I've been thinking (and posting) a lot about voting and one issue that's come up over and over is the chilling effect that downvoting can have on Meta discussions and feature-requests.
Downvoting is an essential tool and is very effective at sorting out "good" ideas from "bad" ones - or, rather, signaling the larger community's feelings on a particular topic.
But I think there's some legitimate concern that well-meaning Meta users can take a huge rep-hit, not to mention getting Q-banned, over one or two exceptionally unpopular ideas.
On Stack Overflow, most heavily downvoted questions are eventually deleted, potentially restoring the OP's rep. Stack Overflow has a grand total of 6 questions with a score of -20 or less.
On Meta, though, these questions are often not deleted because, though the community strongly disagrees with the idea, the discussion is still on-topic and thus "valid". These questions generate upvoted answers, preventing the OP from withdrawing their discussion or feature-request. These questions hang around in the system like an albatross around the neck. Meta has a grand total of 263 questions with a score of -20 or less. Meta is far less active than Stack Overflow, yet has over 43 times more of these unpopular (yet undeleted) questions.
For Meta specifically, does it make sense to have a per-post rep cap limit?
Comparing the reasons for a negative cap to the reasons for a positive cap
According to Revisiting the rep cap (yes, again), there are three (maybe more) purposes for the rep cap:
- To prevent a user with a single popular answer from gaining significant privileges without really understanding the community in enough depth to use those powers wisely
- To avoid a "rich get richer" sort of system
- To encourage heavily active users to step outside once in a while
For Meta specifically (since heavily downvoted questions can still be on-topic and thus are not deleted), I think you could argue that a per-post negative rep cap would:
- Prevent a user with a single unpopular answer from losing significant privileges, even though they understand the community well enough to use those powers wisely
- To avoid a "poor get poorer" sort of system (since the posts hang around, are referenced, and continue eliciting downvotes)
- To encourage active users to take a (limited) risk and make a potentially unpopular (but also potentially valuable) suggestion once in a while
Not much else would change
The post itself could continue to be downvoted, continuing to serve the purpose of signaling the community's feelings on it, but the OP could take comfort in knowing that the bleeding will stop. If the post is abusive (etc), the community and mods could still deal with it using the conventional methods - close, delete, ban the user.
What should the negative rep cap be?
I'm not really sure what a good per-post negative rep cap would be. -40 (20 downvotes/0 upvotes) seems like a good starting point. The negative rep cap would be calculated based on the total score of the question. A question with 5 upvotes (+25 rep) and 25 downvotes (-50 rep) would have a net rep change of -25, so it would not yet have reached the cap. A question with 5 upvotes (+25 rep) and 50 downvotes (-100 rep) would have a net rep change of -75. This would qualify for the rep cap, meaning the OP would lose -40 rep rather than -75.
Please consider holding your vote on these if this is the first time you've seen them. They are listed for reference and I'd like to avoid the "Meta effect" further punishing these OP's.
Look at the following posts for example, none of which are rants, off-topic or useless (they at least serve to reinforce/further define the community's thinking on a subject) but which the community generally disagreed with:
- Should downvote-without-comment rate remain unpublished?
- NSFW proposal on Area51
- Provision to delete answers to your question
- Proposal that FAQ be modified to allow signed questions
- How can I ignore questions from users with less than 20 reputation?
- Stack Overflow should have smileys and friend groups
While I agree that these posts should be downvoted (and I would add my downvote if I had not come across them as part of my research), I do think the authors were asking questions or proposing features in good faith with good intentions. The huge number of downvotes is warranted (in my opinion), but the hit to the OP's rep seems unjustified since, on Meta, we encourage these kinds of questions.
Robert Harvey had a different suggestion for dealing with this problem, which I think would be effective if all users abided by his post. But as Shog9 pointed out, the suggestion is a good idea but largely impractical.
I realize that this post only deals with questions. Answers, however, can be retracted easily if the OP changes his/her mind or makes a concious decision to leave it out there despite its unpopularity. OP's do not have this luxury with questions and so they are a special class, in my mind.
According to this query, of the 259 questions on Meta with a score of -20 or less, 145 questions hit the proposed rep cap. 66 of these were closed (for various reasons). Of those 145 questions, 75 represent a negative rep change of 50 or more (37 of which are closed), and 44 represent a negative rep change of 60 or more (19 of which are closed). The average rep "saved" on the 145 questions (that is, the absolute number of points after adding 40) is about 16-17. I did not take account deletion into consideration when calculating these numbers. I have not looked into all of these questions to their relative value to the community.