Essays on collectivity - 3 - Nash vs. Smith, tit-for-tat
March 28th 2007 01:08
Hey all,
almost over with our reflections on collectivities, we'll talk this time about the Nash solution for a problem once solved by Smith. As i said in the last post, we're in the point where we discover that Smith was write, but incomplete.
The problem in his way of seeing things is that he didn't take time in account. "The best results are reached when everyone thinks only about themselves". The prisoner's dillema is a good case that shows that this way of seeing the problem is the good one, but not if we apply this in a sequence of cases.
This means: if you don't know your partner, and you'll never have to deal with him again, you should betray the other. But if you'll have to interact with him again in another situation, things won't be what we can define as "cool"...
Nash saw what organised groups learned in years of betrayals from both sides. People have memory; once you betrayed, the others will betray you too.
I'm not saying all this out of nowhere, mate. There's an actual case of study for the prisoner's dillema on this. When they play it with many iterations, the best algorithm ever developped for it is one they call "tit-for-tat". Resuming the thing, it works like this:
- the first time you play, you cooperate
- the second time on, you play what your "partner" played in the last round
- you "forgive" sometimes
Following this strategy, you win; nothing has never worked better than this (until today...). Hehe, it's a good example that shows the difference between Nash and Smith. Notice that each rule followed by the strategy has an importance: the first one shows that you're there to cooperate; the second tells that you don't take your problems home; the third one shows that when everybody is betraying, it is good to show again that you're there to cooperate...
And so that's it for this post. In the next one we'll by tying the knots and applying all this back to our conclusions on ethics.
See ya. Uula
almost over with our reflections on collectivities, we'll talk this time about the Nash solution for a problem once solved by Smith. As i said in the last post, we're in the point where we discover that Smith was write, but incomplete.
The problem in his way of seeing things is that he didn't take time in account. "The best results are reached when everyone thinks only about themselves". The prisoner's dillema is a good case that shows that this way of seeing the problem is the good one, but not if we apply this in a sequence of cases.
This means: if you don't know your partner, and you'll never have to deal with him again, you should betray the other. But if you'll have to interact with him again in another situation, things won't be what we can define as "cool"...
Nash saw what organised groups learned in years of betrayals from both sides. People have memory; once you betrayed, the others will betray you too.
I'm not saying all this out of nowhere, mate. There's an actual case of study for the prisoner's dillema on this. When they play it with many iterations, the best algorithm ever developped for it is one they call "tit-for-tat". Resuming the thing, it works like this:
- the first time you play, you cooperate
- the second time on, you play what your "partner" played in the last round
- you "forgive" sometimes
Following this strategy, you win; nothing has never worked better than this (until today...). Hehe, it's a good example that shows the difference between Nash and Smith. Notice that each rule followed by the strategy has an importance: the first one shows that you're there to cooperate; the second tells that you don't take your problems home; the third one shows that when everybody is betraying, it is good to show again that you're there to cooperate...
And so that's it for this post. In the next one we'll by tying the knots and applying all this back to our conclusions on ethics.
See ya. Uula
| 45 |
| Vote |
