Two fairer(?) variations of STV

Steve Eppley seppley at
Thu Feb 20 19:07:41 PST 1997

John Taplin wrote:
> The whole point of the Single-Transferrable-Vote is that each voter
> has one vote of equal value.  This value does not alter depending
> on whether of not his favourites stood or did not stand or if the
> were excluded because few gave them a high preference.  The idea of
> weighting votes harks back to point scoring and we all know that
> such methods can deny election to a candidate with a majority of
> first preferences.

My "unequal fractional transfer" proposal doesn't violate the "one 
person, one vote" principle.  It just splits different voters' vote 
unit differently, depending on how high they ranked the seat-winner.

Traditional fractional (Gregory) STV already splits voters' units, so 
there's nothing new about splitting.  And any candidate who receives 
the quota of first preferences will be elected, so this method isn't 
like those weighted voting methods John suggests it harkens back to.

For an example, suppose voter V1 ranked seat-winner X first. 
Suppose voter V2 ranked X 10th, and V2's 1-9 choices have already 
been eliminated so V2's ballot still has its original full weight.

Both V1's ballot and V2's ballot are part of X's overquota amount.
The question is, how much of each ballot did X consume in obtaining 
the quota?  (The unconsumed parts will be transferred to the next 
uneliminated candidate on each ballot.)  

It doesn't violate "one person, one vote" to say that the ratio of
consumed to unconsumed doesn't have to be identical for each ballot.
And it's not really a completely new concept, since the transfer
fraction in traditional fractional STV isn't identical from
iteration to iteration. 

Still, I'm not sure if this proposal is worth endorsing, since the
incentives for voters to misrepresent their preference orders (e.g.,
rank a likely winner artificially low, in order not to waste much of
one's ballot on a "sure thing") haven't been weighed. 

More information about the Election-Methods mailing list