Expansion of "Cached thought" wiki entry

post by Davorak · 2010-12-16T07:27:07.332Z · score: 6 (7 votes) · LW · GW · Legacy · 14 comments

"Cached Thought" wiki entry has been copied below for you connivance.

 

cached thought is an answer that was arrived at by recalling the old conclusion, rather than performing the reasoning from scratch. Cached thoughts can result in the maintenance of a position when evidence should force an update. Cached thoughts can also result in a lack of creative approaches to problem-solving if one repeats the same cached thoughts rather than constructing a new approach.

What is generally called common sense is more or less a collection of cached thoughts.

 

The above entry focuses only on the negative sides of cached thought. Probably because it can be a large barrier to rationality. In order overcome this barrier, and/or help others overcome it, it is necessary to understand why "cached thoughts" have been historically valuable to our ancestors and in what fashions it is valuable today.

'''Cached thought''' also allow for complex problems to be handled with a relatively small number of simple components.  These problem components when put together only approximate the actual problem, because they are slightly flawed '''cached thoughts.''' Valid conclusions can be reached more quickly with these slightly flawed cached thoughts then without. The aforementioned conclusions should be recheck without using '''cached thoughts''' if a high probability of correctness is necessary or if the '''cached thoughts''' are more then slightly flawed.

Is this an appropriate expansion of the wiki entry? The words are drawn from my observation of the world. How else should the above wiki entry be expanded?

14 comments

Comments sorted by top scores.

comment by Vladimir_Nesov · 2010-12-16T11:37:26.918Z · score: 14 (14 votes) · LW(p) · GW(p)

Given the occasion, I'd like to remind everyone that our LW Wiki fell into lack of development of its content for almost a year now. Nobody improves the pages, and I merely perform passive maintenance. It'll move forward only if it gets some people obsessed with improving it.

comment by David_Allen · 2010-12-16T19:14:01.758Z · score: 4 (4 votes) · LW(p) · GW(p)

I think discussion posts like this one are a good way to highlight wiki entries that need work, and to direct attention to the wiki pages in general.

comment by Vladimir_Nesov · 2010-12-16T19:24:34.214Z · score: 2 (2 votes) · LW(p) · GW(p)

Wiki pages don't need directed attention right now, they need general attention. There should be people with a goal of improving LW wiki, not so much people with a goal of improving specific pages on the wiki.

comment by David_Allen · 2010-12-16T19:37:20.893Z · score: 3 (3 votes) · LW(p) · GW(p)

I'm suggesting that directed attention can lead to general attention, but perhaps I don't understand what you mean by general attention.

So, what kind of general attention does the wiki need?

comment by Vladimir_Nesov · 2010-12-16T19:56:22.895Z · score: 7 (7 votes) · LW(p) · GW(p)

So, what kind of general attention does the wiki need?

Improving quality of the articles. Most articles are stubs, significant portions of articles don't contain any text, they only link to blog posts and other articles. These are the stages of LW article development that I see:

  • Includes links to blog posts that present the idea ("Blog posts" section; this is minimal stub)
  • Links to related wiki articles ("See also" section), and has incoming links from them (a minimal stub located within the wiki)
  • One-sentence summary of the idea
  • One-paragraph summary of the idea
  • Summary that discusses related ideas and links to other wiki articles from the textual description, and not just from a "See also" list at the end
  • Article itself links to original sources, not just to blog posts linking to those sources.
  • Article summarizes arguments for usefulness/correctness of the idea, not just defines the idea.
comment by David_Gerard · 2010-12-17T00:29:50.715Z · score: 0 (0 votes) · LW(p) · GW(p)

I suggest something top-level in the main section to inspire people.

comment by David_Gerard · 2010-12-17T00:29:05.869Z · score: 0 (0 votes) · LW(p) · GW(p)

I see you've supplied the last few mainspace edits that weren't a diplomacy game, so well done ;-)

comment by David_Allen · 2010-12-16T19:01:45.391Z · score: 6 (6 votes) · LW(p) · GW(p)

I agree that the entry appears to ignore the positive utility of cached thoughts, but I think what the entry really needs are descriptions with references to primary sources, Wikipedia style. What aspects of "cached thought" are original to Eliezer and what aspects are discussed elsewhere?

If this is more work than you intend to do, then I think a portion of your suggested change would improve the entry.

''Cached thought'' appears to be related to dynamic programming.

You capture this idea nicely with:

''Cached thought''' also allow for complex problems to be handled with a relatively small number of simple components.

However I am less certain about this:

These problem components when put together only approximate the actual problem, because they are slightly flawed '''cached thoughts.'''

This statement could generally be applied to all thoughts, not just cached thoughts. In general we tend to reason using models that only approximate reality.

For a cached thought to be currently valid, the model that was used to generate it must be valid within the current context.

Valid conclusions can be reached more quickly with these slightly flawed cached thoughts then without.

I would say that conclusions can be reached more quickly with cached thoughts than without. The validity of the conclusions would be highly context sensitive and I'm not sure you can generally claim anything about it.

comment by Davorak · 2010-12-17T02:44:52.055Z · score: 0 (0 votes) · LW(p) · GW(p)

You have a good point that my analogy is too broad. I was trying to express that it would take inordinately long to solve complex problems with out depending on "cached thoughts" and that the cached thoughts can be flawed yet still contain utility. It is inappropriate relance on cached thoughts or treating them like facts rather then guesses that gets people in.

comment by David_Allen · 2010-12-17T05:35:47.192Z · score: 0 (0 votes) · LW(p) · GW(p)

and that the cached thoughts can be flawed yet still contain utility

I suspect that this is true for some contexts, but making this claim would require careful explanation. For the wiki I would like to see statements like this backed up with primary sources.

comment by bogus · 2010-12-16T08:26:28.314Z · score: 3 (3 votes) · LW(p) · GW(p)

Good point. In addition to your proposed expansion, the wiki entry should also reference prior literature, such as Martin Heidegger's What Is Called Thinking and its discussion of the "unthought frontier".

Also, a good rule of thumb is that "related articles" should be discussed in the entry rather than simply listed at the end. In this case, this means that the article should compare and contrast "cached thoughts" with "semantic stopsigns" and "groupthink".

comment by PeerInfinity · 2010-12-16T15:54:35.249Z · score: 2 (2 votes) · LW(p) · GW(p)

random trivia: I recently noticed that "The concept of cached thoughts is the most useful thing I learned from Less Wrong" is now a cached thought, in my mind.

comment by Eugine_Nier · 2010-12-16T07:34:21.036Z · score: 2 (2 votes) · LW(p) · GW(p)

You do realize the wiki pages have discussion pages.

comment by Davorak · 2010-12-16T07:48:17.789Z · score: 7 (7 votes) · LW(p) · GW(p)

Yes I was, but the discussion pages seem less conductive to actual discussions then here where threaded conversations are easy to conduct.

In addition, I had the impression, with little evidence, that it would be inappropriate to put personal observation into the wiki. Which can be confirmed on the wiki main page: "This wiki is not for original research - that is, it is not for directly discussing topics or concepts that have never been discussed on the blog." The idea I took away from this is that there should be some sense of consensus before adding topics/ideas to the wiki and this seems to be the place where that happens.