Arnold Kling  

Robin Hanson Video

What's New at GMU Econ... Viscusi Speaks...

You've heard about him from Tyler, Bryan, and me. Perhaps you've seen his posts at overcoming bias. Now you can see Robin Hanson give a brief talk as a keynote speaker.

In the talk, he says that before you attach yourself to other causes, you should attach yourself to his cause of trying to overcome bias--to become a more objective seeker of truth. The logic is this:

If you have a cause, then other people probably disagree with you (if nothing else, they don't think your cause is as important as you do). When other people disagree with you, they are usually more right than you think they are. So you could be wrong. Before you go and attach yourself to this cause, shouldn't you try to reduce the chances that you are wrong? Ergo, shouldn't you work on trying to overcome bias? Therefore, shouldn't overcoming bias be your number one cause?

Tyler Cowen does not buy this argument. Neither do I. But we could be wrong.

Comments and Sharing

COMMENTS (14 to date)
jlasater writes:

Love the blog. Never commented before. But is it safe to assume that you'll, in a future post, detail your objection(s) to Hanson's argument? Or, have you spelled it out in a previous post, in which case, could you please refer me to it? Many thanks.

Does Robin Hanson ever define what he means by "bias"? I looked for such a definition once on his OvercomingBias blog, but did not find it.

"Bias" typically has a pejorative connotation, so perhaps anybody who is against bias must be seen as wearing a white hat. Is that the source of the the wind in Robin Hanson's sail?

When I was a teenager learning about vacuum tubes, bias was a voltage applied to the grid of a triode. It was the basic, or starting, voltage upon which the signal, or varying voltage to be
amplified, was added. So bias was necessary. A good thing.

I see bias playing a similar role in the self direction of human beings. You get out of bed in the morning because you are biased with a desire to get something done. Should we try to overcome such bias?

Robin Hanson writes:

If you thought biases were usually small, that would be a good argument to not pay much attention to overcoming bias. But if you think biases in choosing causes are usually large, then it is harder to see why you would give overcoming bias a low priority.

Matthew c writes:

I think Hanson is absolutely correct that overcoming bias is a great thing. Unfortunately it became quite clear to me over time that the blog "overcoming bias" was in fact anything but. Part of overcoming bias is questioning your assumptions and taking leave of groupthink echo chambers and exposing yourself to very different ideas and communities of thought. It became clear to me that OB had become a lost cause for that ideal, so I stopped reading and commenting there recently.

I don't blame Hanson for this, he seems very sincere about truth seeking, but some of the others appear to believe in the the infallibility of their own thoughts and assumptions. . . For people unwilling to accept the existence of the category of the things you don't know that you don't know, and unwilling to accept that some of the things you "know" might be wrong -- there is no exit from the cul-de-sac. . .

Here's the kind of thinker I think can actually help us overcome our biases -- someone willing to take some thoughts and run with them, willing to make mistakes and be wrong, willing to explore ideas for their own sake instead of turning his blog into an opportunity to signal his vast intellect and supreme importance.

This comment was the ultimate shark-jumping moment for Overcoming Bias:

. . .I would not bother to voice my objection except that you are the leader of a project that if successful will impose on the entire future light cone decisions that will have the same unbendable and irreversible character that physical law now has.

That comment was the beginning of the end for me at O.B. . .

Thanks to Robin Hanson for his response. The point which I try to make has nothing to do with whether a bias is large or small. It is rather with the meaning of "bias".

Right on the surface, my problem with a blog named "Overcoming Bias" is its assumption that bias is a thing that needs to be overcome, its assumption that bias is a bad thing. But bias is not necessarily a bad thing. In many cases bias is a necessary thing.

So I have an understanding of "bias" which equates more with "predisposition", a neutral term, whereas Robin's usage seems to imply that he sees "bias" as meaning "error".

Zubon writes:

In the relevant context, "bias" most likely does mean "error." You presented an engineering example, so let me give you a perspective from statistics. There, "bias" is a failure to be accurate. A measure is biased if it produces a result that is skewed, like a golfer with a tendency to slice. For example, a survey is biased if the group polled is not representative of the whole. It produces an inaccurate answer that is a flawed basis for decision-making. Worse, if you do not know that you have been given an inccurate result, it looks just like any other statistic, and it fits quite neatly in your plans until your final result is wrong.

Overcoming bias is about being more accurate (with respect to thought). Is "accuracy" a suitably neutral term? If the effort is to motivate action to improve accuracy, exploiting the bias from a negative term like "bias" seems entirely appropriate and deliciously ironic.

TGGP writes:

Holy crap, Matthew C. You left because of Hollerith, who is just a commenter like you rather than a contributor? If you just give up on conversation when you can't convince others that you are right, that leads me to believe that YOU are the one who is biased rather than interested in the truth.

Barkley Rosser writes:

For what it is worth, following up on the link to Mencius Moldbug, Eliezer Yudkowsky can be especially arrogant and biased, although he is provocative and clearly very smart.

I applaud Robin Hanson's efforts. I suspect where some of us disagree with him is that there is a fuzzy zone where indeed the post-modernist critique has it right, where truth is to some degree in the eye of the beholder for a variety of reasons, with Robin being more likely to deny the reality of that zone. Some time ago I gave up more strongly held pomo tendencies. But, paradigms and all that, it is very hard to sort out "bias" necessary to carry out any analysis from the sort of bias that distorts the search for objective truth that Robin valiantly battles against.

Barkley Rosser writes:

Ooops! I meant to say that Robin Hanson is valiantly battling against the bias that distorts the search for objective truth. Sorry about the mistatement.

Matthew c writes:

Holy crap, Matthew C. You left because of Hollerith, who is just a commenter like you rather than a contributor?

No TGGP. Hollerith and the responses to his comment just pointed out to me very clearly the echo-chamber characteristics of the place. The last time I heard that kind of cultic messianic thinking was a brief period of involvement with organized religion in my late teens.

The basic problem with the blog Overcoming Bias is that overcoming bias requires a deep sense of humility and an understanding of the tentativeness of our ability to model reality through concepts. Some of the most prominent and predominant participants there appear to lack an ounce of that humility, and even pride themselves on that lack.

Cynic writes:

I was under the impression that Matthew C was unhappy that folks at Overcoming Bias were dismissive of his 'case' for psionic powers/ESP.

A pro-psionics/supernatural position seems to be the very opposite of humility.

Cynic writes:

". . .I would not bother to voice my objection except that you are the leader of a project that if successful will impose on the entire future light cone decisions that will have the same unbendable and irreversible character that physical law now has."
It seems that the key words here are 'if successful.' Given the history of AI and the distribution of computer expertise we should expect SIAI to fail in its mission, but that doesn't change the expectation that the creation of self-improving superintelligent AI would be an epoch-ending event that could not be reversed by humans without AI consent.

Michael Munger writes:

Boy, it was brief.

And yet still much too long.

Consumatopia writes:

If I switch from my cause to the cause of fighting bias, while my opponents stick to their cause, that asymmetry would seem to introduce a new set of biases. If you could make everyone switch causes uniformly this would be a good move, but otherwise it will be those with reasonable causes who switch while those with unreasonable causes retain theirs.

Also--not all causes come down to factual disagreements. People just care about different stuff, and sometimes causes set some concerns against other concerns. Security vs. liberty. Wealth vs. equality. It's not clear how "bias" applies to normative rather than factual issues.

Comments for this entry have been closed
Return to top