Towards an Algorithm for (Human) Self-Modification
LessWrong is wonderful. Life-changing. Best thing that ever happened to me.
But it’s not really enough to make one a rationalist, is it? I don’t assimilate or even remember all of the knowledge contained in what I read, and I certainly don’t dynamically incorporate it into my life-strategy.
Say you want your computer to be able to open Microsoft Word files. In order to do this, you do not upload a PDF which contains a description of how Microsoft Word works. No, you install the program and then you run the program.
Over several months of reading LessWrong I found myself wishing I had (a) computer program(s) that could train me to be a rationalist instead of a website that told me about how to be a rationalist. I would read an article with a tremendous sense of excitement, thinking to myself, “This is it, I have to implement this insight into my life. This is a change that I must realize.” But I would inevitably hit a mental wall when I saw that just knowing that something was a good idea didn’t actually rewire my brain toward better cognitive habits.
I wanted a rationality installer.
I found myself in the midst of a personal crisis. I came to suspect that the reason for my unhappiness and akrasia was that my goals and my actions had become decoupled—I just couldn’t figure out where, or how.
So I set out to make a program that would help me organize what my actual terminal goals and values are, and then help me causally connect my day-to-day activities with these goals and values. The idea was to create a kind of tree with end-goals at the parents and daily tasks as the children. The resulting application was not very user-friendly, but it still worked.
With the help of my program, I saw that a year ago, I was very happy with my life because all the activities I pursued on a daily basis were very high-utility and directly connected to the achievement of my goals. I saw that I had recently formed a new long-term goal, the existence of which altered my utility function, but I had not altered my life to sufficiently accommodate this new goal. I made some changes in my life which I thought were going to be painful sacrifices, but ended up feeling exactly right once I crossed the threshold. It shocked me how quickly I felt better, how completely I returned to “normal.”
And I thought to myself, hey, why do our cognitive algorithms have to actually be inside our heads? I implemented this one into C++ and it helped me sort out something which was just frustrating and painful and confusing when I tried to manage it on my own.
What other rationality techniques deserve to be coded into “rationality assistant applications?”
(And how much of a desire would there be for such products?)
Are you willing to publish your code? Or is it already published somewhere?
I’ve been interested in “exoself” examples for a while—mental prostheses, todolists and personal information managers. And I’d love to examine an example.
Or if it is too personal, I understand; no pressure.
Seconded—even describing what, exactly, it does would be incredibly helpful. As in, if you don’t want to publish the code, describe the functionality, algorithm, features, etc.
@Johnicholas: Mostly off-topic, but in response to your software interes… I use org-mode and it is abso-freaking-lutely unbelievable. If you’ve been into todo lists, organizers, and the like… please tell me you’ve found that. I learned emacs just for org-mode, and that’s saying a lot about the motivating factors involved.
I’m sure you’ve seen todo.txt and the more advanced spinoff, taskwarrior. If you’re on a Mac, taskpaper is pretty cool. Lastly, timetrap, a dead simple time tracker for seeing where your time goes.
Of course, with org-mode, you can take notes, track todos (including assigning deadlines), log where you spend your time, export to html or PDF (via LaTeX), integrate with gnuplot, ditaa, and tikz for graphics, and loads more. Then again, they always did say emacs was the text-editor that could also cook a pizza for you...
Org-mode made me switch to Emacs from Vim.
The one stupid productivity hack that has stuck with me for several years now is turning on a clock in org-mode for the task I’m working on, and then having to work focusedly on that task until I turn the clock off.
I do use org for keeping random diary notes and wiki-style notes and for doing quick spreadsheet stuff with the lightweight spreadsheet feature.
I found this massive article on workflows and customizations of org-mode pretty interesting.
I’ve heard org-mode is awesome, but I haven’t tried it yet. I hadn’t seen taskwarrior, taskpaper, or timetrap yet, thank you very much!
No problem—the org-mode mailing list is phenomenal and it’s incredibly useful for all sorts of things. Give a holler if you’d like any more discussion about things like this. I’m sort of an open source nut :) Researching programs can be addicting. For something with less of a learning curve, TiddlyWiki is fantastic as well. I just thought of that one. About the only thing (besides exporting and integration with all of those other “modules”) it didn’t do for me was task tracking. As far as taking notes and keeping track of things, it’s awesome. Like a personal notebook in wiki form.
Ok, enough out of me. If you need help setting anything up, send me a message and I’d be happy to help. I wonder if a post specifically on organization programs would be useful...
I’m using TiddlyWiki now. :)
If you’re interested, I can send you a blank version that I modified for use at work. It contains a lot of plugins and HowTos built in. I had some interest from coworkers and took what I’d been using, stripped out all of my information, and wrote tutorials and explanations to help new users get into it.
I haven’t used it in over a year but still have the file. That also means I don’t quite recall all of the functionality that it contains! At the very least, it looks waaaay better because I used the appearance from teamtasks (which has todo functionality anyway), it has contact support from twab, and lots of stuff from some of the amazing plugin sites out there.
It ended up pretty darn cool, though to be fair I was using it to track research experiments for Intellectual Property documentation, not necessarily personal things. Since I found org-mode, I have seen the value of tracking personal life and thus, I think if I had the same revelation a couple years ago, I’d be using TiddlyWiki for such uses as well.
Let me know if you’re interested in my version—I can post it on google docs and give you a link.
Thanks! I will raid it for features and modifications—these tools are so personal, everyone wants to control and customize their details.
Back at work—the file should download for you by clicking THIS. Let me know if it doesn’t.
Thanks much!
This article would’ve been a lot better with examples, yes.
this is why a community that shares your values is so important. I mean a physical community and not just a virtual one. Having others there to point out low hanging fruit and then positively reinforcing your attempts to reach out and grasp it is essential.
There should be a LessWrong Seastead.
I agree. Actually, I do have at least two close friends who I would consider “very rational,” but we have known each other for so long that we can be blind even to one another’s irrationalities. You get used to your friends in the same way you get used to yourself. I think you need not just a community, you also need to meet new people who can look at things from new angles.
All this programming exercise really did was enable me to see various aspects of my life on paper, in a clinical and detached fashion, as if I were looking at the life of a stranger. From that perspective, what I needed to do seemed obvious, just as the solutions to other people’s problems are usually more obvious than the solutions to our own problems.
Personally, as I first read through the sequences etc, I basically developed a case of man-with-a-hammer-syndrome for everything I read. I’d read about positive bias, and start looking for every possibly way to apply that. Pretty quickly, that habit became ingrained, I’d stick the metaphorical hammer into my tool belt, and then I’d read something else that would cause me to develop man-with-a-screwdriver syndrome, and repeat the process.
I do think I made significant progress. The problem is, I have nothing to compare my current state with other than my previous (weaker) self. I’d be very interested in trying computer programs, games, etc and see if/what better results I get.
Suggestion: Would FreeMind (open-source mindmapping software that produces trees) work for representing your graphs?
Request: Concrete example of some links.
Do you find FreeMind supple enough to represent what you want it to represent?
In my case? Yeah.
I’ve used Compendium before, which is pretty good as it is and has some high ambitions - http://compendium.open.ac.uk/institute/
such as dialog mapping.
The feature that I like the most was different types of nodes.
Or xmind (free version available, despite being commercial), or perhaps better yet, gephi, which can interlink, network, and filter data.
EY: FreeMind is great… but it’s always nice to have other options :)
Well, the difference is that when you’re talking about memes, the line between “data” and “program” is fuzzy. All data is potentially active data. But yes, some data is more virulent than other data.
Is there any reason you couldn’t put your program up somewhere for people’s interest?