Mef wrote:...
A while back in one of these discussions I went through and totaled up the statistics on a month of admin activity. Given that the only things that are logged are events that actually require using admin powers (bans / eventing games / approving avatars / etc). That's where the 30% figure and my ballpark estimate for 100 bans/week come from.
OK, so to clarify, the items that this statistic covers are (as from your post):
* Managing KGS+
* Approving user avatars
* Announcing and flagging event games
* Deranking rank cheaters
* Banning
So for a particular month, some time in the past, you counted all of the log messages for each of these categories, and about 30% of that count was for banning. Given that your count of 100 bans/week was also from that sample, we can infer that there were approximately 400 bans in the month.
Minus any error in your approximation, this implies that there were at least 933 logged events from the categories that month:
* Managing KGS+
* Approving user avatars
* Announcing and flagging event games
* Deranking rank cheaters
From here on out, I am going to make a hypothetical analysis, which I believe to be fair. It's all I really can do because, as speedchase implied, the information about your particular sample can only really be known by you and anyone with whom you shared it.Looking at the
KGS Event Calendar, I see two events for the month of March. From the referenced SL page, the maximum number of events in a given month were 8. This is not comprehensive data, but let's say that there were more than the max number of events, say 10 events for the month of your sample. In my experience on KGS, event games are broadcast once, or maybe twice. I don't know what's involved in "flagging event games", but let's be generous and say that there are 10 things that an admin has to do for a given event game. That gives, at an optimistic maximum, 100 logged events in a given month.
This means that there were at least 833 logged events in that month for:
* Managing KGS+
* Approving user avatars
* Deranking rank cheaters
On KGS right now, there are 22 users marked with [-] and 19 with [?]. A deranked user cannot have a rank by his name, so assuming that half of the people logged onto KGS right now with [-] or [?] is a rank cheater, that's 20.5 users that have been deranked. I've been deranked by BigDoug before (not because I was a "rank cheater", but because he assumed I was cheating when I referenced my wife's KGS account), and even though I was not "rank cheating" at that time, it still took a couple of weeks for me to not be "deranked" (which was silly in itself!). I suspect most individuals that are deranked are deranked for longer than that, but let's assume that each of those 20.5 users from above that are "rank cheaters" right now are on average deranked for 2 weeks. Then in a given month, on average, let's say that there are 41 people that get deranked.
This means there were at least 792 logged events for:
* Managing KGS+
* Approving user avatars
It's approaching the end of March, and there are 13 KGS+ lectures in the lecture history. The month is about 74% over (23/31 for March 23rd). Extrapolating evenly to the end of the month, there's 17.52 lectures this month. Let's round it up to 20. I do not know what is meant by "Managing KGS+", but let's say that you have to do 5 logged events for a given KGS+ lecture. This gives 100 events a month for "Managing KGS+".
This leaves 692 logged events for:
* Approving user avatars
Summarizing this analysis points to the following approximate distribution:
If we measure admin utility by the number of events logged in this manner, clearly, the greatest utility is achieved in banning and approving avatars.
This analysis is obviously based on my personal
hypothetical estimates of the categories that you listed, to the aggregate of which banning constitutes at most 30%. Still, even with real "non-hypothetical" data, a sample size over a single month cannot be that much better of an estimate of the true "banning to other-activity" ratio.
A measure of logged events is also not a real indicator of the amount of effort or value that an admin gives to the server. For example, if approving a user avatar takes a couple of seconds of time (I've had avatars approved this quickly), the effort required to do this is clearly small. The benefit added to the server also cannot be measured simply by log messages that this was performed by a particular admin. Who is to say that the common public even agrees with the admin's discretion as to what images are "appropriate"? I've seen various approved avatars that are not family friendly in everyone's eyes (eg. women in scant clothing).
And this doesn't even begin to touch on the point that an aggregate measurement of admin activity does not imply that all admins perform a uniform distribution of the same types of tasks.
tl/dr: My arguments here are somewhat in jest, but only to illustrate the point that the question of an admin's value to the server, and/or the effort they exert, cannot be taken seriously from a 30-day sample of log files. Likewise, just throwing around a stat like that to try to strengthen the argument doesn't achieve its intended purpose.
