If the only cost of overrunning a meter were paying the amount of the overrun, any economically rational actor would simply never pay meters. If he doesn't get caught, he's parked for free; if he does get caught, he pays only what he otherwise would have.
The civil penalty for laws like unpaid parking meters and illegal downloading of copyright works should be slightly greater than the original cost times the inverse enforcement rate. If we catch 1% of parking meter cheaters, they should pay at least a hundred times their overrun to ensure that it's economically rational to pay meters. Since we need only deter humans, not "economically rational actors", we can take risk aversion and incomplete knowledge into account and probably decrease that fine somewhat.
The same principle that applies to parking meters applies to copyright infringement.
However, more generally the final penalty needs to be adjusted by the penalty associated with being prosecuted. If the penalty is 1,000$ / song, but you need to go to court to defend yourself then the effort associated with that court case is also part of the penalty of enforcement. Also, because your fine can't drive you past bankruptcy ridiculous penalties have little meaning (this plays into the last banking bubble. I can sell a trillion dollars worth of insurance far below costs if I don't have a trillion dollars because my downside is limited to the value of my assets.)
Hey look, it actually happened in Mumbai: http://www.freakonomics.com/2007/03/26/how-to-cheat-the-mumb...
That's almost true, but you haven't accounted for marginal utility.
But I guess that if the problem really exists, people would be pro big parking ticket fines, because a minority who get those also annoys the whole population.