Hacker News Reader

The prisoners dilemma is ultimately just a demonstration of game theory.

That is there is more than you win/I lose or I win/you lose but there is win/win.

It’s definately conceivable two algorithms or two ai would be capable of concluding the same. In fact it may be even easier because what’s the actual dilemma/self interest of such a system without the selfishness of human nature?

View on HN

Why would you intentionally operate an AI that wasn't as selfish as possible?

Because game theory/prisoners dilemma/the red black game (whatever you want to call it) suggests you would get a better outcome otherwise.

if you ruin your only competitor, the antitrust folks will come after you

Nah, they still can come after you if you prevent new firms to enter the market. (But if you just do the usual regulatory capture the gov. will be just happy for you, as you gave them a lot of busywork for nothing.)

My pet peeve is how "The Prisoner's dilemma" is the most commonly used example of game theory while being one of the worse examples (because it involves no information exchange and a skewed payback)

Reply on news.ycombinator.com