Cost/benefit analysis of voting

I've always viewed voting in a presidential election as a civic duty and celebration of an important right that is all-too-easy to take for granted, but not as something that is justified purely from a narrow act-consequentialist perspective, because the odds of our vote changing the outcome is so astronomically low. But I've been reading Derek Parfit's "Reasons and Persons", which makes a compelling case why that may be wrong (or at least, similar thinking applied to similar types of choices). That book has been full of stuff that has been changing the way I think in some form, and I'll probably make many posts about topics from it.

Basically, for the same reason that a single vote has a very low chance of changing a presidential election (i.e. it is decided by so many people), it also has a very high potential impact (because it would affect so many people), and so we have to weigh the former against the latter. In the "Ignoring Small Chances" chapter of the "Mistakes In Moral Mathematics" section of the book, it explains how you might calculate this. The expected benefit of an act is the possible impact multiplied by the chance that the act will produce it. So you calculate the value of your vote (to America) between 2 candidates, versus the cost in your time to do so, like this:

((the average net benefit to Americans if your candidate is elected) x (the number of Americans) x (the odds of your vote causing your candidate to win)) - (the cost of voting)

At the time this book was written, there were about 200,000,000 Americans, and apparently a common estimate of the chance of your vote tipping the scales in some states was 1/100,000,000. Using these numbers, the sum will be positive - i.e. voting is worth your time from a pure cost/benefit analysis - if the average benefit to Americans if your candidate wins is more than half as great as what taking your time to vote cost you. And voting doesn't cost you much.