Thursday 21 April 2011

AV - Hard Cheese

I’m a publican in Leicester with an interest in voting systems and broadly in favour of the Alternative Vote  system (AV). In planning a forthcoming cider & cheese festival, I was thinking about asking customers to vote for a Cheese of the Year award and whether it would be worthwhile. With the AV referendum coming up I mused if such a vote should be on a First Past The Post (FPTP) or an AV System. Some interesting points emerged.

For an award vote FPTP seemed like the obvious choice: everyone gets to vote for any one of the cheeses in the Festival and the cheese with the most votes wins the rosette.

If only three cheeses attracted votes and those were:
Simon Weaver Cotswold Brie   62
Stinking Bishop                         65
Keens Cheddar                         81
then on FPTP Keens cheddar would become champion.

This seems such an obviously appropriate system that it seems hardly worth looking at the complication of AV but let’s do the geek thing and check it out. Under AV with three candidates there are six voting possibilities if everyone fully details preferences. Let’s say it went like this:
             
1st Choice
2nd Choice
Votes
Cotswold Brie
Stinking Bishop
52
Cotswold Brie
Keens Cheddar
10
Stinking Bishop
Cotswold Brie
45
Stinking Bishop
Keens Cheddar
20
Keens Cheddar
Cotswold Brie
10
Keens Cheddar
Stinking Bishop
71

On this basis Keens Cheddar has 81 first preference votes, Stinking Bishop has 65 and Cotswold Brie has 62. Cotswold Brie gets eliminated and 2nd preference votes from Cotswold Brie get distributed. 52 votes get added to Stinking Bishop’s total and 10 votes are added to Keens Cheddar’s total. Stinking Bishop then wins with a total of 117 beating Keens Cheddar with a total of 91.

So which is the best system? Is it just the one that makes your favourite cheese the winner? Surely we can do better than that and find objective grounds for selecting a system.

I think it depends on the point of the vote. If the vote is to establish which cheese is more people’s favourite than any other cheese then FPTP is the best system. For a Cheese of the Festival award it seems completely appropriate. AV fails because second preferences are given equal weight, undermining the point of the award – the ranking system seems to stink here. If AV is to be a better system for elections there would have to be something different about the Cheese Award and electoral voting.

The difference is there and it is instructive, if subtle. If our example changes so the the point of the vote is to determine not which cheese to honour but which single cheese to choose for everyone in the electorate to eat then AV now seems to be preferable. The reason for this is that the choice has an effect on the voters as well as on the candidate. Because in this case each voter is equally affected by the outcome (they only get to eat the winner) it seems reasonable that transferred preferences are given equal weight.

Given the preferences in our example, the AV result of Stinking Bishop as a forced food choice will surely be a more rational choice for the group than Keens cheddar. The ranked votes in the example show that people preferring one soft cheese, Stinking Bishop or Cotswold Brie tend to favour the other soft cheese as second preference. The soft cheese voters are split but are in the majority. The vote shows that had either of the two soft cheeses stood alone against the cheddar then they would have won under either system.

If the question put is “which one cheese shall our group eat?” and FPTP is adopted then those people feeling most strongly about eating soft cheese will probably adopt tactics to try to eliminate the split vote and prevail. Discrediting their less favoured soft cheese, trying to get that cheese to withdraw from the contest become options. Some less partisan Softies may vote for their less favoured soft cheese in the hope that it keeps the cheddar out and so on. It all gets very messy.

FPTP is the better system when the point of the vote is simply to find which candidate is the most popular and there is no significant effect on the voters. AV seems to be preferable when the selection has a significant effect on the voters and we want their preferences for and against candidates to be properly considered. Put another way, if you focus on which system is fairest to the candidates then FPTP may seem justified but once you consider the question of which system is fairest to the voters then AV delivers.

So will it be hard cheese for the voters on May 5th?

Grant Cook – Swan & Rushes, Leicester

The Swan & Rushes Cider & Cheese Festival runs from 29th April to May 2nd.