# "New" method: weighted ballots tallied pairwise

Steve Eppley seppley at alumni.caltech.edu
Mon Apr 1 23:55:46 PST 1996

```Here's another method for us to consider.  (If this is one of
Demorep1's, then he deserves credit for its authorship.  I don't
recall the details of all his proposals, though.)

Weighted Ballots tallied via Condorcet
--------------------------------------
Each voter weighs the candidates on a scale such as 0 to 100.
The higher the weight, the more preferred the candidate.  Any
candidate left unweighted is treated as weighted 0.

The tallying algorithm *ignores* the absolute weights and just
gleans the ranking information.  The rest is, of course, a
Condorcet method: plain, Smith, NOTB, Smith+NOTB...

One advantage of this over ranked ballots is that the voters can
express weights publicly, even though these don't determine the
winner.  Another advantage may be more voter familiarity with 0 to
100 (or 0 to 10) scales than with ranking.

The advantage over tallying algorithms which don't ignore the
weights is that here the voters won't have electoral incentives to
vote anything but their real weights.

Example ballot:
100 Favorite
20 Compromise
0 Evil

This is a vote for F over C, for F over E, and for C over E.  As far
as the tallying algorithm is concerned, that's all it is.  As far as
the public is concerned, there's more expression.

A disadvantage is that there can be situations where the winner is
not the same as the candidate who would win by summing the weights,
and this could lead to complaints:

A    B
100   99  voter1
100   99  voter2
0   99  voter3
---  ---
200  297

A wins this election (2 votes to 1), but B would win by summing.
B complains.  How rare would this scenario be?  (A would win this
using ranked ballots too, but B wouldn't be able to prove there's
a problem.)

--Steve

```