13

Identification requests are a constant cause for concern in our community. Some say we should ban then altogether, some argue that we should just be more rigid in enforcing a set of guidelines, some agree with this last group but think we need to revise our current guidelines, and some think we’re perfectly fine the way we are.

Though it may not be determinant in making a decision, some more definitive numbers on identification requests and its users would help the community make a more informed decision, or at least allow the community to know these numbers. It would be good to know:

  • How many users asked an as their first question;
  • How many of these questions have been deleted and why;
  • When were these users created in relation to their first question;
  • How many of these users accessed our site after having asked their first question;
  • What are the scores of these questions;
  • How many of these users sticked around to ask/answer more questions.
16

All the following numbers are as of today’s date (July 21st, 2015).

We currently have a grand total of 1,593 undeleted questions tagged with .

Of these, 1,246 were asked by their owners as their first post, and 1,212 were tagged exclusively with 1.

This means that 347 of our currently undeleted questions tagged with were asked by users who had previously contributed to the site in other tags. Furthermore, there were an additional 321 questions tagged with which were first questions, and were subsequently deleted (11 of which had other tags too)2.

To put it in percentages, approximately 76.1% of the undeleted questions were a user’s first question (tagged only with ), which means:

Approximately 20.4% of our total undeleted questions were asked by (1,212) users that came here to have an anime or manga identified for them.

Of the 1,212 users whose first post was a (currently undeleted) question:3

  • 820 (67.7%) were created at the exact same time the question was created
  • 26 (2.2%) were created up to a minute before the question was created
  • 22 (1.8%) were created between 1 and 2 minutes before the question was created
  • 77 (6.4%) were created between 2 and 5 minutes before the question was created
  • 193 (16%) were created between 5 and 30 minutes before the question was created
  • 22 (1.8%) were created between 30 minutes and 1 hour before the question was created
  • 11 (0.9%) were created between 1 hour and 1 day before the question was created
  • 7 (0.6%) were created between 1 day and 1 week before the question was created
  • 27 (2.2%) were created over a week before the question was created

Of the 1,212 users whose first post was a (currently undeleted) question:3

  • 198 (16.3%) last accessed our site at the exact same time the question was created
  • 158 (13.0%) last accessed our site after the question was posted and up to 1 hour after that
  • 45 (3.8%) last accessed our site between 1 and 2 hours after the question was posted
  • 52 (4.3%) last accessed our site between 2 and 5 hours after the question was posted
  • 149 (12.3%) last accessed our site between 5 hours and 1 day after the question was posted
  • 247 (20.4%) last accessed our site between 1 day and 1 week after the question was posted
  • 168 (13.9%) last accessed our site between 1 week and 1 month after the question was posted
  • 188 (15.5%) last accessed our site over 1 month after the question was posted

Regarding the scores for these 1,212 question:

  • The lowest scored questions are at -5, and there are 4 of them
  • The highest scored question is at 26, with the runner-up scoring 17
  • The average score is 2.5
  • The median is 2
  • The standard deviation is 3.1
  • The sum for all the values is 2,987

Of the 1,212 users:4

  • 1,014 (83.7%) have a single post on our site
  • 121 (10%) have 2 posts on our site
  • 34 (2.8%) have 3 posts on our site
  • 10 (0.8%) have 4 posts on our site
  • 3 (0.3%) have 5 posts on our site
  • 5 (0.4%) have 6 posts on our site
  • 1 (0.1%) has 7 posts on our site
  • 1 (0.1%) has 8 posts on our site
  • 2 (0.2%) have 9 posts on our site
  • 1 (0.1%) has 10 posts on our site
  • 20 (1.7%) have more than 10 posts on our site

The user whose first post was an identification request question and that has subsequently asked/answered the most questions has 77 answers and 5 questions.

Of the 310 deleted questions that were a user's first post, tagged only with :

  • 219 (70.5%) were removed by the system:4 5
    • 182 (83.1% of system removed, 58.7% of total6) of which were abandoned closed questions
    • 29 (13.2% of system removed, 9.4% of total6) of which were dead questions
    • 4 (1.8% of system removed, 1.3% of total6) of which were removed along with the user7
    • 3 (1.4%) of system removed, 1% of total6) of which were stubs for migrated questions
    • 1 (0.5% of system removed, 0.3% of total6) of which were abandoned questions
  • 58 (18.7%) were removed by moderators
  • 31 (10%) were self-deleted
  • 2 (0.7%) were removed via vote

Of the 1,212 users whose first post was a (currently undeleted) question:4

  • 1,112 (91.8%) don't have the informed badge8
  • 100 (8.3%) have the informed badge8

Of the 310 users whose first post was a currently deleted question:4

  • 290 (93.6%) don't have the informed badge8
  • 20 (6.5%) have the informed badge8

1 The 34 questions that had other tags will be discarded from here on, since I’m trying to find questions that were about identifying an anime or manga for sure, and not to identify a piece of music or a certain reference of some kind in other anime or manga.

2 I’ve discarded the deleted questions because some of them are spam or nonsense, so it would only create some noise on the rest of the numbers. As such, I decided to only take into consideration questions that have not been deleted as of this time.

3 Missing 7 users because these were deleted users, or users who didn’t have an account here an whose question was migrated here, etc.

4 Percentages aren't exactly 100% due to rounding.

5 See this Meta post for more info on what abandoned, abandoned closed and dead questions are.

6 "Total" here meaning the 310 deleted questions that were a user's first post, tagged only with .

7 Removed along with a user who had been just created to post that question, but then self-deleted their account.

8 The informed badge is earned when users scroll through the tour page. Used here to try to understand how many of these users have actually tried to find out how our site works before asking a question. For reference, 684 (8.3%) of our total users have this badge — which means those users actually don't fall too far from the overall average.

  • 2
    If it counts, I actually deleted my first question which is an id request and wrote it again later (my first answer is also an id request). Now I have 92 answers and 19 questions desu. – Gao Jul 22 '15 at 0:39
  • 1
    I'm interesting in the trends of some of these. How big a part of the total number of questions were identification-request questions over time? – Madara Uchiha Jul 22 '15 at 6:55
  • 3
    I'd be interested to see the min, max, mean, and median scores for all other questions for comparison. A mean of 2.5 and median of 2 seems very low, but questions on this site also seem to score lower in general than questions on other sites, probably due to our small size and the widely dispersed expertise of our users (e.g. I have no idea how to judge if a Naruto question is good or not unless it's flagrantly bad, so I don't vote on them. If it's bad enough that I know what to do with it, it probably needs to be closed or flagged for moderator attention). It would be interesting to know... – Torisuda Jul 22 '15 at 7:06
  • 2
    ...whether those stats are actually unusually low by our site's standards. – Torisuda Jul 22 '15 at 7:07
  • 2
    @MadaraUchiha Cumulative id-requests over time: data.stackexchange.com/anime/query/261651/… / id-requests in each 2-week period: data.stackexchange.com/anime/query/261652/… (Summary: the identification requests are, indeed, taking over. Since this is from SEDE, it excludes deleted posts, but the fraction of deleted id-reqs vs. non-id-reqs is even worse - 400/1103 deleted Qs are id-req.) – senshin Jul 22 '15 at 16:49
  • 1
    @Torisuda data.stackexchange.com/anime/query/339074/… (Summary: in aggregate, ID requests get about half the score of non-ID-requests.) – senshin Jul 22 '15 at 17:03
  • (See also: queries linked from meta.anime.stackexchange.com/q/1193) – senshin Jul 22 '15 at 17:04
  • @senshin Would you add the data from the queries on a new answer, or edit them into my answer? It would provide some better readability, and we'd have the results regarding a particular date, and not ever changing whenever we run the query again. – JNat Jul 22 '15 at 17:23
  • @JNat Sure; I'll probably get around to that (plus some other related queries) tomorrow or so. – senshin Jul 22 '15 at 17:52
  • @senshin Awesome. I was mostly concerned with the impact of id-reqs that were posted as a user's first post — meaning I was trying to find out trends regarding users that came to us to have an anime or manga identified (do they further contribute to the site?, do they still visit the site afterwards?, how are these first posts scored?, how many of these have been deleted?, ...). – JNat Jul 22 '15 at 18:28
  • 3
    I find the numbers that @senshin provided on the scores of non-id requests to be extremely interesting. While every meta post we've had on the matter suggests our community likes the idea of having id requests, the voting stats suggest that our community doesn't like most of the actual id requests that we get. – Torisuda Jul 24 '15 at 4:43
  • 1
    This is some great information. Is there any possibility you could share some of the SEDE queries you presumably used to gather all this info? – Chris says Reinstate Monica Aug 11 '15 at 11:14
  • @ChristianRau Unfortunately I can't share these, no — they were run in our internal SEDE, where a lot more information that is not shared in the public SEDE is present. For instance, many dates are reset to midnight of the day they refer to on the public SEDE; deletion reasons are also not present in the public SEDE. All the rest should be pretty easy to get from SEDE, though (though, again, I don't have the queries to share as I ran them on the internal SEDE and some parameters' names are different too, and the schema is slightly different). – JNat Aug 11 '15 at 13:06
  • @JNat did take into account users, who were already on other sites active and just created the account here for an identification request? that was the case with me, i already had some experience with scifi, german and history, and i wanted to know the name of a manga, so i came here. since then, i try to help where i can xD so was the data about account creation about network account or site account? – Armin Mar 5 '16 at 23:16
  • Site account, @Armin. I didn't look into whether the users were already active elsewhere in the network or not. – JNat Mar 6 '16 at 12:51
10

Thanks @JNat for the data, here are some graphs to visualize this information better.

Notes:

  • I'm going to refer to "Users whose first post was a (currently undeleted) identification-request question" as ID-Users during this post.
  • Curve graphs are smoothed.
  • I'm listing some observations too, hopefully being unbiased.

Percentage of ID Requests enter image description here

Id Requests are just over a quarter of all undeleted questions

How many of the ID-User's First Questions are deleted? enter image description here

When Are ID-Users Accounts made?

enter image description here

enter image description here

Most Users accounts are created at the same time as the question is asked. Or at least, in the same day.

When do ID-Users visit after asking their question? enter image description here

enter image description here

There's a big dip between 1hr and 24hrs. Do we have a split of ask-and-run users and users who genuinely care about their question?

The majority of users don't care enough to visit again, but there are also users who have revisited the site later.

enter image description here

A lot of users only ask one question, Some ask another, but less than 10% ask more than that. @JNat points a 20+ user that "has subsequently asked/answered the most questions has 77 answers and 5 questions".

ID_Users with the "Informed" badge enter image description here

The tour page isn't being read by the majority of users. But as @JNat mentions: 8.3% of our total users have this badge and this isn't too far off this figure - so in general the tour page isn't being read.

Graph of the amount of ID Requests relative to other Questions (/10) over time:

enter image description here

Source (Shown by user @Hakase)

Graph of the percentage of questions that are ID Requests on the site over time enter image description here

Source (Shown by user @senshin)

  • 4
    Out of the informed users, I wonder how many just scrolled to the bottom of the page for the badge – Toshinou Kyouko Jul 22 '15 at 22:11
  • 1
    Some of the graphs are better represented with accumulated sum, IMHO. – nhahtdh Jul 23 '15 at 2:07
  • 1
    "The amount of id-requests is [...] not over the normal rate of question growth" - this is false, even though the graph you used may make it look true to the naked eye. If it were true, this graph, which shows the fraction of questions that are id-reqs over time, would be horizontal, or close to it. But it isn't. To the contrary, there is a strong upwards linear trend (R^2 = 0.98) on the graph. Since regressions are 100% true always, this means the site will be >100% id-reqs by 2022 or so. – senshin Jul 23 '15 at 5:12
  • 3
    @ToshinouKyouko: I must admit to having done that on some other sites before, although that's half because I generally spend enough time lurking to have an idea of what's on/off-topic and of how SE works, and because I know doing that will give me a badge. – Maroon Jul 24 '15 at 3:16
3

It is almost two months ago that we blacklisted id requests and I was interested in how this impacted our monthly registration quota. So I went to SEDE and compiled this query.

This query returns accurate results for the rows with '4 Months before' or further below. This is because of the way automatic account deletions work.

results

This query shows the amount of new users that registered between the 24th 00:00:00 of the month and the 23rd 23:59:59 of the next month. The second column is the amount of users that were created in that timeframe and also had at least two Posts in that timeframe. The last column is the percentage of users that satisfy the second condition out of all the users that satisfy the first condition.

Looking at the data, it doesn't seem that blacklisting the tag did have any affect on the absolute number of active users. Instead, the amount of inactive users shrank.

Of course, there are some errors in the query, e.g. it does only count posts that were created between 24th and 23rd. A user that was created on the 23rd and created lots of posts since then but no second post on the 23rd is not counted.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .