[EM] Citation for immunity to strategic voting?

[Juho Laatu]

On Sep 3, 2005, at 22:15, Andrew Myers wrote:

> I would like to have a statement
> about strategic immunity that doesn't rely on people judging the 
> difficulty of
> creating a top cycle.

The best I can offer when it comes to freeing people of judging and 
deciding strategies is the following method.

The method consists of two rounds. If the first round produces a 
Condorcet winner, the second round is not needed. Otherwise the second 
round will be held and also the tie breaking method is used if there is 
a top cycle. (Clearly non-winning candidates could be excluded from the 
second round but I won't discuss those rules further here.)

In this method those (sincere) voters that do not want to worry about 
strategies can forget them in the first round. If someone else created 
an artificial cycle or if there was a sincere cycle, those sincere 
voters may come back and consider strategies in the second round. Clear 
use of strategies in the first round could lead to revenge at the 
second round. This fact may to some extent reduce the interest to use 
strategies in the first round.

Having two rounds may of course be expensive and troublesome but at 
least this method seems to get quite well rid of the small strategic 
problems of the Condorcet methods - well, for the first round at least. 
Maybe the second round is also in some respects fair since the 
strategic options can be seen when the "strategists" have exact 
information on how people voted. This availability of information is 
also negative in the sense that it is easier to develop complex 
strategies when detailed information is available.

BR, Juho


P.S. You mentioned also the possible need to effectively co-ordinated 
the insincere votes. I have been interested in learning what such 
strategies there are that could take place in real life elections based 
either on independent individual decisions (=no effective 
co-ordination) or based on some guidance that has been published e.g. 
by one party, and if these strategies will elect some candidate that 
otherwise would not be even close to victory. Most strategies are quite 
theoretical from this point of view, but I haven't been able to 
eliminate all. Current voting methods have many strategic problems, so 
in theory it is enough for Condorcet not to introduce any 
vulnerabilities that are worse than the existing ones. One way of 
freeing people from considering strategies is of course also taking 
Condorcet methods into use and finding out that they work fine.

P.P.S. Sorry about not providing any clear "immunity statements" as 
requested, but maybe these thoughts are at least of some interest :-)