Blitz Queue Trial (Closed)

[Ythan]

Fewer live slots in games? I'm going to be pretty shocked if this is given as justification from dropping blitz but it's not like I'm paying taxes here.

[zoraster]

I don't know of a better way to measure the activity of the site.

[pisskop]

How do you even measure that?

[Ythan]

Things got less active so nuke the fast queue?

I mean yeah, people who would be playing something other than blitzes

because they had to before

probably did stop playing large themes.

[zoraster]

I go through each game manually. It's a process that generally takes me about an hour.

[zoraster]

Things got less active so nuke the fast queue?

I mean yeah, people who would be playing something other than blitzes

because they had to before

probably did stop playing large themes.

Well right now, we're mod limited, not player limited from appearances in Large Themes. We only had two large themes go into sign ups in November I think.

But yes, people's preferences play a role in our decision-making, but only a partial one.

[Ythan]

But yes, people's preferences play a role in our decision-making, but only a partial one.

Oh I'm extremely interested in hearing any other factors you may be able to share.

[zoraster]

Off the top of my head: longevity of the site, activity, breadth of player base, diversity of offerings, quality of offerings, enjoyment of games played.

[Ythan]

Well thanks.

[zoraster]

Sure

[Rob14]

I think we have to consider that the "sign-up time":"average playing time" ratio will be MUCH higher in blitz games by nature of the queue. Once the game starts, it moves quickly, but the queue appears to move no more quickly than any other queue. It could be the case that players are having more down-time between games relative to time spent in games, which means less alive slots across the site.

As an alternative metric, we could perhaps use total player slots in all games started in a month. That's relatively easy; you can just look at the queues. If that were tried and you did not see a similar decrease, then it would indicate the nature of blitz games is causing the "live slots" metric to fall rather than an actual decrease in the amount of games being played/the amount of people signed up for those games.

[pisskop]

I see no more replacements than usual in my games. And if thats an adequate measure of 'active' then.


we should also take into account the holidays, finals, and etc.

[zoraster]

I think that would be an interesting alternative statistic, but it would be less useful for my purposes.

Simply having lots of short games isn't that great. If games aren't going through sign-ups quickly enough to maintain the previous activity, it's not a positive thing.

[Rob14]

It depends how you define activity. I would consider playing 3 blitz games vs 3 non-blitz games should be considered the same activity. One is just more condensed.

The way you define activity, we'd have to play substantially more games overall as a site in order to match the previous "activity", given that blitz games finish quickly. I don't think that's reasonable, but the opposite point-of-view is perfectly valid. I guess it depends what our goal is - games played vs players currently playing.

And then there's a third option for activity that can't be captured by any metrics - time spent by our players on-site playing Mafia. In that sense, one blitz is probably more activity than three non-blitz by virtue of the effort you have to put into a blitz to keep up. Difficult to measure that, of course.

[Ythan]

I mean do we get rid of the Blitz queue because it's faster than other queues if we're talking about its impact on activity?

[quadz08]

I think players currently playing is the more useful goal in terms of keeping the site going in the long-term.

[tn5421]

I think we need to extend the testing time because the current test runs straight through a national holiday which, I believe, negatively impacts site activity enough as it is.

[Ether]

For what it's worth, I did ask around, and two of the people I tried to recruit said they would have been up for a blitz at a different time of year but couldn't do it now.

I'm glad the trial was extended. I am hoping to play one, eventually.

[Rob14]

There is a significant and troubling break from trend in the graph Zor posted even before the blitz trial. We normally see increasing activity in summer, but we didn't this year.

[zoraster]

Keep in mind that this only happens when I run a census, which is its primary weakness. I didn't run one this summer other than August 27th. It's possible that, particularly during peak Team Mafia, that we had an uptick. Theoretically you could figure this out, but it'd involve delving into every game's vote counts and what not, which is a much, much more labor intensive job.

[Rob14]

Yeah, that's somewhat unrealistic to do. I'm surprised you've even done this much data collection over the years.

[Rob14]

Another question to consider is whether the trial itself has had a permanent effect whether or not a blitz queue remains. I would consider running a game using blitz rules even if a blitz queue did not exist. I would not have done so prior to the trial.

[zoraster]

Well, that's certainly something that I want to consider post trial. Even absent the blitz queue, it'd be great if people kept offering some blitz games.

Really though, and I've alluded to this before, the whole queue system needs an update. I don't want to speculate too much on what that will entail before the trial, but there will always be opportunity to run fast games, small games, big games, etc. the question is really just whether we build an entire queue around it.

Right now, my primary concern is that we're not getting enough mods. We've actually stayed steady in Newbies since 2014 (though much lower than at our peak), but we're not converting enough into mods.

[Ythan]

I'm likely to sign up for more games in a slower environment because they'll add up the the same total activity for me.

[BBmolla]

lmao @ large normal