« Marginally Revolved Biases | Main | Professors Progress Like Ads Advise »

February 26, 2007


Your tactic of not telling us the content of the dispute is an interesting way to get us to focus more on the meta issues here. On one side is a pretty smart guy (and his many fans who invite him to give talks) devoted to his topic full time, and so with a vested interest. On the other side are thousands of specialists in related areas who know less, but who might win big from producing related tangible progress.

It is not clear how substantial is disagreement here, in the sense of differing probabilities about this advocate being right. Nor is it clear how high a probability would justify further efforts on his topic. In taking sides, a lot depends here on one's estimate of the chance that, even if this advocate were right, a single related specialist could by themselves produce tangible progress, and the size of the resulting reward. I could imagine this conditional chance and reward being low enough that the lack of attempts could be a pretty weak signal.

Cold fusion?

If most people don't accept consensus based meta-reasoning even implicitly, then such consensus based meta-reasoning is in fact epistemically strong. OTOH, if, as you suggest, "most people do adopt the 'follow the crowd' maxim unconsciously, and only aim to stand out and show overconfidence on selected issues that bring social status" then they implicitly do accept such reasoning, in which case they produce an information cascade which undermines the validity of this sort of reasoning. Ironically, one can rationally disagree with the consensus only when the consensus is, at least implicitly, that one cannot do so.

Alas, my reaction to the whole article was that I simply didn't have enough information to get any evidence at all about who was right, without knowing the actual argument.

PhDs are often wrong. People without PhDs are more often wrong. Geniuses are often wrong. Some PhD-populated fields have very heavy evidence for their mainstream beliefs; others are simply exhibiting herd behavior.

The vast majority of people who defy a field are wrong, but few of these garner enough attention to travel internationally. Hal Finney wouldn't likely find the object-level arguments persuasive if the one were an astrologer.

If the argument is not obviously wrong then this is probably not a field with heavy evidence on the topic in question. Where evidence is light, and people think they can get away with clinging to their opinions, a whole field can go astray almost as easily as a smart individual.

Insufficient evidence. I'd want to know what we were talking about. At least I'm pretty sure it's not me.

My father put it rather simply: "The closest you can come to the truth in a subject in which you are not an expert is by accepting the opinion of the majority of experts during the age in which you live."

There are no cases of a scientific theory backed by solid evidence that failed to gain acceptance in the scientific community in a timely fashion. "Continental drift" is often cited as a theory that took a long time to be accepted, but continental drift, as it was originally proposed, is complete nonsense! It included a mechanism by which continents could move that was completely impractical and contradicted known facts of geology. It was many years later that the modern theory of plate tectonics presented a model of moving continents that was consistent with the facts.

Many correct theories do begin as minority opinions, but when they have evidence to support them, they don't remain minority opinions. If a theory is old and does not have widespread support, it's almost always because the weight of the evidence is against it. It's easy to present a logical, persuasive argument for almost anything. All you have to do is lie about the facts, and you can lead a previously uninformed audience to draw any conclusions you want them to. Makers of "documentaries" about pseudoscientific topics such as UFOs use this technique regularly.

Your friend may one day be vindicated, but the odds are against him and there's no reason to agree with him today.

"There are no cases of a scientific theory backed by solid evidence that failed to gain acceptance in the scientific community in a timely fashion."

"Timely fashion" ranges from a year to a century. Science gets there eventually. It is not designed to get there fast.

Michael, it is just not true that rational people must get stuck in an info cascade.

"Timely fashion" ranges from a year to a century.

What took a century? A "timely fashion" in the face of unambiguous evidence would generally be anything from a few years to, at most, a generation. You often need hindsight to determine when it was that the evidence became unambiguous, but when the evidence is there, it gets accepted.

I'll Paypal you $5.00 if you can come up with a counterexample that satisfies my father.

Doug, who gets to decide what constitutes "unambiguous" evidence? Do you have a definition apart from "the evidence that scientists finally found convincing"?

Doug, who gets to decide what constitutes "unambiguous" evidence? Do you have a definition apart from "the evidence that scientists finally found convincing"?

Good point. For the purposes of my offer of $5.00, I specified that my father does. ;)

I suppose atom theory was kicking around for a thousand years or more in some form before it was accepted. But it was not very similar to the final product.

I too would like to hear the strongest example a critic of mainstream science could come up with of a good theory that took a long time to be accepted. Just use your judgment as to what constitutes "good" and "long". Let's hear about science at its worst, air the dirty laundry.

My history of science isn't broad enough to give a decent answer to this question. All I know about are the areas upon which I myself have touched.

Evolutionary psychology had to fight like hell and was being systematically ostracized in the early days; Bayesian statistics had to fight like hell; Judea Pearl and company are still fighting to reintroduce the concept of causality into mainstream statistics; for years, AI types denied the blatant evidence coming in from neurology that the brain's representation of visual imagery involved actual shape modeling in the visual cortex, rather than just propositional logic; Leo Szilard had a hell of a time getting skeptics like Fermi to pay attention to the possibility of a fission chain reaction; and apparently other fields have this problem too, because Max Planck was moved to say, "Science progresses funeral by funeral."

Now these are selectively salient examples, and I'm sure there are plenty of times when a new idea was taken up without a fuss. But the idea that science is always or even reliably fast on its feet is simply absurd, from a historical perspective.

Science, to work, relies on individuals arriving at correct conclusions ahead of the general community, and their persuading others. It's hard to see how the community could be smarter than its smartest individuals. It would be a heartwarming thought for emergence mystics, but I just don't see how it could be true in real life. The most impressive AI thinkers I know (e.g. Judea Pearl) soar above the general community like zeppelins.

I agree with Eliezer. Complex numbers were accepted only several centuries after they were first introduced. I think Gauss finally made them acceptable.

I'd say that the evidence for the common descent of species was very strong even before Darwin, but it took centuries for biologists to accept what was in front of their faces.

The comments to this entry are closed.

Less Wrong (sister site)

May 2009

Sun Mon Tue Wed Thu Fri Sat
          1 2
3 4 5 6 7 8 9
10 11 12 13 14 15 16
17 18 19 20 21 22 23
24 25 26 27 28 29 30