[EM] The "Turkey" problem and limited ranks

[Forest Simmons]

On Fri, 23 May 2003, Chris Benham wrote:
...

> On the subject of  plain Approval elections being manipulated  by  "fake
> information", I think that there is a danger that a majority's  sole
> sincere favourite could lose because some of  his/her supporters were
> conned/frightened  by  bogus  polls published by a partisan media. These
> polls could falsely promote the possibilty that the MF could lose to a
> horror/completely unacceptable candidate
> (when in fact the real situation is that this is not  a real possibilty
> and that the 2 front-runners are the MF  and  some other less horrible
> candidate), and thus frighten some of  MFs sincere first-preference
> supporters into "approving"  Less Horrible with the result that Less
> Horrible  wins with a  "bigger majority".

Here's an example of a slightly different scenario, from yahoo EM archives
message #10396, of manipulation by fake polls, the most perfidious kind of
election manipulation short of vote buying, ballot stuffing, civil rights
violations, etc.


True preferences of voters:

6000 A>>B>C
3000 C>B>>A
1000 B>C>>A

Poll results reported by the corporate media backing the corporate clones
B and C:

35% A>B>C
40% C>B>A
25% B>C>A

Under Approval candidate A could easily lose to B if enough A supporters
are fooled by the fake poll.

See the full text of message 10396 for further analysis showing that
Borda and IRV are also apt to give B the win in this context, while
Condorcet and Candidate Proxy are almost sure to elicit more sincere
votes and give the win to A.

Forest