Effective policy making: How to recognize and avoid Martingale-like phenomenon?

I was commenting on Cathy's post on big pharma's evil influence on physicians and the comment grew too long, so I decided to make a post our of this. I think the post was a good example to discuss the role of policy making and its perceived role amongst think tanks on the liberal--conservative spectrum, in general.

I can't access the article, but I don't really agree with Dana et al. (2003) 's recommendations to outrightly ban gifts. More generally, I believe that banning first-order incentives (eg. free dinners) will only sprout second-order incentives (eg. awards of recognition for good medical practice certified by BIGPHARMA, or featuring the physician in a popular article by scratching some journalists' backs) that are 'legal'. The same phenomenon is observable in Madoff-like Ponzi schemes. See a post about the Martingale Crisis (link via Laszlo). In one line, the Martingale hypothesis is: "Suppressing system-circumventions by banning them will lead to further circumvention by more innovative system-circumventions".

So, what is NOT the solution? The solution is not to go on banning every incentive system that crops up. I'll go so far as to say that in my opinion, this is symptomatic of think tanks on the far left and far right, who favor top down solutions from the government rather than bottom up solutions by involving the parties concerened. Rather, it is more effective to recognize that incentives are fundamental to ANY negotiation, be it at the level of policy making among the power-elite, or at home between husband and wife, parent and child. If you need people to get on board with a decision/action/law/yadayada, you need to INCENTIVIZE them, and the most effective solution will be to ask each party what they want. Sadly, this is easier said than done. Of course, the importance of being unbiased cannot be stressed enough.

For a more theoretical analysis of such concepts, we could look to the liberal paradox and arrow's impossibility theorem. I hereby incentivize anyone who is willing to translate these theorems from acadamese to English, by promising a beer-day-out on me.