Talk:Wikimedia Foundation elections 2015/Vote Questions

The election ended 31 May 2015. No more votes will be accepted. The results were announced on 5 June 2015. Please consider submitting any feedback regarding the 2015 election on the election's post mortem page.

Coretheapple

Getting that security token mismatch. Please help. Thank you. Coretheapple-vote (talk) 17:01, 30 May 2015 (UTC)

Please post a request from your primary account so we can verify ownership and such. Thank you! --Varnent (talk)^(COI) 17:02, 30 May 2015 (UTC)

Sure. Coretheapple-vote (talk) 17:09, 30 May 2015 (UTC)

I am talking about your none "vote" account. The primary one that you originally tried to vote using. --Varnent (talk)^(COI) 17:09, 30 May 2015 (UTC)

Here I am. Thanks! Coretheapple (talk) 17:10, 30 May 2015 (UTC)

Thank you, we'll post a message here once the vote account has been added to the voter list. --Varnent (talk)^(COI) 17:11, 30 May 2015 (UTC)

Your vote account has been added to the voter list. Please try voting with that account and let us know how it goes. Thank you. --Varnent (talk)^(COI) 18:23, 30 May 2015 (UTC)

I will. Thanks! Coretheapple (talk) 22:59, 30 May 2015 (UTC)

It worked fine. Thanks. Coretheapple-vote (talk) 23:19, 30 May 2015 (UTC)

Great! Thank you for letting us know. --Varnent (talk)^(COI) 23:37, 30 May 2015 (UTC)

William Ellison

Getting that security token mismatch. Please help. Thank you.William Ellison (talk) 02:23, 31 May 2015 (UTC)

@William Ellison: There are instructions on the main page, could you please create a new account as described and let us know? We will add it to the voter list so that you can vote with that account since this account is having issues. Jalexander--WMF 02:51, 31 May 2015 (UTC)

Ok William Ellison-vote (talk) 04:39, 31 May 2015 (UTC)

@William Ellison-vote:@William Ellison: Thank you, the new account has been added to the vote list and you should be able to vote now from that account. Please let us know if you run into any further issues. I apologize for the frustration and I'm glad we were able to get you a chance to vote regardless. For the record, accounts were verified by confirmed emails Jalexander--WMF 05:00, 31 May 2015 (UTC)

Thanks, it worked and I voted.William Ellison-vote (talk) 13:49, 31 May 2015 (UTC)

Volodymyr D-k

Getting that security token mismatch. Please help. Thank you. --Volodymyr D-k-vote (talk) 05:09, 31 May 2015 (UTC)

@Volodymyr D-k-vote: @Volodymyr D-k: Can you please post a message from your primary account to verify identity? Thank you. --Varnent (talk)^(COI) 05:11, 31 May 2015 (UTC)

Of course. --Volodymyr D-k (talk) 05:15, 31 May 2015 (UTC)

@Volodymyr D-k-vote: @Volodymyr D-k: Your vote account has been added to the voter list. You should now be able to vote using that account, please let us know if you have any problems or are successful. Thank you. --Varnent (talk)^(COI) 05:29, 31 May 2015 (UTC)

Espresso Addict

Thanks for this workround; I've tried to vote repeatedly over several days, including logging in & logging out, but always get the error. I should be qualified on en-wiki. I have created Espresso Addict-vote (talk · contribs) as per instructions. Espresso Addict (talk) 05:45, 31 May 2015 (UTC)

@Espresso Addict:@Espresso Addict-vote: Thanks, I've added your name to the voter list and you should be able to vote with that account now. Please let us know if you run into any issues (so far everyone seems to have been able to vote, but since we haven't been able to figure out the root cause yet... ;) ). Jalexander--WMF 06:00, 31 May 2015 (UTC)

Thanks, it worked ok! I hope everyone affected manages to use this fix in time. Espresso Addict (talk) 06:29, 31 May 2015 (UTC)

Vote workaround for User:Crosbiesmith-vote

Hello, I have created a vote workaround account User:Crosbiesmith-vote due to a 'Security token mismatch, cannot log in' error. Can this account be added to the voter list please? - Crosbiesmith (talk) 09:05, 31 May 2015 (UTC)

@Crosbiesmith: @Crosbiesmith-vote: Thanks, I've added your name to the voter list and you should be able to vote with that account now. Please let us know if you run into any issues. Apologies for the inconvenience. --Varnent (talk)^(COI) 09:12, 31 May 2015 (UTC)

Thank-you - that worked. - Crosbiesmith (talk) 09:25, 31 May 2015 (UTC)

SucreRouge

Hi, I am also getting the "security token mismatch" message and I have created another account, SucreRouge-vote. Could you please grant me access to voting with that one? Also, immediately after voting, will deleting the "vote" account have any consequences on the vote? Thanks in advance! --SucreRouge (talk) 12:46, 31 May 2015 (UTC)

Bump. Can't someone please grant me access? SucreRouge-vote (talk) 15:15, 31 May 2015 (UTC)

@SucreRouge:@SucreRouge-vote:, thanks and apologies for the delay ( I think everyone was asleep). I've added you to the voter list and you should be able to vote from the new account. Please let us know if you run into any further issues and I apologize for the disruption caused. Jalexander--WMF 17:51, 31 May 2015 (UTC)

Thanks, that solved it. No worries about the delay! :) SucreRouge (talk) 20:11, 31 May 2015 (UTC)

How are the winners determined: "percentage of support" formula is suspect

This doesn't seem to be a valid way to tally votes. Please help me understand, and if there is any precedent for the Support/Support+Oppose formula, I'm curious to read more.

Here's an example in which one Oppose voter has the same weight as several thousand Support votes, illustrating how the current formula falls short: If Candidate A had been supported by 10% of the voters and no one had opposed him, then he would had a "100%" score according to this formula. By contrast, Candidate B with 75% of voters in support but one single oppose vote would have a lower score, "99.98%", and would have lost the election if there were only one seat available. In that case, 10% of voters would get their way, and 75% would be disserved.

This did in fact happen, and the Board appointed Vrandečić over Sefidari and Ayers, although they both received more supporting votes than him.

The justification for the odd math that was used is a bit tricky, but I'm attempting to follow along here. Please correct any errors you see. Under the "percentage of support" theory, Oppose votes counteract Support votes, and Neutral or N/A votes do not have any effect on the candidate's ranking. These are good axioms, and I use them in my alternative formula, presented below.

The flaw in "percentage of support" is that the denominator changes, consider Support / (Support + Oppose). If only two voters bother to support or oppose a marginal candidate, then the denominator is 2 and these two voters can give the candidate a "percentage of support" of 0/2, 1/2, or 2/2. Each voter can move the score by 50%. If a candidate receives 3,000 support and oppose votes, then any changed vote will be over a denominator of 3,000, and will only have an impact of 0.03% on the score. In other words, it isn't appropriate to compare this "percentage of support" across candidates unless they have the same denominator.

There are two more axioms we need,

• Each vote must have equal weight. [1]
• We must pick a constant factor for weighing Oppose votes against Support votes.

Alternate tally: Every vote counts equally, and an Oppose vote cancels out one Support vote. Calculating the rank is simple, just take Support - Oppose as a candidate's score.

See the stats page for a table that includes this extra column, and see my not-yet-wikified work here for an analysis of how a single changed vote is weighted under each formula.

Adamw (talk) 07:54, 4 January 2016 (UTC)

We discussed this a lot both before and during the election in the Election Committee, and while we thought it would make no real differnce (as had been the case in earlier election) it actually made a big diffrence in this election. I think your expamples are very intersting. I also note that you use T and not S+O. This means in practice that a netural vote is a vote against the person, which could also be discussed. If you use S+O insead on T I beleive you will end up with the same result as was now published.Anders Wennersten (talk) 13:15, 3 January 2016 (UTC)

Thank you for considering! I just did an extra analysis which shows exactly how much one changed vote will affect the score used to rank winners, and looked at three cases for both methods of scoring: One oppose vote becomes a support vote. One oppose vote becomes a neutral vote. One neutral vote becomes a support vote. The tallying methods I compared were S/(S+O) and (S/T - O/T). I used a simple formula to determine the change in score, by measuring the step change when one vote was switched, which in each case looked similar to: ((S+1)/((S+1)+(O-1)) - S/(S+O)), for example.

For the status quo method, I found that a single O->S change for Peteforsyth would have made a 0.18% contribution to his percentage support, but a single 0->S change for Sefidari would make a 0.03% contribution. This doesn't seem like a good property. For my proposed 1:1 weighting, the changes in score are always consistent. An O->S change will make a 0.04 impact, and either O->N or N->S will make a 0.02 impact.

Here's my work, still needs to be wikified.

This does mean that a neutral vote will count as better than an oppose vote, but less than a support vote. That seems correct, though? Ah--and you can see that we already have this effect with the current tallying formula, a neutral vote is better than oppose but less than support, because you're changing S and O, but the impact per vote fluctuates in strange ways. Adamw (talk) 19:20, 3 January 2016 (UTC)

In fact the system you suggest should allow four options (Support, Oppose, Neutral and no vote at all), as Neutral votes also have an impact on the ranking. As in current system Neutral votes are default and users who want to have no impact on the result of a particular candidate (abstention) vote Neutral, we cannot take them into account to determine the final result — NickK (talk) 19:47, 3 January 2016 (UTC)

I see neutral as the same as no vote. More importantly, neutral votes already do have an impact on the ranking, please see the "change if" columns for O->N and N->S. Changing a vote to or from neutral will affect both the S/(S+O) score and the S-O score, so I don't understand the reasoning behind "we cannot take them into account". Adamw (talk) 19:59, 3 January 2016 (UTC)

Neutral is not the same as not voting at all. By voting Neutral a person does influence the result in your system: by adding a neutral vote I decrease (S-O)/T ratio, while if I do not want to cast any vote I should not influence the result at all. Thus by voting Support I significantly increase candidate's ranking, by voting Oppose I significantly decrease it, by voting Neutral I slightly decrease it, and I have no option to leave the ranking unchanged if I want to do it — NickK (talk) 01:02, 4 January 2016 (UTC)

Your additional columns are functionally equivalent to the "Support" and "Oppose" columns, the rank is the same; mainly, they illustrate the arbitrariness of the support/oppose/neutral thingy (I won't call it an electoral method because there is no such method, as far as I know). Choosing the most supported candidates would produce 2 replacements and choosing the least opposed ones would produce 2 replacements. In total we have 6 legitimate possible winners from the information collected and we choose 3 with an unclear rationale.

Not only we ignore the opinion on the specific candidate of all the neutral voters, but we may completely ignore a majority of voters (i.e. up to the 3000–3400 out of 5000) who did not support any of the three winners. Or in other words, a sufficiently organised minority could produce an apparent relative majority and grab all the seats. We currently have no information on whether this happened.

I encourage the responsible persons to release the complete dump of the ballots, as used to be practice, so that people can make their alternative tallies and estimate how much the tally has influenced the result. All supported candidates have the same score, all neutral the same, all opposed the same. Nemo 19:49, 3 January 2016 (UTC)

@Nemo bis: For the record I'm happy to publish the dump, the next week is a bit insane but I'll try to get it out in the next couple days if possible. Jalexander--WMF 08:12, 4 January 2016 (UTC)

@Nemo bis: Thank you for getting rid of that denominator! The math is so much simpler now: S-O. Adamw (talk) 08:24, 4 January 2016 (UTC)

Vote dump

Just as a note, I've published the vote dump for the Board election on meta. It is also linked from the sidebar. I will try to get the dumps for the FDC elections up as soon as possible.Jalexander--WMF 23:09, 13 January 2016 (UTC)