by [anonymous]
2 min read30th Apr 201230 comments

0

There are some issues where a) I have no direct knowledge of the facts, b) there is some dispute over the facts, c) I hold an opinion on what the facts are. You probably do too.

For example, I believe that ~6mm Jews were killed in WWII, and that most people who deny this are anti-semites. If someone produced a theory which conflicted with my views, I would probably be very suspicious of the person's motives. I think this without ever having seen evidence that would satisfy me directly. [I've been to Auschwitz and a few other camps, and I've read a few books about WWII, but the only reason I have to cite the figure of 6mm rather than 3mm, or even 300,000, is that most knowledgeable people use that number]. I suspect most LWers are at a similar state of opinionated ignorance.

Now, I am in no way incentivised to investigate this: my opinions matter roughly zero to anyone, including me. So I don't have any reason to investigate the Holocaust. But there are other areas where the facts do matter to me.

I know very little about medicine. When I need medical assistance, I tend to do what a doctor tells me without criticising his diagnosis. For example, I suffered from eczema a few years ago. I was prescribed a medicine which contained hydrocortisone as the active ingredient. I Googled this ingredient and then took the medication until the affected area cleared up. I noted that there were other steroids available [Clobetasol propionate is one such] which are considered cures for eczema. I did not know why my doctor prescribed me one rather than the other, and the distinction between the two would likely affect me. But I took the 'expert advice' without a pinch of salt. Unlike the WWII question, here the facts are actually relevant to me, and a mis-diagnosis could have caused some complications. But just as I don't check the wiring of my house's electric systems, or check my walls' stability and capacity to support my ceiling, I take my doctor's advice.

There are some less clear-cut examples. For example, the 'smartest' [read: some combination of high-IQ, high-rationality, knowledgeable] people I know tend to identify as either libertarians or utilitarians, rather than as socialists. There are a few exceptions to this, but not many. From this, I could arguably assume that it would be correct to adopt a libertarian or utilitarian, rather than a socialist*, mindset to politics without actually understanding why libertarians are libertarian. Just as I don't need to know why my doctor prescribed hydrocortisone in order to take it, I don't need to know why most smart people I know favour relaxed drug laws in order to share their opinion.

*I'm aware that these are not the only options, but they're the most mainstream 'labels' which are clearly defined. ['Liberal' and 'conservative' mean different things in different countries, but I think those three are relatively constant, at least in the people who apply them to themselves]

New to LessWrong?

New Comment
30 comments, sorted by Click to highlight new comments since: Today at 3:27 PM

There are some less clear-cut examples. For example, the 'smartest' [read: some combination of high-IQ, high-rationality, knowledgeable] people I know tend to identify as either libertarians or utilitarians, rather than as socialists. There are a few exceptions to this, but not many. From this, I could arguably assume that it would be correct to adopt a libertarian or utilitarian, rather than a socialist, mindset to politics without actually understanding why libertarians are libertarian.

Since political questions involve moral evaluation, you need to know not only that they are 'smart', but also that they share your ethical values. Even then, adopting others' political opinions is generally unreliable, since most people, even smart ones, are heavily biased when thinking about politics.

[comment deleted]

[This comment is no longer endorsed by its author]Reply

A "political opinion", at least in my parlance, is a system of beliefs which includes personal preference for a particular political party or ideology. "It is correct to adopt libertarianism" might express a political opinion. "The number of Jews killed in the Holocaust is about six million" is perhaps evidence that the speaker isn't a typical Nazi, but it is too narrow and concrete to be called a political opinion per se.

As for the reasons why the caution applies to political opinions while does not (that much) apply to simple questions of fact even if they have political implications, consider how many obvious fallacies one has to commit to give a wrong answer. "How many Jews were killed in the Holocaust" is a question of fact with pretty straightforward meaning. Every reasonable person would agree that one should look into the archives and population censuses. Practically the only way to credibly justify an incorrect answer is to postulate a conspiracy which has faked all official evidence - and this is what actually happens. Then everything you need is Occam's razor: conspiracies are complex and therefore very improbable.

On the other hand, when considering the question "is it correct to adopt liberalism", both "correct" and "liberalism" are open to wide range of interpretations and there is no universally agreed upon method of answering this sort of questions. The biases have much easier role here in their effort to sway the result towards the prejudiced answer.

Since political questions involve moral evaluation, you need to know not only that they are 'smart', but also that they share your ethical values. Even then, adopting others' political opinions is generally unreliable, since most people, even smart ones, are heavily biased when thinking about politics.

What is the difference between ethical values and values?

You don't trust a doctor because the doctor is smart, or because the doctor is a member of a smart society. You trust the doctor because there are procedures ensuring that experiments were performed to validate these claims.

I'm sure Hippocrates was well smarter than the median modern-medicine doctor, but I'd take a second-quartile modern medicine doctor over Hippocrates any day of the week.

[comment deleted]

[This comment is no longer endorsed by its author]Reply

The good parts of this article are better presented in Information Cascades, and I downvoted for the bad parts - i.e. political buffoonery and misuse of the term "groupthink".

[comment deleted]

[This comment is no longer endorsed by its author]Reply
[-][anonymous]12y130

Regarding "political boffoonery," I think Grognor was referencing your comments about intelligence (allegedly) correlating with political views. The comments about the holocaust and World War II might also fall into the same category.

On LessWrong, we have a social norm to not discuss politics. We also have a social norm against using political examples where other non-political examples could be used. If a political example is absolutely necessary, we try to use a historical example that most people won't have an emotional, gut reaction to. The reasons for why are discussed in Politics is the Mind Killer.

[comment deleted]

[This comment is no longer endorsed by its author]Reply

gentleman's agreement

Well...

[-][anonymous]12y00

I'm unsure of your meaning. Could you explain/elaborate, please?

Edit: JoshuaZ clarified what my mistake was. Unless you were thinking of something else?

A gentleman's agreement is generally used to mean an unspoken agreement. In this context the agreement isn't unspoken.

[-][anonymous]12y40

Oh! I wasn't aware that gentleman's agreements are usually unspoken. I drew the phrase from the LW "About" page. Thank you for the clarification. I'll edit my comment.

Hmm, that page should probably be corrected then, since using the phrase that way is at best mildly non-standard.

I understood it to mean an informal agreement, not an unspoken one. Googling turns up investopedia, wikipedia, and several dictionary entries, which seem to indicate that "informal" is the more common meaning.

I thought so, too. The "gentleman" part refers to trusting that everyone will adhere to the agreement rather than relying on outside enforcement. I.e. it's not a legally binding contract, but a social one.

On a side note, the phrase pushes my gender buttons and I would be happy to see some other wording used on the "about" page.

Under normal conditions (amongst friends) I'd interpret the phrase as part of a deliberate stylistic choice including an overt consciousness of gender issues etc., but I agree it's probably not ideal on LW. Perhaps some moderator would consider changing it (to, e.g., "informal agreement", or something more poetic along those lines)?

Sorry about being unclear; I meant what JoshuaZ said. The "no politics" thing just seems too formal to me to be a "gentleman's agreement".

You're actually not talking about what is called "group think", just about accepting expert opinion (whether or not there is group think in the groups of experts you mention is anther question). Anyway, your and my practice of taking the consensus of experts to be correct without understanding why they believe what they do is both rational and pragmatically necessary.

Also, at least in the States, "socialism" is not clearly defined anymore.

[comment deleted]

[This comment is no longer endorsed by its author]Reply

This may assume a degree of reliability associated with the opinions of smart people that is only very weakly justified. See for example this previous discussion here where I made a similar sort of argument to the argument made in this essay and was given a long list of examples where smart intellectuals have been essentially wrong.

As to concluding it is a good idea to be a libertarian or utilitarian because your smart friends are, here is the potential flaw: the libertarian and/or utilitarian world may be a better one for smart people and not as good for not as smart people. If you don't even know why they are libertarian, you certainly don't know enough to conclude it is not just self-interest on their part.

Reformulating a bit to make the discussion not about the actual labels:

For example, the 'smartest' [read: some combination of high-IQ, high-rationality, knowledgeable] people I know tend to identify as X, rather than as Y. There are a few exceptions to this, but not many. From this, I could arguably assume that it would be correct to adopt a X, rather than a Y mindset without actually understanding why people identify as X.

That breaks down as soon as the knowledgeable people use the same algorithm for deciding what to believe! I assume that's why you put "groupthink" in your title.

The problem arises because we don't have a habit of distinguishing

  • A) "I believe X because I personally investigated the topic",

  • B) "I believe X because leading experts in the field, who investigated the topic closely, believe X", and

  • C) "I believe X because other smart-seeming people believe X".

(A gets the best knowledge but is not usually cost-effective, B is perfectly sensible, and C is the-blind-leading-the-blind, but is the cheapest)

(And that problem arises because everybody has an incentive to imply more knowledge than they actually have, while not being caught in outright lies, so being vague about where your knowledge comes from gets good enough results.)

The problem arises because we don't have a habit of distinguishing "I believe X because I personally investigated the topic", "I believe X because leading experts in the field, who investigated the topic closely, believe X", and "I believe X because other smart-seeming people believe X".

There are languages, real and artificial, in which every sentence grammatically must contain evidential markers, distinguishing such things as "I experienced this myself", "This is a conclusion I inferred from what I experienced", "I heard this from someone I trust", etc.

Always optional in English, unfortunately. It might be useful to cultivate the habit of at least considering the evidential status of whatever one says, and deciding whether it warrants explicit mention. Linking to sources, like I did above, is one sort of evidential.

I had no idea. That is really interesting. What are some artificial languages that have evidential grammar? I knew lojban had evidentials, but I think they're optional.

Láadan has them.

Láadan has a group of function words called "evidentials" that English doesn't have; many other languages do have them.

An evidential tells you why the speaker feels justified in claiming that the words being said are true.

For example, "wa" means "The reason I claim that what I'm saying is true is that I have perceived it myself" and "wi" means "The reason I claim that what I'm saying is true is because it's self-evident; everybody can perceive that it's true, or everybody is in agreement that it's true."

The evidential will always be the last word in a Láadan sentence, and -- unlike the situation in English -- it's required to be there.

I wonder why the evidential is put last. I would have put it early in the sentence.

This way people have to listen to the entire claim before they can disagree with it.

Robin Hanson recently talked about the Bayesian requirement to take in to account the opinions of other Bayesians who have more information than you do. http://www.overcomingbias.com/2012/07/finding-our-beliefs.html

I completely agree with @TraderJoe that the vast majority of what we know is by accepting the word of someone else. I am a physicist, I have never measured the magnitude of the repulsion between electrons, I have never verified that there were excess electrons on two pieces of foil that repelled each other (where I viewed repulsion but had to take on faith that it was due to electrons). I have never measured time dilation, or the bend of light around a planet and seen it was consistent with special relativity. I have measured many superconducting devices and run lots of computer programs that I have written, but in no case did I "understand" all of the measuring equipment or the computer I was running on.

Talk about economies of scale, virtually EVERYTHING I know (except for some math I guess, and is that really "something?") I know because I believe somebody else who told me the answer.