Quote:
Originally posted by Kyo
Actually, I've been over the Prisoner's Dilemma quite a number of times - but always in a purely game theory context. I suppose this would be a case of me not being able to put two seemingly unrelated concepts together.
There are several points to make, however.
1) TIT FOR TAT does not take advantage of unresponsive strategies. For instance, if paired against a strategy that always cooperates, TIT FOR TAT will always score 3 points, when it could always defect and score 5. In real-world terms, I see this as being able to reliably 'dupe' the other party - something which corporations and governments are quite good at.
2) TIT FOR TAT also doesn't do very well against random strategies, because it inherently assumes that the other strategy is attempting to earn the most points. Humans are far from perfectly logical and are prone to indecision and random action.
3) TIT FOR TAT is also a victim of echo strategies. If you are paired with another TIT FOR TAT, except that it throws in a random defection, they will alternate between cooperate and defect forever - earning a lower average than the cooperation payoff (by alternating the 'sucker' payoff and the 'temptation' payoff)
|
That's true. But all of these points are addressed in this book. Going into detail on them, would be going far too much off topic, but the simplest thing to bear in mind, is whether or not populations using these strategies are stable or not.
1. "COOPERATES ALWAYS" would be very easily "invaded" by DEFECTS ALWAYS.
2. TIT FOR TAT does
reasonably well against RANDOM. It is not it's best partner, but it does as well as can be expected. Plenty of other strategies also do terrible against RANDOM.
3. This is also true, and is one of the biggest problems with TIT FOR TAT. A solution to TIT FOR TAT's problem is available, but is unworkable within the prisoner's dilemma, but works in the vast majority of "real" situations.
You might call it 75% TIT FOR TAT. If you assume that rather than just having the clear cut option of defection or cooperation, you can also have any shade of grey in between, a more durable form of tit for tat would be to defect by 75% of the amount that the other player did.
So if the other player defects by 0% (i.e. cooperates) then you also defect by 0%. If the other player defects by 100% (i.e. defects) then you defect by 75%, and the other player will defect by 56%, and then you will defect by 42%, then he will defect by 31% etc etc.
I'm sure that there is a "golden" number, I just picked 75% off the top of my head.
Quote:
[b]To demonstrate a different point, however, consider Shubik's dollar auction. I have little doubt you've heard of it, but just in case, here are the rules:
a) A dollar bill is being auctioned, and will go to the highest bidder. Bids start at one cent, and each new bid must be higher than the previous one.
b) The second-highest bidder still has to pay - but for nothing!
This is obviously an extremely bad situtation to be caught in. Someone will obviously bid 1 cent. If you can get 99 cents of the deal, why not? But then, someone else thinks, "I could have that dollar for 2 cents." They bid 2 cents. But now the first bidder is in the unhappy position of paying 1 cent for nothing, so he bids 3 cents, etc. We keep going until the bid hits $1.00 even. But the previous bid was 99 cents. If he doesn't make a new bid, he will lose 99 cents. So he bids $1.01. And the game continues. Regardless of how high the number, the second-highest bidder will always be able to improve his position by almost a dollar by topping the current high bid.
This experiment was conducted at MIT, and it was found that a dollar could routinely be 'sold' for amounts much larger than a dollar - people were buying a dollar bill for $5!
Cooperation, in this case, would be to just let the first bidder take the dollar - for 1 cent. But that never happened.[b]
|
Well this auction is really a similar thing to a
single round of the prisoner's dillema. In order for cooperation to arise, you need for the same group of players to go through the auction numerous times, the only difference being that there is not "mutual" gain. It
is a case of you vs. me.
But in an abstracted sense, a TIT FOR TAT-like stategy could be found stable, allowing each player to take turns in winning the dollar for a penny....
However, i don't see how this links in with the argument at hand?
Quote:
|
On a different note, about spreading your seed rampantly - wild animals do it, and it seems to work for them. I don't know enough biology to form a more complex formulation, so if you can explain why it wouldn't have worked for humans, I'll take it at face value.
|
I guess it's a different stategy of reproduction.
Do I run around and try and make as many offspring as possible, and then hope that at least, by the law of averages, some of them are to survive.
Or do I "put all my eggs in one basket" and stick around to protect them, to make damn sure that they survive?
It's the same with plants. Some plants produce loads and loads of seeds, and just have them scatter around in the hope that at least a few of them will grow (dandelion). Some create only a few seeds, but these seeds are much more robust, and individually are much more likely to survive. (Coconuts).
If you look at it from the point that 90% of illegitatimate children would die, but only 50% of fathered children would die (figures of the
pulled out of my ass variety, obviously enough), it would seem that rampant procreation is the way to go...
5 sexual partners
= 1 sexual partner.
so anything greater than 5 sexual partners will mean that you are doing better than raising your own kids.
but if you look at the population at a whole, you will see that its numbers would very rapidly decline if this were the case.....90% of the children born would die!
So in this over-simplified model, we can see that it makes evolutionary sense to have a single partner, but different results could be formed by "tweaking the percentages", which may in an abstracted sense, explain why some male animals do not help raise their children.
However, I would point out the difference between helping to raise children, and monogamy, but the two are related.