[EM] Free riding

[Juho]

On Sep 1, 2008, at 0:49 , Kristofer Munsterhjelm wrote:

> Juho wrote:
>
>> This particular example resembles Hylland free riding (it is an  
>> optimized version of it) but there could be also many other  
>> examples, some of which resemble e.g. Woodall free riding. Only in  
>> some special cases this method resembles Hylland. The point is to  
>> evaluate the probabilities and utilities of all candidates and  
>> vote accordingly. If looking for links to Woodall and Hylland,  
>> this approach in a way tries to generalize them under one framework.
>
> If the point is to generalize the various vote management / free- 
> riding methods under one framework, you should explain why they  
> seem different. For instance, to my knowledge, Meek would fix  
> Woodall free riding but do little with Hylland free riding.

Ok, that's a fair request.

If I rank some unwanted candidate Z ahead of my favourite A (Woodall  
free riding) then in Meek's method I'm giving a message that I want Z  
to be elected and kept in game as long as possible, while in methods  
where Z will be eliminated right away I could get away with that and  
get a strategic benefit. Meek's method would make also the new free  
riding strategy work in a different way (probabilities of the  
candidates to become elected would be different).

When using the new fee riding style in Meek's method I would not rank  
the worst candidate (Z) ahead of A but could use some other candidate  
(e.g. D) for that. This is maybe no more classified as Woodall free  
riding (since I want to influence positively on the election of D  
instead of only listing an irrelevant candidate first). Meek's method  
thus fixes Woodall free riding like problems (where some totally  
irrelevant and possibly unwanted candidate is ranked first) also with  
the proposed new free riding case.

In the Hylland free riding case the new free riding approach would  
just find a new location for A on the ballot instead of not ranking  
it at all.

There's a continuum between Woodall and Hylland. If I raise some weak  
low utility candidate at top, then I'm close to Woodall. Then there  
was the case with D above. And if I raise my second favourite at top  
(and drop A reasonably low) then I'm close to Hylland.

Woodall free riding and Hylland free riding can be seen as two  
extreme ends of the new free riding scheme. Meek would in this sense  
fix some problems at the other end (or actually I'd say Meek's method  
is quite natural and neutral while others have a flaw).

The most practical reordering style may not be as radical as in  
Woodall and Hylland but just e.g. swapping two adjacent candidates  
(assuming that Meek's method is used).

The dynamics look pretty clear to me although my explanation above  
was only mediocre (would need more time to find an elegant one).  
Meek's method would make the new free riding system behave better.  
One would not rank bad candidates at top. With Meek's method the new  
free riding style would rather just move (in a rather balanced way)  
the good candidates lower in the ballot that have high probability of  
becoming elected, and would move those candidates that the voter  
_wants_to_promote_ but that need some more help higher in the ballot.

I hope this speculation provided something useful. And I hope I got  
the Meek's method dynamics right.

Juho




	
	
		
___________________________________________________________ 
All new Yahoo! Mail "The new Interface is stunning in its simplicity and ease of use." - PC Magazine 
http://uk.docs.yahoo.com/nowyoucan.html