[EM] Threat/Promise Approval strategy. Summary of strategies.

[Russ Paielli]

Paul Kislanko kislanko-at-airmail.net |EMlist| wrote:
> At the risk of showing my ignorance again,
> 
>>Choice of Approval strategies depends largely on what you prefer to 
>>estimate.
> 
> 
> Somehow I think there should be a way to label posts that are about
> "strategies that voters should employ for a given election method" as
> opposed to "this election method can be exploited by these strategies".
> 
> The same language is used in both cases, so it can be confusing when is
> concerned about the latter but finds the former irrelevant.

You make a good point. The fundamental problem, I think, is semantic: 
the word "strategy" is misleading with regard to voting in an Approval 
election.

In the context of voting systems, the word "strategy" normally implies 
insincerity, but there is really no "insincerity" involved in voting in 
an Approval election. Voters never have any incentive to reverse their 
preference order. Once a voter decides on a sincere perference order, 
the only decision left is where to "draw the line." But that is not 
strategy in the sense of insincere manipulation; it is really just 
optimization. Deciding where to "draw the line" is a fairly 
straightforward optimization problem. One can talk about "optimization 
strategy," of course, but in this context the word "strategy" carries 
unnecessary baggage that only confuses the issue.

Note that I worded my sentences above carefully (or at least I tried to 
do so). I wrote that "there is really no insincerity involved in voting 
in an Approval election." That does not mean, however, that insincere 
strategy cannot play a role in an Approval election. As I and others 
have pointed out in recent posts, any insincere "strategy" in Approval 
has to do with how potential voters respond to pre-election polls. That 
is where any potential "manipulation" can be done.

The issue of vote optimization in Approval has been analyzed quite 
thoroughly, but it doesn't address the question of how an Approval 
election can be manipulated. As far as I know, however, the issue of 
pre-election poll-response strategy has barely been scratched. As I 
wrote before, I am only pointing out a potential pitfall. If it turns 
out that the best poll-response strategy is simply to be honest, then 
Approval is in great shape. Until we know that, however, the 
"effectiveness" of Approval is unknown.

By the way, I realize that one can always fall back on the 
"zero-information strategy." That is not a panacea, however, because 
then the voter is throwing out all information, which may be too drastic 
in many cases. Obviously the "zero-information" guidelines are 
ineffective when only two parties are dominant, for example. And when a 
third party starts becoming competitive, the point at which only two 
parties are still "dominant" could be difficult to ascertain. And that's 
precisely where things get interesting.

--Russ