« Living By Your Own Strength | Main | No Overconfidence? »

December 21, 2008


I've noticed that a mark of academics is that they expect the answers of important questions to be more complex and subtler than others expect. It would be great to collect data to see who is more biased here, academics or non-academics.

I don't think you're wrong generally, but you understanding about the tit-for-tat competition is off. Every competition but the first basically consisted of everyone trying to beat tit-for-tat. So its not really fair to say that these scientists found something about tit-for-tat unattractive, they were just trying to beat the already established competition (its not like they would have been allowed to submit tit-for-tat by a different name!). Now its true that only one person submitted tit-for-tat for the first tournament but 1) there are more simple strategies that were submitted (defect every time, cooperate the first time- defect afterward, cooperate every time... etc.) and 2) Obviously most of other submitted strategies are going to be more complicated because most of the other strategies ARE more complicated. I think if you too a random selection of every conceivable strategy you would probably have just as many complex strategies as you had at the first competition.

what bias or fallacy is related to people's desire to pay for things that they could have for free?
is it the handicap principle?

it seems similar to people looking for complex solutions to problems that already have simple solutions.

I agree with the overall theme of the post but am skeptical about the two examples.

Is it really the case that tit for tat didn't exist at all, before scientists gave other strategies a shot. I would be quite dubious about that assertion.

I guess after tit for tat was already established, scientists have been attempting to either (1) prove something rigorous about the optimality of tit for tat (to my knowledge there is no theoretical proof, it's just that tit for tat has been highly robust and optimal when playing a wide range of strategies).
(2) come up with strategies that are better than tit for tat.

I suppose scientists have given (2) a fair shot, since they didn't think they could make much inroads into (1). Their job as scientists is to further the state of knowledge, not merely implement the simplest state of the art solution.

That said, scientists/engineers in Industry often seem to exhibit the tendency to attempt to innovate for marginal gains, when the need for the hour is to simply have the best known method implemented.

As for index funds, I think the problem is largely that of there being an inadequate market for them in the general public. I don't think the general public has yet absorbed the idea that they (and most fund managers except perhaps the very best, who are hard to identify ahead of time) are no Buffets and that the market can be taken to be efficient as far as they are concerned. Thus people are certainly willing to pay more for an actively managed fund even if it performs worse in reality.

Coming back to scientists, I can easily see how the reviewing process for conference proceedings or journals can push people towards coming up with needlessly complicated solutions. I have written a few papers in computer science, and invariably found that I would lose some points in the "technical depth" section of reviews when proposing a simpler, faster and more accurate solution to a problem than an existing one.

atheism, Freudianism, natural selection, socialism or libertarianism

Hmm... aren't two of those simpler than the alternatives, and two about equally simple? Atheism is a lot simpler than dealing with the problems brought about by postulating a god. Freudianism, that's the odd one out, it's quite complicated. Natural selection is not that complicated, surely no more complicated than creationism. Socialism is about as complicated as capitalism. Libertarianism is lots simpler (and falser) than more nuanced alternatives like Rawlsian liberalism.

Indeed, atheism is simpler, but believing in God is more intuitive for the default human brain, and hence simpler for people. Some atheists might actually choose their position due to show-off bias, but this seems rare to me, as theism can be shown to be false quickly with good evidence - turning atheism into an obvious belief. Show-off bias is unlikely to apply to belief formation when the belief is obvious, even if rarely held.

Atheism might be adopted to show off, not intelligence, but the strength to withstand living in a cold uncaring material universe. Similar reasons might (I think often do) support (profession of) belief in moral nihilism, imminent human extinction, imminent world domination by [political enemy], inevitable failure of cryonics or other transhumanist projects, and many other unpleasant things.


The later Tit-for-Tat tournaments don't strike me as relevant. But you raise a good point about complex strategies comprising most of the available "strategy space."

Paul and Pete:

I call the bias "Show-Off Bias" and not "Complicated Bias," because I don't think it's limited to complicated answers. Smart people will be attracted to answers that are simple, if they are still counter-intuitive--if they still allow the person to show off.

Simple answers can be right for fancy or complicated reasons. For example, many argue that voting is irrational based on collective action or freerider problems in economics and game theory. So smart people might be biased to say that voting is irrational, if doing so allows them to display their understanding of these counterintuitive concepts.

"It seems to me that self-identified smart people are biased towards complex or counter-intuitive answers to problems"

It's peculiar that you chose 'self-identified' here since it isn't referenced again. Did you intend for that to be an important distinction?

Yes, it is an important distinction. You can be smart without self-identifying as smart. And you can stupid and still think you are smart. Show-off bias should only apply to people who think they are smart. Whether they actually are smart is less important.

I don't refer to "self-identified" again because it's awkward and made the sentences harder to read.

"i don't refer to "self-identified" again because it's awkward and made the sentences harder to read."

did it make the sentences harder to understand or just degrade the prose?

self-identified smart people often sacrifice clarity and precision for style. instead of making ideas easy to comprehend they present them in the most impressive, stylized way possible. i think this is far and away the most common and insidious way the "Show Off Bias" manifests itself.

Everyone on Wall Street has an incentive to say bad things about index funds. This is another reason index funds aren't popular, which might be difficult to disentangle from show-off bias.

The problem with your hypothesis is that it is inconceivable that nobody who entered subsequent competitions didn't know that Tit-for-Tat was a candidate for optimal strategy. So what was the point of subsequent competitions? To test the hypothesis that Tit-for-Tat is the optimal strategy. It's not clear that those who entered the first competition didn't know this as well.

This hypothesis seems... hm...

...true :-)

Index funds struggle to fully catch on because they are a bit of a catch-22.

Index funds work great if enough market makers are doing their job correctly for the market to function efficiently (i.e. people not using index funds).

The second everyone uses index funds the second the market complete stops being efficient. By definition index funds cannot filter or select the best companies. All they do is piggy back on others' work.

Corollary: Because a simple answer to a complex problem is always counter-intuitive, the bias of a self-identified smart person toward complex and counter-intuitive answers is not a detractor.

Therefore, smart people should only be given complex problems. Let those with less intelligence (or at least more humility) handle the simple ones, because they will perform more accurately (i.e. with less bias).

Therefore, smart people should have assistants. Or, to quote Don Lapre, "never do anything yourself when someone else is ready, willing, and able to do it for eight bucks an hour". (Actually, in my experience, you can't expect competence for less than twelve.)

Not always.

See this Ted Talk by Murray Gell-Man about how he looks for "simplicity" as an indication of truth:


So how come we don't see counter-showoff bias, where people in communities of people who adopt, e.g., atheism, adopt things like theism to be contextually counterintuitive?

Another explanation for this phenomenon is a complexity is fun bias. If you realize how good tit-for-tat is, you'll stop searching, which means the fun is over.

Finding subtle differences is fun because of the pleasure of grokking. I always get a kick out of understanding something on a deeper level than it's being explained to me.

As for stocks: People also have come up with complicated strategies which are supposed to help them make money while gambling. If there is the potential for easy money with very little work, people will overanalyze.

The show-off bias is what comes in at the end when you share your discoveries, even though they are useless.

Robin wrote: "I've noticed that a mark of academics is that they expect the answers of important questions to be more complex and subtler than others expect. It would be great to collect data to see who is more biased here, academics or non-academics."

Important questions don't have answers. They have increasingly accurate and complex approximations. Academics are more willing to consider more complex approximations. That doesn't mean they have a lower opinion of the simple answers.

(Though it wouldn't surprise me if they did.)

Index funds are not that much different, fundamentally speaking, than actively managed funds. After all, someone has to choose what stocks get put in the index; an index fund simply copies the algorithm used by the makers of the index. It is not a "zero intelligence" strategy of picking stocks at random.

In the case of the S&P 500, the stocks in the index are chosen by a committee of humans, using their own judgment. (They attempt to select stocks that form a representative sample of the economy of the United States as a whole.) If index funds that follow the S&P 500 beat actively managed funds, then wouldn't that make the members of that committee abnormally good at stock picking? ;)

Interestingly, because index funds exist, whether or not a stock is listed in an index affects the price of the stock. If a stock is taken off the S&P 500, all the index funds that follow the S&P 500 have to sell it, depressing the price of the stock. Similarly, a stock that gets added to an index experiences a surge of demand, causing its price to rise.

Philip, yes more precisely academics expect the most interesting or useful answers to be more complex and subtle.


What you say is true if indexes were a collection of stocks picked by really smart people who evaluate what businesses have wonderful prospects etc.

In reality all they look to do is follow a simple algorithm that needs no smarts.

You just start off with the CAPM style reasoning which leads you to conclude that the optimal portfolio to hold (from the point of view of minimum variance 'risk' for a particular expected return), is simply buying the 'market'. This in other words means that you should put some fraction of your money in all stocks and other risky assets in the world, with the weightage being proportional to the market capitalization of the stock or financial instrument and the remaining money in a "risk-free" instrument like US treasury bills.

All index funds do is to approximate at least a country's market, by picking a sufficiently large number of stocks with high market cap. And thereby avoiding the hassle for the individual investor to actually go and buy 500-10000 odd stocks in a market cap weighted fashion.

So for all practical purposes, the index is fairly "zero knowledge", since the algorithm is quite precise and public, which is "create a large basket of stocks that approximates the whole market to the extent possible".

I disagree in part (for there will alway be showoffs). Complexity in explanations are at times a way to "blanket" a problem and fill all the nooks and crannies with a theory before simplifying that theory via a common explanation.

I liken it to writing and to quote Pascal ... "My Reverend Fathers, my letters have not usually followed so closely, nor been so long. The small amount of time that I have is the cause of both. I would not have made this so long except that I do not have the leisure to make it shorter."

Perhaps the reason for complexity is lack of time to make it simpler?

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