Embed Notice

@bonifartius @freemo @louis

If you have some estimate of likely results (in the form of a probability distribution over vote counts -- likely you'd assume an uncorrelated multivariate Gaussian or something similar), then it's fair to say that you can estimate the direct effect of your vote: compute the probability (as a function of what vote you cast) that your vote will be the deciding one, and pick the vote where you like that distribution the most.

In places that have a threshold this usually means voting for a party that has a significant chance of being above the threshold, sadly. Apart from this it often doesn't align with voting for larger parties (nor even, in FPTP, parties that are very likely to get at least one seat in your constituency: voting for a party that's expected to be just below 1 seat is better than for a part that's expected to be above 1 and well below 2, if you like both equally).