How about a comment screener (like Hey)? New comments on your posts are first in an inbox. If you accept it, or whitelist the person, it shows up for you. Others can choose whether to see all or only approved comments.
Login to reply
Replies (6)
Opt in ignoring in a social contexts always eventually degrades to nobody opting in. Imagine you have Alice and Bob, who like each other, and Unlikable Urkle. Alice initially ignores Urkle but Bob replies to one of his comments, stoking her curiosity, so she unignores him. This is why /ignore on IRC and similar systems don't work.
If I understand correctly, Twitter solves this pretty well. If you encounter a muted post, you can see it with a click, without doing any unmuting.
Generally, I mute a lot and find social media barely usable without it. There are a few instances where I still look at muted posts, but it’s not that often. I am sure I miss things here and there, but I’m ok with that.
Yrah, but there's no social aspect to Twitter muting. What you described (as I understood it) is Alice's client influencing which posts Bob sees. In that case, once one person we like violates the protocol, it incentivizes other people in the social group to violate the protocol, quickly rendering it moot.
Gotcha. The social aspect is secondary and optional.
I was mostly trying to think of a way to deal with my current situation where I don’t appreciate many comments on some of my recent posts, and how to avoid getting annoyed by them. Right now we treat comments as approved by default. But maybe sometimes the opposite is more appropriate, even if it’s just in our own client/view.
Instead of having to whitelist people, you could probably have a simple naive Bayesian classifier in your client that you train on the type of comments you don't like. It would hide future similar comments in the future, with you able to tweak the threshold as your mood changed (e.g., you could risk seeing more junk on a day where you were bored). You could still whitelist your friends so they would pass through the filter.
Certainly an option. If I were to build this feature, I would probably start with the most simple option - manual whitelisting, and then feel out via usage what additional filtering might be useful. I do like the idea of manual control, as it puts the user fully in control. But there are probably some limits to that as well.