[EM] Free riding

[Raph Frank]

On Sun, Aug 31, 2008 at 10:18 PM, Juho <juho4880 at yahoo.co.uk> wrote:
> It may well be the best strategy to rank A below D in the example above if A
> will be elected almost certainly since the voter has an interest to
> guarantee that D will be elected rather than E. D and E might be even
> "despised" although of course the probability of finding this kind of a
> working (insincere) strategy decreases when the utility of D goes down.

I think you should rank A as high as possible but behind at least 1
candidate who you are reasonably sure will be elected.

As long as you rank him lower than the expected round that he will be
elected in, there is little risk to keeping him ranked high.

Also, if he isn't ranked in the first round, then you probably want to
give him you vote early.

> If looking for links to Woodall and
> Hylland, this approach in a way tries to generalize them under one
> framework.

It probably comes down to deciding what the cost/benefits are of
shifting him back 1 rank.

A2>A1>....

is just as effective as dropping A1 as long as A1 is elected in the first round.

Maybe candidates with strong person votes should advise voters to rank
them second.  The only time it matters is when they fail to get
elected in the first round, and in that case, the assumption of being
a certainty of getting elected has failed.

Also, some PR-STV methods only look at transferred-in ballots when
working out what ballots to transfer-out.

E.g.

If a candidate gets

0.4: person
0.4: party

in the first round,

and then gets 0.5 quotas of additional votes in the 2nd round.

The 0.4 personal votes (and the 0.4 party votes from the first round)
are ignored when working out the transfers as only the 0.5 quotas of
transferred-in ballots are considered.