@AviD The wikipedia article on STV will get you pretty far along the way, as will the conversation we had here right after the vote.... But to get all the way there, expect to spend some quality time with Arrow's theorem, a spreadsheet, and some aspirin....
@AviD I think, given the great answer and the final overhaul of the question, it can be left as it is. Then again, given that I was the one who did the overhaul, I may be biased.
@nealmcb saw the convo, dont really need the math, just wanted the gist of it (and the high level theory). the wiki article did a good job there.
Though I think the biggest coin dropping was grokking the single part of stv - I was under the impression we all had 3 votes, when in actuality we only have one. the 2nd and 3rd choice are only counted if the 1st is either already elected, or no chance of election. Is that correct?
so what happens, if there is not enough consensus on the 2nd choice for an already elected candidate, such that the proportion cannot be kept after removing excess votes?
I think its similar to RFQ (request for quotes), but in my mind at least RFQ is more appropriate to OTS product, with a fixed(-ish) price, as opposed to a services-based proposal....
@AviD Seriously? I got the person and the event, but you're going to make me go google the year? :-|
@AviD Tell me about it. We're doing a SIEM RFP right now. On the plus side, we've already gone through the responses and this week we have our first onsite for the bake-off stage.
@AviD To give a bit more clarification -- For instance, here we would use an RFQs when we have an exact product we want. I.e., "I need 150 9"x14" chaffing dishes with stands for insert product here burners."
And an RFP for "We want a new SIEM. Here is our 25 page questionnaire/requirements. Build a design and convince us how it meets those requirements."
@AviD yes, there's only one vote. I think the key part from wikipedia and what requires the aspirin edition of excel is this:
> There are variations in applying these STV rules, such as in how to transfer surplus votes from winning candidates and whether to transfer votes to already elected candidates. When the number of votes to transfer from a losing candidate is too small to change the ordering of remaining candidates, more than one candidate can be eliminated simultaneously.
Running different versions of STV elected different third place candidates; I tried several and depending on the algorithm I think I managed to elect every candidate...!
and here I thought it was gonna be a much simpler scheme, such as: > 1. Give 3 points to each vote of 1st choice, 2 points for 2nd choice, and 1 point for 3rd choice. > 2. Sum up all vote points. > 3. Elect the 3 candidates with the highest scores.
@AviD it's not simple; that I think is one of the drawbacks of STV and why I personally voted against the alternative vote (or IRV) - a similar system - when the UK balloted it compared to first past the post (most votes wins). I think it's hard for people to understand that it depends on how you allocate the votes when eliminating or electing candidates.
@AviD the problem is deciding, I think, say my vote. Let's say I cast my first vote for Rory and my second for Nealmcb. When Rory is elected (that happened quickly) does my vote go to nealmcb, or does my vote count towards Rory's 33% required to elect him? That will impact nealmb's score etc. So that's why it depends how votes are allocated.
Also, I that may or may not have been my actual voting preference. It was just an example.
there was discussion earlier about those that voted only for 1st and 2nd (no 3rd). that sounded (maybe only to me, cuz I misunderstood it) like they were talking about a 3rd vote.
thank the jeff that you dont need to understand the election, in order to win it :)
@AviD No, I did not. I assumed more as you did, that I had 3 votes, of 3 different weights. Now, had I actually read what STV voting was, I might have known that from the outset.
havent even delved into the what the various types want to calculate differently. hurts my head.
look, I know "secure elections" are complicated enough, without worrying about calculating the results. protecting your privacy, but also allowing you to verify that your vote was counted, without allowing you to prove to anyone else to avoid vote sales... yeah, it's it's own field altogether.
truthfully, in most cases (political elections, anyway), it doesnt matter how uncorruptible the process is anyway, since the result will always be more corruption no matter who wins.
Proposed Q&A site for computer Science practitioners, researchers, and CS students interested in topics like theoretical computer science, artificial intelligence, programming language design, digital logic, or any other area of CS not directly related to programming.
@Gilles I'll add some possible definition questions in the morning. I hope it survives. I've no real thoughts on the two-level split vs one huge site debate, but I think a place that accepts non-research level questions would be excellent - it would also assist TCS as they then have an obvious place to send those (non research level CS) questions.