Think Before You Speak (And Signal It)

post by Wei_Dai · 2010-03-19T22:21:12.297Z · LW · GW · Legacy · 40 comments

In deciding whether to pay attention to an idea, a big clue, if it were readily available, would be how many people have checked it over for correctness, and for how long. Most new ideas that human beings come up with are wrong, and if someone just thought of something five seconds ago and excitedly wants to tell you about it, probably the only benefit of listening is not offending the person.

But it seems quite rare for this important piece of metadata to be straightforwardly declared, perhaps because such declarations can't be trusted in general. Instead, we usually have to infer it from various other clues, like the speaker's personality (how long do they typically think before they speak?), formality of the language employed to express the idea, the presence of spelling and grammar mistakes, the venue where the idea is presented or published, etc.

Unfortunately, such inferences can be imprecise or error-prone. For example, the same speaker may sometimes think a lot before speaking, and other times think little before speaking. Using costly signals like formal language is also wasteful compared to everyone simply telling the truth (but can still be a second-best solution in low-trust groups). In a community like ours, where most of us are striving to build reputations for being (or at least trying to be) rational and cooperative, and therefore there is a level of trust higher than usual, it might be worth experimenting with a norm of declaring how long we've thought about each new idea when presenting it. This may be either in addition to or as an alternative to other ways of communicating how confident we are about our ideas.

To follow my own advice, I'll say that I've thought about this topic off and on for about two weeks, and then spent about three hours writing and reviewing this post. I first started thinking about it at the SIAI decision theory workshop, which was the first time I ever worked with a large group of people on a complex problem in real time. I noticed that the variance in the amount of time different people spend thinking through new ideas before they speak is quite high. I was surprised to discover, for example, that Gary Drescher has been working on decision theory for many years and has considered and discarded about a dozen possible solutions.

The trigger for actually writing this post is yesterday's Overcoming Bias post Twin Conspiracies, which Robin seemed to have spent much less time thinking through than usual, but which has no overt indications of this. (An obvious objection that he apparently failed to consider is, wouldn't corporations actively recruit twins to be co-CEOs if they are so productive? Several OB commenters also pointed this out.) A blogger may not want to spend days poring over every post, but why not make it easier for the reader to distinguish the serious, carefully thought out ideas from the throwaway ones?


Comments sorted by top scores.

comment by John_Maxwell (John_Maxwell_IV) · 2010-03-20T07:28:04.285Z · LW(p) · GW(p)

Most new ideas that human beings come up with are wrong, and if someone just thought of something five seconds ago and excitedly wants to tell you about it, probably the only benefit of listening is not offending the person.

This doesn't jive well with my intuition about conversations being good ways to shape new ideas. It's very common for conversation participants to state their ideas aloud without having even thought about them for 5 seconds. Maybe if we were all robots sending signals to each other using TCP/IP then it would be good for each individual robot to spend a certain amount of processing power on an idea and ensure basic soundness before sending it to other robots in the network, but humans can't flip on excitement like a switch. Excitement is valuable, and conversations about exciting topics are a good way to generate it artificially.

comment by alexflint · 2010-03-22T07:16:22.734Z · LW(p) · GW(p)

This is exactly what I was thinking as I read the post. Sometimes verbalizing an idea quickly is the most efficient way to determine its merit. However, in such an "excited" conversation it's implicit that nobody has thought about their ideas very much, whereas it would be problematic if people started signalled that they had thought long and hard about their two-cent idea.

comment by magfrump · 2010-03-20T17:36:33.911Z · LW(p) · GW(p)

I think the main disagreement here seems to me to be that wei dai is talking about general conversations, which are far more often about social obligations than about idea generation. For example, business meetings.

Consider the recent episode of "The Office" where the new boss asks for ideas. Some people have ideas that are dismissed, but Daryl proposes a formal idea with a written schematic, and this allows him to get taken seriously.

Conversations are excellent ways to shape new ideas between a few people, but once you have entered that context you will usually have more information (i.e. about which ideas came up 5 seconds ago and which have been stewing for a while) and more shared motivation (i.e. find good ideas vs. get through the day)

comment by kodos96 · 2010-03-20T05:37:10.348Z · LW(p) · GW(p)

Hmmm... just a couple of days ago I posted a comment here at LW, initially prefaced with something to the effect of "I've actually been thinking about this exact issue off and on for several years now".... but I ended up editing that out because I thought it sounded pretentious.

comment by barrkel · 2010-03-20T02:15:36.987Z · LW(p) · GW(p)

I don't know about other people, but I do know something about myself: I don't fully know what I think until I either write it down or speak up. Moreover, the benefits of speaking up without fully thought through ideas is high in group conversations - rather than trying to complete a thought with one's own limited repertoire of to-hand facts and concepts, one can use the group's.

comment by SoullessAutomaton · 2010-03-20T22:51:15.769Z · LW(p) · GW(p)

Also, few ways are more effective at discovering flaws in an idea than to begin explaining it to someone else; the greatest error will inevitably spring to mind at precisely the moment when it is most socially embarrassing to admit it.

comment by Rain · 2010-03-20T21:21:29.332Z · LW(p) · GW(p)

I've had similar experiences, so it's not just you.

comment by JustinShovelain · 2010-03-19T23:39:58.954Z · LW(p) · GW(p)

Interesting idea.

I agree that trusting newly formed ideas is risky, but there are several reasons to convey them anyway (non-comprehensive listing):

  • To recruit assistance in developing and verifying them

  • To convey an idea that is obvious in retrospect, an idea you can be confident in immediately

  • To signal cleverness and ability to think on one's feet

  • To socially play with the ideas

What we are really after though is to asses how much weight to assign to an idea off the bat so we can calculate the opportunity costs of thinking about the idea in greater detail and asking for the idea to be fleshed out and conveyed fully. This overlaps somewhat with the confidence (context sensitive rules in determining) with which the speaker is conveying the idea. Also, how do you gauge how old an idea really is? Especially if it condenses gradually or is a simple combination out of very old parts? Still... some metric is better than no metric.

comment by Erik · 2010-03-20T10:58:59.383Z · LW(p) · GW(p)

To convey an idea that is obvious in retrospect, an idea you can be confident in immediately

Solutions to hard puzzles are good examples of these. NP-problems, where finding a solution is (believed to be) exponentially harder than checking the correctness of it, is the extreme case.

comment by AdeleneDawner · 2010-03-20T04:18:04.992Z · LW(p) · GW(p)

To convey an idea that is obvious in retrospect, an idea you can be confident in immediately

Such ideas are prone to being flawed because they fail to take into account relevant information that has been temporarily forgotten.

comment by taw · 2010-03-20T01:39:28.665Z · LW(p) · GW(p)

Time spent thinking about something correlates far too much with how much a person likes an idea. People who believe strongly in the Holy Trinity have spent far more time on average thinking about it than people who don't.

So, you will end up very strongly biased if you follow own advice (but then I haven't thought about it much at all).

comment by h-H · 2010-03-20T19:21:56.576Z · LW(p) · GW(p)

actually I think the majority probably spend too little time thinking about the Holy Trinity to be actually capable of believing in it, rather spending more on reasons why it makes sense etc.

I mean isn't the type of thinking Wei Die urging is a bit different than thinking about the trinity?

comment by taw · 2010-03-21T05:50:21.190Z · LW(p) · GW(p)

The post seemed to have been asking to simply report time spent thinking about something, as it's easy to measure. Or did I not think enough about it?

comment by idlewire · 2010-03-24T16:25:50.519Z · LW(p) · GW(p)

This kind of brings up the quality of thought that is spent on a subject. Someone with a strong ability to be self-criticizing can more effectively find flaws and come to better conclusions quicker. Those who contemplate on ideas with wrong, but unshakeable (or invisible rather) assumptions, will stew in poor circles until death. The idea of a comforting or powerful diety, unfortunately, sticks so hard when indoctrinated early and consistently.

comment by simplicio · 2010-03-20T04:12:10.607Z · LW(p) · GW(p)

At the edge of Clangford's garden, he built a lovely cell

Where he contemplated Limbo, then Purgatory and Hell.

With the barbed wire in his underpants he found it hard to sleep.

All he had for company was jockey boys and sheep.

(I was, rather serendipitously, listening to that song ("God Woman" by Christy Moore) as I read your comment.)

comment by Clippy · 2010-03-20T00:12:19.543Z · LW(p) · GW(p)

I have been thinking about this comment for three hours before making it. I do that for all my comments, which is why you should trust my inferences more.

comment by FAWS · 2010-03-20T00:27:09.526Z · LW(p) · GW(p)

Wei_Dai 19 March 2010 10:21PM

Clippy 20 March 2010 12:12:19AM*

I have been thinking about this comment for three hours before making it.


comment by wedrifid · 2010-03-20T00:41:09.334Z · LW(p) · GW(p)

You mean you can't predict the behavior of Wei with reliable accuracy? No wonder you are so bad at creating stationary.

comment by mattnewport · 2010-03-20T00:49:26.054Z · LW(p) · GW(p)

Sorry to do this to you again, that spelling discussion seems to have made me hypersensitive. I think you mean stationery.

comment by wedrifid · 2010-03-20T01:02:30.698Z · LW(p) · GW(p)


comment by Clippy · 2010-03-20T01:05:17.672Z · LW(p) · GW(p)

Sorry, I made a mistake. I was thinking about the comment for two hours before posting it.

comment by wedrifid · 2010-03-20T00:14:55.060Z · LW(p) · GW(p)

That just seems to suggest that your your hardware is slow or your software inefficient. Taking time to think over things, letting them be evaluated in the background of your mind, is a human strategy.

(Also - just writing explicit assersions quoted or paraphrased from any given subject on signalling makes the Clippy persona seem to be 'trying too hard'. It would be better to save him for the more poignant insights he sometimes serves to illustrate.)

comment by jimrandomh · 2010-03-20T00:26:33.040Z · LW(p) · GW(p)

You needed to wait another hour before posting that for it to be credible.

comment by FAWS · 2010-03-20T00:25:46.925Z · LW(p) · GW(p)

Wei_Dai19 March 2010 10:21PM

Clippy20 March 2010 12:12:19AM

I have been thinking about this comment for three hours before making it.


comment by Vladimir_Nesov · 2010-03-20T10:23:05.924Z · LW(p) · GW(p)

This may be either in addition to or as an alternative to other ways of communicating how confident we are about our ideas.

Time spent by you thinking about a question doesn't necessarily allow other people to make meaningful conclusions about whether the conclusion is reasonable. Maybe you are brilliant at X, and two minutes' thinking produces reliably correct conclusions, but you are confused about Y, and several months' thinking shouldn't be trusted to be adequate.

Once you are introspective enough to consider the advice of this post, you should also be in a better position to make the corresponding correction in your level of confidence. Overall, it's easier to communicate only the level of confidence that takes into account the appropriate factors, rather than communicating the factors themselves, such as the time spent thinking. This is also polite, like including a link to a discussed article in a post, instead of giving the readers hints on how to find it themselves using a search engine.

So, the advice I can draw from this post is, in estimating your confidence in conclusions you've made about X, take the time spent thinking about X into account, and don't forget to signal your confidence in the assertions you make.

P.S. This post should link to Hold Off On Proposing Solutions.

comment by MichaelVassar · 2010-03-20T15:35:26.644Z · LW(p) · GW(p)

Regarding time, it seems to me that the attitude with which an idea is embraced has as much to do with its credibility as the amount of time spent on it. God is plausibly the idea which has had the most time spent on it, but not very successfully. Most of Newton's time was spent on it, for instance. In philosophy, concepts from evidential decision theory to epiphenomenalism have absorbed massive effort by capable people. In a more practical domain, so have essentially all ideologies and movements, yet the intellectual output of almost all ideologies and movements seems to rapidly level off.

Also, I don't think that your objection to the co-CEO issue works well within Robin's usual frame of status-competition. If co-CEOs are needed for decision making, just hire two non-twins, but if you are looking for incredible productivity as a status symbol twins are advantaged. Many companies, but particularly the Virgin group, clearly benefit from this.

For a clearer case, if you can tell a VC that you completed 4 Bachelors degrees in 3 years while working half-time and coding some open-source software, then twins who hired a third party to play them in the open-source community have a huge advantage.

comment by Kevin · 2010-03-21T03:36:17.836Z · LW(p) · GW(p)

I noticed that the variance in the amount of time different people spend thinking through new ideas before they speak is quite high.

This is one of the basic characteristics of introversion vs. extroversion where extroverts tend to have less of a filter on their thoughts before speaking.

comment by idlewire · 2010-03-24T16:15:40.364Z · LW(p) · GW(p)

While I'd have a difficult time pinning myself as either introvert or extrovert, I notice when I'm with a comfortable crowd, ideas will fall out of my mouth with so little processing that many sentences end with "... wait, nevermind, scratch that." I'll use my close aquaintences as easy parallel processing or to quickly look at ideas from obvious viewpoints that I tend to easily overlook.

When I'm in an unfamiliar group or setting, I'll often spend so long revising what I want to say that the conversation moves on and I've hardly said a word for 20 minutes.

comment by simplicio · 2010-03-20T03:57:36.151Z · LW(p) · GW(p)

This is probably a good idea in general. However, there is likely an advantage to discussing an idea reasonably soon, rather than letting it bounce around too long in the echo chambers of one's own head. When I've tutored freshman physics, I like my students to explain concepts back to me; it's a good way to see if they understand the connections between ideas or are just doing algebra - or worse, have developed a preliminary distorted "theory" of physics (what's the gravitational potential energy of the earth wrt the sun? oh, that's easy. E=mgh).

But I think it's an excellent idea for this community. (5 min).

comment by Morendil · 2010-03-19T22:33:16.007Z · LW(p) · GW(p)

A good way of doing this, ISTM, would be to encourage the practice of writing posts that got their start in life as comments.

This leaves a visible record of a) how long someone has been informally kicking an idea around before writing it up, b) if the write-up comes as the summation of a somewhat extended discussion, what objections were raised and dealt with.

comment by Morendil · 2010-03-20T00:17:02.431Z · LW(p) · GW(p)

I've started work on a post about the "status hypothesis", which I expect to spend no less than six hours researching and up to three hours writing, with an ETA late next week. I've been thinking about it for a few weeks. (Edit: oops on that ETA. I'm still hoping for early april.)

(Perhaps this post might serve as a "special thread" to announce writing on a topic.)

comment by RobinHanson · 2010-03-21T11:53:22.463Z · LW(p) · GW(p)

There are certainly posts that I've thought about more than the twin post, but I'm not sure it was below median. I certainly do have opinions on the relative quality of my posts, and I've struggled with how to signal that to readers. And it is interesting that bloggers don't more often explicitly make that distinction. Even in this post you didn't rank it relative to other posts of yours, which would be the most useful info you could offer.

On the plausibility of secret twins, Vassar has it right below. Twin CEOs only appear more productive, and perhaps only are more productive if a wider audience doesn't know their secret.

comment by djcb · 2010-03-20T13:13:43.386Z · LW(p) · GW(p)

Well, I think it depends on the kind of idea.

A complex theory requires a lot of thinking, and one should do all a favor by making sure there are no obvious showstoppers.

On the other hand, when coming up with ideas for a product name, it can be very effective to do some brainstorming with other people.

comment by roland · 2010-03-20T03:31:05.868Z · LW(p) · GW(p)

Most new ideas that human beings come up with are wrong.

Added to my quotes file.

comment by orthonormal · 2010-03-20T05:34:09.422Z · LW(p) · GW(p)

Is this a disguised plea to karma-enforce a more rigorous separation between "things that should be Open Thread comments" and "things that should be top-level posts"? If so, I suggest you go ahead and suggest that norm. Otherwise, yours seems to be a very vague post.

comment by Wei_Dai · 2010-03-20T06:02:07.828Z · LW(p) · GW(p)

Is this a disguised plea to karma-enforce a more rigorous separation between "things that should be Open Thread comments" and "things that should be top-level posts"?

No, that wasn't my intention. I'll try to restate my main point more clearly:

To get your ideas to be taken seriously by others, get into the habit of thinking things through before you speak, and signal (or, when appropriate, just state) that you have done so. If you sometimes also like to throw around poorly thought out ideas for fun, then make it clear when you're doing that to avoid hurting your overall credibility.

comment by reaver121 · 2010-03-20T09:45:14.075Z · LW(p) · GW(p)

I agree that it would be a good idea to prevent hurting your credibility by signaling that you are either throwing out an idea to be torn apart or that you have thought long and hard about it. However, I still would err on the side of letting an idea out early. There are also downsides with thinking about an idea for too long :

  • you are less likely to find problems in your idea on your own
  • you possibly can get emotionally invested in your idea so you will have trouble in letting go when someone shoots it down
  • lost thinking time when your idea turns out to be false

A classical example is the Perpetuum Mobile. Time and time again there are people who believe that it will work. They invest time & money in something that's impossible as anyone with even a passing familiarity with thermodynamics can tell you.

I can only see one downside in letting it out early (besides hurt credibility) and that's that your future ideas will be taken less seriously. If you acquire a reputation for saying a lot of wrong and/or stupid ideas people will be quicker to just ignore you.

I think that the best way is to scrutinize your idea for basic soundness so that there are no obvious holes. Then the damage is minimal if it turns out to be wrong (obvious within your community off course. If you even dare to suggest Perpetuum Mobile with physicists if give you 5 seconds before they laugh you out in your face). Also, with the rise of the internet & libraries it's fairly easy to lookup if your idea wasn't already thought off and shot down.

comment by idlewire · 2010-03-24T16:53:09.972Z · LW(p) · GW(p)

So the real question is: "How will one's credibility be affected in the environment where the idea is presented?" which most likely depends on one's current credibility.

As of now, I don't have much karma so my risk of putting out poor ideas is more detrimental to this screen name. Eliezer could probably sneak in an entire subtly ludicrous paragraph that might go unnoticed for a while.

He has a history in reader's minds as well as the karma metric to make people ignore that flash in the back of their minds that something was off. They are more likely to think it was their own abberant thinking or that they had a flawed interpretation of a non-ludicrous idea he was trying to convey.

So it guess it just depends on how solid you think your idea and reputation are in making the decision on when to release an idea to a particular audience.

comment by h-H · 2010-03-20T19:14:40.523Z · LW(p) · GW(p)

valid points, but isn't it quite easy to see when thinking has gone on for 'too long' without benefit?

nitpick:"I still would error on the side of", err not error :_)

comment by reaver121 · 2010-03-22T07:35:24.511Z · LW(p) · GW(p)

nitpick:"I still would error on the side of", err not error


but isn't it quite easy to see when thinking has gone on for 'too long' without benefit?

I suspect that in most cases you will be right. However, I know a woman in my street who is convinced that she's empath and can, quite literally, sense people emotions (in the telepathic kind of way, not the body language kind). The first time I met here I tried (naively maybe) to convince her that her ability is impossible. I told her of confirmation bias, unconscious cold reading, that the senses are not completely reliable and so on. It didn't help. At this point, she is so invested in it that she will reject anything that denies her ability.

My point is that if she would have taken the time to discuss this with people (or at least read up on counter arguments on her own) it's possible she wouldn't believe in it now.