I agree that STV can be used for a district of any size. The main
tradeoff in district magnitude seems to be between proportionality and
connection with local / regional areas.
That is, the higher the district magnitude, the greater the
proportionality, but the less connection between representatives and a
particular geographic region.
I see a couple other factors to consider in the trade-off:
- A large list of candidates makes it hard for the voter to study them all, thereby forcing the voter into a more party-based decision process. Similarly, larger numbers of winners make it harder for the voter to keep abreast of them all, and weaken the accountability of representatives to their constituents (a similar argument to your local connection one).
- Some would argue that a viewpoint should have to achieve some critical mass (say, 10% of the voters) before it should gain representation. This reduces proportionality, but it helps prevent splintering and radicalization.
For these reasons, I feel like 7 or 8 is the ideal number of representatives for a district. There could be exceptions - New York City could be one big 13-seat district, for example - but 7 or 8 seems like the right balance between proportionality, connection to the voters, ease of voting, and protection against splintering.
Personally, I like to see PR with a district magnitude of at least 10 seats, and I would tend to prefer more than that rather than less.
I'd imagine you prefer larger districts because you see fewer factors motivating smaller districts than I do.
I agree that a paper ballot for a large-magnitude district might be kind
of expensive. For example, if you had 50 seats, you might have upwards of
500 candidates. I've always imagined those kind of things on some sort of
computer interface (with paper printouts etc., of course). That is, a
user-friendly interface with various panels and buttons and whistles, one
where you can search through the candidates by first name, last name,
party, rank them as you go along, change your rankings, add write-ins or
search from some kind of secondary list of candidates, and so on.
I agree that getting good implementation/results for large-district STV or PAV-based PR almost requires a user-friendly touch-screen interface. That said, I don't think even the most whiz-bang user interface could enable the typical California voter to make informed choices about what 53 candidates to send to congress.
-Adam
---- Election-methods mailing list - see http://electorama.com/em for list info
