>       Let us suppose that so much bad blood exists between Washington and 
> Lincoln that their supporters refuse to rank the other candidate and that 
> Hitler and Stalin have supporters.  Unless one ranks the hundreds of 
> also-rans so that one can make certain that Hitler and Stalin are ranked 
> last, they will be equally ranked with the olso-rans and their supporters 
> might make them win.

In systems that allow "negative votes", like ranking someone "below all 
others", a typical problem is that people may use this feature to push down the 
major competitors of their favourite. For example if we have three major 
candidates and 100 minor ones that nobody takes seriously, then the supporters 
of candidate C1 might vote C1>all_others>C2>C3. That kind of voting might be 
common, irrespective of whether that makes strategic sense or not. If many 
enough voters do so (maybe even all C1/C2/C3 supporters), then one of the John 
Does may win. For these reasons my default approach is that the voter should be 
required to mark all the candidates that he ranks above C2 and C3. The default 
position of unnamed candidates should normally be "last".

Explicit cutoff candidates / tokens could however be used to mark approval type 
limits. One could for example have a system where the winner of the election 
will replace status quo only if the winner gets majority or some supermajority 
or some smaller agreed percentage of the approvals.

I don't se any major risk of electing Stalin or Hitler in typical elections. If 
all other candidates are minor ones and not well known, then Stalin or Hitler 
might get elected. In that case the situation may however already be such that 
people actually want to elect Stalin or Hitler. If people don't want them, then 
there will probably be some other candidates that will be widely identified as 
possible / acceptable alternativess to Stalin and Hitler. If we allow "negative 
votes", then some "Stalin" or "Hitler" might sneak in as one of the John Does.
----
Election-Methods mailing list - see http://electorama.com/em for list info

Reply via email to