1 | ||
1 | ||
1 | ||
1 |
I didn't think that voting had enough of an effect from my own experience. A large of amount of this is caused by there being more variance in time among posts on the front page than score. Posts on the front page might include posts that could be an hour old or 2 days old. But the scores of things might be 1 score or up to 6 score currently. So when both are considered equally time dominates as a factor when I want both to matter.
Also adding "bump" to the sort factor meant that time is basically being counted twice because until a post is commented on the bump value is equal to the post time.
So by doubling the score factor we are really only correcting for the double counting of time. I'm willing to go further than that but for now we'll feel out how responsive voting feels.
That does mean that posts that are well voted will hang around longer so the page age may feel a little older. I'll just have to post more.
Another possible experiment we could do is running a single post like a 4chan thread. I could increase the bump factor and tone down the age factor. This would basically make it 4chan like where threads that also get upvotes will get a little more advantage. Then we could run a thread on an ongoing topic.
Looking at 4chan for topic ideas, one sec... We could run a topic on either traps, pokemon, sketchy homemade weapons, or 2d waifus. Ok maybe there is a more sane topic. Of course that style of thread is meant to be more visual so maybe a particular theme of meme or memes in general. I think it would be fun to test the survival of a specific thread. If you have any ideas feel free to start one and I'll try to contribute to its survival.
Because that kind of thread is visual it helps if everyone knows how to do inline images in markdown-extended. ! [ ] ( url ). <- Remove the spaces.
I think I have a thread idea... look for it.
Edit:
I did make a couple image board style threads that could use some contribution to keep them alive:
LotR memes: https://matrix.gvid.tv/c/Threads/98cc5QpHut
Bad halloween costume ideas: https://matrix.gvid.tv/c/Threads/8Ygivhd3Bw
AI Original Content and WTF AI: https://matrix.gvid.tv/c/Threads/5zm88wZcVo
I'm putting posts for continuous contribution in /s/Threads. People can also start new ones.
FTFY.
thats funny.
How about adding a Time Decay factor. The older a post the more its score degrades in the weighting (not the actual score)
I guess I could. I'm trying to understand the motivation. Basically you want to be able to look at things historically in order?
I think another way to do that is open up the custom sort drop down, slide everything but new to zero, and then just hit next a million times to get to the old content. But maybe there is another reason you are getting at.
I don't think any very old content should be on the front page, makes the site look dead. Front page should be mostly todays stuff with high ranking stuff from the last 1-3 days.
With a time decay you'd have a post with a weight of, lets just sat 10 arbitrarily, on its first day, but with time that weighted score starts to drop the older it is, so perhaps on day 2 its weight might be 5, and by day 3,it would effective only be worth 2-3 points. The post rots away from the front page to be replaced with fresh virgin content. The cycle of shitposting continues anew.
The problem is we just don't have the post volume to consistently prevent old posts. Basically upping the age factor would do the same thing you're suggesting. The problem then is if the age dominates then the upvote button basically doesn't do anything. Then people don't click it. Then the site is just 1s which looks just as bad.
Maybe if I get at the heart of what you're saying is that there is a critical threshold where it just looks bad for a post to be there. So maybe the age curve could be non-linear, where there is a steep cliff right around that critical period.
Essentially if(age>2*day) age+=2; But smoother.
Or another way would be if(age<2*days) age/=2;
Probably pretty low probability I'd do that. But I can keep it in the back of my mind as a consideration.
But yeah, since you are talking about doing things based on appearance (completely fair), I basically need the posts that never get voted on to drop off the page to avoid a page of 1s and 2s. After that there aren't enough posts to not have some old posts.
If I could get people to upvote more consistently I could let age matter more and have a better appearance. It's funny how many people will comment something and even comment "that's amazing" but not even upvote a post.
But the silver lining is that posts can have a longer period of discussion.
How about having no bump score when there are no comments? And I welcome all 4chan inspiration, I think making this site more 4chan-like will really give it an edge. The more 4chan-like, the better.
The issue is that the formula is essentially multiplication. Can't multiply or divide by zero without it really messing things up. I guess you could say, then multiply by 1. I just have to double check that isn't weird when considering what other numbers are going in. Basically we are dividing by a count of miliseconds.
The other possible way is that since bump and age are doing double duty I could set the age factor to zero and have the posts age basically equal to the bump. Basically 100% like 4-chan. Or I can go full hybrid between reddit and 4chan and do 50:50 by making both half. That's basically what I did with doubling score. Its a 1:1:2 parts ratio. Same as a .5:.5:1.
Well, there's also proximity, so there is a difference. If you eliminate age as a factor, that makes the age slider pointless, but this may actually be better as it simplifies configuration. Lemmy allows a sort by 'Activity'. You could remove the age counter and rename bump to activity.
On the other hand, it does make sense to keep them separated. Some users may want a steady stream of news articles which they are gonna read perhaps, but aren't interested in comments and don't want old posts to return to the top. On second thought, it seems better to keep them separate.
I would prefer the multiplication by 1 for 0 comments. This is more neutral: it doesn't affect how the age slider works. At first I considered a power function but this makes the score of highly active threads go through the roof. Instead it makes sense to do bump_factor = bump_weight * (comment_count + 1). What do you think? I think it's fine not to take comment depth into account. Even if you're making a deeply nested comment it's nice to get noticed.
If you want nested comments to have less weight, you could try bump_factor = bump_weight * ( (toplevel_comment_count + 1) + (second_level_count / 2) + (third_level_count / 3) + ... ). But I'm really just writing this as a math practice for myself, I don't recommend this.