Thursday, September 16, 2004

Going On Vacation

Blogging may be light.

Sokal at CBS

As a corrollary to yesterday's post on people not looking for disconfirming information, look at how biases make people suckers.

The recent CBS fraud, where some huckster tricked CBS into publishing fake memos on Bush's National Guard Service, shows how bias makes you vulnerable to cognitive jujitsu where someone takes advantage of your blindness to make you do something you rather would not have.

It seems that either Dan Rather, or CBS, had some liberal bias and/or personal animosity against Bush Jr. Since they all "knew" that Bush shirked his National Guard service, they were easily taken in by a fake memo that supported this "fact". You can see this mental softness in this revealing headline from the liberal New York Times: "Memos on Bush Are Fake but Accurate, Typist Says"

On the other hand, a memo that claimed Bush served exemplarily, for example, would have been scrutinized much more thoroughly.

Right-wing readers were much more skeptical of CBS/Rather's claims, and so had incentive to find something that disconfirmed the memos. Lousy typography, dubious formatting, inarticulate signatures made easy targets. Once people started looking for disconfiming evidence, lots began turning up, and "experts", who had perhaps supported CBS, began to back away from their claims. CBS could have asked the experts to find potential inauthenticities in the documents but they did not, they only asked them to confim authenticy.

We don't know who fed these documents to CBS. It may have been cooked up by Kerry partisans who just want Bush to look bad. It could have been cooked up by Bush partisans who preyed on Democrat (and mainstream media) cognitive weaknesses and fed them the memos as bait. If it was Democrats, they have committed fraud. If it was Republicans, they've executed a hoax.

This reminds me of the Sokal hoax, where he feeds a humanities journal an "article" that was just a mix of mumbo jumbo and far-left platitudes. It was published.

Tuesday, September 14, 2004

Real on Slashdot

Rob Glaser, the CEO of RealNetworks, has a good Q&A on /.. The most interesting response (to a question about pricing):
Based on the data we've seen, we think, long-term, the pricing that will result in the biggest overall market for music will involve some kind of tiered pricing new mainstream songs for 99 cents retail, and up-and-coming artists and back catalog artists at a lower price.
Differentiated pricing indeed. My biggest issue with all the plans to introduce mandatory licensing etc. is that they would eliminate price competition at the mp3 level, which is exactly where it needs to be. Some songs should, legitimately, be available for download for free.

The Wisdom of Crowds

I just finished reading Suriweicki's book on how aggregate decision making is better than individual decision making. Riffing off Charles MacKay's "The Madness of Crowds"--which discusses how mob mentality leads to market bubbles, riots, and other forms of badness--Suriweicki goes through all the cases where group behavior, aggregated, do better than individual behavior.

It's a good book and touched on many topics I've been thinking about for the past year or so. He begins by stating that group averages of estimates, such as the number of jelly beans in a jar, are usually much better than most, if not all, individual estimates. He also goes through the markets, democracies, teams, behavioral finance, and the dangers of group think.

I don't feel he went far enough, though, into why groups make better decisions than individuals, and therefore what goes wrong when they start making worse decisions (such as market bubbles).

Often, when people want to take a swing at Markets, they trot out some behavioral finance experiment that shows people are not perfectly rational. This is a poor choice of argument, firstly because people do not need to be perfectly rational for markets to be efficienct, and secondly because it's much more straightforward to simply point out asset price bubbles instead as examples of gross market inefficiency. Whenever the opponent resorts to talking about "animal spirits", you know they are on the defensive.

To me, the behavioral economics helps explain the value of markets, not undermine then. Individual cognition is a weak and biased thing, it's brittleness easily revealed in simple experiments where smart people can be reliably made to make dumb choices. Markets limit, or actively work against humans' natural cognitive limits, and reward those who overcome these limitations best. Humans are just as cognitively flawed in committees, or by themselves, and only the anonymity and abstractness of the market gets them to try and manage these frailties.

For example, individuals over value stuff they have and undervalue stuff they do not have. So, people will not part with a coffee mug they have been given for $5, but will refuse to buy it for $3. Markets abstract away the feeling of ownership (when you buy some Coca-Cola stock, do you really feel a part of the whole Coca-Cola family) so people make better decisions about what the real value is. This disinterested abstract anonymity is one of the many things that the anti-globo mob, but it is instead a real strength.

Creating procedures that debias or rebias our natural cognitive weaknesses is also central to science. Humans, for example, do not seek out disconfirming evidence. When asked to guess the rule behind a number sequence, and allowed to test other number sequences to see if they follow the rule, most people do something like this:

Tester: The sequence is "2, 4, 6". What's the rule?
Subject: Does "6, 8, 10" follow the rule?
Tester: Yes.
Subject: How about "20, 22, 24"?
Tester: Sure.
Subject: OK [Things hard] What about "1000, 1002, 1004"?
Tester: Yes, that follow's the rule too.
Subject: The rule is that the numbers have to go up by 2 each time.
Tester: Wrong.

The subject usually never tries to disconfirm his initial hypothesis. He never tried "1, 2, 3" (which would follow the rule) or "1, 2, 100" (which would also follow the rule) or even "3, 6, 9" (still OK). By now you've probably guessed what the rule is, but please note how the process of actively seeking out disconfirming evidence is contrary to human nature, and then think how the scientific approach, which is to "come up with a hypothesis and then try to disprove it" actively takles that fallibility. Popper and Kuhn were quite right to point out how Science often happens in very unscientific ways, but in light of the above I think they totally miss the point. Nothing can stop humans being human, all you can do is try to build on your strengths and limit your weaknesses.


Firstly, in a market prices are determined by the marginal buyer, who may be rational (or at least, expert), and secondly because it's much easier The best refutation of efficient markets is price bubbles, and Chicago certainly does not have a good answer into why these occur.

The book also did not touch on Arrow's Incompleteness theorm, which reveals the inherent arbitrariness of *any* political process, and which I consider central to a non-stupid discussion of politics in general. But I thought the ending was right on--democracies don't make good decisions, but the decision to have one is very wise indeed. Squabbling and mudslinging are ugly, but they are better than insular decision making.

Friday, September 10, 2004

Straight talk from Odlyzko

Everyone's favorite telco guru, Andrew Odlyzko, has a good little post on how 3G cellular telephony will succeed even though the TV on your cellphone vision it once promised will never materialize.

He makes a distinction between penetration (what % of a population has a cell phone) and usage intensity (how much the average cell phone is used). Even though cell phone penetration is very high, usage intensity remains quite low--certainly when compared to landline usage intensity.

Flat rate pricing does most to increase usage intensity, and the prevalence of block minute plans in the US goes some way to explaining why usage intensity is high here and rising, as penetration continues to approach European levels. In Europe, although the penetration is higher, usage intensity is much lower, and falling. I don't know what's happening to prices there, but as more and more marginal customers become cell phone owners, one would expect the average usage intensity to fall.

In the US, cellular pricing is moving towards flat rate all-you-can-eat. Right now, most plans have early mornings, nights, weekends, and in-network calls for free. It's easier to list what's not free, which is simply business hours for out-of-network calls. Sprint's recent Fair and Flexible plan goes against this flat-rate pricing trend, and offers a genuinely new pricing product. Time will tell if it's successful.

UpdateIt's been pointed out to me that TV phones are already here. This is true, but WAP is already here as well, and I would not consider that a success either.

Wednesday, September 08, 2004

Usability and Community

Joel has a very nice post on usability in software. One, he claims that webpages are now so usable in general, that there are entire categories of websites (such as content sites) where further investment in usability is unneccessary.

I used to think the web was easy to use wayback in '98, but some time with Creative Good convinced me that, infact, websites are very complicated for the average user and lose lots of money as a result. At any rate, I'm sure that things are better now, but I'm not willing to say there are now worthwhile improvements to be made.

Two, Joel also talks about how interface design impacts the use (or lack thereof) of software that facilitates group interaction in some way. Getting the social interfaces correct make tremendous differences where end-user adoption is critical to success. In a lot of business software, end-user adoption is unimportant since the purchase decision was made by some manager who never actually has to use the damn thing, and employees are largely free to sabotage any management plan at will, including software implementations. But where the purchaser is the user, at home or in a smaller business, ease of use really really matters, and getting a good UI is the difference between getting the benefit of having the software, or not. When the software's primary output is interaction between humans, you must take human instincts into account (curses! not humans!)

(For a very interesting list of well considered design conditions, read this article by Joel where he goes through why his own discussion board is designed the way it is.)

I did a little online community work a while ago and stumbled upon Amy Jo Kim of Naima. I read her book, Community Building On the Web and found it to be lousy. She seemed to be way too taken with fancy gimmicks like avatars and virtual worlds while ignoring the more down-to-earth and meaningful observations like Joel's "I don't let people quote because then they quote too much". Since UI guides (but does not drive) behavior, and you have *some* sort of UI no matter what, you are driving particular behaviors even if you don't mean to. Best be concious about your choices and pick good ones.

Monday, September 06, 2004

That which cannot continue...

Social Security and Medicare (especially Medicare) are transfer programs that take money from young people and give it to old people. As the ratio of young people to old people greys, young people are going to have to be much more productive than before to continue giving money to the old people at the rate the old people have been promised. Or they are just going to have to give more of their income in % terms. Most likely both.

It's worth thinking, for a moment, about what other transfers may kick into place, transfers that will take money from old people and move it to the overtaxed young. Some places, such as Japan, are well down this road already, and it's interesting to see how they have been coping to date.

On obvious-in-hindsight tactic is that young people stay with their parents for longer. As the parent does not charge the child rent, this is essentially a transfer from old to young, and helps to balance out the payments going the other way. Another is the rising cost of education -- old people invest in the human capital of young people, capital that they then put to work throughout their lives. Young people can also ask the old people to donate their labor -- through cooking meals, doing laundry, taking care of the kids, etc.

If you put all that together, it seems that there are many mechanisms young people can use to help defray the looming old-people bills.

Thursday, September 02, 2004

Easy, fun games

In the spirit of expanding the increasingly hardcore gamer-dominated gaming market, Nintendo has created a game where players play bongo drums. Neat, and less socially marginal than Dance Dance Revolution.

Wednesday, September 01, 2004

Sun blogs

Reader DD sent in a note to this piece on the economics of software from a Sun kernel engineer.

It is quite correct, although I think it overstates a little bit the pain of big corporations at the mercy of evil software vendors. For example, while it is true that once a company has purchased a platform and built on it, they are now locked in (and can get price gouged), it is also true that a company can anticipate this and negotiate a big discount upfront, eliminating the price gouging that's coming down the pipe. Upfront competition (like after market competition) is important in creating substitutes, and thus lowering available economics rents, but often goes unconsidered. Hal Varian's "Information Rules" continues to be the best treatment of how all this stuff shakes out.

To me, the real people feeling pain are the poor schmucks left with trying to use the software that their company just installed. The CFO bought on price (check), the CEO bought on CYA (big name vendor? check), and the CTO bought on a laundry list of acronoym features (check check check). No one bought on usability, and hence, true business value. Couple this with your average business, riddled with complex and imprecise rules, and you get failed implementation, and failed operation. And frustrated users.

It's interesting, and neat, that Sun seems to have so many smart, open bloggers.