Return to article

User talk:Bilby

Hi, I don't understand your edit-summary upon removing the table of preferential rounds for the ord. members. Tony1 20:27, 7 December 2012 (EST)

I'm not sure on the best approach. :) The reason was because the count was done with multiple models, but I'm only showing the results of one. That might lead people to assume that the count only holds using that model. But I'm not worried as to the best approach - I'm just as happy to leave it in. I'll do that, and then if people want to raise questions about models, I'll give them the results using those. The difficulty with STV is the sheer number of counting models open to you and variations that are available. - Bilby 20:29, 7 December 2012 (EST)
OK, thanks, Bilby. I must say, I learnt from looking at the table. And in your view, should the board determine the best model, down to that level of detail? Also, what would happen if a tied vote (and arbitrary knock-out choice) did affect the result? Tony1 20:47, 7 December 2012 (EST)
In the end, every model you choose is going to have the possibility of producing results that another model won't. As a result, I figure you keep hands off normally, and just accept the outcome of whatever one you picked. :) Even with the AU Senate I gather that there a coin tosses. If the randomness is produced by the computer only when necessary, you trust the RO (which you have to do, whatever happens), and there is an audit trail, there isn't much more you can do but accept that it came out that way.
In this case I left it to the RO. I explained how the software runs, which has strengths and weaknesses, and explained how to do an export of the votes to feed into his choice of model in OpenSTV if he wanted a different model or wanted to confirm the results. It was all moot, as all the options in OpenSTV, when left to the default settings, came up with the same two winners (bar two or three, which were not applicable to the election). My next update of the software will provide a wider range of options to the committee, so that they have the option of choosing the specific STV counting model in advance, but I guess that means they gain some additional choices, but will probably still leave it to the RO. :) - Bilby 20:55, 7 December 2012 (EST)
If only we had you to upgrade SecurePoll for next year's ArbCom elections. Tony1 23:32, 7 December 2012 (EST)