Polls for possible adaptations to the level design contests.

[IchoTolot]

I started the first poll regarding the number of votes for the rule voteoff. Be sure to vote. 

[Simon]

Shall more than one vote be granted in the voteoff of the single rules depending on the number of levels inside a rule?
Yes
No

"No" means what exactly? Always one vote? Always multiple votes? Different system (e.g., instant-runoff system where each voter submits an ordered list) instead of first-preference?

-- Simon

[IchoTolot]

Shall more than one vote be granted in the voteoff of the single rules depending on the number of levels inside a rule?
Yes
No

"No" means what exactly? Always one vote? Always multiple votes? Different system (e.g., instant-runoff system where each voter submits an ordered list) instead of first-preference?

-- Simon

No should always mean one vote - I clarified the options.

[namida]

I definitely think the number of votes should be roughly proportional to the number of levels being voted between. You can't completely eliminate the impact of the whole "level some people like a lot and some people don't vs level everyone kinda likes" thing in a voteoff of any kind, but I think avoiding forcing a choice of one level out of many (unless it's down to the last two, or maybe three, levels of the entire contest) goes a long way towards reducing it at least.

[Crane]

My original proposal was that if there are 4 or more levels in a list, then everyone gets 2 votes so level designers who naturally select their own level (unless they're feeling extremely generous) can also choose a level that they very much enjoy.

Whether only 1 level gets through or not I left up in the air.  I imagine if there are, say, 8 levels or more, then things might get switched up again.

[Proxima]

I think the important thing is to disentangle the two questions. The number of votes everyone gets to cast does not have to equal the number of levels that make it through to the mixed group. If we separate those, then we can make meaningful decisions on each one, and we won't have a tug-of-war where both want to change and drag the other with them.

My thoughts on both:

Only one level should go through, except in case of ties. If we keep the "3 rules, with a mixed group to determine the overall winner" structure (and that structure has worked well and I don't see a reason to change it), then the ideal is one voting round per group and one for the mixed round. We have to have some system in place in case of ties, but if we go back to multiple levels going through then we have more voting rounds even in the best case, with even more if tiebreaks are needed, and we risk a return to the bad old days of contests being bogged down in an unending stream of votes.

(In case anyone doesn't know or doesn't remember how bad things used to be, the contest that provoked the discussion that led to us changing the system had eleven votes to select a winner out of twelve levels.)

It's true that with only one level going through from each group, the second-best level in each group can't get a prize even if it's second-best overall. I'm fine with that. "Best" is subjective anyway, and the important thing is to have a voting system that does the job of selecting a top three that is in some measure representative of the community's preferences, without the voting being too complicated or taking too long. The phrase "the perfect is the enemy of the good" comes to mind.

As someone who occasionally enters contests but usually doesn't, yes, I would like to be able to vote for more than one level in the rule groups, if there are at least 4 entries. There being a lot of high-quality levels is more common than not, and it's great to be able to recognise this by voting for my favourite two (or even three in a large group); then even if they don't win, I feel at least I was able to give them my support.

[IchoTolot]

I am in line with Proxima here.

The general idea would be:

The number of votes will go up with the amount of entries inside a rule. Still only 1 level per rule will survive though!

This would still keep the number of  voting rounds the same as before - 1 round for the rules - 1 possible tiebreker there - and then the finals between the last 3 levels.

Everything will still be over in 2-3 rounds then.

Crane's proposal would roughly be impletmented - I will try to keep the number of votes at <= 50% of the entries in each rule. So at 4 we can have 2 votes and maybe around 6 entries I will raise it to 3.

I did not specify any strict rule in the voting for this to keep it a bit flexible.

[Crane]

If I had to give a mathematical formula for my proposal, the number of votes would equal the base 2 logarithm of the number of levels, rounded down.  i.e. for number of levels n:

floor(log₂ n)

• At 1 level, it's 0 votes because that's the only one that could possibly go through.
• At 2 and 3 levels, it's 1 vote.
• At 4 to 7 levels, it's 2 votes.
• At 8 to 15 levels, it's 3 votes.
• And so on and so forth.

[kaywhyn]

All great points and feedback, thanks everyone! I completely agree with Icho that it should just still be one level survives regardless of how many votes we get. We haven't had any tiebreakers after the rule voteoffs for tied levels since the contest voting system overhaul, but I also agree there with the edge case that all levels move on if the tiebreaker ends in a tie too. In any case, the community seems to be in favor of knowing the results ASAP instead of having an endless amount of rounds of voting because of ties! The new voting system definitely works quite well with greatly reducing the number of voting rounds and, combined with multiple voting topics open, hence shortening the overall voting time tremendously!

Also Crane's proposal with the base 2 log on the number of levels to determine the number of votes sounds excellent!  

[Crane]

Watching the last vote-off carefully, Rule 3 was dead even between "Migraine" and "Quick Lab Rush" at 4 votes each, but then someone else voted close to the cut-off time to break the tie.  So ties are definitely possible.

[kaywhyn]

Watching the last vote-off carefully, Rule 3 was dead even between "Migraine" and "Quick Lab Rush" at 4 votes each, but then someone else voted close to the cut-off time to break the tie.  So ties are definitely possible.

Right, there wasn't ever a doubt that ties aren't possible at all under the new system. I was simply pointing out that tiebreakers so far haven't been needed, though to be fair it's only been two contests under the new way With more votes depending on how many levels there are for the ruleset, I suspect ties will be more frequent, especially if the maximum number of votes per user is used each time, but of course one can always cast less than the maximum amount of votes allowed if they want.

In any case, like Icho already mentioned, contests would still be over after a few rounds no matter what is decided here. 

[Crane]

Indeed, and regarding the above point, I don't know if someone voted at the last minute just to break the tie or if they legitimately left it late, but hiding the poll results until it closes will also help to avoid any possible manipulation.

[kaywhyn]

Indeed, and regarding the above point, I don't know if someone voted at the last minute just to break the tie or if they legitimately left it late...

Your guess is as good as mine. It could very well be for other reasons too. What's done is done, though. There's nothing that can be done about the results. Instead, let's turn our attention and focus on more important matters, such as things that will be decided by polling.

...but hiding the poll results until it closes will also help to avoid any possible manipulation.

Exactly, that's why I put forth the suggestion that poll results be hidden entirely until the closing of the poll instead of having them be shown to the user once he/she has casted a vote like it always has been done. That's the other option the poll creator can select when making one. This point will be up for polling once Icho makes the call for the current one, takes it down, and puts up the new one

[IchoTolot]

Ok, in a few hours I will start the next poll as this one should be decided - with 9-0 in favor of multiple votes. As for the number of votes I will go with Crane's suggestion as I like the distribution and I haven't heard any heavy points against it.

[IchoTolot]

I started the second poll regarding when the voting distribution shall be revealed.