2 ms·
If you convert "greed and fear" to "rational self-interest, tempered for risk", there's a field of study called mechanism design that looks at how to set up sys
by geofft 11y ago
If you convert "greed and fear" to "rational self-interest, tempered for risk", there's a field of study called mechanism design that looks at how to set up systems and incentives so that everyone's rational self-interest yields a desirable overall result.
I suspect one of the biggest problems with adapting an algorithmic result to the real world is that the resulting system has to be self-sustaining and resilient: you have to provide enough incentive not to change the incentives. Leaving aside its internal merits or lack thereof, it certainly did the Soviet Union no favors to be opposed by a roughly equally-powerful nation that was constantly threatening military action if they didn't change their mechanism. The mechanism itself has to solve that, or it will require benevolent people to keep it in place who have zero incentive to remain benevolent. (See also, the Soviet Union.)
In most other cases you don't have this problem because people actually are benevolent enough. If you're designing a congestion-control protocol like TCP, you just have to make it incentivize nodes to behave; you don't also need TCP to incentivize sysadmins to keep running TCP instead of replacing it with something else.
- derefr 11y ago> I suspect one of the biggest problems with adapting an algorithmic result to the real world is that the resulting system has to be self-sustaining and resilient: you have to provide enough incentive not to change the incentives. An analogous situation within a different optimizing system: it's pretty easy to create a GMO tweak to improve crop yields for the indivdiual seeds you created. It's a much harder problem to create a tweak that's actually advantageous to the plant, such that will be kept in the next generation and spread to fixation.