Report and infraction statistics 2011

The_J

Say No 2 Net Validations
Administrator
Supporter
Joined
Oct 22, 2008
Messages
42,137
Location
DE/NL/FR
NickyJ asked some months ago, if we did not have any statistics about infractions and such stuff. No, we didn't, but I was interested myself in this, so I just did one. I hope this can give us some overview...or at least something a bit interesting to read :).
Please note that there are no names of single users present.

Reports



In the year 2011 all users created together 168 pages of reports, which are in total 7292 reports, with an average of 20 reports per day and a standard deviation of 8.27.
Additionally 36 private messages and 44 visitor messages have been reported.

We got the most reports with 53 on August, 5th, and 50 reports on February, 19th and we had three other days with more than 40 reports,
Lowest number of reports we had on December, 2nd (2 reports), and around the new year (December, 31st and January, 1st, with 3 each). No day passed without any report.

Overview (attention: big! Better to open the image separately).
Please note that the graph is a 7 days moving average graph, so every point is the mean of the datapoint and the following seven days.
Spoiler :
attachment.php


Reports per forum

We got reports from 144 different forums, 36 forums of these with only single reports (mostly modding or DG subforums).
The vast majority came from the off-topic forum, with 3718 reports (51%), followed by Civ5 General Discussion with 1361 reports (18,7%). Only 6 othe forums had more than 100 reports (from low to high: Civ5 - Strategy and Tips, Humor & Jokes, Civ4 - General Discussions, Arts & Entertainment, Site Feedback, All Other Games), lowest with 107 (1,5%), highest with 211 (2,9%).

Most unusual forum for a report: The administrators forum (1 report).

Below a table of all forums with 50 or more reports:

attachment.php



Reports per user

The 7292 reports were a result of 785 different users who reported at least one post.
343 users reported only once a post, 129 users only twice, leaving 313 users with three or more reports.
The most active "reporter of posts" reported 799 posts (11,0%), the second most active 499 (6,1%) and the third most active 261 (3,6%). Six other users reported more than 100 posts, 25 other users with more than 50 posts.
The mentioned top 9 reporters make up 30% of all reports, the top 34 reporters 54%.

Infractions



All moderators together generated 4837 infractions and 1014 warnings. Ignoring warnings, an average of 13,2 infractions were issued per day, with a standard deviation of 16,2.

Infractions per day

Regular users
The highest infraction count for regular users was on Mai, 12th, with 36 infractions, followed by Mai, 22nd (29) and June, 2nd (23). There were 65 other days, on which 10 or more infractions were issued.
We had 11 days, on which no infractions for regular users were handed out (6 in December, mainly end, and 3 in November), and another 15 with only 1 infraction.

Bots
Most infractions were issued at June 16th/17th, with 266 and 95 infractions, respectively. That were the days when it was discovered that bots with only homepage links and no posts are hiding in the birthday calendar, and Camikaze and Knight-Dragon cleaned it out.
We had three days with more than 30 infractions, all in July, and 12 more days with more than 20 infractions.
On 16 days no bots were banned via the infraction system, mainly in the first quarter of the year.

Preemptive banning of advertisers was stopped at the end of October, the IP block list from StopForumSpam was installed at the beginning of November. So please notice the last peak in the infractions at the middle of October.

After the OT rules were clearified and the RD discussions were started, an overall drop in the infractions can be noticed, with a rise in between (probably due to the testing how exactly the rules work now).

Overview (attention: Big! Better to open the image separately).
Please note that the graph is a 7 days moving average graph, so every point is the mean of the datapoint and the following seven days.
Spoiler :
attachment.php


Infractions per type

186 different types of infractions were issued.
From these 136 were only handed out once. In this category mostly fall custom infractions, e.g. combined flaming and trolling.
From the 4837 infractions 2526 (52,2%) were issued for "Spammed advertisement" (so ad bots). Most infractions for real users were (minor) trolling with 524 (10,8%) and trolling (420, or 8,7%).
Three types of infractions were handed out more than 100 times (Spam, Minor Flaming, Flaming).
16 infractions related to piracy were issued, 47 for double logins.
1 infractions was handed out for the abuse of the report system.
Infractions which can lead to a direct temporary ban (advocating (mass) murder, death/violence threads, encouragement of suicide or self harm) were issued 9 times.

Below a table with everything which was issued more than 10 times:

attachment.php



Infractions per user

3093 different users received at least once infraction. 2774 got one infraction (if "Spammed advertisement" is substracted this makes 248; note that bots can accidentially also get more than one infractions, so a slight inaccurary here), 102 got two infractions.
The two top offenders reached each 42 infractions. 15 other users received 20 or more infractions (two permantently banned in the meantime), 44 others with more than 9, but less than 20 infractions. The group of people with more than 20 infractions consists mainly, but not exclusively, of OT users.

Infractions per mod

Most total infractions were handed out by "Mod X" with 1535 (31,7%), followed by "Mod Y" (1203 infractions or 24,9%). Two mods issued more than 200 infractions and five other mods more than 100.
Most spam infractions were issued by "Mod X" (1263), "Mod Y" (600), and "Mod Z" (174). No other mod handed out more than 100 spam infractions. Note that most infractions were probably caused by profile infractions for bots without any posts and that this practive has been stopped.
Three mods, which have infracted at least one poster, did not ban any bots.
In contrast to this, 2 mods nearly exclusively infracted bots.
Please note that not every permaban is done with an infraction, so that some are missing here. At least one of the posters who seems not to have infracted any bots banned a couple, but did this in the other way.
There is a change in the places if the spam infractions are substracted.
The number of infractions for real users is lead by "Mod Y" (603), followed by "Mod X" (272) and "Mod A" (211). Only two other mods handed out more than 100 infractions for regular users (189 and 111, respectively).
 

Attachments

  • reportsPerForum.jpg
    reportsPerForum.jpg
    35.7 KB · Views: 538
  • InfractionTypes.jpg
    InfractionTypes.jpg
    48.7 KB · Views: 981
  • separated_overview_infractions_moving_average_+7_mod_public.png
    separated_overview_infractions_moving_average_+7_mod_public.png
    396.3 KB · Views: 475
  • overview_reports_moving_average_+7_mod_public.png
    overview_reports_moving_average_+7_mod_public.png
    305.2 KB · Views: 503

Methodology



Reports
All pages in the reports subforum which contained threads from the last year were downloaded.
Poster and subforum were extracted from the thread titles. This ignores the cases in which one post had been reported by more than one person, and also ignores the cases when threads were merged. Since this does not happen too often, this amount of data was considered negligible.
The date of the report was extracted from the date of the last response. This is not necessarily the same date, but this is again a) either negligible or b) has the same distribtution over all posts, and so probably compensates possible shifts.


Infractions
User name and type of infraction were extracted from the thread title, the responsible moderator was set the same as the thread creator.
The date of the infraction was determined from the last post. This ignores the cases when infractions were reversed, and does not take into account responses to the infraction threads. Since this does not happen too often (again), it can be neglected.
 
Thanks! That was an excellent read. :goodjob:

It's interesting how reports and infractions dropped off in mid-December. Christmas spirit at work? ;)
 
If somebody is interested in something else (which can be answered without user names), then let me please know :).

But user names would be the most interesting thing, especially the names of the moderators. Although I can probably guess who Mod Y is anyway.
 
I think that usernames should and will be left out.
 
CivV General Discussion is adorable. What was the reported post in the administrator forum? :O
 
I was unaware that CFC still gave out infractions for flamming, I thought they had just cut it down to "everything is trolling" :p

Very interesting stats, I approve of this. Although I still think being able to view all of the public actions of a moderator would be highly beneficial to everyone.

EDIT: I think you missed a few troll accounts.
 
The most active "reporter of posts" reported 799 posts (11,0%), the second most active 499 (6,1%) and the third most active 261 (3,6%). Six other users reported more than 100 posts, 25 other users with more than 50 posts.
The mentioned top 9 reporters make up 30% of all reports, the top 34 reporters 54%.

Holy crap, I felt like I reported a lot of posts, but I might have reported a couple dozen, nothing nearing 800.

Poster and subforum were extracted from the thread titles. This ignores the cases in which one post had been reported by more than one person, and also ignores the cases when threads were merged. Since this does not happen too often, this amount of data was considered negligible.

I'm surprised multiple people don't report the same posts more often.


Friendly fire incident. :mischief:

There's at least a couple times where I've reported what I meant to post, or vice-versa, from not paying much attention to what I was doing.

Very interesting stats, I approve of this. Although I still think being able to view all of the public actions of a moderator would be highly beneficial to everyone.

I'm available for hire to write a script to scrape all this info for you, if you're interested.
 
Although I still think being able to view all of the public actions of a moderator would be highly beneficial to everyone.

Why? I think that will result in mods getting reputation as being an active infractor/banner etc. which might lead to annoyance towards those mods which will interfere with their "regular" CFC activity. People getting biased against them just for what Mods are supposed to do and such...
What also might happen, thinking the worst of people, is that these "bancounts" become a statussymbol for mods. leading to "overenthusiastic" Mods.

doesn't seem desirable to me..
 
I'm surprised multiple people don't report the same posts more often.

A lot of people subscribe to the idea that more than likely, someone else must have reported the post already so why bother?
 
Why? I think that will result in mods getting reputation as being an active infractor/banner etc. which might lead to annoyance towards those mods which will interfere with their "regular" CFC activity. People getting biased against them just for what Mods are supposed to do and such...
What also might happen, thinking the worst of people, is that these "bancounts" become a statussymbol for mods. leading to "overenthusiastic" Mods.

doesn't seem desirable to me..
Moderators already have reputations, so that isn't going to be a new 'problem'. The problem is it is impossible to see if that is how they always operate or if that was just a one off thing. The mods can also already see their own lists of actions since every infraction has a thread made for it. If a mod did see a public listing of their actions as a way to grow their e-peen and go on a total power trip then it would clearly show in the list which could very well cause many users to complain. I would also hope the CFC staff would do something about it too before it becomes a problem...

A lot of people subscribe to the idea that more than likely, someone else must have reported the post already so why bother?

The only mention of this I've seen is a lot of people reporting the same spam posts
 
What was the most reported post that wasn't infracted? E.g. if a post had 10 reports but you didn't infract. What post was that? I'm assuming that isn't PDMA since, well, there was no action :mischief:

What fraction of infractions came from reported posts vs posts that mods just stumbled upon and found infractable?
 
:hatsoff:

What was the reported post in the administrator forum? :O

Most probably a test report (did not check it).
If changes were made to something in the system, the admins make somewhere a test report. We also have some from the moderators forum.

EDIT: I think you missed a few troll accounts.

Since a big bunch of true troll accounts are also DLs, the number is not totally correct. And I've also kicked some as "spammed advertisement", so there's again a loss.

I'm surprised multiple people don't report the same posts more often.

It's only my impression, there's no number behind that.
But in general, unless something is really outrageous, there are rarely double reports.


What was the most reported post that wasn't infracted? E.g. if a post had 10 reports but you didn't infract. What post was that? I'm assuming that isn't PDMA since, well, there was no action :mischief:

:dunno: no numbers for that.
Recently a post was reported 3 times by 3 different members, and it was not acted on it, because the thread was necroed and the post was already way too old.
That was probably the most.

What fraction of infractions came from reported posts vs posts that mods just stumbled upon and found infractable?

Can't say that, it would mean to download all the reports, and I don't feel like stressing the server (because it would be 7200 report threads and the same amount of the related posts).
And like mentioned, a lot of bots have been kicked without having ever posted, so there's already a given discrepancy.
Additionally, if a post is reported, most mods will take a look at the whole page and will infract more if necessary -> you can't calculate a relation out of that.
 
But user names would be the most interesting thing, especially the names of the moderators. Although I can probably guess who Mod Y is anyway.
Wasn't me! :p

The_J, what about the custom infractions - the ones where one post had two or more offenses attached to it? Did you try to make sense of those incidents? I know there were some that were rather confusing to sort out.
 
Back
Top Bottom