Recently Popular

[Noxid]

Would it make sense to use a logic similar to what already picks Most Popular, but not include fav's from watchers in determining score?

[Frank LeRenard]

I feel like this is kind of a long-standing problem with the arts in general, isn't it? That there are loads of extremely talented people getting overlooked. I suppose having a committee or a single person out there browsing and picking their favorites might work, assuming said committee or person had what the community would consider respectable taste. I'm sure you'd always have to try to hit that balance of "I think the community will really like this one" and "I think the community should see this one". You'd also have to look at EVERYTHING: not just art, but music, photography, writing (stories and poems!), sculpture, etc. It sounds kind of hellish if you ask me.

[XoPachi]

Well like I said, a small start won't hurt and I'm not pressing for immediacy or first time perfection. Basically any system where it's not the same artists or same content being boosted would be lightyear ahead of where we are now.

I like where Taw is going, personally.

[Kayla-La]

The problem I have with automated systems is that it then can't discern quality or any other problems (unless it were queued and then a person manually approved it). I feel like it would just end up becoming random lower-quality art in an automated system where many people will then end up learning to ignore that section. For me, the point of this idea is less 'show any art from people with less watchers' and more 'show quality art from people who just aren't being noticed'. I understand 'quality' is subjective to a degree, but I don't think this is an impossible problem.

I understand the worry of bias, but I do feel there needs to be some sort of human involvement to make it a worthwhile feature. After all, it's -people- who are going to be looking at the art in that section. I think my favourite idea is a team of different people with different tastes, and maybe a SFW version and a NSFW version? A community vote/call for suggestions for the best way to go about it might help get more attention than just this thread.

[Antumbra]

Instead of a staff picked DD, maybe add a feature that allows people to promote an artist under a certain watcher number and have them voted on by users? Then those could be sorted by popularity and eventually if that artist gets popular enough, they can't be nominated again.

[XoPachi]

The problem I have with automated systems is that it then can't discern quality or any other problems (unless it were queued and then a person manually approved it). I feel like it would just end up becoming random lower-quality art in an automated system where many people will then end up learning to ignore that section. For me, the point of this idea is less 'show any art from people with less watchers' and more 'show quality art from people who just aren't being noticed'. I understand 'quality' is subjective to a degree, but I don't think this is an impossible problem.

I understand the worry of bias, but I do feel there needs to be some sort of human involvement to make it a worthwhile feature. After all, it's -people- who are going to be looking at the art in that section. I think my favourite idea is a team of different people with different tastes, and maybe a SFW version and a NSFW version? A community vote/call for suggestions for the best way to go about it might help get more attention than just this thread.

I was also thinking of some sort of curator rotation system where the people would switch hands at regular periods. I didn't put it in the OP because a lot of problems immediately arised from that when I thought about it a bit longer. Mainly that it would assume Weasyl is even active enough where we could regularly have a new group of people each interval to choose new art. There's also the issue of how we'd choose new curators in the first place every time making sure we select

EDIT: Strange, I don't know why my post cut off like that. I thought I finished my thoughts...
Can't remember what I was going to say.

[Frank LeRenard]

Instead of a staff picked DD, maybe add a feature that allows people to promote an artist under a certain watcher number and have them voted on by users? Then those could be sorted by popularity and eventually if that artist gets popular enough, they can't be nominated again.

That's a cool idea.

I'm thinking too of something like a rotating system based on keeping new users appearing on the front page. Like each time a particular user's submission makes it to the 'currently popular' section, an integer flag gets bumped up by 1, and when it reaches a certain number that user's account is barred from the 'currently popular' section for a reasonable period of time. So that way, users who are already popular will keep getting exposure, but it will also allow for other faces to start gaining some more views as well.

And no, I haven't thought deeply about that idea to discover all its potential flaws.

[Bornes]

Would it make sense to use a logic similar to what already picks Most Popular, but not include fav's from watchers in determining score?

That may be a way to fix it.

Maybe consider [(% of watchers who faved art) + (% of online users at time of upload who faved art)] instead of a direct number? That way a person with 50 watchers who gets 5 faves in five minutes probably won't get a higher 'recently popular' score than a person who has 5 watchers but 2 faves in 5 minutes?

Edit: Or instead change the whole thing.

Instead of counting faves, count watchers within a certain time frame of a submission. If X artist gains Y amount of watchers within Z amount of time of an upload, that upload is featured in 'recently popular'.

[Hendikins]

The biggest issue I take with it is that in most cases, the majority of what is popular are uploads by already established users with wildly massive followings across multiple sites. That goes triple if they specialize in NSFW.

I'm still on holidays at the moment (and the weather in Hong Kong has been pretty miserable the last couple of days) so I'm not going to respond to everyone here, but I do think this could be dealt with to some extent by weightings:

Determine the average number of views/favourites for each rating, and weight accordingly

We can already determine information including the average number of views/favourites for each rating, as well as percentage of views/comments/favourites per rating. If NSFW material is say, 2x more likely to be faved than General, that's fairly easy to neutralise.

Add a weighting based on ratio of watchers to views/favourites

This one is fairly simple. If you notify 3000 people instead of 30, you'll probably get more activity. Including a weighing based on ratio of watchers:views/favs would help neutralise that effect. 50 favs from 3000 watchers wouldn't be significant at all, but 50 favs with 30 watchers is very significant.

Ultimately it comes down to whether people want "Recently Popular" to be in absolute terms or relative - and the latter does increase discoverability of artists.

[rbartrop]

Whatever you do, someone isn't going to be happy (no big surprise there), but here's my two cents:

Weighting by rating:
It's no surprise that certain subjects get more attention than others, and I've been there where a piece gets ignored because it's not one of those subjects. On the other hand, there is a sfw filter, and plenty of tools for removing anything else that people don't want to look at. What would be useful, I think is ensure that Recently Popular still returns a page full of results, even if it does have to search a little deeper.

Weighting by watchers:

This is a good idea. I'm all for discovering new talent, and if they're already a hugely popular artist, and there's really not much point in having an artist come up if they're already on your watch list. For that matter is if feasible to have a filter that checks your watch list, and filter them out of Recently Popular?