Clarke Tax Voting

Mike Ositoff ntk at
Fri Jul 17 19:25:09 PDT 1998

Well, I finally went to the library & looked up the preference-
intensity measuring method that I described, the one in
_Topics in the Theory of Voting_, by Straffin or Strafin.

The description that I'd given here doesn't even have the slightest
resemblence to the actual proposal in the book. It's a good thing
that I emphasized that my description might not be reliable. If
it were a suggestion, I'd withdraw it, but it wasn't a suggestion--
only an attempt to roughly recount the book's method--very roughly :-)

The method, called the "Clarke Tax" is very simply stated. You
pay for that part of your bid that made a difference in the outcome.
In other words, you pay the amount of your vote that was needed to
make the winner win--you pay the amount the winner would have lost
by had you not voted.

Obviously it would be very rare for a Clarke tax to ever be paid,
because it's so unusual for 1 person's vote to change the
election result. But just as Plurality, Approval & related
point system strategies are based on that unlikely occurrance,
that possibility, howerver unlikely, gives pracatical incentive
for sincere voting when the Clarke tax is used.

As I understood the chapter, Clarke described his system for
economic situations rather than voting, and it was Tideman who,
with another author, first suggested it for voting.

The Clarke tax is vulnerable to coalition overbidding, but
they suggest that an upper limit on bids would do much to 
get rid of that possible problem. Also, coalition overbidding can
backfire badly, especially if there's more than 1 coalition trying

They acknowledge that it seems unfeasible to fairly adjust for
different incomes & other economic factors. That's why I'd prefer
the play money version, which would charge payments in a free
currency whose only use is that. Given to voters equally for
a particular election in which a number of different issues &
races will be voted, or given to them for a 1 year or 1 decade
period of elections, etc.

I don't suppose that the play money version would elicit sincere
votes, but it would give minorities an opportunity to get something
that they want very much, by sacrificing their already low ability
to influence other issues or other elections. And since one's
currency would only be used up for a winning candidate, there'd
be no wasted votes--and as I said, currency would rarely be

I don't rule out something like that for the future. I obviously
haven't studied the strategy & merit of Clarke tax with play
money, but though unlike the Clarke tax with real money, it
doesn't seem to elicit sincere, strategy-free votes, it may
have merit in its own way. Somethingt to study the strategyk
of, and maybe consider someday. But my proposals are
Condorcet(EM), Smith//Condorcet(EM), Schulze, & any rank-balloting
method with the candidate-withdrawal option (Plurality of course
the simplest for that).


Incidentally, let me mention that Condorcet(EM) could also 
benefit from the candidate withdrawal option. It would be needed
only when order-reversal is successfully used. It would be
able to thwart that offensive strategy every time. In the
familiar A,B,C example where B is Condorcet winner, but A voters
elect A by order-reversal, then C could withdraw, ensuring that
Condorcet winner B wins. No one would ever have to do other than
vote a completely sincere ranking, or shorten a ranking.

Of course if some voters like me truncated on principle, and
the supporters of the person we didn't vote for tried order-reversal,
it might backfire as I've described, and when it elects someone the
reversers like less than the Condorcet winner, the person elected
would have no reason to withdraw. That would be a fitting penalty
for the reversers--but since order-reversal couldn't possibly gain
anything, it wouldn't occur, and therefore, the withdrawal
option would actually never have to be used. Because we'd have
it we'd never need to use it.

Mike Ossipoff

More information about the Election-Methods mailing list