[EM] Beatpath and SSD aren't manipulable. "Manipulable" is barking up the wrong tree.

MIKE OSSIPOFF nkklrp at hotmail.com
Mon Nov 7 22:15:57 PST 2005


Warren--

You wrote:

I would expect  it [beatpath] is extremely manipulable.

I reply:

Critics of pairwise-count methods speak of how they're vulnerable to two 
offensive strategies:

Truncation and offensive order-reversal.

And, for all Condorcet methods other than Condorcet(wv), they're right. All 
Condorcet methods that don't use winning-votes are a strategic mess, just as 
you suspect. But wv is different. You're ignoring the distinction between 
different kinds of Condorcet.

The criteria SFC and SDSC describe why complying methods don't have a 
problem with those offensive strategies. Of course SFC and SDSC aren' t just 
about offensive truncation and order-reversal. Those strategies pretty much 
don't exist in Plurality and IRV, but they both fail SFC and SDSC. SFC and 
SDSC measure for some desirable properties that can be gottten with rank 
methods and, not surprisingly, nonrank methods don't pass those criteria.

There is one very important strategy problem that even Condorcet(wv) has, 
and that RV doesn't have: Incentive to bury one's favorite. RV passes FBC 
and Condorcet(wv)  (and all Condorcet versions, and every 
Condorcet-Criterion method, and nearly all rank methods) fail FBC.

Because of FBC, you're correct to say that, with our existing voting 
sistuation, and with our existing voters, RV is better than Condorcet(wv) in 
public elections.

But you werent' referring to a public election. You were referring to an 
organizational election, where there's no reason to believe that FBC is 
necessary as it is in public elections.

There, RV loses its biggest advantage over Condorcet(wv).

If Condorcet is going to be proposed for public elections, it should be SSD, 
because of SSD's natural and obvious motivation and justification. Likewise 
for large organizations, though maybe there CSSD would be better. Beatpath 
has the advantage of a very brief and simple algorithm and computer program. 
That advantage could be decisive for organizations. One reason I send the 
Beatpath algorithm out to organizations is because I've been too lazy to 
write the longer CSSD program.

Which is better for organizational elections: Beatpath, Approval, or RV.

It depends partly on what you want. If you want to maximize social utility 
(SU), and there's nonspatial disutility, then RV can be expected to do the 
best job in small organizations (maybe Approval is just as good for SU in 
public elections, where its roughness averages out). Approval is next best 
for SU in committees where there's nonspatial disutillity.

If majority rule is what you're after then Beatpath is better.

Warren, you said that you believe that strategic voting on a large scale is 
llikely, or at least that it may be likely in the election to which you 
referred.

If that's so, then you don't want RV in that election, because with RV, 
strategizers can take advantage of sincere voters. Under those conditions 
Approval does a better job. As I've said, with Approval, in 0-info 
elections, if we disregard small-committee effects (which are difficult to 
strategically allow for anyway), a strategic ballot is identical to a 
sincere ballot. Vote for the above-mean candidates.

Another solution with many strategizers is Beatpath.

The choice between Beatpath and Approval depends on whether you want the 
best majority-rule enforcement, or emphasis on SU. It depends on whether 
you're more interested in  preferences or emphatic preferences. It depends 
on whether you're more interested in counting the direction of pairwise 
preferences, or counting the strength of pairwise preferences.

These are subjective considerations. Beatpath is a good choice for Debian.

We're talking about Debian's organizational election. As I always say, in 
public elections I prefer RV to Approval because it somewhat mitigates the 
damage done by incompetent voters when they are forced to vote their 
incompetence full-strength, all or nothing.

Mike Ossipoff

_________________________________________________________________
FREE pop-up blocking with the new MSN Toolbar – get it now! 
http://toolbar.msn.click-url.com/go/onm00200415ave/direct/01/




More information about the Election-Methods mailing list