Interestingly, I did come to rationality through politics.
In high school, I discovered Ayn Rand and became rabidly libertarian. I later grew less interested in politics but remained libertarian. In early college, I started thinking about politics again and discovered econo-blogs. The 'thinking like an economist' quickly grew on me and eventually led me to Overcoming Bias and then Less Wrong.
People go funny in the head when talking about politics. The evolutionary reasons for this are so obvious as to be worth belaboring:
Just wondering if there are any citations to go along with this bit.
I'm puzzled by Eliezer's comments on the Nixon example. How is "Republicans are not pacifists" going to make Republicans feel unwelcome? That would require either (1) that it be seen as an insult to Republicans or (2) that it be seen as an obnoxious stereotyping of Republicans. But (1) lots of people -- the great majority, I think -- are not-pacifists and generally no one thinks any the worse of them for it; indeed my impression is that far more people think that pacifism is foolish than think that not-pacifism is evil; and (2) since (a) the whole point of the example is that generalizations like these should generally be taken as approximations only and (b) it is in fact a very good generalization that Republicans are not pacifists (note: I have no statistical evidence for this, and am willing to be refuted; but I don't expect to be), it seems unlikely that many Republicans will feel insulted by being told that Republicans are generally not pacifists.
Now, I happen not to be a Republican (or a pacifist or a Quaker) myself, so I don't entirely trust my intuitions about how members of those groups might react to the example. So:
Am I missing something? Is anyone reading this a Republican who would feel insulted or over-pigeonholed by seeing the example Eliezer cites in a textbook or lecture?
(For the avoidance of doubt: I agree with the general point Eliezer is making; it just seems to me like he could have chosen a better example.)
As soon as you make any statements with political or ideological content, the way people understand them will be determined by their informal signaling implications much more strongly than by their literal content and its formal logical implications. This is especially true if such statements are made abruptly in an unrelated context. Sometimes people really mean innocently only what they literally say and are surprised at the reactions to the unintended signaling, but more often signaling is a part of the speaker's intention (though of course its effects can be mispredicted).
In this case, the Nixon example is meant to send a clear ideological signal (whose details I won't spell out to avoid making potentially contentious statements). Moreover, the use of the example in a totally unrelated context, in a document intended for a community of technical experts, signals that ideological agreement is expected in this community and places dissenters in a position where they have to take it in silence (thus confirming their low status) or protest loudly and expose themselves to ridicule (or worse).
Some historical background might be relevant: Nixon's political opponents, at the time, used the conflict between his Quaker affiliation and his aggressive politics as evidence of dishonesty. Citation courtesy of a recent Boing Boing article:
http://www.collectorsweekly.com/articles/the-meanest-u-s-political-campaign-buttons/
In 1968, candidate Richard Nixon promised to bring the troops home from Viet Nam. Two years later, as the war escalated, anti-Nixon forces accused the president of hypocrisy (above, right) in light of his religious background—Quakers are pacifists.
It is, as Eliezer points out, a distracting example, regardless of whether it is an insulting one.
OK, so that would mean that Nixon supporters might find the example distractingly offensive, which might well be a reason not to use it. But Eliezer said Republicans, and the two aren't quite the same.
(After Watergate I think there were rather few Nixon supporters -- meaning, in this context, people who would be offended by the suggestion that there might be anything suspect about Nixon's integrity -- even among Republicans. When did the "Nixon diamond" start being used as an example in AI?)
Sorry — I didn't mean to imply that an example had to be offensive in order to be distracting. Simply bringing up a matter of partisan conflict in the recent past can be already distracting, even if it's not personally offensive to any single reader.
A classic, and one well worth linking to out in the wider world.
I'm not saying that I think Overcoming Bias should be apolitical, or even that we should adopt Wikipedia's ideal of the Neutral Point of View.
As someone who's been around it for years, I think NPOV is actually the most amazing thing about Wikipedia - greater than letting everyone edit the website, for example. It's the only way we can get everyone editing the website without killing each other. But it's also the quintessence of how to serve the reader. Can be hard on the writers, of course.
Trouble is, NPOV is often in direct conflict with the "reputable sources" and "no original research" rules. In areas where reputable sources are mostly unbiased and disreputable sources are mostly crackpots, everything's fine. However, in those where respectable opinion is nowadays remote from reality, reporting what reputable sources say is not going to produce a NPOV account, and if editors attempt to make sense of the available information on their own, this becomes illicit "original research."
Also, it can be misleading to note the contrast between the modern seemingly neutral academic tone and the tone of the old scholarly works that strikes the modern reader as unabashedly opinionated, as you have noted about the old Britannica. What looks like neutral and dispassionate tone in modern academic and reference works is often every bit as opinionated and biased, except that this is achieved in more underhanded ways. And frankly, I prefer the old-fashioned open and explicit way.
How is that different from standard encyclopedia practice? It at least initially appears that it's one of the few things standard encyclopedias do better than wikipedia.
I think I really meant it did it much better than pretty much anything on the web. Which I suppose isn't saying much, but I think it's saying more than nothing - Britannica is a culturally reified gold standard that approximately no-one actually cracks open past high school; Wikipedia is something ordinary people use every day when they just would not have used a paper encyclopedia.
Though, of course, Wikipedia calls itself an encyclopedia and has always looked up to Britannica. As I note below, it would be interesting to actually compare the neutrality of current Wikipedia with current Britannica.
Oh, I certainly agree with this. I'd just trace it back to Wikipedia's roots - and I suspect that as fewer people are familiar with Britannica, Wikipedia will lose that more and more.
I suspect that as fewer people are familiar with Britannica, Wikipedia will lose that more and more.
That's ... a thought on Wikipedia's future that hadn't occurred to me before. You can see it already, of course. Even just "We're here to write an encyclopedia" requires a prior notion of an encyclopedia other than what you're already doing. But then there's anecdotes of kids already saying "encyclopedia? Is that like Wikipedia?" I wonder how Wikipedia will self-define.
I'm under the impression that most encyclopedias had articles written by a single author, chosen as an expert on the subject. As such, the article on "Christianity" would be written by a renowned Christian theologian, or historian of Christendom; with no contribution by (for instance) any Muslims, Buddhists, neopagans, or atheists. While the author would be expected not to come across as overtly partisan, there's not necessarily any correction for biases inherent in that author's point of view.
This is a different approach from Wikipedia's NPOV, where work on an article is expected to converge towards a point where contributors from multiple points of view find it equitable. Wikipedia doesn't demand that any given editor write a fair treatment of the POV opposing her own (although "writing for the enemy" is offered as an ideal); rather, she's merely expected to work with other editors from different points of view toward a common goal of a good consensus article.
One of the controversies surrounding Wikipedia is that it demands that experts submit to the NPOV policy and to having their work edited by non-experts as well — rather than asking other editors to defer to their expertise. This is in contrast with competing projects Nupedia (now defunct), Citizendium, and Knol, which propose to attract expert editors by giving them a greater voice.
One of the controversies surrounding Wikipedia is that it demands that experts submit to the NPOV policy and to having their work edited by non-experts as well — rather than asking other editors to defer to their expertise. This is in contrast with competing projects Nupedia (now defunct), Citizendium, and Knol, which propose to attract expert editors by giving them a greater voice.
It is worth noting here that Citizendium and Knol both failed as well - Citizendium was overrun by pseudoscientists (who know how to work credentials) and Knol became more or less a crank and spam dump. In the attempt to appeal to writers, they utterly failed to appeal to readers.
It's also worth noting that Wikipedia has lots of experts editing in their fields anyway. It turns out that if you're the encyclopedia that everyone actually reads, people will go to some effort to get their field properly documented.
The expert problem is still a problem, of course. Wikipedia can't keep idiots out of experts' faces, but then Wikipedia can't keep idiots out of anyone's faces.
Britannica authors, for instance, are Britannica authors first and moneymakers second and opinionated people third or fourth. Wikipedia authors are highly opinionated, and are compromising (which fixes some objectivity problems but is a poor substitute for objectivity or neutrality). So the Britannica author is likely to be less expert than the sum of wikipedia authors but much more neutral.
This is surmise when the material for the test is available. Factual accuracy has been compared more than once between Wikipedia and Britannica - has anyone attempted to compare neutrality? How would we actually do this? I suspect no-one has as yet, but it's one I'd quite like to see if we can interest a third party in doing!
(Old editions of Britannica are notoriously bad for this, by the way. They were written to quite definitely inculcate a given cultural viewpoint and are loaded with opinion. This was a major problem in seeding Wikipedia with 1911 Britannica material in the early days - the stuff just didn't pass NPOV muster. I would be unsurprised if more recent editions did much better; Wikipedia's notion of NPOV is really obviously a current cultural construct.)
Well, to compare neutrality we can do one of four things: a: rely on the impressions of people who've used both (survey people who've claimed to have read both) b: trace the prior likelihood of either group of authors being biased on the material they're writing about (the profit motive vs writing about what you're passionate about) c: ask contributors what they've seen done that damages the neutral point of view d: come up with a neutral definition of neutrality.
It sounds like you want to do d: how might we start on such a thing?
Oh, and obviously yeah - "neutral" will depend on your culture. Objectivity might or might not, but neutrality must. So this makes d trickier.
I'd expect something like (a): get experts to say whether the articles constitute a reasonable survey of the field. We'd probably need (d) as well, because the way I just defined (a) is one definition of a good neutral Wikipedia article.
Is Britannica trying specifically for neutrality? It claims authority (whereas Wikipedia explicitly claims none which is why we're so obsessive about references), but I'm not sure it kowtows to such culturally relativistic notions as "neutral". The Wikipedia article on Britannica notes that EB has been lauded as increasingly less biased with time, but then Wikipedia would note that. Glancing at britannica.com, I can't find a claim as to what it represents editorially except that it's Britannica, you know, Britannica, so I'm not sure what would be a fair test to them.
This is getting wildly off topic ... I suppose it's vaguely related to politics by reframing while trying to be aware of and flag the reframings so as to avoid mind-killing ... this is the sort of thing my autodidact's knowledge of postmodernism comes in handy for.
I wouldn't mind "ask experts who do not post to Wikipedia or write for Britannica" to rate the articles for accuracy, neutrality, etc. I would expect them to call Wikipedia more comprehensive, to call Britannica more neutral, and I have no idea which would be rated more accurate. If they did indeed call the Wikipedia articles more neutral, I'd have to update my understanding of the field.
My experience: I fixed mistakes in two articles, then got thoroughly distressed and stopped participating. I'm an anesthesiologist, as background. The first article was on a painkiller, and I found my changes overwritten by a drug enthusiast who believes/writes that narcotics are non-addictive. I did not push the issue. The second article was on anesthesia, and I linked to a reference document published by the American Society of Anesthesiologists (the premier research organization of anesthesiologists in the US.) A nurse anesthetist editor was very proud of his ability to prevent any documents from the ASA from being linked to on the page while maintaining a link to the American Association of Nurse Anesthetists, and made it clear that it was his "turf" and that he was highly political. I did persist briefly to see what would happen: he tracked my real identity and threatened me. I immediately lost all interest.
I don't believe Britannica is trying for neutrality per se. I believe it's trying for objectivity, which is related but nonidentical. On many topics wikipedia attempts objectivity as well rather than neutrality (evolution vs intelligent design, for instance).
3)If someone is considered an 'expert', the pressure to be right will be enormous. Thats because experts have special status for being knowledgeable about a topic and getting answers about it right. Every mistake is seen as reducing that expertise and proportionatly reducing the status of the expert. Being wrong to someone considered a non expert is even more painful then being wrong to an expert.
In many fields (politics and finance readily come to mind, but I'm sure there are many others), being considered an "expert" creates an immense pressure not to say anything that contradicts the prevailing opinion, and being proven right eventually usually doesn't redeem you for the status you lost by expressing an unpopular opinion.
Politics is my life and career. One of the ways I am trying to circumvent irrational behavior in others with regard to politics is not to disclose one's political affiliation ever or be very ambiguous about it, and learn to speak the language and seemingly share the concerns and beliefs of my conversant. Then, ask subtly leading questions about the consistency and empirical validity of those beliefs.
Yes, it's not the most honest thing to do, but I believe this disarms the irrational, tribal psychology at work and enables a more fruitful discussion. "The slow blade penetrates the shield" and all that.
Today's post, Politics is the Mind-Killer was originally published on 18 February 2007. A summary (taken from the LW wiki):
Discuss the post here (rather than in the comments to the original post).
This post is part of the Rerunning the Sequences series, where we'll be going through Eliezer Yudkowsky's old posts in order so that people who are interested can (re-)read and discuss them. The previous post was Outside the Laboratory, and you can use the sequence_reruns tag or rss feed to follow the rest of the series.
Sequence reruns are a community-driven effort. You can participate by re-reading the sequence post, discussing it here, posting the next day's sequence reruns post, or summarizing forthcoming articles on the wiki. Go here for more details, or to have meta discussions about the Rerunning the Sequences series.