Borda Count for evaluations

This is a question about evaluations in hiring or sports judging. Normally score is the best choice for voting because people have an incentive to distinguish. In this case there is a pathology where and evaluator will give moderate scores to all candidates to avoid decision responsibility. Because of this, it seems that Borda would be a better choice because it forces the evaluators to give different values to all candidates. This avoids the problem of evaluators trying to avoid responsibility.

Two Questions:

  1. Borda is Monotonic and pretty good overall but is there a problem I am missing?
  2. Is there another better system which would do this?

Note in specific implementation:
First rank gets score of 10, second gets score of 9 and so on. All unranked candidates will get 0 so that evaluators can give multiple people a 0.

Hmmm. Your “Borda+” system is better than normal Borda because it is immune to the DH3 pathology, and partially resists the “teaming” effect because voters opposed to a monster faction can just refuse to rank any of them (and actually, that monster faction would split votes).

(Hey, just noticed a contradiction in the website. The DH3.html site I linked to says that BTR-IRV (IRV but eliminate the majority-opposed of the lowest 2 candidates) is vulnerable to DH3. On the DH3 page is a link to BTR-IRV, and BtrIrv.html says BTR is immune to DH3!!)

I don’t know if it’s enough, but if you have an even number of possible scores, the voter is forced to go for or against a given candidate. At its logical extreme, it’s a scale of (0,1), but an example would be a scale of 0 to 9. There is no midrange value; a very neutral voter must choose to give a 4 (below the midrange) or 5 (above the midrange). So maybe try a scale of 0 to 3 in Score Voting for enough differentiation while forcing judgements? A perfectly neutral person must give a 1 or 2, which shows some bias.

I do not think lowering the gradation will actually help the issue. On a 3 score system they can always just put the middle score so it would get worse. I have told the person I am doing this consulting for that Borda will make the difference between 1 and 2 the same as 3 and 4 ect so they lose some information but they said they care less about that than the issue of evaluators giving everybody a moderate score.

As a side note it is interesting that here is finally a situation where Ranking is better than Scoring. Once in a blue moon I suppose

What about STAR? That gives incentive to differentiate.

The incentive in STAR is driven off of preference in outcome. This is different. We are trying to force evaluators to distinguish. In politics people tend to be more passionate.

Borda doesn’t do well with strategic voters, so if people try to manipulate the evaluations, that could be problematic.

Is there at least a way to combine Score with Borda? Say, a ranking system that permits some % of equal-ranking at the middle ranks. If there is none, then decent options are likely to go down to the bottom because they are only slightly worse.