The FREAKest Links: “MySpace 101” and Home-Cooked Samosas Edition

Looks like all that time spent on MySpace could start earning you college credits. Via Andrew Lavallee at the Wall Street Journal: More and more universities are incorporating curricula on social computing, allowing students to study subjects like online communities, social networking and user-contributed content as part of graduate and undergrad programs.

This month in Scientific American, Cornell economics professor Kaushik Basu revisits his twist on the classic Prisoner’s Dilemma. Called the Traveler’s Dilemma, the game involves two travelers who find that identical belongings of theirs have been damaged by an airline. The airline manager agrees to compensate both, but first asks them separately to write down the object’s price as any dollar integer between 2 and 100. If both write the same number, he’ll pay each that amount, but if they write different numbers, he’ll assume the lower is the actual price.

Saritha Rai at the New York Times profiles India’s now-booming business for dabbawallas, local delivery men who have joined up with Web service providers to form an extensive network in which they deliver home-cooked lunches to thousands of busy professionals — a nice example of traditional customs merging with modern technology to confer a public benefit.


jonathank

I almost referenced the SciAm article in response to Dubner's post about complex systems. The article is great. My favorite line is "Some analysts have argued that many people are unable to do the necessary deductive reasoning and therefore make irrational choices unwittingly." Wow! You mean people don't act optimally. Stop the presses! I assume this line was meant in the spirit that theory needs to be adjusted, but still.

lermit

Did that take long?

.lermit

procrastinating_econ

If the airline manager's only criterion is the lower price. This isn't exactly prisoner's dilemma situation because first he has given both travelers a range of 2-100. Regardless of what the other passenger chooses, either passenger SHOULD choose 100 as a utility-maximizing agent!

And you won't believe how many grad schools I have been rejected from! Okay, maybe this is too easy a problem.

davinic

It seems that the TD setup really fails the dynamics that PD provides: PD pits the two players against each other, while TD does nothing to make either player care about the other's outcome.

>>Continuing with this line of reasoning would
>>take the travelers spiraling down to the >>smallest permissible number, namely, 2.

The way the scenario is set up, there is no incentive for either party to try and "best" the other. Each part would be completely happy with $100, or $101 and $97, respectively (if the low bid was $99), because it is more than each paid for the item.

Now, if the setup were that, if they differed, then only the lowest bid would be paid, then $2 is obviously the answer. But as written, I fault the premise more than "people are unable to do the necessary deductive reasoning and therefore make irrational choices unwittingly".

baggie

Surely the answer to the Travellers paradox is that real people work with probabilities. E.g. playing the Nash equilibrium = 100% chance of $2. Playing say the number 90 gives you x% of some reward around 90 and (1-x)% chance of $0 (playing against a Nash player). As long as you judge x > 0.03 (e.g. 3% of the population plays generously) you will be ahead by playing high in the long run. I suspect utility plays an aspect too. $2 is not enough to win to make a difference. Make it $20,000 (e.g. a million dollar pot with steps of 10,000) and you will start to see a lot more Nash players.

pparkman

What I wonder is that no one seems to think that the players in the TD ought to tell the truth. In the long run, honesty is always the best policy. While you may get a few cents more or a few cents less in any one situation, the next time this game is played, you may be the airline manager. And the contempt for the truth that these exercises bring only leads to a society where everybody cuts corners. If you compare the cultures around the world, you will notice that the most desirable places to live are those where the citizens are the most honest.

see

If I assume the other player is selecting randomly, then $99 is my best choice; my average payoff is ~$49.06. If he assumes I'm random and selects the same, we both come out with $99.

If 95% of the time the other player will be picking the game-theoretic optimal choice of $2, and 5% of the time he will select completely randomly, then the expected value for selecting $99 is ~$2.45, while the expected value for selecting is ~$2.20. So a mere 5% chance that he will select effectively randomly makes picking $99 a more rational choice than picking $2. How often do you think other human beings are irrational? 5% of the time?

If I assume the other player is answering honestly, I know honest answers are still going to have some variance; appraising isn't an exact art. If I give any response above him, I'll come out $2 short of his honest answer; if I go even a mere $3 below him, I come out with less money than if I'd gone way above him. If I think he's honest, I might as well treat it as a random number and go with $99.

If I assume he's totally naive, I think he'll go $100; $99 gives me a return of $101.

If I assume he thinks I'm irrational in any of those ways, he's going to go $99 in them. So if I assume he's perfectly rational but thinks I'm less-than-perfectly rational, my best choice is then $98 . . . and we go down the spiral again if we both assumed the other is perfectly rational but might suspect we're irrational.

But, unless I learn he's rational from a series of similar events with him, I have no reason to expect he's perfectly rational. So I'm betting $99 anyway since that maximizes my payoff if he isn't . . . and if he's truly perfectly rational, he's going to come to the same conclusion, and we'll both get $99.

Read more...

schadenfreude

Actually, with a random selection, the optimal play is 95 or 96 dollars. Taking a guess of 96 vs a guess of 99, they are equivalent when the other player chooses 2 thru 95. When the other player chooses 96, playing 96 gives a 2 dollar advantage. When the other player chooses 97, playing 96 gives a 4 dollar advantage, for 98,99 and 100 playing 96 gives a 2, -1 and -3 dollar advantage, respectively. So the net payoff is about forty cents higher for playing 96 versus 99.

schadenfreude

Correction: (96,97) vs (99,97) gives a three dollar advantage for playing 96. Which brings the expected value of playing 96 down to thirty cents above playing 99.

frankenduf

of course pparkman's point is the ethical answer- the deeper dilemma is then when is it rational to be unethical- this question cannot really be gleaned from travelers and suitcases, but this may be a fundamental critique of capitalism- i.e. does it create an environ where it is rational to be unethical

ctcarton

In the TD problem, if you're assuming that the other person is using the exact same decision making process as you are (and you don't invovle probabilities or randomness in your decision) then 100 is the obvious answer. Any game theory analysis that says otherwise is obviously flawed. But that's the point he's trying to make in the article: that current game theory has this flaw.

The problem is the recursive nature of assuming that the other player is using the same logic as you are and then basing your logic on that assumption. It's like a math problem that, depending on the intial conditions you choose, can converge to different possible solutions or diverge completely and produce no solution at all.

A proper analysis would have to consider multiple initial conditions until it found the one for which the solution converged on the most favorable result. Or, considering that the number of initial conditions could be infinite, you might have to settle on a result that's 'good enough'. Sort of like a hill climbing algorithim where you have to pick different starting points until you find a good solution.

Read more...

nothings

Actually that the logic involves a series of discrete steps moving backwards, eliminating a choice from the model ofthe other participant, is very reminiscent of the hanging judge paradox.

This is also a very light treatment of the subject, but it seems deeply flawed in how it works by picking a train of thought: assume they start by picking $100. You could also try: assume they start by picking $80, and the same result will obtain as long as the model assumes both people pick the same starting number. In fact, if your model assumes that they pick two different numbers but they're known, it still collapses to the nash equilibrium. But obviously the start conditions aren't known (it's perhaps probabilistic, as suggested above), and in this case there is no "selfish rational model" to follow that leads to the NE. So it seems weird to me that they consider this to upset the selfish-rational model.

Read more...

nothings

Whoops, left out one more remark: the other thing is that as was discussed in this blog previously, there isn't a linear relationship between money and perceived value! This single issue could possibly explain the entire thing, leaving the actors perfectly rational and essentially selfish.

The gain or loss of a reward of $2 at $100 is nearly imperceptible. The gain or loss of a reward of $2 at $2 is proportionally enormous--but it's still an irrelevant amount of value compared to $100.

Thus it can be perfectly rational (under this hypothetical value model) to offer $100; you lose $2 no matter what the other person picks, but the range of possible payoffs extends up into a realm with actual value, whereas picking $2 guarantees the payoff is near valueless.

This model nicely explains the decision to pick a number in the $90s; it keeps you up in the high-value region, and attempts to pick up a few bucks in the process.

Read more...

shaungallagher

To me, the reason why a person would gravitate toward 100 is quite clear: It's the golden rule at work! It's one of the most basic principles of cooperation. In fact, so universally understood is this rule that I think it's justifiable for both parties to presume that the other is aware of it. You might even call the golden rule a sort of "implicit communication" -- it's a way for both parties to arrive at a strategy without any direct communication between them.

Can it be taken advantage of? Yes -- but that involves a breach of cooperation. And the thing about game theory is, you should never assume that your "competitor" is more virtuous than you are. Thus, the more you're tempted to swindle, the more you should assume the other person is, too. And if that's the case, then "pure" logic wins out, and you might as well just chose 2.

SteveEisner

Wow. This article made me feel dumber for having read the whole thing. It's extremely frustrating to follow the author's hand-wringing journey through the "rejection of rational behavior". The comments on this blog are far more insightful.

Sure, I get the idea that classic game theory (or "pure" rationality, or whatever) will typically lead to the Nash equilibrium, and that's not happening here, and wow isn't that mind-blowing. But this is just a game theory version of Zeno's Paradox: a theory doesn't match the real world, so which one is wrong?

Instead of trying to figure out why people aren't "purely" logical according to contrived definitions, he should reconsider the scenario he's set up. The situation sets you up as co-conspirators against the airline, and it's actually a win-win situation that you *both* get paid if you *both* over-estimate. Sure, you may get less than your opponent but you *both* do far better by cooperating than if you compete.

At some point you can try to game the system by bidding one dollar less than your "opponent" but why bother? He'll be doing the same to you. Anyone would realize that you can't go very far down from 100 before it's obviously a path to mutual destruction. And then, the only number you could both agree to tie at is 100, so it's a clear choice.

As others have pointed out, it would drastically alter the course of the game if the $ difference between winner and loser were "real money". The author alludes to this as well. See also: http://blog.plover.com/risk.html

Read more...

SteveEisner

(By the way, I don't mean to imply that the original author is "dumb" - clearly he is better versed in economic theory than I am. But his presentation of this as a failure in rationality is not convincing.)

JesseAlred

I am a veteran teacher in Houston seeking a dialogue with Teach for America teachers nationally regarding policy positions taken by former Teach for American staffers who have become leaders in school district administrations and on school boards. I first became aware of a pattern when an ex-TFA staffer, now a school board member for Houston ISD, recommended improving student performance by firing teachers whose students did poorly on standardized tests. Then the same board member led opposition to allowing us to select, by majority vote, a single union to represent us.

Having won school board elections in several cities, and securing the Washington D.C Superintendent's job for Michelle Rhee, Wendy Kopp's friends are pursuing an approach to school reform based on a false premise: that teachers are the cause of sub-par academic performance in urban schools, They disregard major factors like the degree of parent commitment, students habits and economic inequality. 

The corporate-TFA nexus began when Union Carbide initially guided Wendy Kopp's efforts to create Teach for America. “I assembled a board of directors composed mostly of the corporate chiefs whom I had met through the head of union carbide,” she acknowledged later.
A few years before, Union Carbide's negligence had caused the worst industrial accident in history, in Bhopal, India. The number of casualties was as large as 100,000, and Union Carbide did everything possible to minimize its responsibility at the time it embraced Ms. Kopp. TFA recently started Teach for India. Are Teach for India enrollees, who presumably love their country and its people, aware of the the Union Carbide/TFA relationship?

When TFA encountered a financial crisis, Ms. Kopp  nearly went to work for the Edison Project, and was all but saved by their managerial assistance. The Edison Project sought to replace public schools with for-profit corporate schools funded by our tax money. Think Haliburton in your neighborhood. Ms. Kopp's husband, Richard Barth, was an Edison executive before taking over as CEO of KIPP's national foundation, where he has sought to decertify its New York City unions.

In 2000, two brilliant TFA alumni, the founders of KIPP Academy, joined the Bush's at the Republican National Convention in 2000. This gave pivotal cover for Bush, since as Governor he had no genuine educational achievements, and he needed the education issue to campaign as a moderate and reach out to the female vote. KIPP charter schools provide a quality education, but they start with families committed to education. They claim to be improving public schools by offering competition in the education market-place, but they take the best and leave the rest.

D.C. Superintendent Michelle Rhee's school reform recipe includes three ingredients: close schools rather than improve them; fire teachers rather than inspire them; and sprinkle on a lot of media-thrilling hype. Appearing on the cover of Time, she stood sternly with a broom in hand, which she was using to sweep trash, the trash being a metaphor for  my urban teacher colleagues. MS RHEE, MY COLLEAGUES WHO WORK IN SOME OF THE TOUGHEST SCHOOLS IN THE NATION ARE NOT TRASH.  They are American heroes, Ms. Rhee!  

TFA teachers are highly effective educators. My mentor, when I started teaching, was a TFA teacher, ironically, Ms. Rhee's interim Director of Human Resources, and he saved me in that first, difficult year. But when TFA's leadership argue that schools, and not inequality and bad habits, are the cause of the achievement gap, they are not only intellectually dishonest, they feed  the corporate influence which has blocked social changes we need to bolster our middle class, they aid the people who say the public sector can do nothing right, and thus should never regulate businesses or provide national health insurance or protect a worker's right to organize.

Our society has failed schools by permitting the middle class to shrink. It's not the other way around. Economic inequality and insecurity produces ineffective public schools. It's not the other way around. Ms. Kopp claims TFA carries the civil rights torch for today, but Martin Luther King was the voice of unions on strike, not the other way around. His last book, Where do we go from here?, argued for some measure of wealth distribution, because opportunity would never be enough in a survival of the fittest society to allow most of the under-privileged to enter the middle class.

TFA teachers, please remember, its your hard work in the classrooms that gives TFA executives moral credibility and a political platform, its not the other way around. Please contact me at JesseAlred@yahoo.com.

Read more...

jonathank

I almost referenced the SciAm article in response to Dubner's post about complex systems. The article is great. My favorite line is "Some analysts have argued that many people are unable to do the necessary deductive reasoning and therefore make irrational choices unwittingly." Wow! You mean people don't act optimally. Stop the presses! I assume this line was meant in the spirit that theory needs to be adjusted, but still.

lermit

Did that take long?

.lermit

procrastinating_econ

If the airline manager's only criterion is the lower price. This isn't exactly prisoner's dilemma situation because first he has given both travelers a range of 2-100. Regardless of what the other passenger chooses, either passenger SHOULD choose 100 as a utility-maximizing agent!

And you won't believe how many grad schools I have been rejected from! Okay, maybe this is too easy a problem.