Tuesday assorted links

1. People are averse to machines making moral decisions.

2. Declining Chinese productivity growth.

3. Paying the gender stereotype tax in poker.

4. New NBER paper, possible overturning of the Autor, et.al. results on the China shock?  I am not able here to read through it, however.

5. “We find that at least 31.2% of the citations to retracted articles happen a year after the article has been retracted. And that 91.4% of these post-retraction citations are approving.”  Link here.

Comments

2. Since productivity is defined as the ratio of production output to input, I can understand it falling as China reinvents itself as the leader in tech, from AI to robotics to clean energy cars. No, not the copy-cat technology Americans complain about, but original technology. Why would productivity growth fall? Because it takes awhile for input into technology to generate output. Anyway, that's one explanation, although not the explanation that would be appreciated by many China critics and pessimists.

As I have pointed out many times, we are entering a new phase of trade with China, from one dominated by China firms making products for American firms (such as the i-phone) to one dominated by China firms making products for China firms to compete with products made by American firms including products made in China for American firms. Do Trump and his advisers appreciate this sea change, or are they implementing policies based on the old China rather than the new China?

Given his preference for coal, steel and aluminum, it certainly seems like the Trump administration is focused on preserving 1970s jobs rather than helping us compete internationally in the modern economy.

@#2 - productivity falls when measuring "total factor productivity" (the GDP not due to capital deepening and extra people) just after recessions, I've heard, so after the Great Recession of 2008 it's natural for TFP to fall. Exception: it rose during the Great Depression of the 1930s, by most estimates, but usually it falls during 'ordinary' recessions. Thus much ado about nothing.

Bonus trivia: On the advice of Rayward, I'm reading Adam Tooze's "Crash". It's surprisingly bad. It's like a journalist's account of the Great Recession. Still, it has good facts which I cut and paste for my form file, but it's verbose and full of priors (the writer is left wing, and quotes for example, as authority both magazine articles and stuff like, P. Augar, The Greed Merchants: How the Investment Banks Played the Free Market Game (London: Penguin, 2005), which I also will read but it shows a journalist mentality by Tooze.

An historian can only provide an account of the past. He cannot replicate the past because the past is no longer accessible. Is Tooze's account of the financial crisis accurate? More or less accurate than the conflicting accounts I've ready by many economists? Do historian's have a stake in their account of the past? Do economists have a stake in the ir account of the past? One phenomenon that's become obvious is that many economists have become advocates (of a point of view, philosophy, etc.). Some for reasons of financial gain. Are historians advocates, and if so, is it for financial gain? As you know, I am an advocate, but it's no secret that I am; it's the nature of our adversarial system. Can a lawyer be relied on to give an accurate and unbiased account of the past (e.g., an auto accident)? As between historians and economists, who are more likely to provide an accurate and unbiased account of the past?

If you think the book is bad, read the review of it by Fareed Zakaria
the NYT.

Did he even LIVE through it like we did?

Irrationality is always more costly than rationality, and collectivizing people is irrational. If her opponents tried to view her as an individual and not as a "female poker player" she could not use their willingness to stereotype against them.
This really illustrates a point that many libertarians have been making counter to the alt-right: that stereotyping people based on race or other collective groupings is not actually a rational optimal strategy, even if the stereotypes are on average true. It's always to your advantage to try to look at other people as individuals and recognize how they differ from the mean. (Or even just see that there are many means and many collectives).

That was in response to #3.

I can easily imagine a bunch of dudes in the sausage fest of the professional poker world making wrong assumptions about female poker players and getting pwned for it, but that said, are we 100% sure this is irrational? Maybe the typical woman player doesn't bluff that often? Men and women do display (on average) differing levels of aggression; I don't think it's irrational for male players to take that into account when playing women. The irrational part is "never fold to a woman because its emasculating."

The link is 2 female poker players typing male poker players into 3 categories.

They are doing the same stereotyping through a finer screen. Everybody stereotypes; that's learning.

Finer screening is better than coarser, always. Right down to the level of picking up on individual nuance.

The women are not stereotyping, they're classifying some men based on observed behavior in order to decide on a strategy for playing against them.

Sure. If it's your ox, it was gored. If it's your foe's ox, it was blah blah blah handwaving.

You nailed it, China Cat. Pro player here. The game is nothing if not an exercise in intelligent stereotyping. What's it mean when a person like this makes a play like this in a spot like this? If you're not stereotyping, you're losing.

Also, poker imitates chess in that (most) women are not up to the current level of the game at higher stakes. That said, the female players who *can* cut it would mop the floor with the likes of old-school hustler Annie Duke.

It's not a good strategy even if the probabilities are correct, because a person (like this subject) aware of the strategy of, say, assuming women bluff less often, can exploit that strategy, even if the premise is technically correct.

If I know you know Women on average bluff less, then I can modify my strategy to exploit that.

Yeah, but only if you know that they know, which you may not know. You can't know if they know, you know?

In the broader analogy, you always know that the other person "knows" about the stereotype, because that's what makes them stereotypes - a stereotype is by definition a popular perception that is commonly believed. It's plausible that there are a lot of people out there who are "contrarian" or "non-conformist" precisely because it (in a way) is a way of exploiting other people's belief in stereotypes.

Maybe...it kinda depends on how knowledgeable and experienced the other person is. Plus, it's complicated by the fact that there's skill involved. To take another analogy, in baseball, teams study opposing hitters' tendencies all the time and throw them certain pitches or don't throw them certain pitches, and shift their fielders around towards the parts of the field where the batter frequently hits the ball. Knowing this, because he's able to see it with his own eyes, at-bat after at-bat, the hitter should adjust and start hitting the ball to other parts of the field, and thus these big infield/outfield shifts shouldn't work. Most hitters are not successful at this, though...because it's simply hard to go against your natural tendencies. I think even if you know the stereotypes that exist around your demographic, sometimes it's still hard to avoid living up to them.

Right, well, in that analogy the hitters actually are being studied individually. They aren't saying "black hitters always go left", they're looking at the statistical record for that player. There's going to be a lot more cases where individuals don't conform to a group average than when they don't conform to their own personal record. The bigger and more generic the group, the more such cases there will be, and thus the stupider the strategy it is.

You are addressing a tangential matter. My point was this: if adjusting ones aggression levels was easy, we wouldn't see differences in the first place. It's just a trait that's pretty well wired and it's difficult to force one's self to behave in ways that are contrary to that for long periods of time, absent a lot of practice.

That said, I still think you're a little off base (no pun intended). There are plenty of instances where managers don't have good information about a specific player's tendencies: rookies, new minor league call-ups, international signees, etc. I'd be willing to bet that managers still use shifts against those guys based on what information they do know about them, just not as dramatic or exact as one's where they have better data, and it may not be much more than "this guy's a lefty power hitter, and most lefty power hitters...."

It is a good strategy as long as you believe she won’t exploit it. If you believe she will adjust, then you should adjust to exploit her adjustment. Whether she hypothetically can adjust or not isn’t particularly relevant to your strategy if you believe she won’t.

If you read poker fora, this discussion is repeated ad nauseam. The "adjust your strategy to exploit your opponents' weaknesses" approach is called "exploitative" and the "if your opponent adjusts then you have to counter-adjust" line of reasoning leads to a game-theory optimal or "GTO" style of play. Of course, even though Nash taught us that a GTO strategy exists, I don't think anyone actually knows what it is for multiplayer no-limit Texas Hold 'Em.

Exactly. And if your opponent is playing anything close to game-theory optimal strategy, the correct response isn’t to make your own play unexploitable - it’s to find a new opponent.

Depends a lot on what you mean by "costly." Cognitively and computationally, groupings, heuristics, and stereotypes are drastically less costly than figuring out every individual. Note, of course, that the way these female poker players attempt to exploit male players is by a simplified grouping.

If you have 100k hands against someone online: sure, by all means, solve them and exploit where they aren't playing GTO. If you're playing live poker, that's not realistic, and everyone is going to be relying on simplified player types to some extent. Has nothing to do with libertarian and the alt-right and everything to do with basic poker ideas like image and leveling.

A person who always chooses the least costly strategy can also be called "lazy". Stereotyping is a lazy strategy.

The more hands you are able to observe someone playing, the more information you have and the less you will rely on stereotyping. But on the first hand it’s not an option to have formed any opinion based on “looking at her as an individual”, because the looking has not yet taken place. Your options are either to stereotype, or to play every new player exactly the same. Stereotyping is more profitable.

How often do you really ever only play one had at poker against someone?
The difference between playing "women never bluff" on hand one and then adjusting your strategy after wards and playing "random" on hand one has to be marginal.

The individual-specific playing information you have on hand 1 is non-existent. After 30 minutes, it’s very poor. After an hour it’s poor, after 2 hours it’s mediocre, and so on.

So what you end up doing is starting hand 1 saying to yourself “he/she is X gender, Y age, Z race, dresses like this, is talkative, has a good luck charm, knows the dealers etc. and based on that I’m going to tentatively assume tight passive, not creative, but I have fairly low confidence” or whatever. Then over the next half hour you’re going to play like 15 hands, and he/she is maybe going to play 4 of them, and maybe you end up seeing his/her cards at the end once. Maybe what you see is consistent with your initial read (based on stereotyping) or maybe it isn’t, and you adjust from there, but your confidence level doesn’t become very good for many hours, and you’re probably still relying on stereotyping to some degree until it does.

So to answer your question, yes, the difference is marginal on the first hand against any given opponent. But if I consider that marginal difference with all of my opponents over however many hands I play in my life, it matters a lot.

But then, if your opponent knows you're likely to stereotype them on the first hand (and on a number of subsequent hands thereafter), then they are going to adjust their strategy and not play according to type. So it's to your advantage to also not play according to type which means therefore not play as if you expect them to play the stereotypical way. In other words, stereotyping is like having a "tell". The other person is going to *expect* you to stereotype, and play accordingly.

No, that’s not really the way it works in practice. What you are doing is assuming your opponents are very thoughtful. In actuality, they are probably not-very-thoughtful degenerate gamblers. If they’re not, you should go find a table where they are.

Most weak players do not adjust to how they read how you read their image. If they were capable of that level of thinking they probably wouldn’t be weak players to begin with. In other words, you’d pick up other signs that your initial read was wrong and this player is aggressive/creative/whatever and you would abandon that initial read. And yes, they probably could use that to their advantage for a short time before their skill level became exposed (which is what the women in the article did.) But most of the time this won’t happen, because the player you stereotyped as weak actually is weak and doesn’t make those kinds of adjustments.

Also, even when your opponent is actually a good player laying in the weeds, you’re protected to some degree from the counter-adjustment threat by the fact that your good opponent will have a hard time being confident enough that you will adjust that he/she can counter-adjust.

Re 5. I wonder how much of that result is driven by the time it takes from submission to publication and how much is the peer-review not paying attention to the situation.

5. Why are there so many retracted articles? Are most retractions due to failure of replication? Sloppy research? Faulty models? In the cited study, the authors used a data base of over 3,000 retracted articles. 3,000!

3. Then there's Elizabeth Holmes. Who paid the gender tax for her bluff? I suppose nobody believed she was bluffing because she is a woman.

#3 all demographics are (often profitably) stereotyped in poker, and all good players try to exploit their own image - it’s not something that’s unique to women. I found it striking how personally these two seemed to take their opponents’ strategies. “I get bullied by men who believe women don’t belong at a poker table.” I think she’s probably misreading motivation here. Poker is a competitive game - someone who is trying to run you over is typically doing so in an effort to win.

Don't most professional poker players kinda take the attitude of "if your money's green and you're not as good at poker as I am, you belong at the table?"

If you can't tell who the fish is when you look around the table, it is probably you.

3. Of course, the difference between bluffing and lying is simply a matter of degree. We are all familiar with the overly self-confident male, who soldiers on no matter the odds or the reality. At some point, doesn't even he realize that it's just lying? Most likely, but he doesn't care; indeed, the lying and the bluffing become one and the same. Donald Trump comes to mind, and so perhaps does Elon Musk. They continue to succeed by gaining legions of followers who are so invested in the man and his bluffing and lying they must rely on self-deception. The thing about Robin Hanson (signaling and self-deception) is that his observations about behavior are so obvious but few notice; indeed, few notice even after Hanson has made it clear. My observation is that people are fascinated by the bluffers and the liars, and only wish they were as capable of bluffing and lying as the master poker player, Donald Trump, or Elon Musk.

In poker, over the long haul the player who has the best grasp of mathematical possibilities, ie. odds, will be the winner, regardless of the bluffs of the other players.

Well, an important part of the odds is, for example, the odds your opponent is bluffing. Or the odds he opens 79o from the cutoff. These are not things that can be accurately calculated - they have to be estimated on the fly.

We are all dead in the long run. One can play for a while an quit the game.

> In poker, over the long haul the player who has the best grasp of mathematical possibilities, ie. odds, will be the winner, regardless of the bluffs of the other players.

Looking at the number of global national ranking players per million pop vs the country's average GRE verbal and quant scores,

http://i64.tinypic.com/20f6me.png

For poker, that metric is very very significant wrt GREVerbal and negatively so so significant wrt to GREQuant. On average in poker, Verbal IQ is more important that Quant IQ, the higher the Quant IQ the worse the performance. The people from China, HongKong and Macau (an international casino city) are bad at playing poker. Those from Singapore, Taiwan, Vietnam and Korea are also not that good.

#3

How soon before THIS paper is retracted?

#4:
"You are eligible for a free download if you are a subscriber, a corporate associate of the NBER, a journalist, an employee of the U.S. federal government with a ".GOV" domain name, or a resident of nearly any developing country or transition economy."

With other words, thanks for supporting the creation of this paper, assuming you are a net tax payer, but unless you're already sucking on the taxpayer teat, you will not get free access.

#1 But what if our autonomous machines were programmed to follow Kantian ethics or Aristotelian ethics?

Kantian morality would be predictable and much easier to program than an approximation of Aristotelian ethics, but most people aren't willing to accept the consequences of deontological morality.

#3 is great, if true. A testable hypothesis about sexist attitudes.

Except, why would we take a poker player's story at face value?

Re: retractions and citations. Yes, the social psychology literature is still littered with Stapel citations.

#1. Ties into Tyler's proposition that the most important future thinkers/intellectuals will be religious/theological.

Comments for this post are closed