I'm trying to think of evidence that would count against Eliezer's claim here.
Eliezer's conditions for "the majority is always wrong" scenarios are:
And his claim is that when these conditions hold, "the most dominant alternative will probably be the worst of the lot — or at least strictly superior to none of the others."
How strong the "most dominant alternative" would have to be in order for an example to count against this claim? Or how far down the "niche alternatives" do we have to go to find an example of one that's worse than the dominant one?
For instance, in religion in the U.S., Christianity is the most dominant alternative, and there is a popularity effect for religion. However, it is not clear to me that Christianity is "the worst of the lot" for the individual believer — compared to, say, Scientology, or a suicide cult like Heaven's Gate?
One might say that "Christianity" is not a single alternative, though, and that there is no "most dominant alternative" among the Catholic, Baptist, Methodist, Mormon, etc. churches.
Returning to desktop computers: Eliezer's coworker Marcello uses the example of Macs in contrast to Windows PCs: if the Mac were "worse" than Windows, no one would use them. However, perhaps instead of "no one would use them", we should say "you wouldn't have heard of them." Obviously there are worse choices than Windows for getting stuff done on a computer; they're just more obscure than the Mac — so obscure that mentioning them as serious alternatives comes across as a bit perverse. But their popularity is still above zero.
For instance, hardly anyone but Richard Stallman uses gNewSense GNU/Linux on a Lemote Loongson MIPS computer. I've been a Linux user for years, but if my choices were between the Loongson and a Windows PC, I'd take the Windows PC; the Loongson is really, really slow. But some people do use it; even the set containing only RMS is not the empty set.
One might say that since RMS has clearly thought about what sort of computer to use, and chose the Loongson, that the Windows PC is not "strictly superior" to the Loongson. It's not superior in the metric that RMS cares about, namely software freedom.
In which case, the claim seems to reduce to, "If someone chooses an alternative other than the majority choice, they must have some reason for doing so, because they're clearly not doing it by simply taking the obvious, easy, popular choice." Which sounds like a much weaker claim than talking about the majority choice being "the worst of the lot".
But ... how about medicine? The dominant form of medicine in the U.S. is the one that we usually just call "medicine" ... sometimes known as "Western medicine", "allopathy", or various other names. There are more practitioners of "Western medicine" (physicians, nurses, etc.) than there are homeopaths or any other particular school of "alternative medicine".
As with many other market choices, there is a popularity effect, mostly driven by money and advertising: it is easier to find a doctor than to find any particular variety of altmed practitioner. There are highway signs for hospitals, not for homeopaths. If you want a Reichian therapist or even a practitioner of Classical Chinese Medicine, you probably have some searching to do; whereas Western medicine doctors take out ads on buses and billboards.
However, I doubt many of us would conclude from this that Western medicine is thereby shown to be "strictly superior to none of" the alternatives.
So, is medicine a counterexample?
I think to make it work we should add a third condition:
If this condition is not satisfied, and people have different priorities for the different dimensions/criteria, the existence of multiple alternatives needs no further explanation, and we can't derive any conclusion about "betterness".
There's a lot of fuss in your comment, and a lot of quibbling. But the point you make at the end is a good one. I agree that medicine seems to be a counterexample to this.
To explore this in hopefully a more direct way:
Options for medical treatment:
Actually, looking more closely the way I see it is this: Standard medical practice is often ineffective or overly expensive (at least in the US)--doing less or using ONLY the placebo effect may be a significantly more cost-effective way to treat yourself.
If we think about homeopathy as a way to capture the placebo effects for minor ailments without having to pay for "real" medicine, I think it does seem like "Western" medicine is not strictly superior.
Today's post, The Majority Is Always Wrong was originally published on April 3, 2007. A summary (from the LW wiki):
Anything worse than the majority opinion should get selected out, so the majority opinion is rarely strictly superior to existing alternatives.
Discuss the post here (rather than in the comments of the original post).
This post is part of a series rerunning Eliezer Yudkowsky's old posts so those interested can (re-)read and discuss them. The previous post was The Error of Crowds, 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, posting the next day's sequence reruns post, summarizing forthcoming articles on the wiki, or creating exercises. Go here for more details, or to discuss the Sequence Reruns.