so glad yer here....could you add another button to the profile navigation bar called, "promoted"? I would like to see who me threw our sweet-meats to....

Lucky, the new sort order doesn't seem to be working for me. Right now the unique views of the first 7 vids in the queue are: 7, 4, 12, 13, 5, 11, 6. I get it, those vids have been requeued. Just the (visible) counter hasn't been reset.

Also, I have a more philosophical question: If every video just gets requeued endlessly until it is published (which sooner or later it will, unless it's REALLY terrible) then how is this site acting as a sifter? Sifting is supposed to remove the chaff from the flour, so shouldn't a fair percentage of our vids be dying in the queue? Originally, that's how the site worked, and it's something that I don't think is working properly anymore.

^word to arvana

i totally don't understand requeue (it really does work if you constantly requeue, eventually you'll get there, i guess at the expense of taking up a slot in your queue, but hey, i am not a vote whore so i don't really care if i lose a few days)

then, when my "bad" video got through, it got another 8 votes very quickly. Some people just don't look at unsifted i guess.

loads of vids in the top 15 needed a requeue

if we didn't have requeue therefore we would lose good vids....?

it's a mess, the queue no longer filters the cream, the definition of cream is now so broad with so many users that "quality control" is going out of the window in favour of "catchy title control"

i don't pretend to have a solution, but it's a problem. i no longer feel that sense of validation when a vid gets out of the queue, it's like "whatever" rather than "hey! the community liked my choice!"

there is no longer such a strong connection between queue-escape and quality.

and, lucky WANTS TO BE ME which is why he changed his avatar to basically MY AVATAR.

nice one lucky, i always knew you had the hots for me

Yours is a very valid point, arvana. This is why I was wondering out loud if we shouldn't be actively downvoting queued vids that shouldn't be published, as opposed to just downvoting vids that you completely hate (which could be how we could continue to treat downvoting of already published videos).

If not that, then maybe we should set a maximum time from initial submission after which you can no longer requeue and it will be discarded? Sure, the submitter could just resubmit after that, but then they'd be starting from scratch with zero votes again.

Sorry for the confusion regarding unique view count. I just realized I should just add the separate unique member view counter to that hover text information. I'll do that shortly.

I think it's too late to suggest a change to Sift Culture that is as deeply ingrained as our downvote behaviour.

My suggestion would be to limit the number of requeues, in the same way as promotes are limited now. Say a maximum of one requeue every 2 days.

lucky blanks me again!
it's getting like ant's downvotes.

i wouldn't mind, but he won't admit he secretly loves me.

talk to me lucky! we look so good together in our matching jackets.

The thing is the queue lifetime is 2 days, so we can't do that, but even if we could we would still have the same problem of infinite requeues. I think it'd be best to only allow requeues up to like 4 days (double the queue lifetime) after the initial submission at which time the video will be discarded.

Another more complex algorithm I am considering which doesn't limit the number of requeues would be something like this: Instead of a requeue always resetting the uniques to zero, it will decrease the unique views based on how long ago the vid was submitted. For example,

• requeue at under 24 hours old = the unique views will be reduced to zero to give it a whole new chance at the top of the queue
• under 36 hours old = 25% (e.g., 12 uniques before requeue, 3 after) so it's near the top but not at the top
• under 48 hours old = 50% (e.g., 12 uniques before requeue, 6 after) so it's not quite as near the top
• over 48 hours old = 75% (e.g., 12 uniques before requeue, 9 after) only a small bump forward in the queue toward the top

I like it!

I don't think lifetime in the queue is really all that relevant. Perhaps limiting the requeues with the number of unique views would work better. In other words, if I requeue something 10 times but it still only has a few unique fews, it should have precedence for a requeue over a vid I just posted (never requeued) with 20 views. More directly, you could use the votes/views ratio. In fact, you could use that as the discarding criterion instead of time altogether. If a post falls below a ratio of, say, 0.1 (9 out of 10 people wouldn't upvote), then it's a pretty good bet it's not quality. (Might need to weight downvotes lighter in that case.)

Lucky, I downvoted the proposal above because it strikes me as overly complicated. It's an interesting idea, but I think will end up causing lots of confusion.

Also, I thought I should point out that over the last week, I have used requeue quite successfully in getting new views on my vids. Every time I requeued, I did so during high traffic times, and I was always able to get at least another 3 views on the post within a couple hours. I think the implication is pretty clear. Most people still prefer using the sort by newest option.

It may seem complicated, but in actuality, common users need not know the algorithm for it to work. The bottom line with my "back off" proposal is that a video will start at the top and as time goes by, each requeue will place a video first near then further and further back (but always nearer than before the requeue). This means that your video has had a few chances at or near the top and was never published, so eventually it will have very little chance to get published.

This insures there is some actual sifting happening. I.e., it won't be guaranteed that every video will get published if you just bide enough time. It's like a votes-to-views ratio except takes into account longevity and staleness.

Am I missing something? I don't see how the number of unique views is taken into account for whether a vid gets discarded. I mean I see what you're doing with the unique views to help push a post back to the top of the queue, but ultimately a post will get discarded regardless of how many views it gets. Since the goal is to improve the quality of vids escaping the queue (isn't it?), it seems to me the only way to accurately gauge that is after some number of people have actually viewed a particular vid. I don't see how the amount of time a vid spends in the queue is any measure of its quality.

Let's say I've got a vid in the queue with only 3 unique views. I requeue within the first 24 hours and it goes back to the top, but after another 24 hours, it still only has 6 total unique views. Now I requeue again, and the unique view count is set back to 3. So, I'm right back where I started. And now as time goes on, it gets harder and harder to get eyeballs on the post, despite the fact noone has bothered to look at it. Ultimately, the post gets discarded with hardly anyone seeing it. That doesn't make sense to me.

I know what you mean about users not needing to know the algorithm (you don't need to know how it works, just that it does work), but they're still going to pepper you with questions.

The other thing about complexity is that it's harder to code. Simpler incremental changes are generally better, but maybe that's just my coding style.

Bottom line - I'm up for trying anything to improve queue efficiency.

The thing is that if we did it as you're suggesting, every subsequent requeue would always do nothing to help a good video get published, just give it more time to sit far back and not get viewed.

With what I proposed, a video will have a couple of good chances to get published because it will get more exposure initially. If after the initial big jumps toward the top it still isn't published, it's probably not good enough so it will slowly fade backward into obscurity.

The number of unique views isn't directly being taken into account for whether a vid gets discarded, but it is used indirectly by determining how well it places after a requeue. If a video has not had many views, then after its first requeue it will be popped much higher up than another video that is: 1) also on its first requeue but has more views; 2) on it's second requeue and with fewer views.

So, in essence, if your vid hasn't been viewed much in the first few hours, it gets priority so more people can have the chance to see it. If a while later still not many people have viewed it, it will continue to get high priority. But, if a lot of time passes and not many people have viewed it or gotten it published, it's likely just not interesting enough to garner any attention.

^ it makes sense, its not that complicated (by algorithmic standards) and it is appropriate and probably necessary for our current system.

Its not like any poster/submitter/re-queuer has to know whats going on in the background, just that there is something reasonable/practical/sensible and meaningful going on.

damnit, i jost got several paras into a detailed alogryhtm discusion and lost the post by some miss type (not sure what key or how but the browser reloded and lost it). got a headach. ill come back to this.

>> ^lucky760:
The thing is that if we did it as you're suggesting, every subsequent requeue would always do nothing to help a good video get published, just give it more time to sit far back and not get viewed.

Well, I wasn't suggesting any particular solution, just that I think the solution should consider the # of views instead time. I think I see what you're talking about though. And surprise, surprise, I disagree. That assumes that everyone (or at least the majority) are using the default sorting option, which I don't think is the case (see my earlier comments about the vids I requeued this week). But even with the fewest views option, if a vid only has a couple views, it doesn't need any help getting on the front page - it's already there!

With what I proposed, a video will have a couple of good chances to get published because it will get more exposure initially. If after the initial big jumps toward the top it still isn't published, it's probably not good enough so it will slowly fade backward into obscurity.
The number of unique views isn't directly being taken into account for whether a vid gets discarded, but it is used indirectly by determining how well it places after a requeue. If a video has not had many views, then after its first requeue it will be popped much higher up than another video that is: 1) also on its first requeue but has more views; 2) on it's second requeue and with fewer views.
So, in essence, if your vid hasn't been viewed much in the first few hours, it gets priority so more people can have the chance to see it. If a while later still not many people have viewed it, it will continue to get high priority. But, if a lot of time passes and not many people have viewed it or gotten it published, it's likely just not interesting enough to garner any attention.

I definitely agree with you about the effects of your proposal. But it says to me that the title, thumbnail, and/or username then is at least as important as the content itself in judging its quality. I can't really say I agree with that, but I wouldn't argue the point, because that's not really my call. And I remember older talk posts discussing the importance of good titles. Just want to make sure I'm on the right page.

What constitutes a "unique" user, praytell???

You're definitely a unique user, choggie.

I believe every video deserves at least one chance to requeue, so if you're determined to change the requeue rules then that's what I advocate. Believe me, it's hard for me to even endorse that, but a one-requeue rule won't be the end of the world. Adjusting queue time from 2 to 3 days might be in order if we adopted a one-requeue limit. Something to consider. Hopefully a one-requeue limit will encourage more people to check out the unsifted pages, because they'll know that no queued video has an indefinite shelf life, so if they want to see something that they like to be published they better get their asses over to the queued videos and start voting.

Hopefully, but probably not. I wish there was a way to encourage people to visit the unsifted pages more often.