What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 127 Stack Exchange communities.

I'd like to contribute to the reform of the election primary and this paper sets out my thoughts.

I see three opportunities for improvement to the existing election primary which is basically to end the current practices of:

  • Negative Voting
  • Voting rights for Members with low level of activity
  • Publishing the Voting in Real Time

Negative Voting

Some candidates were practically lynched in the election primary. Although in all elections regardless of their nature there are winners and losers, the effect was exacerbated by the possibility of voting negatively in the election primary.

Most candidates are aware that to present themselves as candidates in an election will spark negative opinions about them. In this case, the negative vote had a greater symbolic impact, because the fact of having only 5 votes compared to 100 for another candidate is really different than having -95 compared with 100. In both cases, the same voters have expressed their opinion, but the current voting system reinforces an unnecessary sense of negativity in the election primary. Voting against someone is not the same as just not voting for them, although ultimately the same effect is achieved in most cases.

If we analyze the results from the primaries we can see that removing the negative voting would not have affected the ranking so much except for the 2 last candidates.

| Candidate              | Up & Down | Up Only    | Down Only |
| SLak                   | 710       | 810        | 100       |
| Robert Harvey          | 641       | 680        |  39       |
| Lasse V Karlsen        | 528       | 570        |  42       |
| Michael Mrozek         | 492       | 533        |  41       |
| Justin "jinguy" Nelson | 480       | 525        |  45       |
| Tim Post               | 476       | 512        |  36       |
| Kev                    | 394       | 449        |  55       |
| Dave DeLong            | 300       | 390        |  90       |
| Paul Dixon             | 274 *     | 311        |  37       |
| George Stocker         | 259 *     | 321        |  62       |

Voting Rights Members with Low Level of Activity

The current rule allows any user with 150 Reputation points to vote for a moderator. While this can prevent any attempt to influence elections by opening fake accounts, there is still a problem.

With the current system, a user of a community can get 100 reputation points simply by linking his account. In addition the system of question migration allows someone from community A to obtain reputation in the community B without even taking part explicitly. The remaining 50 points are not complicated to get in a very short period.

Where it is problematic is when a candidate is very popular in community A, presents themselves to act as a moderator in community B. A large number of users that are not active or only very slightly active in a community will be able to influence the vote!

I have proposed the creation of a new badge called Citizen . The definition of the badge is as follows:

Citizen — Has posted at least a question, answer or comment with at least one upvote 20 times a month for the last 6 months.

In order to get votes from real participants in the community, I recommend that we limit voting to the owners of the Citizen badge.

Published voting in real time during the election primary

In my humble opinion, this is the most serious problem. A public vote in real time is problematic because it promotes two types of negative behaviour that militates against good results in the election primary.

  • The Herd Instinct
  • Strategic Late Voting

You will see that it is for this reason that the publishing voting in real time is virtually unused anywhere.

The only case I know personally where it is used is during TV games where television viewers have the opportunity to vote for candidates who appear on screen. The show producers decide from time to time to publicize the intermediate results -if they don’t fit the editorial line set. That why some television shows publish these interim results, and sometimes not.

The Herd Instinct

The first objection is related to the herd instinct. A nice humorous illustration of this effect may be viewed in this very old hidden camera video.

People behave like sheep: the judgement of the mass is influenced by the votes already in place. Consciously or unconsciously, voters identify with the candidates and their current score. Many people choose to vote where there is the least risk, that is to say vote for the most popular. Other types of personalities identify with rather less popular candidates, for pity or out of a contrary spirit.

You can see this type of behaviour over many years simply by observing the answers to questions in the community. In absolute terms, in organizing elections we are interested in finding out the opinions of unbiased individuals and this is not achieved where voting is influenced by how the election is proceeding.

Strategic Late Voting

The late voters are rewarded by their late participation; they have more information than others and can act accordingly. Early voters have not seen the real impact of their vote.

Late voters can influence by voting strategically regardless of their view of one candidate’s competence compared with another’s. For example, they may think that their preferred candidate has no chance of winning and use their vote for a different candidate – this type of voting becomes self fulfilling prophecy. This effect is heightened by the possibility of a negative vote in the election primary: a voter might be tempted to put his vote against all candidates too close to his favourite.

This type of strategy has nothing to do with expressing a preference.


  1. The negative vote has no useful function in the election primary, only the negative perception attached to it remains. I therefore propose that this feature is removed from the election primary before the next election.

  2. I propose that we limit the voting to true members of the community.

  3. It appears obvious that the only solution to eliminate the effect of herd instinct and the advantage of late voting in the election primary is not to announce any results until the counting has closed in the election primary.

These adjustments will allow the community to get candidates elected on a solid foundation, as in any system whose base is democratic.

share|improve this question
#1: Agree with this. #2: I think the Citizen badge is a bad idea, and I think I discussed why on its page. #3: I agree... but this only applied during the primary, as the final vote didn't show you what the votes currently were. –  Powerlord Feb 1 '11 at 20:51
I think that Paul's and George's upvote + downvote and just upvote stats are swapped in your table. How could they have fewer total votes if downvotes are removed? –  Brad Larson Feb 1 '11 at 21:42
How do the last two people in the table have higher up/down values than up only values? Did you flip the columns? –  Michael Mrozek Feb 1 '11 at 21:44
@Brad+Michael: thanks I fixed that. –  user150926 Feb 1 '11 at 21:51
Amen. Except for #2. 20 times a month for 6 months? So a month on vacation loses your vote? or holidays for students? IMHO, it would be hard to come up with a numeric participation quota that most people would agree on. But raising the rep bar for voting? maybe. #1 and #3 - definitely. –  RD1 Feb 1 '11 at 22:10
I've mentioned elsewhere that I regarded the primary votes as an ongoing opinion poll on the candidates. In such as case the public nature of the votes and down-votes were acceptable. –  ChrisF Feb 1 '11 at 22:14
I have some serious problems with this post -- mainly, you make no distinction between election and primary when the primary voting is intentionally quite different than the final election. –  Jeff Atwood Feb 1 '11 at 22:47
"reinforces an unnecessary sense of negativity" - why, because you think it's unnecessarily negative to consider someone a poor choice as a moderator? Or simply because you dislike negative numbers...? FWIW, by far the most negative thing I saw in the SO elections was a candidate being attacked by someone who thought that mentioning where he lived amounted to a declaration of cultural superiority. This happened well before any voting started. –  Shog9 Feb 1 '11 at 23:20
As I recall (too late to test the understanding now) I could've changed my votes in the primaries up or down at any time. There's no lockin until the election proper starts. So I could cast a vote early, then come back at the end and reverse my vote. So what's the problem? Gaming the system? That always happens in elections, if people are concerted in their efforts. –  jcolebrand Feb 1 '11 at 23:38
Also, now that my edits are public (being under 3k sucks ;) ) ~ You'll notice that most of the downvotes were pretty consistent in #, except SLaks (the leader usually attracts the most downvotes) and Dave DeLong (not sure what's up there, apparently the community spoke more heavily that he didn't belong there???) ~ I just think that you're making some assumptions and not considering the whole story. Statistics is a science for a reason you know. –  jcolebrand Feb 2 '11 at 0:02
Jeff: I mention primary. The problem is that after the primaries, trends are set for both sheeps and strategists. –  user150926 Feb 2 '11 at 7:06
@pierre in the USA, very few voters know about or even care what happened in the primary. The primary is for insiders, and the primary results are known in public -- and the final election results are often RADICALLY different than the primary results. Remember too that primary is SKIPPED if there are 10 or less nominees. –  Jeff Atwood Feb 2 '11 at 8:28
@Pierre, now I see why you didn't run on Programmers SE, you were too busy analyzing the process :) –  Tim Post Feb 2 '11 at 8:39
@Jeff: I think it makes a difference that primaries are internal to the parties, and that they come months before the actual election. This is not the case for the SO elections, where they are more like a two-rounds voting system, but with different rules for the first and second rounds. –  Jan Fabry Feb 2 '11 at 8:44
@Tim: I wish I could! I'm even ready to take the necessary time to do it if they provide me with the data. –  user150926 Feb 2 '11 at 10:15
show 2 more comments

7 Answers

Negative voting during primary and hiding results during primary

The election primary is a more akin to a panel at a convention than a secretive voting booth.

  • The nominees make statements
  • People respond and start discussions
  • People publicly demonstrate their allegiance to their nominees - thus we need upvotes
  • Just as importantly they actively campaign against nominees they don't feel will fit well - thus we need downvotes

With such a large field of nominees, it's important to have an open dialog and discussion, with equally open voting tallies. As nominees adjust their platform the votes can and do move, and it's important that everyone be aware of these shifts.

Now, keep in mind that the voting itself (ie, who voted for who) is still secret, but the vote tally is updated in real time so that as the "panel" progresses people can see how the crowds are shifting and re-evaluate their own position.

It's important to have downvotes for candidates in the same way that it's important to have downvotes for answers. Some answers do not fit the question well. Some nominees do not fit what people want as a stackoverflow moderator. If downvotes make you feel uneasy, consider it rating each candidate on a scale of one to three. The fact that it's negative is hardly important as long as we record that there are a number of people who disagree with the nominee's position.

There may be some who believe that seeing the tally affects the voting significantly, but fail to consider that this is the basic premise that stackoverflow was originally built on. There are countless suggestions on meta asking that voting results on answers be withheld for a certain time period to avoid this effect, but as of yet no one has shown that it's a significant problem. Further, in this and past elections the voting has a significantly stronger correlation to the reputation of the nominee than it has to the time they entered. Those who entered later (when others had higher vote tallies) got just as many votes, and more, than those who entered early and got a head start on voting.

In other words, there is no evidence of the "problem" regarding showing the tally during the nomination period.

share|improve this answer
Thanks Pollyanna: I can't provide evidence of the problem since I don't have access to the data. With the data, it's very easy to determine if we were affected by such behaviors. However, in order to avoid any contestable results, election organizers should ensure those behavior are not be possible. That's my point. 4000 voters, it's enough to avoid any strategists. I think it becomes more a problem with less than 500 voters. Let's see what happen on smaller communities. –  user150926 Feb 3 '11 at 9:56
add comment

Quite honestly, I found the primary process to be extremely useful and informative. Negative votes are the strongest indicator to a candidate that the community disagrees with something in their agenda, or desires clarification regarding past actions. I was hoping for something a bit more coherent than the comment system for subsequent discussion, but it did the job.

You can't document a negative. That is to say, you can't document up votes you did not receive as an indicator to better refine your nomination. Negative votes are just as useful as up votes during that phase. Down votes tell you "I don't like something about your nomination", whereas a lack of up votes might be ascribed to the popularity of other candidates, rather than something that might be missing or misunderstood in your nomination.

I suspect that the results of the election will be a bit different than what many might predict based on the results of the primary. I'm extremely curious to see the results, even beyond knowing if I've been elected or not.


To the other points, I think the idea behind the citizen badge is good, but I'm worried about encouraging people to ask questions when they don't really have questions to ask. The same for providing answers that just echo existing answers. There are also many people who spend more time editing and re-organizing than they do answering or asking questions.

I think the badges that we have are good for seeing who actively and constructively participates on any given site.

As for the herd instinct, I'm not entirely sure that it's applicable. While I did notice several nominees who had quite a few down votes, it was easy to determine that they received them due to low or no participation on meta, or something obvious in the text that they wrote. I'm sure that there are some who just follow the pack, but I think that they are a minority.

Vote count during the actual election is a mystery, we have no idea who is leading and who has been eliminated at this point. To follow the herd from the primary into the actual election, you'd have to go back and write down how many votes each person got, select the top three and then cast your votes accordingly. I have faith that most of my fellow programmers are far too lazy to do that :)

Final Update

According to the primary, I would not have been elected. It showed me as 'electable enough', but I was consistently in the fifth or sixth place according to votes.

This tells me that when it comes time for the actual election, your previous community building activity and the text that you wrote when nominating yourself is what counts.

share|improve this answer
Tim: you know I would be very interested by your opinion on the two other points. –  user150926 Feb 2 '11 at 10:13
@Pierre - I updated my answer –  Tim Post Feb 2 '11 at 14:21
add comment

I deeply disagree with the entire premise of this post, to the point that I think it's confusing and harmful to the average meta user; I am actually tempted to delete it because it is so bad.

You make NO distinction between election and primary here, when the voting systems and function of these phases are entirely different.

In the primary, the goal is to decide who is in fact electable. That's why:

  • you can vote on every nominee
  • you can vote up or down
  • vote scores are visible at all times

There is obviously some intentional bias here, because we want the strongest, most electable candidates to proceed forward. That way the election is between ten strong candidates that you have to actually THINK ABOUT to decide who you vote for .. rather than mindlessly and reflexively voting up the highest rep person, which is what happens far too often.

In the election, the goal is to have a private, fair vote. That's why:

  • you get only three votes (1st, 2nd, 3rd)
  • there is no "downvote"
  • all vote scores are private until the very end of the election.

I think this is not a critique of the election system so much as an indication that you don't understand how elections work.

share|improve this answer
What electoral system permits downvoting of candidates? Also aren't primaries usually secret ballots (which this was) but the outcome is revealed upon completion of the primary? This was more like an anonymous Caucus. –  Kev Feb 2 '11 at 0:36
Jeff: I disagree with what you've written, particularly because it is obvious that the OP was referring to the primary (since the election votes are already kept secret). You're just attacking a strawman with your post. –  George Stocker Feb 2 '11 at 1:32
Displaying the current vote tallies during the primary wouldn't seem to help with determining the most electable candidates. It just encourages mob voting behavior and strategic voting. Mob voting lets peculiarities of timing (who got the most votes earliest, etc.) sway people's opinions. –  Brad Larson Feb 2 '11 at 1:34
@Jeff First off, I honestly don't think that this post would ever qualify as: "I am actually tempted to delete it because it is so bad" deleting it would be, with all due respect and abuse, since he is entitled to his opinion and a relatively simple edit would make the post refer (explicitly) to the primary and would avoid any confusion. –  Trufa Feb 2 '11 at 3:14
@Jeff: I mention primaries explicitely. The problem is that after the public primaries, the trends are known from both the sheeps and strategists. –  user150926 Feb 2 '11 at 7:07
@trufa see my edit, then. It took a lot of editing because this question is so ill-conceived. –  Jeff Atwood Feb 2 '11 at 7:12
@pierre in the USA they have primaries that work the same way (everyone knows the voting and outcomes), and elections that work the same way (all the voting is private). –  Jeff Atwood Feb 2 '11 at 7:13
This paper is the result of many hours of research, covering the fields of behavioural psychology, sociology and politics, and discussion with professionals in these fields who I am in regular contact with during my work in the media sector. –  user150926 Feb 2 '11 at 7:15
@Pierre And honestly, this is an election for a moderator on a Q&A site, who is volunteering to do it, not an election for the next president of the world. I agree with Jeff on this simply and solely because I believe that a lot of time we spend too much time complicating thing on Stack Exchange, instead of just finding simpler ways of doing things. I strongly believe, as great as the new election process is, it is overkill. It is great being a moderator, but in the end, it's doesn't gain you anything major outside of the Stack Exchange world. –  Diago Feb 2 '11 at 7:43
@diago remember too that most of his complaints are about the PRIMARY, which is frequently skipped entirely when there are <= 10 nominees. It's only really meaningful on SO when you have 50+ nominees.. –  Jeff Atwood Feb 2 '11 at 8:29
@Jeff Agreed. I also saw it work brilliantly on Super User were candidates that did not deserve to become moderators, never made it to the actual election. –  Diago Feb 2 '11 at 8:33
@Diago: I agree it's not correct to compare SO elections to US elections. I'm just proposing stuff I found may be interesting for SEN. I don't claim to be right. I just provide some feedback that I think may be useful for internal team discussions that may occurs. –  user150926 Feb 2 '11 at 10:12
@jeff, it did, I was thinking more in the lines of a big bold note at the beginning saying "this only refers to the election primary!" But I understand you concern. I actually withdrew my vote mainly because of @Diago 's point of view. Glad you did't delete no matter how "ill conceived". –  Trufa Feb 2 '11 at 11:43
@jeff - if you do insist on the second stage of the moderator election process being called a "Primary" then you might have noticed that in your own state the results of the "Primary" are not tabulated until after the close of poll. –  Kev Feb 2 '11 at 17:56
add comment


I see three opportunities for improvement to the existing system which are basically to end the current practices of:

  • Negative Voting

Negative voting isn't a problem.

  • Voting rights for Members with low level of activity

Your definition of sufficient activity is just as arbitrary as the one currently in effect. Me, I say that voters without both the Beta and Fanatic badges should be excluded.

  • Publishing the Voting in Real Time during the Primaries

This has some merit, in that it will tend to affect the results. However, you only get to see votes during the primary phase - once the most popular candidates have been selected, the real voting begins... I think you greatly overstate the importance of "late voting": there's no ultimate winner decided at this stage, so at most you're gonna be able to bump a marginal candidate out in favor of someone else. And since there's no limit on votes, an earlier voter could have just as easily shifted the balance in the other direction - the only "disadvantaged" voters are those who don't use their votes!

IMHO, you crammed waaay too much stuff into one suggestion, as evidenced by your failure to properly analyze each of your own complaints. I'm voting to close this - I suggest you split these up and re-ask them (with a bit more thought put into each) so that they can be discussed in detail.

share|improve this answer
ok, tl;uiria? Apparently I'm not savvy. –  jcolebrand Feb 2 '11 at 0:03
@drach: too long; unfortunately i read it anyway –  Shog9 Feb 2 '11 at 0:04
add comment

I'd like to point out that for the record, I got into the election from the primary with only one vote over #11. That's really close, and I'm not sure the votes would have gone the way they did if people couldn't see the vote totals in the primary until it was over.

A few improvements in line with yours:

  1. Only show vote totals after the primary is over.
  2. Only allow candidates to vote for their top 10 choices (where 10 is the number of candidates in the general election). As it stood, everyone could vote for everyone, so you had a potential outcome where all 30 candidates could have received an equal number of votes (unlikely, but possible).
  3. Put participation in meta front and center; as well as present participation for moderator functions (closing, opening, deleting, undeleting, and editing questions), and reduce the importance of overall reputation:

Overall reputation represents how the community thinks you do with asking questions and communicating answers. It has nothing to do with what a moderator actually does (close, delete, undelete, and edit), so those actions should be quantified, because they're relevant to moderation (user x closes 500 questions, where user y has never closed or opened a question).

By keeping reputation on there, it gives the impression that it's important for a good moderator, and it isn't.

share|improve this answer
You're a good case for the detrimental effect of downvotes on a candidate. The only reason I got anywhere close to you was because you had a disproportionately high number of downvotes. This might have been due to people negatively reacting to your relatively brief campaign statement (those with shorter statements tended to rack up greater numbers of downvotes, I noticed). However, things tightened up toward the end of the primary and I hope people weren't placing strategic downvotes after looking at the totals. Therefore, I agree with the removal of downvotes and the vote total displays. –  Brad Larson Feb 1 '11 at 21:52
Yea, I kept my campaign pitch short because I've done a lot on Meta and Stack Overflow, and my entire theory of moderation and my history is available on Meta, so it seemed superfluous to make statements when my actions are available for view. Given the outcome of the primary, that was a mistake. –  George Stocker Feb 1 '11 at 23:12
I really want to see reputation in general removed from the criteria of selection, and instead see other metrics hoisted in front. I concur. –  jcolebrand Feb 1 '11 at 23:34
+1 I couldn't agree more. And in particular your point about primary votes being kept secret until the primary is over. I think they got that really wrong. –  Kev Feb 2 '11 at 0:43
I agree mostly, but reputation is an indicator of how effectively you communicate. –  Tim Post Feb 2 '11 at 8:34
@George: If you want some stats to back this up, I found in my answer here that those with the shortest campaign statements tended to rack up the most downvotes: meta.stackexchange.com/questions/77557/how-can-i-improve-myself/… . It's too late to make a difference now, but I found it interesting. –  Brad Larson Feb 2 '11 at 21:19
@Brad Upvoted. Excellent work! –  George Stocker Feb 2 '11 at 21:23
add comment

I agree with the removal of downvotes and hiding the running total of votes for candidates until after the election has completed, but I disagree about upping the requirements for voters.

If you want to up the requirements for voters, why stop at a particular badge? Why not restrict voting to 3k users? They have clearly shown a commitment to the site and have started to become familiar with editing / closing questions. How about only 10k users? They know the site even better, and have been exposed to more of the moderation-related capabilities, so they should be even better educated as to who would make a good moderator.

Any sort of arbitrary threshold that someone must cross to vote (beyond those put in place to avoid gaming of the system) seems slightly reminiscent of the old Jim Crow literacy tests that used to be employed in the South. Rather than being racist, high bars for voting could be seen by new users or outsiders as being elitist or as supporting a "good old boys" network on the site (note some of the comments in the primary about people supporting candidates because they weren't "part of the Meta crowd").

Also, the criteria you put forth for the Citizen badge in your previous question seem a little complex. Having to explain to someone why they can't vote in the election because they didn't meet some part of the criteria you've put forth would not be easy.

The comments that troubled me most about how votes were being cast (voting for someone because of their platform of choice or because they were helped by their answers, rather than the qualities that show good moderator skills) came from users with high reputation and activity on the site, as well as new users.

share|improve this answer
Speaking as a very new member of SO with but 245 reputation, I must agree that a threshold of 150 reputation seems too low (I will endeavour to make responsible use of this privilege). But the proposed threshold for the Citizen badge, as well as being complex, seem very strict indeed to me. Maybe just increase the threshold to 1 or 2K reputation? –  Tim Bergel Feb 17 at 20:44
add comment

Great post! Some parts I agree with, some not:

Negative voting

I disagree, because I think the ability to vote against someone is important. I had plenty of upvotes in the primaries, plenty of no-votes, and few downvotes, but I meant the downvotes -- I thought those people would make abysmal mods. I'm not comfortable relying on "well, hopefully lots of people just choose not to vote for them"; I like the ability to explicitly say "I don't want this person to win". I do agree that people getting massively negative scores is disappointing, and possibly recording the true score internally but displaying max(0, true_score) is a good idea. But speaking of displaying scores:

Publishing voting in real time during the primaries

Completely agree, and you're not the first to discuss either people changing their votes based on the current totals, or people downvoting everyone they didn't upvote (strategic downvoting). The main argument for showing them seemed to be that we do the same with regular posts, and it works out, but I think the main difference is regular posts exist forever, while the election has a fixed ending; knowing when votes will be locked in gives late voters the advantage you described, and I'm quite confident herd voting played a major role in the overall outcome (which I guess means early voters have an advantage as well -- it's the people in the middle who are useless)

Voting Rights Members with Low Level of Activity

I don't know if the definition of the Citizen badge is perfect, but I definitely agree that 150 rep is too low -- it's less than the rep cap itself, so you can spin off a sockpuppet with that much rep in a single day (as many as you like, if necessary)

I'll also jump on what George said:

Limit votes to 10 per voter

I think the ability to vote on everyone is probably reasonable, but this would certainly limit strategic voting

Highlight meta participation and moderator function usage

Very much agree; focusing on reputation is completely the wrong approach. Editing, closing, and helping on meta are probably the best metrics, since they're similar to the sorts of things mods do; posting questions and answers are almost unrelated by comparison

share|improve this answer
"posting questions and answers are almost unrelated by comparison" except that how can I edit if I have no domain knowledge to know what I am even editing. Disagree. Definitely related, though should not be overwhelmingly so. –  Jeff Atwood Feb 1 '11 at 23:08
"so you can spin off a sockpuppet with that much rep in a single day" errr.. not really; as soon as the voting anomaly scripts run, all those votes would be invalidated. –  Jeff Atwood Feb 1 '11 at 23:08
@Jeff I edit posts all the time without knowing exactly what they're talking about; it's usually formatting, spelling, etc. And obviously I don't know how the voting anomaly scripts work; I assumed they looked for one person voting on another person's posts a lot, which is totally ineffective against this kind of fraud (voting one time on a bunch of people's posts) –  Michael Mrozek Feb 1 '11 at 23:50
add comment

You must log in to answer this question.