promethea.incorporated

Rationalists, we need to talk about models

wirehead-wannabe:

h3lldalg0:

Epistemic status: not a professional, but I was the sort of kid who asked for the DMS-IV on her fourteenth birthday.  

This is a rant, I’m not grabbing sources at the moment, I’m just making some observations about things that are starting to bother me. 

I really like the Less Wrong/Rationalist crowd, and I count myself in this group. I can’t shake the feeling, however, that there’s an unfortunate mix of contempt and fascination with the softer sciences.  On the one hand, contempt for the academic system in general and for the value of soft sciences in general.  As though once you see that a field has issues and generates a certain amount of bullshit, it will be trivial for a smart person to fix.  

At the same time, subsets of this group seem to be in love with models like Kegan’s development levels, the MBTI, “amateur sociology,” and other things like that.  To the point that I find myself in conversations where these things are taken as far more universal than I think is epistemically responsible to claim.  Where, when I say something like “Kegan’s development levels might be locally useful, on an individual basis, but I think there’s risk of overfitting if you try to apply them everywhere,” I get the equivalent of “You haven’t done enough reading on this/this isn’t a problem if you’re applying it correctly.”  The second response is nearly verbatim.  It’s from a conversation where I tried to point out that Nonviolent Communication techniques could be the wrong tool to use in some situations.  

I won’t pretend that I’m above this kind of navel-gazing. I like taking personality tests and getting the results, seeing if they match my experience.  I like speculating on sociology, or wondering casually if certain fictional characters are sociopaths.  I’m also not above reading an interesting thing in a pop psych article and trying it out, seeing if it has value for me.  But I think that without large swaths of evidence (we’re supposed to be all about base rates and evidence, right?), value found in ad-hoc models written by random bloggers or psychologists from the 1980s should be locally applied.  Attempts to apply them universally should be treated with skepticism.  

More than that, I see a lot of people ignoring the potential for anchoring.  It seems reasonable to me that if you are attached to certain model, you have an incentive to make all of the evidence fit.  Or rationalize evidence away.  Or ignore people who disagree with your pet theory’s claims.  This seems to be a really dangerous habit for so-called “rationalists” to get into.  A theory or a model is only as useful as its results.  It’s only as empirical as its predictive value. 

So long as we acknowledge the infallibility of sortinghatchats I agree with this.

2 weeks ago · 84 notes · source: h3lldalg0 · .permalink

1. almostcoralchaos reblogged this from jadagul
2. vulpineangel liked this
3. angrybisexual liked this
4. zamboni-whisperer liked this
5. cytelric liked this
6. digitalfixation liked this
7. daniel-r-h liked this
8. michaelkeenan0 liked this
9. ozylikes reblogged this from skeletontemple
10. ozymandias271 liked this
11. zeteticelench liked this
12. rosetintedkaleidoscope liked this
13. neurocybernetics liked this
14. inferentialdistance reblogged this from wirehead-wannabe and added:

    Different sub-sections. One group wants to play with all the neat toys (i.e. models) in the softer sciences. The other...

15. shellcollector liked this
16. phenylalaniiine liked this
17. plain-dealing-villain reblogged this from dagny-hashtaggart and added:

    Ironic, considering that Taleb is terrible about this.

18. michaelblume liked this
19. theauspicioussuitcase liked this
20. turingtestflunker liked this
21. laropasucia reblogged this from neurodiversitysci
22. creativepup liked this
23. skeletontemple reblogged this from mitoticcephalopod
24. derplefurf liked this
25. ghostopossum liked this
26. davidsevera liked this
27. finestoftheflavors reblogged this from dagny-hashtaggart and added:

    Well, well, you see, we can’t really help it. As INTPs, it’s in our nature.

28. neurodiversitysci reblogged this from dagny-hashtaggart and added:

    Because social sciences are all “fluffy” to them, they can’t tell the difference between valid and invalid social...

29. neoliberalism-nightly liked this
30. spiralingintocontrol reblogged this from gruntledandhinged and added:

    A lot of people have responded to this with “but nobody really takes the MBTI seriously,” and that’s… kind of fair? But...

31. lisp-case-is-why-it-failed liked this
32. spiralingintocontrol liked this
33. injygo liked this
34. coselick liked this
35. unknought liked this
36. give-some reblogged this from socialjusticemunchkin
37. give-some liked this
38. not-a-lizard liked this
39. automatic-ally liked this
40. theungrumpablegrinch reblogged this from drethelin and added:

    - George Box

41. h3lldalg0 liked this
42. dagny-hashtaggart reblogged this from h3lldalg0 and added:

    Makes sense. I’m reminded of Nassim Taleb’s discussion of nerds (i.e. people with a compulsively systematic cast of...

43. polyarche liked this
44. neurotic-biotic liked this
45. chamomile-geode liked this
46. lyycernment liked this
47. h3lldalg0 reblogged this from dagny-hashtaggart and added:

    There are definitely people who are good about this. Maybe 10% of the conversations I have about psych involve this set...

48. jadagul reblogged this from dagny-hashtaggart and added:

    I pretty much agree; probably the first (and most successful) “serious post” I wrote on this blog was on this subject.

49. Show more notesLoading...