- As a general rule, if you must choose between too actions, one of which you know how to do, the other you don't but have ideas about, do the one you don't since you may forget your ideas by the time you get back to it.
This must be part of the calculus of context switches.
- Refilling the water is a wonderful example of how the computer should monitor the uncertainty in the state of various objects, and use this to determine checks that it asks the user to do.
- In order to ask Brian how do I get my bank information, which BTW is a dumb question, I have to write the software that communicates with him.
- This is as good a place as any to work on my stuff.
I have to improve my system.
Basically, I should divide everything into a list, which I carry around me, consisting of things which I will do and I won't do.
And of course watching movies must be against my rules.
- The list of bad things is, in this order: socializing, watching movies, wacking off, being affected by advertisements from media webpages, sugar drink, working too much, not spending enough time per day working on my project, leaving my area, movements, purchasing items.
Now all I need to do is find adequate punishments.
This might be physical pain, such as leaving those nipple things on for a certain length of time.
That should certainly qualify.
Also, not using the computer for an hour, but this might be bad.
They have to be followed.
I can't think of any that are good enough right now, but they will come.
Exercise would be a good one.
50 pushups, etc.
- Probably the most important thing to do is to always make concessions.
For instance, no pornography.
I suppose, obviously, one would destroy the critic system if this were this case.
I also will have to institute punishments for violating the rules.
This means that, frequently enough, I must take the entire survey.
I don't know why I feel good right now.
I think its mental.
- Need to find a new formula for isolation.
Here's the plan.
Justin needs taking care of as long as I am in Pittsburgh, but he's in a sense using me and isn't gaining any discipline.
Therefore my emphasis must be on discipline from now on.
I will use my software that directs people where to go to do that.
Now, if I move to a new place, I cannot have any friends.
They have never done me well.
I can try to take a job in Boston, this will put me closer.
Justin will not be able to move.
I should actually keep the actual location a secret and communicate only by email as a pretext.
- This movie certainly did.
I saw Star Wars.
I think it has been too much.
I never want to see another.
This involves not hanging out with Justin.
I don't know what to do here.
- Justin said in his sleep "Matricies of cardinality do carve themselves into my psyche.
- We can repeatedly ask someone to do something that they should.
- Could tell manager to do this.
- The theory of navigation also applies to social groups.
They don't have to see what we do every moment, rather, they simply measure us every now and then carefully to determine how far we've been able to move into position.
Of course, this applies for us as well.
- Should model who uses my software because they may become dependent and in order to solve a goal we may have to be concerned about upgradability and maintainability of this software.
This is a serious unanticipated problem.
I can't believe I don't know exactly what to do.
- Show this log file to mstevens to give him an idea of what I do mainly.
- Try to recognize realistic REASONS for collaboration.
Try to estimates someone's REASONS for doing something.
Record REASONS in addition to plans and tasks.
This is necessary to predict what they will collaborate on.
Find out why, if possible, they do or do not do things.
Automated prediction of this may be too complex.
- If you do something at the request of another person, usually, the system should let that person know that it has been done.
- Again, more thinking.
Essentially packaging will always be _my_ thing.
So, its best that if I want people to work with me, I get cracking on finding out ways, empirically, of getting people to do what I want.
- do everything with this.
- What do I need to do to accomplish this?
The use of Cyc is premature, and it doesn't work.
Perhaps I could ask Sameer to get working on this.
We also need to create a formal command system, and give out rankings, or perhaps not.
- That way, we can determine what people will want to do, and get that working.
We need a basic system for ensuring this happens.
Basically, our goaling system must be done first.
Then, we need to ask other people.
Then we need to identify mutual goals, and tasks that need doing.
In other words, I should simply provide people with the information they need to do certain things.
But if I don't provide it correctly, they may not wish to do it at all.
- We need people to pull their weight.
This means Justin has to do some work for the project.
Of course, his work on patterns in context is what he is best at.
- Get commitments from people to do certain things.
Don't bother with Brian of course.
- Another thing is I could befriend just one person whom I could rely on to do this.
Another thing is I could live in the woods near a bus stop, and that way, make semi-annual shopping trips.
- Risks with the shelter system.
This system has several risks.
First, how do we transport to and from it?
Maybe it should be near a rail line and I could hop the railroad.
I might get caught if AI improves.
So, what about this.
I suppose this clearly presumes some kind of contact and dependence on the world.
Of course it does.
The question is where to build so that I don't get caught.
I also need defenses against other kinds of problems.
- Ask Mike to order me some DVD disks.
This will help in freeing up hard drives.
Schedule to do this on one day.
- It is a risk that I don't know how or with what I'm going to do the spidering.
- Forget it, maybe or not, all we need to do is write software, the rest will come.
- For every person or group we meet, we must do plan recognition.
The problem is getting all this software together is tough.
So maybe we should concentrate on finding a few high profile people.
- code-monkey can do automated refactoring of libraries based on perhaps similarity of functions.
- Would be interesting if we used subversion to store everything, and then could do massive updates by running scripts for data processing on older versions.
- Should ask mstevens about generating power on Justin's exercise bike, and how you'd do that.
- The only way I'm going to be able to keep up appearences is by being ignorant of them myself.
In other words, the AI has to do the planning.
- A good rule for determining how successful deception is, is to shoot two scenes, one involving the information which a witness has seen which leads to the inference they do not wish to disclose, and show it again, without special knowledge.
See whether anyone suspects something in the second.
- In other words, the system will do plan recognition, as well as hide information, from the user, that they don't need to know.
- Note that go is very useful indeed, since, in some sense, ideas are a territory and they need to be connected in order to make inferences, so the attacking of certain ideas breaks inferences.
Therefore, all you have to do is put a white piece on a square in order to break the connection.
- When meeting a person, a basic assumption of their actions models is necessary, and the result is that we act in a way that gets them to do what we want.
- Andy what are you doing?
How are you going to survive.
There's always Brian.
Does Brian make you suck his dick?
Don't do it, its not worth themoney.
- Okay, back to reality.
We do need to get the club together.
- One thing to do would be to find other people in similar situations and get them to make their projects part of the FRDCSA.
To do this, we need a really cool website.
- Also, here is some speculation on questions.
I observe that Justin takes up a lot of my time.
So we want to give him more autonomy.
My thinking processes are slowly becoming shot.
The weight of the world.
So I need to come up with a schedule.
Also a nagging system.
But I don't want to get stuck at Justin's with nothing to do or to watch a movie.
I think that I should get in contact with Stallman, and get things lined up that way.
Then I should accept a move home as a pretense for leaving.
I move home, but I have another job lined up.
After I move home, I job search some.
Then, when gets ineffective, I move to the new job, with parental approval.
Of course, all along, I've had a third job waiting for me to arrive.
- I haven't felt very good all day.
I need to get back to doing some programming.
And developing that system that reminds me when to do things.
- The American people do not wish to be controlled by the stratagems of corporations.
- Here is another dream.
I only remember little parts, but one of thenm is that our offices are like interrogation rooms.
It was something about do you believe in God and so forth.
- Design tests for people - we need real time simulator for group involvement, and record of who passed which tests.
Mainly however, we need people to work on software.
We have to avoid laywers as well.
It must all be conducted rather clandestinely.
One thing we can do that will aid the cause is the training of people.
We have to keep Justin out of the loop and functioning well.
He is mentally ill and will misperceive many things and ultimately hurt us.
His ethics are right on the money of course, and we shall not deviate from those.
We must be against the killing of beings of any kind.
- do not say things such as those which made Kim act antithetical to our goals.
- Could also do training - you could send an employee to get trained by us.
- We may use slashdot as an efficient mechanism for learning what people find useful and interesting - should do mining of its contents - especially its ratings system.
- We can do a few things - we can study public policy.
We can also ask people to head up given project areas - to be the manager of such and such component of the AI.
- So today, spencer said he would work with me over the summer.
Therefore, I need to do basic project management.
I could use microsoft project.
- I don't know what I would have done without your help.
Oh wait, yeah I do.
I would have had to run away...
- Teach Renee how to do income tax on line.
- Teach Lena how to do income tax on line.
- In order to get clear working, I'm going to have to do certain things like ensure it has all of its run time dependencies and I'm not interested in that.
- Does RT do dependencies?
- We should do what the ISPs no doubt do - machine learning on user patterns.
- Because you can model things in different ways - the computer has to have the flexibility to adopt to the representation, as we do.
- Develop a planning domain for manager to be able to do certain things, like, obviously, for instance, when I go to sleep, start all computation intensive processes on my personal machine.
- Now, next, I just realized the following.
If one is against certain things in society, for instance, military work, one has to recognize that people are paid to do things, and that the need for money forces people to do these things to help the people who already have money.
So, the ability to work together for the free in order to fight corporate control is actually important, because corporations are just wrappers for special interest groups.
- AI: Andy, are you there?
AI: Would you like to do your work now?
- There were some things that I needed to do before going to Walmart, but I forget those.
- Should also do the following - should, based on user searches and natural stratification, ensure that users look at many different systems.
- One interesting thing that audience should do is monitor the state of communication.
- We need manuals on what to do.
- Ask existing leaders of the free software movement the hard questions.
What are we going to do about Palladium.
What are we going to do, etc. about all these other things.
- Also, do a lot of stuff in Java.
- Use my writings to build more complex structures.
Try to do this automatically.
This is where using existing java applications is really going to help!
- I suppose a simple model for success is having the money to pay people to do what you want them to :) Therefore, it seems, one major criterion that we are looking for are people that aren't interested in making money or being paid beyond basic poverty.
|("comment" "1170" "such people are probably poor to begin with")|
- Need to do "conceptual search" of websites.
- Now, nitpick ought to do reconnaisance as to who will and who won't collaborate software wise.
- Set up a system to handle all the predictors we set up.
What do you call that?
Someone that can foresee events.
- So, some of the things I really need to do are write an "Alarm" system, capable of sending messages at certain preset times, durations, etc. So should write that.
- One thing to do is to try to find projects already being worked on and contribute to those instead of my personal projects, or merge them.
- Interesting observation - just now I accidentally engaged in bad behaviour - I chatted for far too long.
I really need to do a variety of things, esp, laundry, get food, and so on.
The stupid thing is it is always just before the deadline that I come up with the plan.
So for instance, it's too late to take the bus now to do laundry.
I'll have to wait till morning.
So all of this is known in advance.
How do I win.
Note that Jana didn't seem happy to see me walk into the information retrieval course.
Anyways, we have to log who we ran into.
- Justin just sent me a whole bunch of conspiracy stuff.
Its interesting, but it seems unlikely in many cases and well, creates simply too simple a picture.
It also conflicts with other conspiracy theories.
Therefore, I think that one good thing we could do is to axiomatize this stuff, in which case you might be able to identify the genres.
- The request - can I go to a bathroom, should be translated to a query, and then a plan to do so should be searched for.
In addition, various other considerations might apply.
Also, a dialog manager should be used to elicit these requests properly.
For instance, if I say, "can I go to the bathroom", and the system says "no", and I say "why", this should be programmed so as to give a justification.
- Use Cyc, eventually logicmoo to model world state.
Make sure that we have to ask manager "permission" to do everything, and also all requests.
manager, I am hungry.
This initiates manager to plan for the world state.
This calls the Cyc Agent for world state, which it sends to manager, who then sends the modified domain file to the verber agent, which then generates the plan, which is then returned to the manager agent.
- It is important to realize that Justin's situation is somewhat permanent (if he lives) - therefore, he is reliable in that sense.
Now, if gourmet is finished and he is otherwise improved, in health and so on, well, then you sort of have a stable guy.
But I think he's just gonna want to do his own thing.
We'll see what happens.
- The few people that do know me despise me. I am completely socially backward, with delusions of grandeur.
I am a crackpot AI researcher whose pathology is similar to mentifex.
I write crappy OO-perl software that just sits on my website.
- It will be nice to go ahead and get this system operational so that Justin can have some good reading.
Should assemble a bunch of nice texts for the lad - not ones that focus on survival so much - since he's inclined to abuse that - but one's that aim to build intellectual stamina - he should rate them!
He has a fast enough computer to do voice recognition.
Seth should provide us with the "Justin Coslor is the master of the universe stuff".
To get programming I will work on manager and "fun" stuff as to focus only on gourmet is a mistake - since I dally.
- Need to record the position of every person I meet.
I think, afterall, that I do work for money, and not because of what I am doing is good or bad.
Obviously, now I know it is in a restricted sense good, but at the same time, I think its rather like disturbing.
- Looking back on how things have progressed, do you agree now that it would have been better for me to leave UofI for a while, as I had requested?
- Add a boss::Config CLI front end to predator so we can do things like search archived and packaged software by name, etc.
- My project can do a lot of good, but unfortunately there was nothing that my project could have done to help those people.
The many parts of the project that specifically deal with crisis management and personal safety would not have been able to do anything in this case.
In the future, ad hoc wireless networks of digital intelligence augmentation systems can't really forewarn the population to move upwards and inland, nor issue remedial instructions or deal with crisis as mentioned in the Automatthew project.
No, there was nothing to be done, because we ARE PIECES OF SHIT.
That's why we float away into the ocean like bloated turds, you know?
Cause WE'RE FUCKING PIECES OF SHIT.
- (Definitely we are going to need to do the following.
Before meeting my parents I am going to have to finish the MKS, and this will include a bidirectional speech interface that interfaces to audience and uses dialog management in order to do these things.)
- (Meet with Colin over dinner to do two things, talk aobut his research, and secondly, get wireless networking for Pitt, and ask him for pointers on places to bum out at Pitt.)
- (really, should just put a debian folder in with each internal codebase so that packages can be generated directly from these, or simply study how people do incremental packaging of online sources)
- study NLP so that I know how to do sentence detection
- Here are some of the things we need to do next: fix the way pse craps out all the time, get Clairvoyance working and with some basic document management systems, including for instance authorized reports to ensure back to me that various people are learning various things (based on testing), find some way to get those recipes normalized, get shops up and running and taking inventory of everything I have, add this information to verber, get MDR (minimum detection route) planning operational, get new tagsets for AWB/Predator working, fix problem with script determining perl dependencies, make packages of my systems and upload, fix mini-dinstall problem, write tutorial on agentification, agentify or otherwise get command line bugzilla working, fix up manager with sleep learning capabilities, create a sample course for clairvoyance, and run backup.
You're such a lazy guy, you know that?
- have someone show me how to do blogging
- for data scrubbing, one could do clustering of likely matches using a robust semantic distance measure and then validate matches, and hopefully come up with sets of edit rules that are valid as per se William Cohen's stuff.
- do NL translation of verber plans.
- Clairvoyance should determine based on how quickly a user is losing attention the level of interest of the material, and then determine ways in which to teach the user interesting things based on what they do find interesting.
- If there is something to be learned from the axioms of magic, it is that we should determine what people's wills are, how to do this is important - to figure out what they want.
- An operational decision that seems to make sense is we should structure operations much the same as they are really constructed.
In other words, we ought to have in the CIA various divisions as they do.
- The trick is to simply forget about that one girl - they do this to compete for resources - its pretty stupid.
- Let's start by packaging some of my important software, such as clairvoyance.
We do not need pbuilder, etc, to do this.
- Interested in exact formulation of that rule, whereby we calculate which of two items to do based on various things, money etc. For instance, do I buy a cross over cable or do I manufacture one by hand, given that the need is such and such, mrwright needs it back, and that the crimper is in johnstown, and that the lowest price we could find is... etc.
- I should probably write a letter to the people there explaining what Justin is going to do.
I need Justin to write me a letter assuring me he will not do that to me again.
Under no circumstance is Justin Coslor permitted to bear false witness against Andrew.
I should write a letter in which I ask a psychologist for ways of minimizing the
- I think that I could ask someone at the bug squashing party to write a classifier for the unilang file that does the automatic stuff (although I don't know anyone who can do that, we'll have to find them.
It would be a cool way to get stuff done, to have the code online and these people would improve the code at the meeting, and we would reward them in someway, such as giving them more code in exchange, etc. The FRDCSA could simply be the wrapper, or the project information site, for everyone's projects, and in this way we would do this.
We could alternate who's projects we worked on each time, or something, if necessary :)
- do not mix clean water with dirty water.
- Not necessary.
New algorithm asserts that do job with maximum necessary.
- Trying to recruit Jana to work with me was a notable failure.
She said "how old are you" - so as to put thing into context.
"Human's are irrational".
I can't stand all this BS, actually, that is a mostly statement - but she was using it to say that the human process is irrational.
I say, what does that have to do with whether or not computers can solve subproblems.
This whole notion about computers not being able to solve problems is the height of irony.
Now, to get on with it, I need to hide from this kind of stuff.
However, they make it hard for chickens to hide in the food pen of earth.
Well one thing that I hope is not too late is to get castrated.
Another thing, I talked with William Cohen, that seemed to go well.
- Was able to load the init file but not get transcripts working.
Even more importantly, was able to create a working server by running with -b option.
This should be sufficient.
I don't know wtf is going on with this computer that it can't run that, but, it seems nice to know how to do that.
- Need to sit down and plan out everything I need.
Here is some thinking in terms of being homeless.
A combination of movements is going to most effective.
The goal of course is to minimize at a rather large cost the number of times one is discovered sleeping in places that they shouldn't.
Therefore, a schedule of sleeping in different places is advised.
For instance, always being at my computer is a problem.
Also, this problem with laundry is bad.
But I shouldn't do laundry until the evening when fewer people are around, or no? Anyhow, I should find a local laundromat, ok, so let's make a todo list - asking for the nearest laundromat.
- I think one thing that makes sense is to ask generous people for "favors", i.e. find a task that you know they like doing, and give them everything they need to do it, perhaps money as well, and just ask them to do it.
The idea of working on other people's projects has paid off as mbrewer, tomcauch and mrwright have been enthusiastic about these, esp. Mrwright today.
- Some of the things that we want to do: get Rolf to fix the opencyc daemonization problem - as this will make radar's KB system practical.
We can always interoperate between CYC and other ontology systems I would imagine.
- Justin is not serious about working on the svre, no more than he is serious about turning me into the cops.
Moreover, if he is not serious, there is little way to make him do this.
Of course he should work on the still suit first.
The way we should approach this is I should give him $200 and tell him I don't want to see him again until the still suit is completed.
This way it forces him to work on the project.
But remember what happened the last time.
I think therefore we should consider ignoring him.
- Some of the things I am thinking about.
It will be nice when we have the gourmet system done.
It will then simply be a matter of heading to my selected food location at the proper time, preparing my food with minimal effort, and eating.
In order to get gourmet done, we have to schedule a meeting for gourmet in which we sit down and do most of the planning.
This depends on the meeting scheduler.
Perhaps there is such software that we could use.
- We need an expert system for the construction of systems.
Let me explain.
Let's think about how to do the following task: sort incoming gtk-gnutella files into various topical entries.
In retrospect, its stupid to make such a classification.
Luckily for me, I will be taking lots of classes and hopefully I can rid myself of this stupidity.
Back to the point though.
The system should be able to elicit some kind of understanding of the task to be done and to automatically do such a task.
Of course, this simply means the construction of a large system which can do these things, but the important point is the rapidity with which it develops interfaces for a task.
- + I have determined that a naive way of trying to get people to work with us where we simply ask, is entirely wrong.
I will do this more intelligently from now on, i.e. more forcefully and scientifically.
- Interesting idea, I never saw anything of my fathers that would lead me to believe that he was not in the furniture industry.
So, supposing he was involved in secret work, wouldn't the ideal way to do this be to code information in all those furniture design books?
- What do you have to use.
Given the extent of your present need, I refuse to believe that you would pass on free assistance all because of some minor ambiguity!
- What do you have to use.
Given the extent of your need, I refuse to believe that you would risk not receiving aid all because of minor ambiguities.
- Maybe that summarizer sucks, or maybe we should filter them first using.
- I need to do ontological modelling...
- The thought of having to outside into the world again is weighing on me. I think its best that I get to stay underground.
But how to do this?
How to get this to work?
- I seem myself to be falling into the fallacy that AI implies that other people cannot do anything, which is of course total bull shit.
I love the way its so difficult to understand it.
Other people simply can't make the connections.
- Want to do something similar to http://boost.org/, this is very related to our reviewed of packages.
- There is a new thing to do.
This is, namely, we cannot bother to explain what we are working on to anyone, this has always failed.
We can explain what they need to do.
We should also look at finding things that people want to do.
- Could even have radar do prediction of which sites have new releases!
- Use FCA (Formal Concept Analysis) to ensure maximum independence of our systems - that we do not depend on software when we don't need to - actively look at the package network using network analysis tools.
- I suppose it is natural that software systems develop command based - because those who have to develop the system are typically people who are command based, that is, they are telling the system what to do.
Whereas, if users developed the system (which they can't), the system would do everything for them.
- Must predict every action that the opponent can do.
- The shops purchasing system should have the ability to basically do this: find these items, have them approved by this person, have them billed to this account and have them shipped to this location.
Report any anamologies.
- Tell them that I have been working on an open source nutrition management system, but that these games are very cool since they program the user to do these things automatically, which will speed up and make more usable my software since they won't have to check everything.
Neither will they have to study which of course no one will do.
- So, what I should do is write a letter apologizing to Al about taking
- at CASOS.
I would very much like to have money pooring in. How do I
- me go.
But I don't want to have to do that again, it was pretty
- an officer.
Therefore I pleaded with him to not do anything.
- Questions for debian and linux people: where do you find out about upcoming lectures, etc.
- humor: why do all top CS departments refer to everyone by only their first name?
Because everyone's last name is Cohen.
- Adopt http://triplestore.aktors.org/SemanticWebChallenge/ to do Linux information.
- Henry Kissinger to Jed, I can assure you your honor the Chileans will move, but push it past the 23rd and nobody this side of the Nile can do anything about it.
I'm mean, I've got Hugo calling me more often than I call his women!
- Where possible, force programs to do the work
- Ask sameer the following question: I know my general project is too inspecific and "grandiose" for you, or any other practically minded computer scientist, to really take note of, however, I have been putting an effort into studying software engineering, and have attempted to divide, as do the engineers, the problem (i.e. real world) domain from the solution (i.e. program) domain, and I call the problem domain projects, and the solution domain systems, and I was wondering whether you would be interested working on some of the systems, since we have put an effort to make them conform to standard software engineering considerations, and so therefore some of them are very practical.
Perhaps, you will find the purpose of all of them to be boring, in which case, would you consider starting some projects?
- There is an inherent problem in say "they can't do", since it may be the case that "you don't know they can do"
- Here is the game plan which defeated their logic.
The theory goes that, upon being reintroduced to my system, and upon finally following through with its designs, I will understand my approach once again.
The notion is that I can't see my approach for the reasons mentioned before.
I suppose the fallacy I am engaging in is thinking that its my fault that I don't know the ways to do things, when, in reality, no one has.
- Make not of which projects are for download, which are not, and which we don't know, and who, if not open, do they go to that we can tell.
- do recognition of where files are actually downloaded from to predict locations.
- Should be able to do text/web searches using things like this: undergraduate advisor brad myers /chinese name/ site:cmu.edu
- As everybody knows there are different forms of intelligence, so when we realize that I am genius, we must also realize, that, socially speaking, I am a retard.
But I love both of these and will not change, since I do not find those things interesting or worthwhile.
- I have an idea now for a useful system.
The system monitors social networks, and tries to find ways to persuade large networks of people of some points.
For instance, if we want to overtly convert people to become FRDCSA members, it does extensive social network analysis and learning/simulation in order to send messages/etc that obtain the given objective.
Or it can try its hand at recruiting covert resources.
(Modelling of weaknesses/etc will be an important goal.
Notice that the detection of helplessless is important.)
Obviously - this is a HARD problem in many senses, but what we will do is gather a system together that does situational awareness - a kind of social intelligence flight simulator, and then begin improving its learning and interaction capabilities - all to the point where a precise, data-driven model of other people can be obtained.
- Need to start classifying examples of which files we want/don't want.
To do this, well need to have some sort of labelling mechanism.
- Part of the problem is that although Sameer is intelligent, he doesn't really know about certain things like the item above, nor do I.
- We can use some of those graph theoretic notions of centrality from the CASOS papers in order to make determination of usage.
Really, I should spend alot of today either reading the Schaums Software Engineering, or planning what to do (as long as it includes Schaums).
But knowing myself, I will not, prefering to program radar and predator.
Which is in a way, the most intelligent choice possible.
I do have to definitely plan for when work is to return on shops.
Let's say to be stupid that I will do it after sending out AIX 1.0.
- How do people model products they are developing?
- Document all programs soon, since I am already forgetting what they do.
- Incorporate the knowledge from books into the very algorithms of our programs.
do this almost automatically - i.e. convert manuals into programs automatically.
- Should make a video of me programming so that we can show people so they understand what it is that I do!
- The failure at home was a result of a failure to understand was I was able to do, and not necessarily only what I wanted to do.
I wanted to do what I was good at, and also, things that I didn't fail at.
The double standard is that other people can do the things and they think I am just being lazy, but its much more difficult if you can't do the things.
- A new result: That, as a natural consequence of what we have said, any action is going to be of reduced potential if the group has fewer contributors, since the potential depends on what others do, so the user cannot effectively finish.
Sort of like linear vs. nonlinear planning.
- Kevin, would you really rather work when you could have everything provided for you where all you can do is research?
- Should implement a policy to watch the fs weight, so we don't get so overburdened with new data ...
Another thing we should do is set up one computer to generate DVDs by downloading software and caching webpages about software from the web, using say the Dell.
- I love the way you think I'm going to come back to you crawling on my knees.
The fact that you constantly thought I was asking for your assistance shows that that is the self-righteous attitude that you have.
Perhaps it would please you to know that I would rather be murdered a vagrant than come back home to you!
do you think I will be treated at home any better than I was BEFORE I ran away?!
- Perhaps we can have Al explain these things to Kevin.
Kevin doesn't understand the transitive closure violation that is implicit in a program's ability to write a better program, although he understands what happens when the program is able to mine the resources of the net.
do we need to teach him?
I think so so that he is informed in his approach.
Moreover, there is also his argumentative opposition to human beings being selfish, which although I would agree tends to be the case it is not true that all human beings are selfish since some of them are sincere in their desires, and his 2nd-law of thermodynamics argument fails.
- I suppose we can do reference counting on the data.
- Should also monitor for packages the reasons why experts recommend or do not recommend them, such as say for instance, why glimpse is not available through Debian (maybe there is something equivalent or better?
This is related to the gourmet equivalency functionality and could be explored as its own problem.)
- General thoughts: i am still disapinted in my lack of svre.
Until that glorius day, I shall think that It will be most difficult for me to do any work, yet I shall try.
I am feeling sick in the head.
- How do we search already installed packages?
- Some of the things we ought to do: for our FRDCSA release on the web, we should make sure to spider to download a lot of software and populate the archive this way.
We can take software from our different sources.
The first version should simply have the tools of the FRDCSA and not mention packaging all of this software.
Then when we have more money, and we are further along, we can release all of the packages and sell it again, perhaps giving a discount to those who purchased the first disk.
- Convince yourself that you know how to do this, cause your going to be the one doing it, please.
- I also dreamt that I was in the hardware store again.
Walking around looking after people, talked with the owner who was teaching me what to do.
- Whatever I do, they will react poorly.
They think this is wrong since they think, well clearly he could just do this and I would react fine.
But, they are wrong, since it is precisely those things which they want me to do that I am unable to do, hence, whatever I do, they will react poorly.
I cannot do what I cannot do.
For instance, they might say, be nice to other people, don't hurt their feelings.
I wish I didn't have Asperger's syndrome, but Asperger's syndrome causes me to hurt other people's feelings, against my own wishes and best efforts.
- You call it academic?
do you really understand the value of academia?
Who do you think wrote the program that performed the difficult task of planning and scheduling all the attacks in Iraq?
Most of the military is using the so-called academic systems to protect you?
- the formal definition of incentives, as a mathematical object.
How many more points must I refute you on before you realize that you have made at least one mistake.
do you know that one mistake is enough to conclude anything, right or wrong?
- Plan events so that they coincide so you can do things on their anniversary.
- Why do I feel better?
I think its because of the air conditioner, really.
I feel much better now.
- You could say, do you acknowledge that I am the stronger chess player?
do you acknowledge that I have studied mathematical logic in greater depth?
Ok, I know you are phenomenonally intelligent, but to be honest I don't think that you have been able to understand some of the finer points of mathematical logic.
I regret that I seem to have contacted you too late.
I was hoping that you would have survived West High without developing the mentality of a leader of a small pack.
And most of all, I regret that you do not care about the plight of other peoples and therefore I cannot care about you.
But if you want to tinker with some of the projects be our guest but as far as making a serious contribution I don't think you have the ability.
- You could say, do you acknowledge that I am the stronger chess player?
do you acknowledge that I have studied mathematical logic in greater depth?
Ok, wellwell I don't know how useful you could be to our group because I think that, although you are wonderfully intelligent, and have the capability of being an extremely powerful thinker, I don't think that you have understood some of the finer points of logic.
- Of course he doesn't accept Aspergers, because he doesn't accept who I am.
He has this image of who he wants me to be, and so do I. I can live with myself.
I can't stand him.
- Bad joke, we could call the places where Aspergers do their thinking "Asperger Concentration Camps"
- So, this is the reality.
A memcon with Kevin Reeves.
The first answer is always correct.
His first answer was: I like my life right now, I'm enjoying it.
I don't care if I die, so I don't care about helping other people.
(thanks) But mainly, the thing I noticed was that, based on limited data he jumped to a conclusion about what my purpose and motives were.
I agree that I haven't analyzed these things.
But that he doesn't understand what we are going to do with it.
He also is familiar with AI, this is good, and so we have to change our tune.
He is skeptical of my motives.
(Naturally, so in some sense this was a mistake.
He focused more on the death issue, probably as a result of his situation.
So, should we have been more careful?
Yes, definitely we are caught in an Asperger trap.
Thus Kevin has perished.)
- Could add support to the program for calculating, minimax style, scores for conversation.
Also, we would want to be sure to add support for calculating the suboptimality of the opponent.
Another thing is that, as the conversation gets larger, the search space increases.
Are we to fill out the entire thing.
Should we do "language modelling"?
How does this work?
Not very well I presume.
Another thing we can do is add an agenda to the system, with talking points that are specific to each thing, with various relationships between the discourse objects.
- do you think a snake would eat a burrito?
- Another thing is that, no one gets access to my thoughts files, but we describe the system we want designed, they do so, and then we integrate it.
- I've got the chance to do something here and you people are acting ignorant.
- Some of the evidence for why my theory of AI has validity: no one ever told me about AI.
I simply figured out one day how to do enormously useful things with computers, things that would satisfy my goal of improving the human condition.
Its not like I read it from some book and said, hey that's a good idea.
No, rather, I was thinking about how to do some very difficult social reasoning, when it hit me, that there did in fact exist algorithms which could perform this reasoning, and that they are realizable.
- Commission a study of what specific things to do/say to avoid giving this impression.
For instance, we can say we are only interested in having about 20 people.
Specifically construct things which we say which are designed to refute each and every axiom of a pyramid/etc scheme.
They try to act like they aren't.
Well, just because we try does not mean we are.
- I do feel bad though, but it is a result of these problems.
I mean I wouldn't ordinarily mind lifting the cart to get the duck tape, but after I fell off my bike, it now hurts to do so, so I now mind.
Same kind of issue.
I am now overweight and that affects things.
However, if these problems were already solved, I would not be overweight, right.
I know that proper diet is very important, and if followed, one would not get over weight.
I do enough work, even though it is not physical, that I should burn calories.
There should be guidelines on how to do this, but there aren't.
Which is annoying.
No one should have to go through this.
Understanding can be built up in a different way.
Logic would help, that way people wouldn't wonder whether they really understood things, since they would know.
- That I need to do more physical work.
- Should put certain sites on forbidden list.
Have our political analysts do that.
- So you know, you could be in a mech machine, and it would automatically avoid collisions, allowing you to focus on higher level stuff.
This is the problem with many games I imagine, is that they think it somehow is fun to have to do the stupid shit manually.
No what is fun is if you have automatic collision avoidance, etc.
- Here is food for thought for my critics.
People say you really understand a subject once you can explain it to another person.
In my case this is entirely false, since, although I may be able to understand the subject, I do not understand the person.
- Good influences: today has not been very good.
I mean I have some good ideas, but right now there is just too much work to do.
- Dad has to realize the inevitability of what I have to do.
- Perhaps identifying what we need is important.
What else do we need.
A debian defveloper, a lot of land, etc.
- Another thing Kevin and I need to do is to, having set this up, find an autistic child to be the "one", and to have all of our resources.
- Ideally I don't do anything except float in my underwater VR chamber.
- I should get a new system up and running, do something about being saturated by possibly useful information (perhaps simply record in and put it aside.)
Suffer from FSM problems.
- How do I check what packages are installed?
- Its easy to nag people to do things.
Its much harder to institute methods for doing these.
I asked Mom to do one thing for me, to contact Kevin Reeves, and she forgot.
Its obvious that either you don't care or you are not very well organized.
Either way, it reinforces my work on AI.
- Should not bias Kevin with our agenda, should simply tell him what we are trying to do, and see if he is interested and what he proposes.
Should tell him we are building an Artificial Intelligence.
Perhaps ask him to come to Pittsburgh for that event.
But, then again, that would only give him the really wrong impression, if he did come to Pittsburgh, we should really just hang out.
- looking at a screenshot of the game "enter the matrix", I was immediately struck that there are absolutely no cognitive aids to the game.
So the user pretty much just sits there not knowing what to do.
The neat part of the "matrix" film was the "intelligence" that people got from their operators.
Well, having a Soar like cognitive architecture with a dialog engine would be a neat way to make the game more interesting and goal based.
I guess games are just inherently stupid.
do Abstain Responding From-Now-On)
- You asked whether AI was dedicated to Ann.
What do you think AI is all about?
- Beethoven can write music, thank God, but he can do nothing else on earth.
- system should calculate to determine what other people are going to do, for instance, like fatso with the air conditioners.
- when reading manuals we should specifically create links between topics so that coming back to the item we know what to do.
- I am still a little confused on how one manages to upgrade a package to the latest version.
do they keep it in cvs?
- If you really know as much as you pretend about what I need to be doing, then why don't you go do it for me.
- But if we don't know anything about the myriad of tools they may be using, how do we do this?
Well, we can add partial coverage?
- here is something interesting.
the website, or whatever ought learn about the user and determine whether they know such things as: it is useful to read research papers from related projects and use those ideas.
Also, all information we present must be consistent with some plan of action along the lines of what we want the user to do.
- Here is how things ought to work.
Suppose we implement a stable perl module for reading configuration files.
Suppose further that there is some indication that we ought to standardize all of our systems to use this to read their conf files.
Then the user authorizes an order to do so.
The time required is estimated and the programmers involved are contacted and the process begins, almost like a scroll bar in progress, by the simple observation of programmers working.
- "The real problem is not whether machines think but whether men do."
- B. F. Skinner
- why do we need politics when we've got each other.
- file http://www.isi.edu/~lerman/projects/weatheragent/README.rtf includes instructions for obtaining Ariadne wrapper, perhaps we must do this from a CMU site.
- Send Florin Manoloche a link to my site.
Say, I was mistaken about my imminent position at CMU, I apologize.
I wanted to thank you for your help.
Here is my work as it has turned out.
It was the cyc-mode I was writing at the time you were so generous as to host my server.
(I needed the memory since my laptop was too small).
My mission is to make packages out of existing software for Debian, and maintain a repository of rough packages.
I think this will aid research since it will make it easier for people to compare contrast and use various packages.
It will make it as easy as saying apt-get install this or that.
So, just wanted to say thanks again, and to show you where I am right now.
Things are going well, we now have a project of 5 people and we are focusing on recruitment.
Obviously we look up to CMU as our role model, that is of course the deciding factor in our decision to do this in Pittsburgh.
- Am interested in making a distributed trust management system for knowledge.
So for instance, one gets their knowledge from different sources.
They audit their sources.
The program computes whether to trust the audited trail of information, to determine whether it has been tampered with and which sources are reliable, which specific information is credible, etc. The information contains assertions which are asserted by the party to be true.
So for instance, we want a system that does not allow an industry to corrupt our gourmet systems knowledge of diet.
So, their are two ways to do this.
The first is that health information is entered into the gourmet system from research studies and sources are cited.
And the second way is that gourmet actually learns for itself what specific diets are good for what people.
- its not enough to spell check, one must do grammar checking as well
- Well sameer, I don't think that I can contribute much to any organization.
Its for the better that I have not gained any serious academic employment, since, I really lack the ability to be effective.
Now, under highly controlled circumstances I can be quite effective.
I also know that you are not accustomed to thinking of my ideas as being particularly implementable.
We are reaching a point where we will get funding and so on, and I am simply worried that we won't adequately be able to determine whether or not it is better to do my work, or to not do it.
It seems to me that my work would aid our scientists, but at the same time, any other people who would wish it to be distributed.
So perhaps we should offer it to the government as classified, since that way they would have some control over it.
- Justin could consider to become a writer of propaganda, such as a book about a good robot, bad people who didn't understand their own importance, or what they could do.
- So some of the things that we might add to our system are the ability to track the trustworthiness of various developers and sites, based on their previous capabilities.
For instance, we really want to ask a host of questions: what is this persons past record of packages.
How well are they kept?
How long do bugs last against this person...
Etc. What misinformation does he have.
Who does he work for.
- some things that we could do to opencyc would be to simply get it into a position where we could run some basic scripts to start it up and to load it full of knowledge.
Then, it can answer questions by other agents, etc. Simply write our own scripts to manage the states.
- Picture the reaction of blind people to getting the pair of glasses.
Why do I need these?
- I am getting in trouble for things that I do not have the power to change, and unfortunately, I am also being attacked for attempting to create the power to change it.
- Well, I did a lot of cool reading today, however, I do feel a little down, maybe cause I don't know how my mind is coming.
I think I have t og o ahead and contine to write my software.
- For convenience predator 1.0 should just do things with scripts, but we should none the less make objects that represent steps.
- do you know how to go to the bathroom?
Well, I know how to make AI.
- if we fail generally too long, what did we do wrong?
we'll have lessons lerned.
- It had to generate.
I know it had something to do with this thing being in theway, the moniotr arm
- do more studying of core concepts such as case based reasoning
- Traditionally, planning systems are all about finding ways to solve the combinatoric problem of satisfying a few goals.
Why these goals are selected is not usually of any consequence.
What we really need is a system that understands what happens if, for instance, you do not pay your rent.
Another thing, is that it make sure that the daily average of some required thing exceeds what it needs to be, hence ensuring that you don't fall behind.
So, our system is for instance able to reason as to what consequences will happen if certain actions are taken or forsaken, and on the basis of this, determine what it thinks are important goals.
- My work with the critic system has made me extremely bitter.
I keep trying to fix nonsense bugs, but the system is too unwieldy.
I'll think its something and modulate this back and forth, trying all possibilities, until it becomes obvious that the problem is somewhere else.
This is a lesson, mainly, that we will have to break tasks down into more manageable chucks and verify them.
Divide and conquer.
The ideal way to do this therefore involves listing tasks, decomposing them, and verifying the completion of each subtask manually.
Inductively, so that when the system is run, we methodically check all possibilities until we have eliminated them.
I.e. we must be able to prove correctness of plans, etc. But in order to even implement this ability, we will have to apply these rules manually until we have some systems to do this.
This means interactively building up a system from small, but verified components.
- one thing that would be useful: how to specify a failure condition.
for instance, we need to get the airconditioners, now, any date set is arbitrary, what we really mean is that we need them before it gets warm.
Since we know it will get warm soon, when we don't know, we imagine that we need them ASAP.
So how do you quantify this need?
How do you judge it?
In a perfect system, it would determine belief in when it would be cold, and factor this against belief in how disruptive a particular amount of heat would be.
But how do you quantify that?
Well first, we can say that if it gets hot one day, then we have failed, to a certain degree.
While we cannot specify the amount of damage and hence cost, would we estimate it?
What we should perhaps set is "circuit breakers", or warnings that we have reached a particular class of problem.
For instance, environmental stability has been breached.
- How do I ensure that my plans are complete, in the sense that I will finish them?
- should get a minimal shops system written and deployed very soon.
it just logs sales, with a nice GUI, and prints receipts, etc. Should only do basic things, be very forgiving, and then we should scan in their entire inventory.
get this all planned out.
deploy it on a given day.
- must keep in mind, for machiavelli, or manager, that management is not about telling someone what to do, it is about causing them to think in a way that makes the do it automatically.
- we have to do something about when we patch and adapt other users software.
- I'm getting a second gust of wind and its helping me to figure out how to do things
- look into the configure programs since they seem to be pretty similar to what we need predator to do.
- here is another thing to do, simply operate on the ones in this convenient unilang format
- The new institute should never mention my name, and should have nothing to do with FRDCSA.
- scan all their inventory with prices, etc. use various online suppliers to do the reordering.
- can use pse to generate very current agenda data structure, so for instance, it pops up questions: how do we store ftp ls-alr files?
- GAME OVER!
That's all we have to do.
Approve it with Al, figure out a place to put everyone, get it all set up.
And there you have it.
- well, lets give a go ahead for the system and just have it sort all of these anyway, and then see what we can do with them.
- The question of whether I can create AI or not really depends on what you do, whether you are willing to let me do it, or whether you're going to continue screaming in my ears when I'm in the most important "chess match" of my life.
The fact that you think I should get into chess shows how little you understand.
AI is a thousand times more interesting than chess.
I'm insulted that you don't trust the chess expert to know these things.
- could use recorded screen shots or video, etc, to help a person categorize what they were working on, in order to aid recall in segmenting time according to what the user was doing.
Also, probably a simple keyboard recording, or system activity should be able to do this automatically.
Of course it should, you know.
- I just manually redid all of these FRDCSA.xml entries, which is total stupidity, since it would not have been very hard to write a program to do it.
I am thinking like someone else.
The thing is now we have wasted all that time and we don't even have a program that could be used with auto-package for automatically generating these!!!!
- In a separate part of the dream we were filming some sort of movie about a town square.
and something about blocking someone from taking over the square, and this was done effectively with long white wide but thin and hollow bambo poles.
Then we got on a bus and as expected few people chose to sit by me. Then I realized my mistake at having taken the bus since it was to snow and we still had to drive through PA. The last scene involves what happened to me when they explained to me how to play the game.
They wondered why I was interested, and I said a bunch of stuff.
There was an older lady there, I think it was Ira, and she asked me some stuff about what we were going to do with the game, and I said that if we could build nuclear weapons with it we would.
I said a whole bunch of stuff, and one particular child was very impressed and wanted to learn more.
So we were going to teach him to think like us.
- one thing we can do is write a system to find synonyms using wordnet for acronym generation.
For instance, if we needed a word that meant culinary but started with a U.
- I can do that now in fact, and.
- autopackage can do semiautomatic extraction of dependency information from readme files (from experience, I had errors resulting from unsatisfied dependencies).
- we could have a list of "criteria for coming home", which lists all of the do's and don'ts, what constitutes material breach of this list, etc.
- screw you for putting me in the position for having to do that to you.
- This HTN should be put on the projects web site as well, although, who do we really fool in putting this information up?
And moreover, perhaps we would then be subject to attacks.
Therefore, we should find hosting of our project relatively soon, through pair.com, or whatever.
- use the metadata for the cache as input to critic, it was funny, cause I was thinking how arbitrary it was to simply save the cache metadata as an XML file, and that the cache metadata file ought to do something, and then, quite naturally, we have the concept of feeding it to critic.
- how do we prevent a user base from tying our hands?
- with gourmet, also do expected cost
- should use interface diagrams that should how each system works, like the big projects do, to convey the principles of the system, and put these on the project web page
- note, meetings with Justin must continue to be restricted.
Note he still continues trying to tell me what to do.
He says I need exercise.
- I was just passing out right now, and I saw something, where they had split the thing into two parts, one like sg, and the something else, and I thought, wow, that was smart, and then I remembered it was me who came up with it.
The idea happened so fast I have no idea what it refered to, but I am pretty sure it was correct.
I think basically it had to do with the notion of losing in chess, and you'll find that at a certain point the person hasn't realized the refutation for their line.
- similarly even filing legal stuff is not worth our time.
basically, I should monitor the status of my papers, but otherwise, I should have nothing to do with gov
- notice how Ken has routinely tried to make me think I can't do this.
- obviously if this is what my mood is like I can't do it.
- You probably think I have had no success with AI, where is it, you probably say.
First of all, you cannot understand it, you cannot see it, so you cannot reach a conclusion.
But do you like to jump to conclusions?
I have had much success, great success, but this has no doubt been mitigated by the fierce enemy tone and complete lack of any assistance.
- should do work on a general purpose game playing system.
simulate social situations, etc.
- I said right now I am filling up the ocean with a teaspoon.
I also said that it is easy to fill up the ocean, all you have to do is let the rivers open.
The rivers are damned however!
There are lakes of software, just waiting to be released!
- I should have it record a log of all commands issued when making a particular package, and then do logfile analysis and have commentary from the person trying to group these into abstract tasks that can be computed in someway from the package contents.
- as part of our packager, we can also do an automatic rating system which analyzes the extent to which one can expect to have the given software perform a useful function, and also as it stands wrt our goals.
- in reading the online logs, I find that the arguments are quite silly, and mostly resort to name calling.
What would be useful for an online system is our system that identifies logical fallicies in arguments, and would also force arguments to be formal.
To do so would require large libraries, but this is not impossible.
So that, an idea can be refuted intelligently, instead of just bickering.
- contact scott gar or whatever from advogator and tell him that we intend to create an organization whose sole purpose is to recruit programmers to dish top notch code to support the open source effort.
This will be entirely funded by another business which exists solely to do so.
Our advantage will be in the quality of our software engineering processes, our extensive software libraries, and so forth.
We are extremely interested in this but agree that we cannot find people who are qualified and take the threats seriously enough, since, after all, they don't really care about intellectual freedoms, since if they did, they would get a helluva lot more organized.
- Tell Al the main problems so far are we don't have gourmet done and inventory management done, so we can solve the food problem.
We are tempted to use an off the shelf system, but they never do what you want.
- if we are ever interested in hacking, I suppose we should create a library of hacking software, but moreover, we should attempt to reproduce vulnerabilities, and actively crack test systems.
Having done this, we could then break into other systems according to a plan.
It is highly unlikely we will ever find the time to do so.
- if an agent volunteers to do something, their success or failure should go into the capabilities matcher
- The FRDCSA thesis holds that given the amazing range of software that now exists, there is an enormous opportunity cost associated with beginning a project before checking whether equivalent or related functionality exists.
In order to do this requires integration of existing systems.
A subthesis holds that given the amazing range of free software and open source software, that we can rapidly integrate these systems by only writing a small set of tools, designed to be subsequently modularly replaced as we find the real technologies.
The aim of these tools is to package software for Debian GNU/Linux, since this system has very good distribution of software.
- $CPAN::Config, that is how we should do config, have $radar::Config, etc.
- do I want to read the documentation about these systems?
- have frdcsa-perl do two things, first, reduce to most basic ones, but also, check whether it exists before searching for it.
- and they way to do that is to not have the conscious participation of the messenger.
- the only way to do this would be to communicate through entirely secret channels.
- so this is how you like to do things?
fine, I cannot help you.
I am already in complete seclusion.
I will reemerge when finished.
I'm sorry, you leave me no choice.
- absolute isolation will be achieved but only when I have sufficiently finished the planning systems that it calculates how to really do this.
- I just realized here that complex social phenomenon, and the role of ignorance and mistaken reasoning, do not hamper logic, but are instead simply cogs in the wheel of complex behaviour generated by simple acting machines.
Of course, the complexity being high, we are in a superior position to begin calculating the states of the machine.
The machine itself is, unlike individual people or even society, extremely complex.
The methods brought about in mathematics begin to approximate some of the structures of the machine.
In essence, it is the machine of platonic reality.
Therefore our goal is to get at understanding its functioning, so that nothing it does surprises us.
- But the opposite is also true.
The moment any one steps forward and helps AI, we will achieve what we are trying to do.
And this is the last thing that anyone there would do.
- I also last a bunch of dat from mz log.
I suppose it should log things on the go, or does it do that alreadz_
- extend our simple script in frdcsa-perl to do lookup on our partial list (perlmod)
- need to finish critic program to do this
- I figured out how they do that stuff for boss, where they work on it and test it and also can install it.
this is simply done with something like CVS checkout.
when they need to they build it to fit in a directory, and otherwise, then they just package it as a source.
- do trash take out
- email peter jansen with questions about the gourmet system: what do you think of this information, can you suggest anything related to making this better, health information, using KB to plan health, legality of product barcode?
- widestudio to do quicker GUI development?
- obviously do CPAN stuff before gourmet and Reviewer
- before complaining to authors to do more, first package and release their software
- in general we need to contact open source developers to assist with projects, filling roles in projects, etc. I should join as a developer for some other projects to do this.
- I think I should no longer do any recruiting whatsoever.
Put your best foot forward.
- What a trap.
They do not realie how important this is. If they realized how important this is, they would not be treating me with scorn, shame, etc. They would not be acting like I was full of it, and
- here is something thieves do, appeal to their supposed good nature
- What a person says, and what is the truth, are completely independent.
Sameer said he didn't do it.
I don't trust him.
- This leads further to another idea for autopackage.
That is, when autopackage wants to do something, like install binaries in /usr/sbin/, or do certain scripts, it must be authorized to make these changes.
In this way, autopackage can implement security features, by validating potential installs against software that recognizes exploits, like that recently controversial system I read about on slashdot a few days ago.
I.e., lots of possibilities here in terms of organizing installs.
And furthermore!, we can run debs through autopackage to determine how they check out wrt security..
- tell Al, that one method of doing things is that I write scripts to do what I need.
when the scripts perform what they are supposed to do, I have the resources to go ahead and refactor the functionality of the scripts into the project which coordinates a set of related capabilities, and write them with lasting structure, making a library.
Then all applications can use these features.
As far as automatically building libraries from not just my own code, but other people's code, that, too, is a capability we are investigating.
- Perfect example of how what I am doing is applicable.
However, my methods are not necessarily complete, and they do require assistance.
We're morons if we don't get Kevin Reeves involved with us.
Yet, at the same time, we should not disturb my isolation.
- Well, we run into a little problem here and there.
How do we effectively search, aquire, package, and use software.
- Things usually succeed around the same time.
So, this means, when I achieve AI, I will most probably have reached the last step before discovering other people's AI.
What could this last step be?
Perhaps it is to give up work on my AI, since that is the last thing I would do.
But I don't know whether this is true or not.
- I've noticed some programmers do not care much for doing things of practical value.
But they don't know what is, so, we can give them items that they think are fun but are of great importance.
We should walk backwards, lest they suspect this approach.
- A natural thing to do is use manager to initiate shifts between fronts and goals that are generated by pse
- Digest is related to the surjective file system search, but this is relative I believe to current goals, and thus the problem of deciding what is relavent is not solvable given limited knowledge.
How do we ensure that if we have seen something, and later have a need for it, we are able to retrieve it?
- First of all, the field of AI is alive and well.
Whereas people who don't understand how AI works, and what it does, were most probably let down by the 1980's AI depression, the fact remains that people who were successful with AI have been extremely successful with it.
And moreover, many people object to my AI, saying I don't know what I am doing, and I think this has more to do with their ignorance and my difficulty with communication than with my actual methods and sources.
- Here is a question of probability theory that I want to know the answer to.
Suppose you have a person who died, and you don't know whether they were died of murder, suicide, or an accident.
Knowing nothing else, should not the probability be equiprobable of each?
This means that they are independent.
Now, given this, since murder and suicide are deliberate, what is the probability that she died deliberately.
Knowing nothing else, should this not be 50/50?
Well, I would imagine that we do know something else and that the options are 2/3 deliberate.
So, I would like to understand how you solve this riddle.
I think basically, deliberate in this case is a dependent variable.
- I am upset because I have had to watch movies and go driving so much lately.
I really need to sit down and start programming again.
Now is a great time to do so.
- do research for Justin on a computer and an expresso machine.
Look at Ben's bargains.
- Most of the time life sucks and then you have sex and then its ook for a second, but all you do is create more life.
AI is like permanent sex because it is so effective tha.
- How to do this exactly I don't know but is worth looking into.
- we could do NLP analysis ofuser manuals to extract linguistic systle guides.
- There are several imporatntc considerations in seeting up shops to run your store.
The first is to Teh first consideration is installation of the computer system runing SHPS . If your To do this, you need to insert the CD into a system.
WARNING: YOU WILL LOSE all DATA ON THE SYSTEM.
shops REFORMATS YOUR SSTEM.
Only use a PC thatyou don't need any more data an.
- Another thing to do is to get the monitor stand, alsthough I still don't know how to attach the laptop, and also how to.
Well, I think I can run to lowes to see if I can't build myself a stand, although for the time involved it might be better, you know.
I could also run to best buy.
They are both closed.
- The main things I need to, in addition to a lot of work at the store - however, once we're done there we should be well to do - is the following.
- I think the smart thing to do is maintain strict radio silence.
I said I was coming back, etc. etc. They can't get a hold of me. This means no AIM.The basic reason he wants me to come home is because he lost a child.
- Actually, and you have no reason to worry about this, there is again nothing he acn do.
You should focus on programming.
Maube you should send a letter saying that you are not ready toc ome home.
- Thi sis a problem of course in that omy parents want to see me. Why?
Why can't they do what is best for me? Must it come to blows again?
- Is there a method to do this/
- Basically, programmers in our group ought to normally do what I say.
- do something about darkness faced with when having to come back from the bathroom at night.
- Message to Al, progress is forthcoming.
System is doing well.
Want Al to consider contacting and recruiting Kevin Reeves for a startup company -do not mention my name, or no?
- One way to do this is to implement tests that the activity is going on For instance, a tool in emacs to set triggers for events, can then tools can mine event logs to determine that the user is, for instnace, writing in an unstructured way.
Or, for instance, a simple case would be a web site proxy that makes sure the user is not visiting bad links.
But in general, the notion that checks must exist for every bad habit listed will be usefulfor eliminationg bad habits.
- The trouble with Asperger's syndrome is it is very difficult to detect.
It is also related to Tourettes syndrome, and that is partly responsible for the unintentional mistakes that we make.
Now, in assessing my difficulties, I realize that it is quite simple, and roundly beneficial, if I were simply to write programs which perform these tasks flawlessly.
Computers perform chess flawlessly as compared to most people and they can do a great deal of good in other areas as well.
No strawman fallacies waged against AI can convince me of the overwhelming evidence I have assembled as to the true capabilities of computers, and it is a point of fact that there has been no task which has been demonstrated which has been proven to be impossible for a computer but not humans.
- Here are the things to do with respect to gourmet
- Id efintely should have tnmtoghing to do with him.I got a bync of discontentment of biological models.
- do youk now what I am afraid of.
The c clamps won't show, so unNo itmkI don't not have fun with.
How man weapns do you need.
- +HEre is an anlogy that implies the model of the brain used b AI people is wrong.
Of course, the annogy is dictated by feeling and nothing to do with anything else.
- I think that this is imbuing stupid old fashion to me. what a shame to have to do so now.
- \I am sick and tired of getting the go around.
I don't believe him anymore.
He asks me what I need him to do, I tell him do meal planning and he laughgs.
Although , of course, if it were anything more complicated than meal planning he would say he didn't understand it.
- His attidutes are dangerous and subversive.
They mean nothing.
They really do.
Who could defend such a hopelessly retarded position as his.
- Of course the goal here today is to make it impossible for them to say or do anything that would be harmful.
To this end, I will meet them at CMU, and will not get in a car with them or anything.
We will walk to where we need to go for dinner, etc. I should have some materials printed for them to read.
- Thought Ben afflect was programing and he didn't know his stuff and was using a "card trick" which had something to do with 0.
- Its nice to do tests where otheres have over looked.
- One thing we want to do is record dreams, and then see if these can serve as valid statisticaly predictors.
- Should ask Justin to do some nutrion research, but have to phrase it in a way that is useful.
He could do the artwork for it, for example.
- One thing I forgot how to do is to refine goals into subgoals.
Also, you might want to consider that when something dclares a dependency on a goal, it might actually declare one on a subgoal.
- I am seriously overweight now, so is justin.
We really need to figure out this cooking thing.
I don't know how to do it any faster.
I keep trying to get it done but it sseems to take days and days.
- Moreover, we ought to look into complex NLP techniques as well.
But this implies that we are already finished, so there has to be some sort of better way to do this.
- So clearly, we have a lot of work to do in order to formalize things.
- If we do formalize everything, this has the advantage that we can translate it among our tools.
- I mean, do you simply have a set of things that it can refer to.
- Yeah, so anyhow, it calculates what is most effective for you to do.
This page is part of the FWeb package.
Last updated Sat Oct 26 16:51:18 EDT 2019