[EM] serious strategy problem in Condorcet, but not in IRV?

[James Green-Armytage]

Dear Markus, 
Thank you very much for your input. A couple replies / questions...

>> Sincere preferences
>> 30: A>B
>> 25: B>A
>> 23: C>A
>> 22: C>B

>When IRV is being used, then I suggest that the 25 BAC voters should use
>bullet voting so that the 30 ABC voters would have to give their first
>preferences to candidate B to keep candidate C from winning.

Yes, I agree that the BAC voters have this option. It is similar to the
strategy that the B voters use in my first IRV example. Repeating that for
the sake of clarity:

46: A>B
26: B>C
5: C>A
23: C>B

C wins, and the A voters have to choose between voting for B first or
letting C win.
As I said, the example you have used is easier to manipulate, so the
strategy under IRV that you suggest is much more plausible than the one I
suggest in this example directly above. 

I agree that this is a serious problem for IRV as well as for Condorcet,
as both lead to a game of chicken. The IRV strategy, however, is still
slightly less likely than the Condorcet strategy, because the B voters
have to begin the chicken game by taking a loss, whereas in the Condorcet
equivalent, the B voters immediately move to a more desired outcome due to
their strategy. 

Thus, the viability of the strategy under IRV depends on B voters
effectively playing psychological games against the A voters, whereas the
strategy under Condorcet can be accomplished by sheer stealth, and in fact
works better if the A voters don't see it coming.


>I would point to the fact that IRV is vulnerable to "pushing over"
>(i.e. ranking a weak candidate higher than one's preferred candidate)
>while those methods that meet monotonicity aren't.
  
Do you mean something like this: (IRV is the method being used)

Sincere preferences
47: Bush
6: Gore > Bush
21: Gore > Nader
26: Nader > Gore

Gore wins IRV.

Strategically altered preferences (2 Bush voters switch to Nader > Bush)

45: Bush
6: Gore > Bush
21: Gore > Nader
26: Nader > Gore
2: Nader > Bush

Gore is eliminated first, and Bush wins. Is this a good example of pushing
over? Certainly the result is pretty screwy, but I'm not quite sure if the
opportunity to use it will come up as often as the opportunity for the
kind of offensive order reversal / burying in my initial example. What do
you think?


>Similarily, even when the parties in the USA weren't allowed to _instruct_
>their supporters to vote in a certain manner, it wouldn't be possible to
>forbid these parties to _educate_ their supporters about the consequences
>of voting in a certain manner.

Hmm, okay, well, what would you suggest should be used to keep this sort
of strategy in check if Condorcet was ever to be used in serious public
elections where a lot is at stake?




my best,
James