Well, they’ve tied with him once - perhaps they will get in the same situation in the next game, and play the same strategy so they both win again. That would be the most overall benefit considering the two of them (a bit like prisoner’s dilemma, where I consider the “solution” to be the one where both prisoners always co-operate, as that is the greatest net benefit).
However, the fatal flaw with this is that if both players get it right, by not betting the maximum Semret has lost his chance to tie, and simply ends up coming second. So it’s a really bad move for that reason, IMO.
Ayup. I’m a friend of Semret’s, and he said he didn’t realize his error until Alex pointed it out afterwards. He’s pretty philosophical about it, which I would not be.
FWIW and on the topic of the thread: he didn’t have a problem with the board-skipping, but he did think that Arthur’s dominance is mostly about being phenomenally good with the clicker.
I’m sure felicity with the clicker cannot be separated from his success, but I would hesitate about ‘mostly’. It seems to me that his success is ‘mostly’ due to his knowing more and about more diverse topics than the other contestants. I often see him being beat out on things that you would expect everyone to know, and come in with the correct answers on more obscure things.
If he’s better on the clicker, that’s probably largely a function of experience. He’s been on the show how many times now, when his opponents have only been on once or twice. Practice matters.
Well, I think you answered your own question. We see it in sports all the time, do we not? Anyone can kick a 30yd field goal, or score a penalty kick in soccer, or pot a straight ball in pool, but when it really matters, pressure gets to you. I can easily imagine his thought process: Chu is known to bet for the tie, so I’ll bet for the tie, too - then if the tie happens, he’ll be nice to me again next time! Without realising the fatal flaw I have already mentioned.
Add a step in there, and it becomes an even more reasonable mistake: Chu is known to bet for the tie, and it’s worked out well for him, so I’ll bet for the tie, too.
It’s hard to separate, especially since knowledge (or at least, confidence in your own knowledge) can give you a leg up on ringing in. I forget if Ken Jennings actually said it or it was just a theory, but I remember talk that the best players often just ring in first no matter what, then hope they can come up with the answer.
And man, how good was that Brad Rutter vs. Mike Dupee matchup on Friday? Mike came all-so-close to knocking off arguably the best Jeopardy player of all time. Really fun to see him frantically trying to ring in first on that last category, knowing he had to get within half of Rutter’s score.
I wonder if we’ll end up with another Rutter/Jennings showdown again.
In the prisoners’ dilemma, they are not allowed to collude ahead of time. No matter which choice the other one makes, you will come out ahead by turning the other one in. And you can’t directly influence the other one’s choice. So you might as well go for it all.
In the case of Jeopardy, you get to keep all the money you have if you win. Unless you’re going for the big payoff and want to bet it all, there’s really no difference between beating someone by one dollar and tying them. One dollar is insignificant. The payoff for the tie will likely be worth more than a dollar. There’s really no comparison to the Prisoners’ Dilemma at all.
You can’t collude ahead of time in the Prisoner’s Dilemma, but you might be able to collude afterwards. If you have the opportunity to play the game repeatedly, you should always play nice until and unless the other player turns on you, because that’s better in the long run. And in real life, there’s almost always going to be some other “game” with the same players some time down the line.
I am aware of how prisoners’ dilemma works. My argument is that yes, logically you always do better by turning the other one in. But given both players know the rules ahead of time, although they cannot collude, in my view they should both always co-operate, since if they do it gives a better outcome for both of them than if they both turn the other in. I agree this is not a perfect analogy for the way Jeopardy! works (in fact there was a short-lived British gameshow called Goldenballs* in which the final round was based precisely on prisoners’ dilemma), I was just saying there is some similarity in that it can work your favour to co-operate in the short term at a small cost, for long-term benefit (as in iterated prisoners’ dilemma). This is essentially what Chronos is saying. I like to extend the principle to a single game of prisoners’ dilemma.
*I don’t want to continue this hijack any further, but in Goldenballs the two final contestants had to “split” or “steal” the prize money they had collectively won. If they both opted to split, they would each receive half the pot. If one opted to split and the other opted to steal, the person who opted to steal would take the whole pot. If they both opted to steal, neither won any money at all. Having explained that, I now see it is subtly but crucially different from the traditional prisoners’ dilemma, in that by stealing you receive nothing if the opponent also steals, but if you split you also receive nothing if the opponent steals. In other words, if your opponent decides to steal, it doesn’t matter what you do.
The way the gameplay worked, the two players were allowed to discuss what they were both going to do beforehand, but made their final choice in secret. Naturally this lead to some entertaining games where people would do their best to convince their opponent they would definitely split, before going ahead and stealing anyway. Of course, I hope that in some of those cases the winner agreed to hand over half the prize money after filming anyway - whether the rules of the game allowed this or not.
Now, my point is that you only get one shot at this (unlike iterated Prisoners’ Dilemma), so you have to get it right. I would do my utmost to convince my opponent that I would split (and then do so), because although he would then know he can take the pot by stealing, if we both thought like that then we would both end up with nothing. Therefore stealing is in fact a losing option. The only way to ensure money is won (by somebody) is to split. I would hope that I could be sufficiently convincing to achieve this outcome, but I was never on the show so who knows?
In a strictly logical sense, if you’re only ever going to have one interaction with that person, then you should in fact try to convince your opponent that you’re going to steal, not that you’re going to split: That way your opponent has no incentive to also steal, and knows it.
Except that, when dealing with real human beings, that’s going to backfire, because our social instincts evolved in an environment where isolated games are almost nonexistent, and so people will try to punish anticooperative behavior even when that punishment won’t have any benefit.
That’s exactly what happened on the aforementioned Golden Balls. I heard about it before, but it popped up again last week on RadioLab:
[spoiler]One contestant (#1) insists that he will steal and will not budge on negotiating to a split. He flat out states he is stealing and that if the other player (#2) chooses split, he (#1) will split the money after the show. Arguing ensues. Contestant #1 does not waiver.
No, I also read articles on news sites and the like. Not only is every single article supportive of him, so are most of the comments. I suppose there may be quite a few people who don’t like him, but they’re definitely not close to the number that do.
Well, I know I’m repeating myself, but most of the commentary seems to revolve around his betting and Daily Double strategies… and those strategies are secondary to his basic skill at answering questions at the right second.
Fishing for Daily Doubles is ONLY an option if you’re constantly answering fast and first… which Arthgur Chu does.
And playing for a tie only works if you’re always in the lead at Final Jeopardy AND confident you’ll know the answer. Again, Arthur Chu HAS steadily been in the lead for Final Jeopardy.
My guess is, he’d still be winning steadily even if he followed the plain vanilla strategy most players use.
My bet for the last three standing will be Rutter, Mark Lowenthal and Ken Jennings. It would have been nice to see a rematch between Jennings, Rutter and Jerome Vered (from the Ultimate Tournament of Champions), be the latter got beat by Tom Cubbage last month.