When I noted the queue, I saw there were new problems in the queue, and all the old problems that were there got bumped down. Theoretically, if 2 or 3 people "control" 2 spots each in the queue, that narrows it down to 5 for everyone else, especially if their problems bump others' down....

My idea for this would be to not allow problems to get bumped down. Any queue-weight calculations resulting in problems moved to the queue should get spots 11, 12, 13, so on so that they don't bump the existing problems out of the way.

Currently levik only uses two queue weights, so under that suggestion, if a person submitted 10 problems, after the first 2 (queue weight 1) no more of his problems would show up until everyone else's first two showed up. If someone else submitted 10 problems, but spaced them at 2 per month, he could prevent the first person's problems from ever showing up. Also, once all the QW=1 problems were finished, all the QW=2 problems would become QW=1 and we would have the same problem as before QWs were instituted.

To properly implement your suggestion, levik would need to create a hierarchy of QWs: the first two get QW=1, the next two get QW=2, etc. And still the first problem mentioned above will not be prevented. I'm not sure he wants to go to all that trouble for a dubious gain.

>> It is still possible to stall the queue under my suggestion, but I think it's a bit less likely than you're making it out to be. It would require a number of people submitting problems at a trickle's pace such that they always had no less than one and no more than two pending submissions at any given time. It would also have to be done by enough submitters to fill the visible portion of the queue.

I was not suggesting that the entire queue would often be stalled, but that someone who submitted more than two puzzles at a time might be disadvantaged to the point where his third puzzle might never rise to the top ten. Your solution might work if levik changed the QW system as I mentioned would be necessary in my last post (and which you apparently assume is how it is already).

>>Re: your second paragraph: Isn't that what he does already? Or am I missing something?

As I said in the first sentence, currently levik only works with two queue weights: the oldest two problems by each submitter are QW = 1 and all the rest are QW = 2. Since that has the effect of limiting each submitter to two puzzles in the top ten at any time, it is all that needed to be done to achieve that goal. To add a complete hierarchy of QWs, whether recalc'ed every night as is done now, or even following your suggestion, would involve much more coding, and the gain would seem to be so minimal.