Welcome to the Judaism community on Codidact!
Will you help us build our community of learners? Drop into our study hall, ask questions, help others with answers to their questions, share a d'var torah if you're so inclined, invite your friends, and join us in building this community together. Not an ask-the-rabbi service, just people at all levels learning together.
Should imported posts retain their original votes?
It seems likely that we will import at least some content from Mi Yodeya. There's a separate question for that; in this question I want to focus on a suggestion brought up in an answer there: keep the original votes for the imported posts instead of resetting to zero. I'm separating this out because that answer addresses two points, this and what to import, and there's one comment so far saying a vote was for one part of that and not the other. I want to avoid such confusion when we decide this.
Assuming we import content, and setting aside the question of what to import (please discuss that on the other question), what should we do with the votes that posts gained on Mi Yodeya when we import here? We could:
-
Keep the votes as they are. Those votes reflect valuable evaluations that have already been performed; why throw that away?
-
Keep the votes but apply some fudge factor, so that votes earned natively here will count more. This reflects that Judaism Codidact is a different community, although with many of the same users, and "native" work is a stronger reflection of this community.
-
Keep the votes but separate them somehow, so a post might be +3/-1 in native votes and +15/-2 in imported votes -- find a way to show all that. (Note: this would require new code.)
-
Other things I haven't thought of yet.
If we import votes we will not be able to associate them with voters. We don't know who voted. You might end up voting again for something you voted on there. The answer I linked to questions whether this is really a concern if our goal is to bring in valuable content.
2 answers
What distinguishes us from a blog is mostly the scores on the posts. That's one of our main advantages for people seeking quality content: they have a rough measure of the average interested internet user's opinion of the post.
We are going to import content. To throw away useful information about that content is doing a disservice to our readers. That's why we should keep votes.
Is there any downside to keeping votes?
Almost none. The only downside is with the signal there is also noise. This is true of all signals. This noise takes a few forms:
-
Double voting: a user may have voted for a post on Mi Yodeya and vote again here. Inaccurate signal.
-
Calibration: scores at Mi Yodeya are generated over a certain user base and time period. Scores on native posts will be generated over a different user base and time period. Thus a score of 10 may have different meanings depending on the origin of the post.
-
Differing standards: a post on Mi Yodeya was judged on its scope and quality standards, while posts here are to be judged on our scope and quality standards.
The question for us is: is this noise significantly outside the margin of error of our native signal generation?
To me the answer is clearly no. For two reasons: the noise isn't that bad, and our native signal isn't that noiseless.
-
Double voting is a realistic concern for only a handful of users and if they are so inspired by a post to reach for the upvote button twice, it's probably worth it.
-
Scores at Mi Yodeya hovered in the 1-15 range mostly. A few weeks on this site already has a number of posts up at +7. I see no reason to expect the ranges to overlap reasonably soon. (See below about extreme cases.)
-
No one as far as I know has yet suggested any scope or quality standard which would be stricter than Mi Yodeya's. In other words, if a post was of good quality there, we can be confident it will be here too. (The inverse doesn't hold.) If we exclude some specific set of questions (eg. product recommendation questions) we can always remove questions with that tag later.
In addition to all of that, our native signal generation will not be perfect. There will be random questions that get more views for some reason, even if it's not because of SE's Hot Network Questions. And there will be good posts that fly under the radar for whatever reason. It happens. You can't get worked up about a random few vote points. Plus many of our initial users will not want to go back and reread all the Mi Yodeya content they already read once to regenerate the original signal.
Thus it is worth bringing over the scores with the posts.
My only caveat is some extremely highly voted posts on Mi Yodeya which, while not really harmful there, we may as well scale down a bit. I'll propose different fudge factors in different answers, though the details don't really matter all that much (noise margin of error, etc.)
Vote here if you support the general idea.
0 comment threads
Another proposal: Award at most one vote on import per post.
If we're importing a post that had a net-positive score at Mi Yodeya, indicating that the community there considered it valuable, it would seem wrong for it to show up here with a zero score. So give it one upvote "from the community."
Beyond the simple signal from the sign of the Mi Yodeya score that this post has value, the signal sent by the scale of the score at Mi Yodeya is more confounded. It could come from degree of publicity (including happening to be featured on the network's sidebar) as much as from the community's judgement of its value. So, there's much less value in importing that scale here. Add to that the aforementioned problems with mixing Mi Yodeya votes with local votes.
One place where we might want to pay attention to the relative scores of posts is in the import of multiple answers to the same question. There, to capture Mi Yodeya's collective judgement about the relative value of the answers, it might make sense to give an import vote only to the highest-scoring answer, or perhaps the top Nth percentile of answers.
1 comment thread