Saturday, April 22, 2017

Universalizing Tactical Voting

I regularly come across two objections to tactical voting, i.e. voting for Lesser Evil rather than Good in hopes of defeating the Greater Evil candidate.  One objection is just the standard worry that individual votes lack instrumental value, debunked here.  More interestingly, some worry that tactical voting is positively problematic, morally speaking, on grounds of its putative non-universalizability.

On one version of the worry, tactical voting involves (something approaching) a contradiction in the will, insofar as even if those who most prefer Good constituted a majority, they could get stuck in the inferior equilibrium point of all (unnecessarily, and contrary to their collective preference) supporting Lesser Evil.  On another version of the worry, tactical voting involves (something like) a contradiction in conception, insofar as it involves responding to how others plan to vote, which might seem to depend upon those others voting non-tactically, i.e. not waiting to first learn how you plan to vote.

To avoid either worry, it suffices for tacticians to follow Regan's co-operative utilitarianism (which is, in general, the correct theoretical solution to any sort of coordination problem).  To wit:

(1) Identify those who are willing and able to cooperate with you (by following this very decision procedure) in pursuit of the best collectively attainable outcome.

(2) Determine the best available total plan of action for the group of cooperators, given how non-cooperators will actually behave, then play your individual role in said plan.

In the first problem case: If a majority of the electorate is willing and able to cooperate with you, then correctly following the above procedure entails (i) that they will recognize themselves as forming a majority, and (ii) will follow the best plan available on that basis, viz. electing Good.  Your group of cooperators will only vote for Lesser Evil on the condition that they are not in a position to collectively elect Good (who is, by stipulation, better).  So: no problem.

The second case is a bit trickier.  I take it that we are to imagine difference sub-groups, each of which attempts to cooperate in voting tactically to best achieve their (differing) goals.  Perhaps 40% unconditionally support Greater Evil, and 30% each support Good and Lesser Evil, whilst still preferring the other over Greater Evil.  How should the G and LE groups vote (given their values)?

Since they have different goals, they do not count as fully mutual cooperators.  All we can do is consider each group separately.  In each case, what the members of a group objectively ought to do will depend upon what the other group does.  But that's not a problem, since there is after all a fact of the matter as to what the various individuals will end up doing, and so the above tactical decision procedure will (separately) determine what they all should have done (given their values).  It can be summarized as saying that each group should vote for whichever candidate the other group votes for, whereas determining which candidate that is is a matter outside of moral theory.  (It may instead be determined by whichever group is better able to position their candidate as the 'default' alternative to the Greater Evil.)

While it would of course be unfortunate if it turned out that the Lesser Evil folks were able to outmaneuver the Good folks here, that would just be a sad fact about the world and not any sort of indictment of tactical voting per se, or so it seems to me.

Is there some residual problem that I'm missing here?

(The practical problem -- how to act given incomplete information, etc. -- is a different one from the theoretical problem I focus on here.  It's obviously highly non-trivial to actually identify and co-ordinate with like-minded others.  But a rough-enough approximation might be achieved by answering opinion polls with your truly most-favoured candidate, and only in the final instance voting for the most popular lesser-evil candidate.  That way if a sufficient number of tacticians really support Good, then this will show up in the polls and they can vote accordingly.)


  1. Hi Richard,

    How about the following issue?
    Let's say A is good, C is a big evil, and B is a lesser evil.
    After looking at the polls, the group following your procedure realize that it's almost certain that by voting for B, they will not be able to prevent C from winning. However, by voting by B, they send a message to other people - namely, that they support B -, which is worse than a message that they support A. Of course, there is also the issue of whether someone will read "we can almost defeat C", if they get close, etc., but that has to be weighed against the issue I just mentioned, also considering that the combined votes for A and B also may well also send a message that an alliance (which would be even less evil) might get close. So, there are multiple issues.

    In general, it seems to me that in order to assess the instrumental value of a vote, the voter needs to factor in not only the likelihood that a candidate will win and how good or bad that winning is, but also the positive or negative result of sending a message (i.e., in terms of how the votes will be interpreted by others) of support for one or another candidate (in addition to factors such as personal cost), and implicitly perhaps, (some, many of) their policies.
    Granted, voters can spread the word that they're voting tactically and that they don't really support B, but that can be costly, especially if they need to reach lots of people who generally don't get involved enough to take a look at that sort of detail, but look at the final tally.

    Another issue when it comes to spreading the word is the cost of making one's vote for any of the leading candidates (or maybe any candidate) public.
    For example, in a highly polarized election, that may well have high personal costs for a person, in terms of retaliation (social isolation from several people including family members, insults, difficulties in one's job or loss of opportunities, etc.). Lying and saying (for instance) one didn't vote or cast a blank vote might make such retaliation much less probable, but then, that would keep the tactical voter from spreading the word that they voted tactically (increasing the negative value of their support for a bad candidate), and additionally, it may well have a significant psychological cost anyway: lying to people one is close to on issues one knows they care deeply about and would outrage them if one told them the truth, knowing that one is risking being despised by several people at work if someone finds out, etc., also may be quite distressing (depending on the person's psychology, of course).

    1. Hi Angra, sure, I guess it could (in some circumstances) be most tactical to vote for A even when A is the assured loser. It's an interesting question how much "sending a message" with one's vote matters -- I'd assume that the answer is generally "not much", so it's going to be much more difficult to vindicate the expected value of voting on individualistic grounds without the chance of the massive impact of actually swinging the election result.

    2. I agree it won't matter much, but the very, very tiny, positive impact of sending a message with one's vote for a good candidate (or against bad candidates) seems very likely to happen, whereas the huge impact of changing the outcome is almost certain not to happen in many elections.

      Here's another case: the two leading candidates (i.e., the only ones with realistic chances) are bad, but it's very difficult to assess which one is worse. For example, depending on a number of other factors (such as the support they'll get in the legislature, how many judges they'll get to appoint, etc.), the relative impact of each can change drastically, so it can go either way. In those cases, I'm not sure how your procedure would work. Do you think it's still better to vote for one of them (the one that might look very slightly better, if there is one at all) than voting for a good candidate?

    3. Perhaps not; it would depend on the details. Most likely it would not be worth voting at all in such a scenario. (A better way to send a message is to write to one's local representative, after all...)

    4. Yeah, maybe it's better, though I'm not sure what counts as a "local representative" in many places and with respect to most elections (e.g., over here). But leaving that and other issues aside, even if one writes that letter, mail, etc., there is still the issue of what to do when a general election comes.

      I tend to agree about not voting if voting is not compulsory. If it is, there's a question of whether a blank vote or a vote for a third candidate is better.


Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)

Note: only a member of this blog may post a comment.