Dropping science on your moderator bias..... um...w0rd!

DeVore

So, a while back Conor and I were talking about measuring various health indicators of the site and moderator actions was one of them. just recently, he came back to me with a bunch of stats and two of them I would like to share with you in graph form.

One in the form of the absolute number of moderator actions (Infractions, Warnings, Bans)




But if you think of it, the rate of posting now is considerably higher then even a year ago, so we should graph them, not by number of actions but by actions-as-a-percentage-of-posts. So here that is:




I think you can see where this is going. So, I hate to rain on all you mod-haters out there but the numbers simply dont lie.

DeV.

seamus

Just to pre-empt the obvious question - there's no ban information before July 2008 because that information wasn't recorded until then.

AckwelFoley

How long is it been since Karoma has been a mod?


:pac:

A-Trak

How're warnings recorded and how do they differentiate from infractions?
(Sorry if there's an obvious answer to this I'm missing...)

DeVore

They are similar to infractions, effectively they are more red cards then the yellow cards of infractions. Its quite subtle though.

DeV.

pickarooney

DeVore wrote: »

They are similar to infractions, effectively they are more red cards then the yellow cards of infractions. Its quite subtle though.

DeV.

Huh? I thought the yellow card was the warning and the red was the 'infraction'?

Gordon

I think for the purposes of the graphs: "Infraction" means Red Infraction and "Warning" means Yellow infraction, and Dev is saying that there are more "Infractions" than "Warnings". I think Dev is drunk on stats..

I could be wrong though, I don't have the raw data.

DeVore

Could be, sorry, I never really give them out, stuff that lands on my pile tends to have gone through that process

DeV.

AckwelFoley

There should be no warns and infractions, just one or the other.

I feel that receiving either will do the same thing.. to me it was a wrap on the knuckles and didnt matter either way which it was as i go the message.

2cents ^^^

Wibbs

Interesting stuff It looks like infractions were dead fashionable for a while. How many of the bannings were shills and spammers? I'd say that would change the ban stats a fair whack? In a few forums anyway. I'd reckon personally I've banned waaay more spammers than other users.

The other bit I take from those stats are the users. All of us in each community are for the most part chilling, chewing the fat, having a laugh and informing others, making the day go easy. Rarely causing any guff on thread. When it does happen, then other users report it and again most of the time its resolved no bother.

Terry

snyper wrote: »

There should be no warns and infractions, just one or the other.

I feel that receiving either will do the same thing.. to me it was a wrap on the knuckles and didnt matter either way which it was as i go the message.

2cents ^^^

As a mod, I find the different ones come in handy when trying to determine previous bollix acting. It's faster than going through previous posts because it's very easy to find why someone has been given a red or yellow card.

I look at the yellow cards as minor infringements, which act as a warning, and the red for use when someone has really been screwing around.

Wibbs

snyper wrote: »

There should be no warns and infractions, just one or the other.

I feel that receiving either will do the same thing.. to me it was a wrap on the knuckles and didnt matter either way which it was as i go the message.

2cents ^^^

I dunno, I think a yellow card is more "official". Just speaking personally I'm not that sold on them and prefer either on thread chillax or a quick PM explaining why. Rather than "ohhh you've been a bold boy/girl". Most of the time the vast majority of people just slip up/have a bad day/post in wrong forum. Actual stirrers dont last long.

trout

It's hard to measure the impact of on-thread interventions, or friendly words in the ear via PM

Stats are always good to look at though ... makes me feel warm & fuzzy

xw2lj9uspm1eyh

seamus wrote: »

Just to pre-empt the obvious question - there's no ban information before July 2008 because that information wasn't recorded until then.

Thanks for stating that I was gonna ask did Gordon get his banhammer then I guess he has gotta wait till he bans 5,000 people before he get his progress plotted on a graph.

Tallon

Mmmn, nice to see stats

Bit strange that only mods thanked the OP

Gordon

mendusa wrote: »

Thanks for stating that I was gonna ask did Gordon get his banhammer then I guess he has gotta wait till he bans 5,000 people before he get his progress plotted on a graph.

Graph meeeeee!!!!

AckwelFoley

Terry wrote: »

As a mod, I find the different ones come in handy when trying to determine previous bollix acting. It's faster than going through previous posts because it's very easy to find why someone has been given a red or yellow card.

I look at the yellow cards as minor infringements, which act as a warning, and the red for use when someone has really been screwing around.

In my experience that doesnt pan out though, one mod could give you a red and another a yellow for essentially the same thing.

I preferred either your "shut the fcuk up" method or then a infraction and then wave "day day" for persistant acting the dick.

Most/many users would respond in positively to a quick pm - but thats time consuming. I do rather regularly pm posters, generally infrequent members that were baited by trolls and give them a "heads up" to delete a message they will receive a ban for, and always get a positive response.

AckwelFoley

Wibbs wrote: »

Interesting stuff It looks like infractions were dead fashionable for a whileHow many of the bannings were shills and spammers? I'd say that would change the ban stats a fair whack? In a few forums anyway. I'd reckon personally I've banned waaay more spammers than other users.

.

Very good point, but easily calculated id say.

I would assume spammers is a constant, if you could calculate from 1 week twice in either end of a year how many were spammers, and subtract the average of the weekly results from the rest of the figures it would give you a better "number" but the flow of the graph would remain the same

trout

snyper wrote: »

I would assume spammers is a constant

I wonder how true that is ... I've seen waves of spam, and then nothing for days / weeks ... I'd say over time it's increasing too ... but there are probably spam filter stats too.

I'm a junkie for stats ...

Mr Magnolia

Could rule changes in adverts also have affected the stats?







Also, anyone else feel violated?

:pac:

Économiste Monétaire

Be super nice, DeV, and attach the data, too .

OldGoat

Mods, bah, damned shirkers. Never earn the coke and hookers. Back in the day you'd be banned for breathing heavy and now it's all nice whispers in yer ear an Politeness. The numbers don't lie!

DeVore

Économiste Monétaire wrote: »

Be super nice, DeV, and attach the data, too .

Can do, what do you want it for though? (I mean that both in a defensive way, and in the "maybe I help you if I know what you are looking for" way).

DeV.

Économiste Monétaire

DeVore wrote: »

Can do, what do you want it for though? (I mean that both in a defensive way, and in the "maybe I help you if I know what you are looking for" way).

DeV.

Nerdy things, like looking at/correcting seasonality, projecting the future path of bans, the correlation between infractions and unemployment, et cetera.

trout

*nerdgasm in 5,4,3,2 ...*

DeVore

I dunno if its in enough detail to do that kind of analysis, its pretty much just what is in those graphs but knock yourself out

DeV.

pickarooney

snyper wrote: »

There should be no warns and infractions, just one or the other.

Yep, and just call them warnings. No more of this 'infraction' rubbish. A red card doesn't actually do anything so it's symbolically redundant. Warn and ban - simple and to the point.

Sparks

Interesting that the drop in bans, warnings and infractions happened at the same time as a large sustained spike in traffic.

Gordon

Sparks wrote: »

Interesting that the drop in bans, warnings and infractions happened at the same time as a large sustained spike in traffic.

The eyes of Sauron were on the hacker of 01/10.

FarmerGreen

DeVore wrote: »

I think you can see where this is going. So, I hate to rain on all you mod-haters out there but the numbers simply dont lie.
DeV.

Yes, users are beginning to understand the rules.

4zn76tysfajdxp

Gordon wrote: »

Graph meeeeee!!!!

copacetic

interesting stats, wonder is the fact that more infractions leads to more bans part of th the cause or just the effect?

It certainly seems that warnings and infractions lead to bans, or at least don't seem to diffuse situations?

DeVore

Gordon wrote: »

The eyes of Sauron were on the hacker of 01/10.

The hacker has more to worry about than Sauron.

Yes user are beginning to understand the rules, and it seems like a lot of people like having some form of rules around here.

So, where are all those people who were swearing this place had become colditz??? That it was freer in the "good old days"

Whereeeeeeaaaaaaareeeeeeeyouuuuuuuuuuuuuuuuuu???


DeV.

tallaght01

An awful lot of us just stopped bothering.

But the reason you have such wild fluctuations is because of the scale you're using. You're talking about differences in twentieths of a percent being represented as huge swings. On a different scale it would be less stark.

It doesn't bother me a bit. It's your business, and you should run it your way. But graphs like this are very difficult to interpret.

Some people went a bit mental in 2009. BUt aside from that, the figures have remained relatively stable, at first glance.

DeVore

You obviously don't understand the graph then. That percentage is the percentage of bans etc to total posts. Since the number of posts per month is very very large, moving that by 0.2% takes a swing from 200 bans/warnings to over five times that at 1000 bans/warnings.

So quite pointedly, what you are saying is not only inaccurate and wrong, it is in fact exactly the opposite of what the reality is. For the graph to move even on notch on the y-axis, say from 0.10 to 0.15 takes an increase of quite a bit....about 200 extra a month. Considering were currently running at about 500 for warnings, that means a fairly drastic increase to see only one notch movement.


I've supplied the raw data behind it too, so you can see for yourself.

DeV.

tallaght01

DeVore wrote: »

You obviously don't understand the graph then. That percentage is the percentage of bans etc to posts. Since the number of posts per month is very very large, moving that by 0.2% takes a swing from 200 bans/warnings to over five times that at 1000 bans/warnings.

So quite pointedly, what you are saying is not only inaccurate and wrong, it is in fact exactly the opposite of what the reality is. For the graph to move even on notch takes an increase of quite a bit.


I've supplied the raw data behind it too, so you can see for yourself.

DeV.

I think the point is that you were (rightly) trying to get away from raw numbers, as you're not comparing like with like, so there's no point in now reverting back to raw numbers to exlpain the differences in percentages.

So you looked at percentages, and the changes are tiny.

But, like I said, you don't owe me or anyone else an explanation of what your volunteers do. But that graph isn't as convincing as some seem to think.

I just noticed it, and tought I'd point it out.

For the record, the bio+med mods and the SDMA mods (the forums I read) are lovely. Even our dearly departed Locum-motion

DeVore

*beats head off keyboard*

I AM TALKING ABOUT THE PERCENTAGES!!

I've tried to clarify my post. I really can't do any more. The Changes in the percentages are small because the divisor, the number of posts is huge. It's quite correct to show it at that scale because to move the graph on that scale takes a tremendous movement in the number of bans/warnings.

DeV.

tallaght01

DeVore wrote: »

*beats head off keyboard*

I AM TALKING ABOUT THE PERCENTAGES!!

I've tried to clarify my post. I really can't do any more. The Changes in the percentages are small because the divisor, the number of posts is huge. It's quite correct to show it at that scale because to move the graph on that scale takes a tremendous movement in the number of bans/warnings.

DeV.

I know the point your making, so you can spare your head.

The point is that you're either talking the about raw numbers, which don't compare like with like, or percentages. So you chose pecentages. The percentage changes are tiny, but represented on a scale that suggest they're very big.

The point of using the percentages was to avoid having to use the raw data. Lots of things can be represented on % scales that have big numbers. But tiny %changes are still tiny changes.

Anyway, like I said, just a passing comment. You obviously don't agree. But I'm sure you'll agree it's not worth a long debate.

flahavaj

I think the big dip in mod can be directly related to the tital u-turn in behavious on the PW forum the last 6 months or so. What used to be an absolute minefield and a regular pain in the mods/admins respective holes is now an oasis of tranquility.

I'm only half joking too. That forum and its complete transformation should be held up for all to see as an example of the many benefits of even handed, friendly, ego-free moderation.

The Recliner

pickarooney wrote: »

A red card doesn't actually do anything so it's symbolically redundant. Warn and ban - simple and to the point.

Not entirely true though

A red card carries a point which contributes to a temp site ban if the user earns enough of them in a short time so they ahve their uses and diffferences

Eliot Rosewater

You could also use these stats as a kind of benchmark to see how naughty or nice of a poster you have been.

For the last month, there was a moderator action (warning/infraction/ban) for 0.32% of all posts (from the spreadsheet). Thus, an average poster with 1000 posts should expect to have roughly 3 wrist slaps. I, with 1600+ posts, would expect to have 5. (I've only 1, though I have been warned by mod posts "on thread" a number of times and gotten in more silly spats than that datum infers.)

Of course that's terribly inaccurate: I realise that infractions were only introduced a few years in, and that a lot of bans go to spam posters who aren't really community members, and shouldn't count when evaluating average normal user behaviour. It's still interesting though, warts and all.

DeVore

You would have to have posted your 1000 posts IN THE MONTH IN QUESTION to expect 3 disciplinings.

What you COULD say is that if you have 1000 posts, on average over your lifetime on boards, you should expect to have gotten 3 disciplines in that time. (I think this is what you are saying).

DeV.

pickarooney

...and the average poster with 10,000 (I mean, of those with 10,000 ish posts, take the average) should have 30 disciplines and probably have been sitebanned for cumulative cardings, but realistically I think the more you post the more likely you are to know the rules and not get punished.

Eliot Rosewater

Very true!

Further to the above I decided to ploy the total "documented" mod actions (warnings + infractions - bans) in one graph. It doesn't say anything more than the original graphs do, but, given there's only one plot, it makes the point a little clearer. I omitted the months where ban data wasn't available.



(Plotted using old school Gnuplot )

(I've a hangover today from dirty Aldi beer, hence the effort put into this!)

Ghost Train

Well here's a graph of warnings plus infractions as a percentage of number of posters

didn't include bans because if advertising and spam are are part of the ban numbers I don't see how it reflects on moderation

FarmerGreen

It could of course be due to posters temperament being affected by the weather
http://www.statusireland.com/data/charts/Mean-Monthly-Rainfall-Dublin-Airport.jpg

Nevyn

There are posters which have never been banned or infracted/warned.
I have no infactions at all.

And there are some who in the first few months get a few and that it, there are those who
after being on the site a few years then wander into the more stricter forums and then hit a new learning curve and then you have those who troll for the hell of it and have a a huge ammount of band and infractions and seem to just like running the gauntlet.

So trying to average out infactions/bans per poster just doesn't work, it usually tends to be a % of 'bad apples' rather then the average poster.

Nevyn

FarmerGreen wrote: »

It could of course be due to posters temperament being affected by the weather
http://www.statusireland.com/data/charts/Mean-Monthly-Rainfall-Dublin-Airport.jpg

I think what you are seeing is the fall off in posts from those in college,
every year this happens and then you have the seasonal September effect
and also as the evening get darker and people stay at home more they will look for entertainment which has a low cost.

But I get bans being down, infractions seem to head people off at the pass as it were and often infractions are given for a first offence, where as before the infraction system was in place a ban would be given.

I do think that boards has started to see it's own small verison of the
Enternal September* effect as broadband has reached most households,
but other then the drive by posters a lot of the new sign ups get the learning curve in between 3 weeks to 3 months.



*

Usenet originated among universities, where every year in September, a large number of new university students from the Northern hemisphere acquired access to Usenet, and took some time to acclimate themselves to the network's standards of conduct and "netiquette". After a month or so, these new users would theoretically learn to comport themselves according to its conventions. September thus heralded the peak influx of disruptive newcomers to the network.[1]

In 1993, the online service America Online began offering Usenet access to its tens of thousands, and later millions, of users. To many "old-timers", these "AOLers" were far less prepared to learn netiquette than university freshmen. This was in part because AOL made little effort to educate its users about Usenet customs, or explain to them that these new-found forums were not simply another piece of AOL's service. But it was also a result of the much larger scale of growth. Whereas the regular September freshman influx would soon settle down, the sheer number of new users now threatened to overwhelm the existing Usenet culture's capacity to inculcate its social norms.[3]

Since that time, the dramatic rise in the popularity of the Internet has brought a constant stream of new users. Thus, from the point of view of the pre-1993 Usenet user, the regular "September" influx of new users never ended. The term was first used by Dave Fischer in a January 26, 1994, post to alt.folklore.computers:[4]
“ It's moot now. September 1993 will go down in net.history as the September that never ended. ”

Terry

flahavaj wrote: »

I think the big dip in mod can be directly related to the tital u-turn in behavious on the PW forum the last 6 months or so. What used to be an absolute minefield and a regular pain in the mods/admins respective holes is now an oasis of tranquility.

Pfft.
It's entirely down to my presence there. I might not post much, but I'm always watching. Always watching.

I'm only half joking too. That forum and its complete transformation should be held up for all to see as an example of the many benefits of even handed, friendly, ego-free moderation.

Dude's not wrong.
There's only about 1 reported post a week, and that's usually just someone asking for a spoiler warning in a thread title.

Gimmick and Bounty Hunter, along with the regular users, deserve all the praise there. It also help that quite a lot of the people who used to cause trouble there are now sitebanned.