I've never written a bot. I thought perhaps that the process of parsing instructions (we could put more than one variable in the equation), then writing code to solve, might be more effort than it's worth to them. Especially if we're doing something that is somewhat unique so their reward would only be one page. However, phildawes is apparently implying that once the bots have parsed the problem and know what to solve for they could go to a page that implements Mathematica and submit the problem to be solved, therefore saving the time required to write problem-solving code. I didn't realize they worked like that.

agreed, computers are good at solving math problems or running Turing machines (since they are them). You would be better off with a turing test (http://en.wikipedia.org/wiki/Turing_test) that only a human can answer - and if any does get a computer to solve it you get them to start a company because they have solved a fundamental computing problem.

Computers are good at simulating Turing machines. Computers aren't good at determining whether Turing machines halt; in fact, it is impossible for a computer to determine in the general case whether a Turing machine will halt.

Fair enough, though it might have the effect of not letting humans in either ;-)