Where should we keep seats_total for at large constituencies?

[GeoWill]

For elections with a Division, seats_total lives there. For 'at large' constituencies (gla.a, pcc, mayor) this doesn't work. I think it makes sense that seats_contested is only a property of the ballot, but it would be nice to have defaults for seats_total somewhere.
At the moment the default is essentially 1: https://github.com/DemocracyClub/EveryElection/blob/dde5a199e5a3a4de3deb381e168a082a7152d8d0/every_election/apps/elections/utils.py#L230-L231

So the simplest fix would be to special case gla in the same place. Alternatives would be adding it as a field to the Organisation model, or perhaps the ElectionType and ElectionSubType models.

I think my current preference would just be for an extra if statement in the get_seats_contested method.

[chris48s]

Random thoughts on this topic that you didn't ask for:

• For mayor/PCC, they inherently elect one representative: There can only be one mayor, there can only be one commissioner (PCC) so I think it is fine to just assume 1 for those election types and the simplest place to do that is in code. Those two are already fairly heavily special-case in the wizard/builder code.
• Its kind of odd to store a seats_total on the organisation as it is really a function of the Organisation and something else. e.g: Watford council has one mayor, but it also has 26 councillors so Watford Council / seats_total=1 is conceptually wierd. Also the GLA has 25 total members (14 geographic, 11 additional) and although they're elected differently functionally they're all London Assembly members so London Assembly / seats_total=11 would be properly odd.
• Another possible model if you want to store it in the DB instead of code would be to say if there's no division, the number of reps can be defined on the ElectedRole or ElectionType? That might be a slightly nicer way to express "there can only be one mayor of a place"?
• Sym can probably fill in better on this, but I think the whole "elects directly to an organisation" thing for Mayors/PCCs was basically to avoid the IDs being mayor.watford.watford.2018-05-03 etc. I've previously thought that another possible model for this would be that mayor/PCCs have an "invisible" divisionset with one division in it (but that is probably big).
• gla.a is its own unique special snowflake and is special-cased everywhere. Again, I think the only reason it is like that is really to make the ID be gla.a.2021-05-06 not gla.a.a.2021-05-06. Again another possible model would be that it exists as an 'invisible' region-like division (its basically the same as SP/Senedd regions except there is only one of them).

[GeoWill]

I assume

Probably big means a big change?

Given that where to store seats_total isn't a problem when there's a division (this covers almost all cases) and mayor and pcc elections can be easily handled in the code I'm not sure it's worth overhauling our model just for gla.a. What makes it special is that it's an at large constituency that elects more than one representative. Adding an extra line to set seats total on the gla.a ballot object in the builder seems like the simplest way forward.

If that's not satisfactory for some other reason, then I agree putting it on Organisation is weird. I think ElectionType makes more sense than ElectedRole in terms of mental model. For me the attraction of this approach is getting rid of the other places we keep seats_total - but if we did that we'd need to add it to ElectionSubType too, which is more of the same.