15

Stack Exchange is currently looking for volunteers for their 1 reputation voting test. The Arqade mod team has decided that we’re fine with opting into this experiment, but we believe it’s a decision that should be made with input from the entire Arqade community.

What’s 1 reputation voting?

The short version is that 1 reputation voting means that any registered user would be allowed to vote on posts, both up (normally requires 15 rep) and down (normally requires 125 rep), and downvoting answers will be made free (it normally costs 1 rep to downvote answers). Stack Exchange believes that this change will improve the user experience, which will make it more likely that new users will become regular users of the site. They also believe this increased engagement will assist long term users as well, since their posts will see more engagement. For more details, I encourage you to read the original meta post.

What’s the experiment?

Stack Exchange has spent many months behind the scenes planning, and doing research, and they’re now at the point that they need to move beyond theory and start getting real-world feedback on 1-reputation voting. To that end, they want to implement the change on a few sites so they can start collecting data on how effective the change is at improving the user experience. They also plan to collect data and feedback on the downsides to the change as well, such as if it leads to a misuse in voting by spam accounts.

The data and feedback collected will be used to determine if:

  • The change has the desired effect
  • If any new community and mod tools will be required if the change is made permanent
  • If the pros outweigh the cons

When would the experiment start?

Assuming we opt-in and are selected, then the experiment would start sometime within the next few weeks to a couple of months. The vague timeline is because Stack Exchange plans to do data collection before the experiment starts so they can set some baseline statistics on the things they’re testing.

Why does the mod team want to opt-in to the experiment?

The mod team believes that if the change is successful in improving the user experience, then that’s fantastic, and that it’s something we should consider implementing permanently. Do we have doubts and concerns that it’ll work? Absolutely. It’s entirely possible this whole thing will blow up in our faces. However, we can’t know for certain which way the change will go until we actually try it out. In other words, in order to know what the outcome of this change will be, we need to conduct SCIENCE!.

Could this result in a bunch of extra work for the mod team?

Absolutely. We fully expect to see an increase in our workload while this experiment is in effect, but we have a fairly large team, and we believe we have more than enough capacity to handle the increased workload. In fact, one of the things being tested is just how much this changes increases the workload of the mod team. If it’s too large, then SE may decide that either additional moderation tools or moderators may be needed if this change was made permanent.

What if this turns out to be a total disaster?

If this whole thing blows up in our faces, then the CM Team has assured us that we can roll back the change fairly quickly. They plan to work very closely with the mod team to assist us with any problems that may arise, and they’ve told us they will prioritize improving tools/settings quickly to combat any issues.

So what are you looking for from the community?

The main thing we’re turning to the community for is to determine if Arqade wants to take part in the experiment. We’d also like to hear any concerns and questions you have, but the main thing we’re wondering is if the community thinks we should participate in this experiment.

9
  • 2
    I like the idea of leaving answers open to allow for more detailed responses, but could we also have some unequivocal / canonical Yes and No answers for making the decision itself? (To avoid the possibility of the top-voted answer being a nuanced take that doesn't clearly take one of the two positions.)
    – Schism
    Sep 26 at 2:35
  • 2
    For all that matter, I would be strongly against this on any site at all, for the reasons already described multiple times by other users in the original post. That said, I will also add that this seems a blatant case of Rand own argument - Any site willing to volunteer for this experiment is probably a site less likely to suffer negative consequences. - so ironically this may have less negative consequence exactly because testing that here proves nothing about the effect it would have on sites like SO.
    – Ꮢedacted
    Sep 26 at 8:16
  • 2
    @Ꮢedacted certainly a valid point - I suppose that is why (or rather a reason why) they (SE) would like multiple sites to partake in the experiment. They have stated "We would like to have some of the larger sites participate and will likely lean towards them as the impacts of these changes will probably be seen more quickly than on very low-activity sites." Ideally they want more active sites to volunteer to get a more conclusive result. Arqade averages shy of 6 new questions a day in the last 2 weeks. The top 10 sites (sorted by questions a day) have 30+.
    – Timmy Jim Mod
    Sep 26 at 13:29
  • Free downvotes still means that users get -1 rep for each downvote received? Because that will be so wrong.
    – pinckerman
    Sep 26 at 14:20
  • 1
    @pinckerman I'm not sure I understand that. Why is paying for downvotes in any way correlated to being penalized for having your post downvoted (which is -2, by the way). You earn rep for good content and lose it for bad content. That's how it's always worked.
    – Catija
    Sep 26 at 14:49
  • @Catija I fear bad usage from random users. If there's no penalty to downvotes what blocks a user to downvote for revenge, maybe someone who criticized their bad contents? Having to "pay" for a downvote was always a good reason to think twice, I used to downvote when something really deserved it.
    – pinckerman
    Sep 26 at 14:59
  • 3
    @pinckerman Those fears are reasonable - making decisions based on fears without testing is not. 1 rep users would either not be able to downvote still or the downvotes would only be free for them, since you can't go below 1 reputation. Either is not great. But we will be able to see how frequently downvotes get used as a result of this test. Many people don't like downvotes, so it's completely possible that cases of misuse will be much lower than expected and if they fall into categories we can look for, we can invalidate inappropriate ones more effectively by having this information.
    – Catija
    Sep 26 at 15:07
  • @Catija - quick couple of questions, apologies if it's been answered elsewhere, but voting without meeting the rep requirements currently shows a message to the user that "their votes have been recorded". Has this data been incorporated into any analysis leading up to this experiment? Are there any plans to apply/backdate these votes (at least the ones from logged in users)?
    – Robotnik Mod
    Sep 27 at 0:48
  • 2
    The feedback is not connected to a user, @Robotnik it's only tied to the post and whether it's up or down. It's intended to identify whether the reception of a post is aligned with the wider viewership. We can look at the information but we're hesitant to make decisions based on it because it doesn't represent 1-1 voting per user. The same person can technically send feedback on the same post multiple times and even both up and down feedback from the same person.
    – Catija
    Sep 27 at 1:13

2 Answers 2

17

If you believe we should take part in the experiment, then vote this answer up. If you don't believe we should take part in the experiment, then vote this answer down.

As per request from Schism

14
  • 5
    The voting methodology which you've chosen here has a significant inherent bias towards positive votes and, thus, approval. As I'm sure you're aware, in order to downvote requires significantly more reputation. Thus, a much larger number of users are able to upvote than are able to downvote. This is further skewed by the ability to upvote, but not downvote, being granted to users with just the association bonus, although I'd hope such users who are not active on the site would choose not to vote in order to allow such a choice to be made by site regulars.
    – Makyen
    Oct 17 at 17:18
  • 3
    A more equitable method of polling would have been to have two answers, one for approval and the other answer for rejection. Then the capability to vote for the two positions would be equal, rather than heavily biased towards approval. If the goal is to allow more open voting, which is what this proposal is about, then count only the upvotes on those two answers. If the goal is to have the voting be, mostly, by established users who have already participated in the site, then count only the downvotes. Counting both up- and downvotes gives a mix with extra weight for established users.
    – Makyen
    Oct 17 at 17:18
  • 2
    @Makyen As far as I know, you only need 5 rep to participate in Meta. Users who disagree with the experiment are welcome to post an answer. The lack of answers from them might indicate that there is no user with 5 rep who is interested enough in this topic to share their perspective. Oct 18 at 0:27
  • 1
    @galacticninja The 5 rep is just to post a question or answer. Upvoting requires 15 rep, downvoting 125. The 'up' answer has a headstart of multiple weeks by now, experience tells me that no disputing answers will be incoming by now.
    – Mast
    Oct 19 at 21:22
  • @Mast I agree with Makyen’s suggestion and reasoning to only count upvotes to allow more open voting. About the "adding another answer for rejection" part, what I'm saying is that there seems to be no user with at least 5 rep who is interested enough in this topic to write an answer rejecting the experiment (which could then receive upvotes). So, IMO, it doesn’t really matter if we count only upvotes or not. Oct 20 at 3:52
  • @galacticninja The reason no answer was written for rejection is likely largely because the method of determining acceptance/rejection was predetermined and that method precluded the need of writing such an answer for those most likely to write such an answer (i.e., those with more participation, who are more likely to participate by writing such an answer, strongly tend to have more reputation, and, thus, the ability to express their rejection by downvoting).
    – Makyen
    Oct 22 at 18:25
  • The reasons for someone having written such an answer would include A) someone with > 125 reputation feeling enough need to explain their objection, rather than just downvoting; or B) someone with < 125 reputation, who is unable to downvote, but who feels strongly enough about the situation such that they feel a need to express themselves, despite the fact that they have been explicitly told that they are disenfranchised by the voting method that will be used. Either of those reasons to write such an answer are far less likely than just someone with < 125 reputation wanting to vote to reject.
    – Makyen
    Oct 22 at 18:25
  • Unfortunately, given the disparity in reputation for up-/downvotes, it's not possible to determine what the "real" level of support is for the proposition. We can, however, determine the rate of downvoting for the 15,496 users who are able to downvote, which is 0.058%. Extending that rate to the 77,017 users on Arqade who can upvote, that's an equivalent of 44.73 downvotes, assuming there's the same level of participation for users who can and can't downvote (which probably isn't accurate). That shows that the choice of how to vote was, potentially, sufficient to completely swing the result.
    – Makyen
    Oct 22 at 18:25
  • Note: the above doesn't include the votes from users who are unable to vote. Those votes are available in the 10k tools. However, those anonymous votes are substantially biased by the fact that users who already know their votes won't be counted will tend not to click on the vote buttons.
    – Makyen
    Oct 22 at 18:26
  • 2
    I'm not trying to say "let's have a do-over for this vote". I'm trying to raise awareness that the manner we choose for having a "democratic" vote on the sites has inherent biases, just due to how voting works. IMO, it's an especially apt point to make for a vote that is intended to allow/reject the expansion of the voting privilege. My making this point here is not intended as support or disagreement with the experiment. Given that I'm not active on Arqade, my opinion on this site's operation probably shouldn't matter, except to the extent that other sites are affected.
    – Makyen
    Oct 22 at 18:29
  • @Makyen I’m not sure what you’re trying to say when you mention “the method of determining acceptance/rejection was predetermined”. Do you mean that this answer was somehow predetermined by the community or the mods? I don’t think that’s the case. Are you suggesting that this post (by a mod) should not have asked for both upvotes and downvotes to avoid discouraging answers rejecting the experiment? Can you clarify? I actually agree with the latter. Oct 23 at 6:33
  • @Makyen As far as I'm aware, there are no predetermined answer posts here. Anyone who has at least 5 rep on Arqade can post an answer to support or oppose the proposed experiment. I agree with Mast’s idea of having two separate answers—one for yes and one for no—which would have been a fairer way of polling. But it seems that no one is interested enough in this topic to write an answer against the experiment. Oct 23 at 6:34
  • Is this a joke? You can't be serious that it sounded like a good idea to use the actual subject of discussion (voting in its full effect) to determine the outcome of the discussion? Clarifying that the current situation is biased, makes the voting to determine the outcome biased by definition. This is like asking rich people to vote if poor people should get more money from the rich because they have less.
    – Mixxiphoid
    Oct 26 at 19:50
  • @Mixxiphoid I don't understand. Can you clarify what you meant? How do you propose we determine if the Arqade community wants to opt in to the experiment? Oct 27 at 8:01
9

After reading Catija’s (community manager) clarificatory comments, I’m reassured that the moderators and community managers have the tools to prevent the abuse that I had expressed concern about. I support the voting experiment as a good idea, as long as they can intervene if things go wrong.

What follows is what I initially wrote prior to Catija’s comments:


I believe we should, if SE allows us to opt out of the following:

  1. Removing the 1 reputation cost to downvote answers
  2. Reducing the reputation required to cast votes in meta to 1

(I've expressed a similar sentiment in my answer to the MSE post.)

Elimination of the 1 reputation downvote penalty would invite abuse and devalue upvoting

The experiment aims to increase participation and engagement, but eliminating the 1 reputation penalty for downvoting would do the opposite. It would encourage abusive behavior (such as downvoting other answers to gain an advantage), devalue upvoting, and might make new users perceive Arqade as a hostile place where you are more likely to be downvoted rather than upvoted. It would discourage participation if there were more downvotes thrown around without penalty. The current top answer to the Meta Stack Exchange post “Why do you lose reputation for downvoting answers?", among several other answers, explains this point well.

The motivation behind it is to put emphasis on up-voting or not voting at all. This way, down votes will carry more weight and it will also prevent users from abusing the system by down-voting excessively.

According to what the founders, Jeff Atwood and Joel Spolsky, discussed on an episode of the Stack Overflow podcast, they wanted to find a way to discourage users from down-voting for less legitimate reasons (say a pro-Java developer down-voting everything remotely related to .NET or the like).

Reducing meta voting reputation thresholds would allow bad actors to influence site policy

SE communities use meta discussion posts to reach consensus and set site policy, e.g., "should questions of this kind be closed or reopened?" If we lower the reputation required to vote in meta to 1, users who have not participated enough on the main site could affect site policy by voting on meta discussion posts. This is not a good idea for several reasons. For example, it could facilitate “brigading” from other communities (such as those outside the SE network) to influence site policy.

“Brigading” […] covers all coordinated voting behaviour to make something or someone seem more or less popular than they actually are, and now it means all coordinated abusive engagement behaviour online.

- https://www.institute.global/insights/tech-and-digitalisation/social-media-futures-what-brigading

The voting experiment sounds like a good idea to me, as long as we can opt out of those. I’ve noticed that voting has become less frequent over the years, which may discourage new users as it is now harder for them to gain reputation. We need to encourage more voting to make Arqade a more active and lively community.

11
  • 6
    I'm not sure if opting in to just part of the experiment is possible. That said, we gave the CM team a heads up before posting this, so hopefully they can weigh in on this.
    – Wipqozn Mod
    Sep 26 at 12:53
  • I like your request, but I don't think it's feasible
    – pinckerman
    Sep 26 at 14:14
  • 1 is already part of the experiment
    – Kevin B
    Sep 26 at 14:35
  • Voting seems to be heavily reliant on the actual Exchange being used. Arqade seems to have moderate voting participation, The Workplace has questions with ridiculous amounts of votes, and Stack Overflow has almost no voting participation. Doesn't seem like a one size fits all thing will apply :\ Sep 26 at 14:47
  • @nightsurfer I guess that's why the SE team want a bevy of different sites to test with.
    – Robotnik Mod
    Sep 27 at 0:34
  • @galacticninja I would be keen to see separate results of experimentation around modifying different aspects of the voting thresholds. Similar to what you suggest, but "as well as" not "instead of". Spend a couple of weeks with the voting thresholds reduced but the downvote penalty in place. Then spend a few weeks with the downvote penalty removed. Keep tweaking things until we nail down what works/what doesn't.
    – Robotnik Mod
    Sep 27 at 0:39
  • @Wipqozn Can we request for a community manager to shed some light on whether we can opt out of what I described above? And if not, could they explain what tools are available to moderators to deal with the potential abuse scenarios I mentioned? The votes on the two current answers here suggest that the Arqade community wants the site to participate in this experiment. Sep 27 at 0:43
  • 1
    @nightsurfer I think Arqade actually has low voting participation. I feel like I don't really vote much, and yet I'm one of the top active voters in Arqade. I believe Arqade would benefit from this experiment if it could curtail the potential abuse I mentioned above. Sep 27 at 0:50
  • @Robotnik That sounds good. I think it would be helpful if a community manager could clarify what parts of the experiment can be tweaked while the experiment is going on and what tools are available to moderators to curtail the potential abuse I mentioned above. Sep 27 at 0:54
  • 3
    I'm not sure I can give a complete explanation of what can be tweaked, @galacticninja because that's unlikely to be able to sufficiently predict the outcomes here on Arqade. I can say that the new tools have identified significant cases of potential suspicious votes on SO in particular, which is one of the few sites that struggle with it. I think the most valuable thing I can say is that changing the rep required to vote is a 30 second change that doesn't require a build, so we can react extremely quickly if it truly becomes an issue.
    – Catija
    Sep 27 at 16:59
  • 8
    My ability (as a CM) to remove problematic votes isn't absolute - we, too, have limits in some ways for good reason - but there are very few situations I've come across where I couldn't remove targeted or otherwise problematic votes by either invalidating them or removing a user who is not participating acceptably from the site. In probably 200 vote invalidation requests I've gotten over the last year, I can remember 2 where I couldn't remove the votes - and to be clear, the outcomes of this test can include improving my tools to handle even those rare cases.
    – Catija
    Sep 27 at 17:03

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .