Wikipedia talk:Top 25 Report/Archive 2

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
Archive 1 Archive 2 Archive 3 Archive 4 Archive 5

Rule of three again

  • Cotinine: A nicotine metabolite with a two-day spike would seem like a Reddit thread, but I couldn't find one.
  • Canal+: Er, well, they broadcast Game of Thrones in Spain. Other than that... eh?
  • Automobile: It's been splattering itself all over the view charts essentially at random; Wiki articles don't behave that way

Serendipodous 12:54, 27 April 2014 (UTC)

    • @All of them, no news articles, no logical reasons for a spike ... I'd remove them. Ed [talk] [majestic titan] 20:51, 27 April 2014 (UTC)
    • The search of my usual sources agrees. West.andrew.g (talk) 02:35, 28 April 2014 (UTC)


Looks like you have Frozen (2013 film) on this week's list twice.(Spots 17 and 19)192.234.241.146 (talk) 16:35, 29 April 2014 (UTC)

Thanks. Always a pain when that happens. Serendipodous 17:32, 29 April 2014 (UTC)

April 27 - May 3 2014

I am writing a blog article about Donald_Sterling. I wanted to include a link to his week at the top of Top 25 Report.

I found the correct Top 25 Report: For the week of 27 April to 3 May on a Talk Page.

Why does Last week's report skip a week?

Thank you.

Mitch3000 20:08, 12 May 2014 (UTC)

Sorry; neglected some necessary housekeeping. It's fixed now. Serendipodous 20:19, 12 May 2014 (UTC)

Rule of three again

Need a second opinion this week:

Serendipodous 22:26, 25 May 2014 (UTC)

Assuming you've done the usual "Reddit TIL" checks (you always do), I don't have much for you. Genghis Khan, the copyright system, and the illusion all seem like ripe topics for this kind of thing. Is there any Internet property the watchlisters can imagine that operates in a similar vein? Could a site like Metafilter or Digg drive this much traffic? What about the most popular of Twitter accounts? When we see Reddit TIL thread traffic, do we also tend to see increased talk page activity? West.andrew.g (talk) 02:57, 26 May 2014 (UTC)
Genghis Khan was indeed due to a TIL thread. [1] _dk (talk) 04:17, 26 May 2014 (UTC)
I can't find any reason for computer virus, checker shadow, or Durning; the last major thing I can find for Durning is in November. [2] For the CAS, there was a CNET video? Ed [talk] [majestic titan] 17:32, 26 May 2014 (UTC)
I couldn't find a date associated with the video, but absent a direct link to Wikipedia, and given the fact that not a single person has commented on it over at CNet, I find this to be an unlikely culprit. The Reddit TIL for Durning also had a measly 34 upvotes. Puzzling. West.andrew.g (talk) 23:18, 26 May 2014 (UTC)

Seems to be a pattern these days, but I need a second opinion again

Serendipodous 16:33, 3 June 2014 (UTC)

  • They all look like non-human spikes to me. No other coverage of these topics, no spike in editing during the view spike.--Milowent • hasspoken 17:00, 3 June 2014 (UTC)
  • No explanations here, either. West.andrew.g (talk) 13:29, 4 June 2014 (UTC)
  • Ditto. I would be interested in knowing why (a) bot(s) got interested in the obscure railroad. Just strange. Ed [talk] [majestic titan] 18:22, 4 June 2014 (UTC)
  • If the Denial hits all came within a couple hours it's probably some wannabe hacker trying to be clever with an attempted Denial-of-service attack. --NeilN talk to me 17:36, 8 June 2014 (UTC)

no more updates?

Shouldn't it have had an update 48 hours ago?--Stemoc (talk) 03:57, 9 June 2014 (UTC)

24 hours ago. And my computer crashed, so I lost a day's work. Serendipodous 09:08, 9 June 2014 (UTC)

Another nightmare

  • I went through the stats and recent edit history of all six, and think all six should be excluded. 1, 2, 4, & 5 are the easiest cases; short spikes with no corresponding increase in editing history. No. 3 is a bit tougher, but the spikes are not smooth and there is no recent editing. No. 6 (Flounder), there is no increase in news I can find, nor a jump using a Google Trends search, so it also seems to have a non-human cause.--Milowent • hasspoken 04:59, 30 June 2014 (UTC)
  • Concurring here with Milowent, but I'm more skeptical of flounder. Most bot-driven view counts look more like, well, the others here, no? Why does this one ramp up, then slowly drop? Strange. I wouldn't object to you excluding it, but perhaps this one is worthy of a footnote at the bottom. Ed [talk] [majestic titan] 06:22, 30 June 2014 (UTC)

Who Is Behind This Great Article?

Hi guys, I'm only new to Wikipedia so please forgive me if this is already is common knowledge. I just came across this article is very interesting and is one of the things that my mind would ponder in the past even before I joined Wikipedia. So basically I'm just asking who takes credit for all this work? (And well done to whoever it is!) The Free Editor Anyone Can Cite (talk) 23:00, 20 July 2014 (UTC)

@The Free Editor Anyone Can Cite: -- My machine and code is used to acquire the raw page access logs and do the weekly aggregation. Longer and non-editorialized data is posted to WP:5000. From there, a user will come and eliminate some entries caused by non-human activity and write the English text. This week's version was done by User:Milowent, but User:Serendipodous has a long history of contributing. West.andrew.g (talk) 23:42, 20 July 2014 (UTC)

Amazon

I've been thinking for a while now that Amazon should be removed from the list. I don't think it reflects genuine reader interest. Biosthmors (talk) pls notify me (i.e. {{U}}) while signing a reply, thx 22:47, 19 July 2014 (UTC)

For the record, that was the conclusion reached by the Wiki meetup I went to. Serendipodous 00:53, 20 July 2014 (UTC)
Wiki meetup? Glad to hear you are evangelizing our work out in the real world. Thanks, West.andrew.g (talk) 22:56, 20 July 2014 (UTC)
  • Amazon would be #1 on next Top25 (Subaru Justy has to be bot-driven to get its 4 MILLION + views, or it would be #1), but I'm inclined to remove it as well, as Serendipodous has been watching it for a bit already. Anyone who thinks it should remain, please chime in. I want to exclude a good explanation in the exclusions section.--Milowent • hasspoken 20:45, 21 July 2014 (UTC)

More football

  • The top 25 will still have more football in the next report, I realized that I got a day off in the prior report -- it actually covers July 6 through 12, not July 7 through 13 as reported. So the July 13 statistics (the final day of the world cup) will keep a few articles in the mix.--Milowent • hasspoken 20:49, 21 July 2014 (UTC)

Notes for next report (anyone is welcome to add comments)

  • Amazon.com - continue to exclude? Views dropped to 616,453 this week. Still seems quite high compared to facebook (336070 views) - should that be expected? Spike on July 24 not duplicated on de/es/fr.
Drop it. No one believes it any more. Serendipodous 12:09, 27 July 2014 (UTC)
  • Land mine - View spike on July 25 seems odd. Exclude?
Reddit thread: http://www.reddit.com/r/todayilearned/comments/2bmixu/til_landmines_planted_on_the_coasts_during_the/ _dk (talk) 14:15, 27 July 2014 (UTC)
  • Tom Lehrer - popularity due to reddit thread [3] (4081 upvotes as of now)
  • Rose Leslie may make top 25 after all exclusions - what is basis for her popularity this week?
Reddit thread: http://www.reddit.com/r/todayilearned/comments/2be6py/til_that_rose_leslie_ygritte_grew_up_in_a_castle/ . Serendipodous 12:09, 27 July 2014 (UTC)
  • Exclusions: Subaru Justy (absolutely massive viewcounts continue), 100, Java, History of Bălţi (excluded many times in past, its back this week), Key of Solomon (classic one day spike, no editing during spike, no spike on es or fr (de doesn't have an article on it?)

--Milowent • hasspoken 11:52, 27 July 2014 (UTC)

  • Two more reddit threads! Thanks for the links. Will maybe have to work in this new NYTimes piece on reddit [4] into the commentary.--Milowent • hasspoken 03:13, 28 July 2014 (UTC)
  • The new report is now up. Lots of doom and gloom, if not for Reddit, that would be even more of it.--Milowent • hasspoken 03:27, 1 August 2014 (UTC)

Notes for next report (july 27 - aug 2)

Actress Laura Prepon had a big view spike on August 1 (212,526 that day) not duplicated on the fr/de wikipedias. She gets a bit less than 5,000 on other days. Any thoughts as to cause? I don't see a reddit thread, but sometimes I can't find them too easily.--Milowent • hasspoken 05:37, 6 August 2014 (UTC)

  • Jack Riewoldt - Australian footballer -- is another one, I don't see good explanation for the spike.--Milowent • hasspoken 06:41, 6 August 2014 (UTC)

Can't find anything in Reddit or the news, and they don't look like natural spikes. Wish I could say more. Serendipodous 19:31, 6 August 2014 (UTC)

  • Thanks for weighing in. Its a tough call, but I'm going to keep Prepon and remove Riewdoldt (who would be at the bottom of the list anyway). Prepon is a very likely case to be subject to a spike like this. Perhaps it will spur someone to weigh in. I tried to figure out how to view the hourly stats around Aug 1 using Wikiviewstats and couldn't do it??!--Milowent • hasspoken 12:47, 7 August 2014 (UTC)
Wikiviewstats doesn't do hourly anymore. Shame. I think whoever started it realised he didn't have the room on his hard drive. Serendipodous 15:27, 7 August 2014 (UTC)
Darn! You can get hourly for the current day, but after that the information is gone. I know it could be extracted from the logs for a given day, but I'd have to learn a lot to do that.--Milowent • hasspoken 15:58, 7 August 2014 (UTC)

I used to persistently store *all* per hour data. It takes up a *lot* of space, made weekly aggregation take forever, and wasn't getting utilized enough to justify these costs (I now store daily view counts). I'll note that I still have a database table that records a (page, hour, views) tuple in a DB table any time the number of views is >= 600 in an hour. This has proven useful, for example: (1) In showing the scale of DDOS attacks, and (2) reporting at fine granularity on crazy public interest events, like the Super Bowl halftime show and celebrity deaths. I could query this in isolated cases of interest, but I'd prefer this not to become part of our weekly workflow in preparing the report. Thanks, West.andrew.g (talk) 15:21, 9 August 2014 (UTC)

Great Seal bug

The Great Seal bug was actually linked from a reddit TIL thread [5], so I think it shouldn't have been excluded. :) _dk (talk) 04:33, 23 August 2014 (UTC)

Thanks! And now I know how to search for redirects on Reddit. Which is good. Serendipodous 06:23, 23 August 2014 (UTC)

We got a problem

Grok.se isn't working this week, so I'm left in the dark on a couple of topics:

  • LASIK: Nothing on Reddit; nothing in the news that I can see
  • TIDEL Park, Coimbatore: Nothing on Reddit; nothing in the news
  • Daboase: Nothing on Reddit, nor can I see how a TIL could emerge from it. Nothing notable in the news either
  • Corona (satellite): Might be related to this video; otherwise can't see anything.

Thoughts? Serendipodous 14:08, 7 September 2014 (UTC)

  • Hmmm. Grok.se is showing stats for me at least at the moment, but for TIDEL Park, Coimbatore I see their stats are not showing any view spike. For LASIK I see a huge spike on Sept 1 and Sept 2. But no edits to the page recently, and no matching spike for de:Laser-in-situ-Keratomileusis. Daboase shows no spike on grok.se; a one sentence stub created in one edit on July 2012 and never edited; perhaps subject to bot with a misspelled "database" request? Corona (satellite) was subject to 3 minor edits on Aug 31, the day of that article's spike. But I see no source for the popularity, it remind me very much of NASA Helios from last week, I couldn't figure out a source for that either.--Milowent • hasspoken 12:14, 8 September 2014 (UTC)
grok.se looks great, until you see that the graphs stop on Sep 3. Serendipodous 13:31, 8 September 2014 (UTC)
Indeed, I've looked at the graphs for a dozen pages (the four above, Lobster, Greenwich, Roundhouse, Portal:New Zealand, Talk:Whitchurch, Shrophsire, Alive!, Llŷn Peninsula and Wikipedia:Redirects for discussion/Log/2014 September 3) and not a single one has had any views recorded since the 2nd of September. Thryduulf (talk) 13:44, 8 September 2014 (UTC)
  • Ah, that would be a problem then! This weeks Top25 will run through Sept 6. There have been times when it drops a few days in the past, I know.--Milowent • hasspoken 14:03, 8 September 2014 (UTC)

Regarding LASIK, I've just been clearing out my spam folder and found several emails offering me deals on LASIK treatment (unless I live in Canada). Could this be related to the viewing increase - either people looking up what it is, or maybe a link in the email I didn't spot? Thryduulf (talk) 14:15, 9 September 2014 (UTC)

Possibly; though that might also be evidence that it's due to an automated spambot. Serendipodous 14:19, 9 September 2014 (UTC)
Probably the latter. Clickthru rates on email spam are quite low, and then getting to the wikipedia page for LASIK would be a small subset of that. But the effect of mass emails on wikipedia views are something I don't think we know much about.--Milowent • hasspoken 14:53, 9 September 2014 (UTC)
I suspect that Corona is either (a) a spambot for the beer or more likely (b) referred to from Gizmodo and the other sites with the video. I'd include it but note that it's suspicious. Ed [talk] [majestic titan] 15:53, 9 September 2014 (UTC)

Sept 7 - 13

Any thoughts on Chernobyl disaster spike cause, as well as Killer whale? Brittany Murphy must be due to cable movie that came out about her.--Milowent • hasspoken 05:04, 16 September 2014 (UTC)

Chernobyl is a Reddit thread. Quite a sobering one too. Many people sacrificed themselves that day to save the lives of millions. Serendipodous 06:45, 16 September 2014 (UTC)
Killer whale is also a Reddit thread. Serendipodous 06:48, 16 September 2014 (UTC)
  • You are awesome! Thanks. I did a quick search at reddit, but I swear their algorithm must be odd because this isn't the first time I haven't been able to quickly locate threads.--Milowent • hasspoken 12:19, 16 September 2014 (UTC)
I find it easier to use Google's site search function. Serendipodous 12:30, 16 September 2014 (UTC)

Sep 14-20

It's one big pile of bleah this week.

Payday loan: Could be any number of reasons for this to be on here: a crackdown on payday loan advertising in the UK? A payday loan scam in the US that bilked $162 million from struggling workers? Or just a spambot trying to bilk Wikipedians?
NASA Helios: Fairly seasick viewing history does not suggest human views.
Wireless security camera: Again, a slightly nauseous viewing history.
20 Fenchurch Street: One-day spike. No associated activity that I can see. Fairly interesting topic though.
Mali (GPU): Hard to determine a pattern, because we only have one day's worth of info.
Warner Robins: a redirect. Usually means a Reddit thread. Not this week, apparently.
Human resources: Artificial-looking 2-day spike. Also not the sort of topic that would inspire wide interest. Serendipodous 09:41, 21 September 2014 (UTC)
  • Have you emailed Oliver? You were copied on the email thread about last weeks entries, but I wasn't sure if you were getting them. He should be able to review these, he can access information we can't regarding hit sources that can help ferret out bots.--Milowent • hasspoken 19:05, 21 September 2014 (UTC)

On the Non-Reporting of Mobile Views

A significant statistical issue has come to my attention. Quite simply, the WMF does not record/report per-article mobile views, and thus they are unavailable for my aggregation....

The complete write-up is at User_talk:West.andrew.g/Popular_pages#STICKY:_On_the_Non-Reporting_of_Mobile_Views.

Please consolidate all discussion at that location. Thanks, West.andrew.g (talk) 18:40, 4 September 2014 (UTC)

Sept 21-27

An absurd number of potential exclusions this week.[6].

It could well be from the DYK on our own main page! _dk (talk) 13:40, 29 September 2014 (UTC)
Yeah, it tracks with the spike Serendipodous 13:58, 29 September 2014 (UTC)
DYKs rarely get views of this level, but the DYK probably got noticed elsewhere.--Milowent •hasspoken 18:25, 29 September 2014 (UTC)
FYI, it appears the DYK directly spawned a reddit thread[9], so I will credit DYK's role in the TOP25 report I am currently preparing.--Milowent • hasspoken 15:34, 2 October 2014 (UTC)

--Milowent • hasspoken 12:50, 29 September 2014 (UTC)

Drop me an email so I can schedule it? Ironholds (talk) 13:38, 29 September 2014 (UTC)
I've trolled 20 pages of Google listings for the days cited and I can't find anything notable on Sun Yatsen. As for China? Take your pick. The terrorist attack in Xinjiang? The protests in Hong Kong? The border dispute with India? The ambivalence about ISIS? "Hong Kong" and "Xinjiang" don't really track with the spike for China, so it's not likely to be those. Serendipodous 13:53, 29 September 2014 (UTC)
Yeah, I'm going to solve for this point of ambiguity by just...you know. Going through the unsanitised request logs ;p. Ironholds (talk) 13:57, 29 September 2014 (UTC)

for Less, 800k requests a day from the same IP *readies 403 Hammer * . Email me with the others? Ironholds (talk) 15:33, 29 September 2014 (UTC)

  • I just emailed you a list of the 10 (culled from above) I see the most concern about making a call on. Thanks.--Milowent • hasspoken 18:23, 29 September 2014 (UTC)
    Not got any emails. Ironholds (talk) 23:15, 29 September 2014 (UTC)
Clearly "email user" is not working well for me lately, I need to look into that. I just sent new one to the email address I know is correct. Sorry, and thanks!--Milowent • hasspoken 20:58, 30 September 2014 (UTC)
  • New report is up, though it may need a proofread. After working through the exclusions I was a bit worn out! But that led me to quite the Easter Egg -- Scotland, England, and Marriage (all excluded), have clearly similar spikes for 26-28 Sept, like it was purposefully coordinated. Who would go to the bother of doing that, though I am duly impressed.--Milowent • hasspoken 17:22, 2 October 2014 (UTC)

Did Oliver give an explanation for all the exclusions this week? It would be best if we could say what they were, even if it's just one word. Serendipodous 18:42, 2 October 2014 (UTC)

  • Not every one, but he responded to all I asked about. We got farther down into the raw 5000 than I expected. Here is what he said, what do you think we should add:
  • Marriage appears totally legitimate. (Milo adds: Note that I still excluded it, he wasn't asked to look for what I found)
  • Payday loan - (BAD)
  • Infant swimming appears totally legitimate, but via reddit. (And then I found it ...)
  • China (bad)
  • Rome (bad) - same source as China
  • Ford Ranger - bad, spammer
  • Sun Yat-sen, Last spike, Yazid I, Positive and negative predictive values- all bad.
--Milowent • hasspoken 19:45, 2 October 2014 (UTC)

Maybe we should include a brief (like, 5-word) summary of how the exclusions were derived. Given that to my computer-illiterate brain, Oliver's explanations make about as much sense as High Latin would to the average serf, I think that would be best. Serendipodous 21:38, 2 October 2014 (UTC)

I guess I'd summarise as: we took a tuple of IP address, user agent, referer, and page, for each page that appeared to be an outlier. In almost all cases, we saw that the numbers were thrown off by a small number of unique IP/referer/user agent combinations, indicating bot or spammer activity - we're talking ~350k requests from the same IP, with the same user agent, in a week.
Not exactly five words, though ;p. Ironholds (talk) 01:39, 3 October 2014 (UTC)

28 Sep to 4 Oct

OK. Last call: List of Bob's Burgers episodes- one day spike on 2 October. Bob's Burgers season premiere: 5 October. Bob's Burgers current rank in the raw 5000: 1424. Legit, or not? Serendipodous 11:10, 6 October 2014 (UTC)

  • That does not look legit at at all. Over 1 million views in one day, with only ~2300 on the days on either side. Plus I've never ever heard of this series so it can't be that popular.--Milowent • hasspoken 12:34, 6 October 2014 (UTC)
It is very popular, but I've never seen an "episodes" list without its accompanying main article nearby. So yeah, I think it should go. Serendipodous 12:40, 6 October 2014 (UTC)

Mobile Views on WP:5000

I see the WP:5000 has been updated by Andrew to include mobile views with a new column showing the percentage of mobile views. Aside from the minor point of being another tool to ferret out automated views (they get very low mobile #s I see, buh-bye List of Bob's Burgers episodes, or absurdly at 100% in the case of Angelsberg), the view counts are also much higher overall. And the percentages of mobile views can vary noticeable based on content, so On the Internet, nobody knows you're a dog arrives high on the list due to a reddit thread but gets only 28.69% mobile views (i assume reddit is not a mobile crowd for best use), but Ebola virus has a whopping 71.23% mobile. A sobering article like Deaths in 2014 has only 22.55% mobile. Nice work!--Milowent • hasspoken 22:59, 9 October 2014 (UTC)

Frigging reddit threads. Can't believe I missed that one. Serendipodous 23:17, 9 October 2014 (UTC)

What happened?

There used to be almosts. What happened to them? They were interesting to add. (If they come back, there should be five listed. Or could I add them for the next week?) A Great Catholic Person (talk) 04:04, 14 October 2014 (UTC)

What happened to them was that every time I made an adjustment to the list, I had to make an adjustment to the almosts, and since the almosts were almost always either articles that had already been in the top 25, or would appear in the top 25 later, their value seemed questionable. Serendipodous 06:41, 14 October 2014 (UTC)
Almosts were an occasional thing, if I have time I'll look for this week, but time is short and I want to mention the new inclusion of mobile view stats in our data.--Milowent • hasspoken 14:40, 15 October 2014 (UTC)

Oct 5 - 11

Oliver found Rich Gang was due to a large number of null referrers. Thor Heyerdahl is a Google Doodle. Anencephaly is probably down to this popular news piece. Age disparity... is a Reddit thread. So is Stephen Jay Russell. Serendipodous 16:01, 15 October 2014 (UTC)
Wow, you're good .....--Milowent • hasspoken 17:53, 15 October 2014 (UTC)

Amazon.com

back on the list. 5.86% mobile views. Exclude? Serendipodous 14:22, 19 October 2014 (UTC)

  • I agree with the removal. No doubt the Amazon article has a decent number of legitimate views (which probably includes most of those mobile views), but nothing near to making the Top 25.--Milowent • hasspoken 20:16, 21 October 2014 (UTC)

Oct 19 to 25 2014 now up

  • New report now up. I made a mess of getting it up, doing a move instead of a cut and paste, but I think all is well now.--Milowent • hasspoken 17:48, 27 October 2014 (UTC)
I was a bit iffy on include/exclude The Book of Life (2014 film); but it did get more than 2% mobile views, so its exclusion should be mentioned in the list. Serendipodous 18:20, 27 October 2014 (UTC)
Good point, I added a mention. I had concluded yesterday it was definitely bogus, my only question being whether it was intentionally inflated to influence our report. Theoretically that could happen at some point.--Milowent • hasspoken 18:35, 27 October 2014 (UTC)

Irregularities wiith the TOP25 this week

@Serendipodous: I see some problems with te TOP25 report this week:

  1. Facebook, #6, is repeated as #8.
  2. Taylor Swift is #13; it's comment is: See #13.
  3. Avengers: Age of Ultron is #18; its coun t is bigger than #6.
Fixed. Serendipodous 07:25, 6 November 2014 (UTC)

And it does seem odd that you would excleude Tanksgiving this week - I find it hard to believe that over half these 1,000,000+ hits are not "due to premature interest in the upcoming holiday". עוד מישהו Od Mishehu 05:33, 6 November 2014 (UTC)

Checked with the analytics people. It is, in their words, "totally artificial". Serendipodous 07:04, 6 November 2014 (UTC)
  • Re Thanksgiving, it is also relevant to mention here that at the WP:5000 report you can now see the percentage of mobile views that an article gets. A very low mobile percentage (2.84% for Thanksgiving) is an extremely strong indicator of artificial views, aside from getting confirmation from WMF staff. Also, its view rise here mimics that of Cyber Monday and Black Friday (which is a disamb page) which are right next to Thanksgiving on the raw 5000. Evil corporate forces at work!--Milowent • hasspoken 13:21, 6 November 2014 (UTC)
I'm actually thinking we might rise the minimum for inclusion to 4 or 5% mobile; certainly if numbers like this keep happening. Serendipodous 13:43, 6 November 2014 (UTC)
This probably makes sense. We know some percentage of views of an article like Thanksgiving are going to be legitimate, but when the inflated artificial views dominate, you know the article wouldn't be close to the Top 25.--Milowent • hasspoken 17:33, 6 November 2014 (UTC)

Nov 2 to 8, 2014 Now up

New report is up; a rather boring week overall. The only thing that personally interested me is that film Interstellar was #1 in views, but was beaten at the American box office by Big Hero 6 (film) which was only #24. Of course the latter is a film directed more to children, who probably use wikipedia less frequently for movie information, that's my guess.--Milowent • hasspoken 21:09, 10 November 2014 (UTC)

Missing word?

The entry for Myles Munroe reads "Bahamian Evangelical preacher who caused a ruckus with his opposition gay rights this year, and then capped it by dying in a plane crash this week.". Should that be "opposition to gay rights"? I'm not awake enough to be certain, hence posting here rather than editing. Thryduulf (talk) 01:54, 18 November 2014 (UTC)

@Thryduulf:  Fixed - thanks! GoingBatty (talk) 02:41, 18 November 2014 (UTC)

STICKY: WMF now reporting mobile/zero pageviews

Just one month ago I took to many of these same talk pages to explain that WMF statistics were under-reporting per article views by approximately 1/3, because mobile traffic was not being included in those totals. Further details were included in a Signpost article. I'd like to commend the wMF for quickly rectifying that situation, as files including mobile and wp-zero traffic are now available. The Wikipedia Zero project currently sees very little traffic, but I'll be including it in all my reports regardless (recall that mobile views were also minor just a few years ago).

You'll notice the WP:5000 and WP:TOPRED now breaks down the (increased) totals into "mobile" and "wp-zero" percentages (the complement being the "desktop" views we had previously). This will be the case from the OCT-14-2014 report onwards. In addition to the higher totals, another immediate benefit is that articles with very low mobile participation are often indicative of bot/misconfigured traffic. Though an intelligent malice spammer can evade this by altering user agent strings, I anticipate this be of great utility moving forward.

I know the WMF has reached out to stats.grok.se about updating their user-facing tool. I greatly look forward to having this new data on board, and aside from the fact its going to make year-end aggregation a bit messy, I'm excited to see what we can learn from deeper dives into the data. Thanks, West.andrew.g (talk) 23:46, 9 October 2014 (UTC)

Thank you for all you are doing regarding this. In addition to this report, page views are frequently very important for deciding the outcome of redirects nominated at WP:RFD, so more accurate data is really beneficial. Thryduulf (talk) 16:01, 13 October 2014 (UTC)
...eeep. How important? Ironholds (talk) 16:48, 13 October 2014 (UTC)
Whether a redirect is used or not, and if it is whether it appears to be used by humans and not just spiders, can determine whether a redirect is kept or deleted. Especially for things like typos, it is a good indication of plausibility, and generally if there are a lot of hits for a redirect with no internal links then that suggests it's linked from somewhere outside en.wp which is a good reason to keep it. Thryduulf (talk) 08:22, 14 October 2014 (UTC)
Aha! Okay; cool. I'm writing the new pageviews definition and this was not a use case I had considered. I'll work it in :). Ironholds (talk) 13:08, 14 October 2014 (UTC)

United States

In this week's entry for the United States, you say "country with by far the most English speakers". Not India then (see List of countries by English-speaking population)? Hamish59 (talk) 12:55, 18 November 2014 (UTC)

Hm. It was 125 million last time I checked, though that new number apparently includes those with only a small amount of English. if we go by the figures presented in the notes then 5 percent of men and 3 percent of women are fluent, whcih would make 48 million fluent speakers. Serendipodous 23:16, 18 November 2014 (UTC)
Fair enough. Hamish59 (talk) 15:23, 19 November 2014 (UTC)
Apparently, the number was changed last week without discussion or additional citation by an anonymous IP with a history of disruptive editing. So I've reverted it back. Now let's see what happens. Serendipodous 16:14, 19 November 2014 (UTC)
Ah, that explains it then. Hamish59 (talk) 20:24, 19 November 2014 (UTC)

Big in Japan - New Top 25 Report (November 16 to 22, 2014)

Now posted.--Milowent • hasspoken 13:47, 24 November 2014 (UTC)

Dead black men and science fiction (Nov 30 - Dec 6, 2014)

New Top 25 now up.--Milowent • hasspoken 23:37, 8 December 2014 (UTC)

Can this be extended

I presume the top 25 report is based on absolute traffic. I wonder if it might also be nice to give a chance to the lower end traffic and examine spikes based on the factor by which it peaks above the previous week's average. Requires more computation, but might be quite interesting. Shyamal (talk) 05:25, 7 December 2014 (UTC)

  • Shyamal, the Top 25 is indeed derived from the raw top 5000 articles reported at WP:5000, but we manually exclude articles affected by bots after examining relevant data. Occasionally we add some comments in the Top 25 report on articles below the Top 25, as I just did in the newest report, i.e., noting how many articles exceeded certain view thresholds. Anyone with an interest in doing additional analysis or special reporting is very welcome to do so. I believe we would find that the sorts of things which often cause articles to spike into the Top 25 (e.g., a big news event, a movie release, a song or album release, a celebrity death or celebrity news story, Google Doodles, popular reddit threads) also occur with similar frequency lower down the list. See also the February 2013 special report in the Signpost on article popularity trends. It would be fascinating if someone created lists examining the popularity of articles in a certain opic, such as a list of the Top 25 Indian movies in a week, or top 25 actors, etc. Or the Top 25 albums, comparing that to sales charts. And I am sure many other fascinating things could be done with the numbers as well, given sufficient effort.--Milowent • hasspoken 23:55, 8 December 2014 (UTC)

New Report: North Korea Is Not Pleased

New report now up. Serendipodous, am I able to lure you back for the next report? If not, we need to recruit additional help if anyone is willing to assist with the Top 25 Report preparation.--Milowent • hasspoken 21:38, 23 December 2014 (UTC)

Don't worry; I wouldn't leave you in the lurch. Serendipodous 16:22, 24 December 2014 (UTC)

Exclusion

Serendipodous 'Tis the Season was the Google Doodle for several days. I thought we normally include it as normal traffic. Talk:'Tis the Season has more details. Widefox; talk 13:25, 4 January 2015 (UTC)

Yep, and that's why we post the exclusions on the bottom of the list. Because every now and again we make mistakes :-) Serendipodous 14:28, 4 January 2015 (UTC)
Thanks for fixing it. Widefox; talk 23:01, 4 January 2015 (UTC)

Odds and Ends from Reddit

  • I noted that the report's grok stats spiked on Dec 7 to 3575 views for the day (which excludes mobile views, remember), it was probably driven by this reddit post noting the report's existence: [11]. Even though redditors can be snarky about everything, they appreciate wikipedia in all its odd glory.
  • Via another reddit post [12], I was reminded to check the "most visited pages of the year list" at wikitrends [13]. You can see how bots have rendered this list much less meaningful, as #1 (Online shopping), #2 (Java), and #3 (Alive) are all articles long excluded from the Top 25 for being proven to be driven by non-human views. We also know #7 (Amazon.com) was boosted for a time by bots, as was #11 (English alphabet), #12 (Alphabet), #22 (IPv6), and no doubt many others.--Milowent • hasspoken 13:50, 8 January 2015 (UTC)
My much longer year-end list (with some clues about bot presence) will be published in the next couple days. Thanks, West.andrew.g (talk) 17:02, 8 January 2015 (UTC)

update for this week

I am curious what top pages will be thanks User talk:dghavens 00:07, 12 January 2015 (UTC)

It's basically all Charlie Hebdo. Serendipodous 00:24, 12 January 2015 (UTC)

January 4 to 10, 2015

@Serendipodous:, Some mighty fine writing this week, especially number one. Keep up the good work. SchreiberBike talk 02:52, 14 January 2015 (UTC)

Thank you :) I really struggled with this one. Serendipodous 09:26, 14 January 2015 (UTC)

The 10,000 most popular Wikipedia articles of 2014

See The 10,000 most popular Wikipedia articles of 2014 (link goes to a talk page with hopes of consolidating discussion). Thanks, West.andrew.g (talk) 14:59, 21 January 2015 (UTC)

January 4-11, 2015

January 4-11, 2015 Top 25 is now up. Since we were late this week, and the Signpost was very punctual, there is no Top 10 in the Signpost this week.--Milowent • hasspoken 22:14, 22 January 2015 (UTC)

January 18 to 24

is now up. Now I am going to bed. Serendipodous 01:32, 28 January 2015 (UTC)

Jan 25-31 now up

New report posted.--Milowent • hasspoken 15:35, 4 February 2015 (UTC)

I'd probably send Morgan Freeman to the analytics people. We should probably send Lil Wayne there too. Serendipodous 20:22, 4 February 2015 (UTC)

Feb 1 to 7 report up

An ace job by Serendipodous, your treatment of the Superbowl is great, despite revealing you're a Seahawks fan!--Milowent • hasspoken 21:49, 10 February 2015 (UTC)

I'm not a Seahawks fan :) But I am REALLY not a fan of the idea that "you didn't win, ergo, you don't matter." Serendipodous 10:01, 11 February 2015 (UTC)
By the way, the Patriots' Super Bowl touchdowns count as passing touchdowns for Brady and receiving touchdowns for Brady's teammates. GoingBatty (talk) 04:32, 12 February 2015 (UTC)
See this is why I never liked American football. Too many friggin rules. The commentary always sounded like a cattle auctioneer explaining quantum mechanics. Serendipodous 04:54, 12 February 2015 (UTC)
Thanks for updating the report. It's not a rule, it's one of a bazillion statistics that the commentators spout out during games (and fans use to play Fantasy football). Keep up the good work with the report - I always enjoy learning what articles are the most popular (and why), especially when its sprinkled with good humor. GoingBatty (talk) 05:11, 12 February 2015 (UTC)

Feb 8-14, 2015 - February is for Lovers

now up.--Milowent • hasspoken 01:58, 18 February 2015 (UTC)

Seems odd that no mention was made of Presidents Day in the entry for List of Presidents of the United States. - Presidentman talk · contribs (Talkback) 23:02, 20 February 2015 (UTC)

New report is up

Have at it. Serendipodous 18:47, 23 February 2015 (UTC)

No Deaths in 2015? Thryduulf (talk) 23:29, 23 February 2015 (UTC)
Not this week. Serendipodous 07:05, 24 February 2015 (UTC)

Feb 22-28 now up - Attack of the Movies

Well, I know more about the Oscars than I'd planned now.--Milowent • hasspoken 18:42, 3 March 2015 (UTC)

This week's list is up

Have at it. Serendipodous 20:51, 9 March 2015 (UTC)

March 8 - 14: Its not cricket.

New report up.--Milowent • hasspoken 05:52, 17 March 2015 (UTC)

March 15 - 21 is up

Have at it. Serendipodous 23:45, 22 March 2015 (UTC)

March 22 to 28 - All over the place

New report up.--Milowent • hasspoken 18:50, 31 March 2015 (UTC)

Error in table

item 21 (The Walking Dead (TV series)) description is "see #16" but it should be "see #15" (The Walking Dead (Season 5)) because #16 is Zayn Malik. Mdob (talk) 13:07, 3 April 2015 (UTC)

  • Yes, thanks for catching that, feel free to correct an error like that when you see it. I had skipped a number in an early draft and had to renumber and add a new #25 this week, so I missed that renumbering.--Milowent • hasspoken 19:22, 3 April 2015 (UTC)

March 28 to April 4 Traffic report is up

have at it. Serendipodous 21:08, 6 April 2015 (UTC)

April 5 to 11 report in progress

Will have some Fast and Furious theme since that dominates the top spots. But anybody have idea what propelled Frankie Valli to probable #20 and Prince Rupert's Drop to probable #24?--Milowent • hasspoken 19:45, 12 April 2015 (UTC)

Prince Rupert's Drop" probably comes from this video on imgur. Serendipodous 20:05, 12 April 2015 (UTC)

As for Frankie Valli, I have no idea. We may have to exclude it, but throw it out to the public and see what they can find. Serendipodous 20:21, 12 April 2015 (UTC)

Run with it. Not the typical scripted entry like "Tablet computer". Maybe we missed a documentary film on HBO or something '''tAD''' (talk) 20:24, 12 April 2015 (UTC)
Shot an email off to User:Ironholds. At least he can tell if it's genuine or not. Serendipodous 20:37, 12 April 2015 (UTC)
Thanks!--Milowent • hasspoken 03:59, 13 April 2015 (UTC)
  • I've put up the report and included Valli. The Four Seasons are on tour, and as their fanbase is on what I call the "old-person internet" (email and other traffic not as easily traced) perhaps that is the reason.--Milowent • hasspoken 21:41, 13 April 2015 (UTC)
Well, according to Oliver, that was the right call. Serendipodous 07:30, 14 April 2015 (UTC)

April 12 to 18 report is up

Have at it. Serendipodous 10:33, 20 April 2015 (UTC)

April 19 to 25 - Bruce, Nessie, and genocide

New report now up. Scrutiny before I submit the Signpost article is, as always, appreciated.--Milowent • hasspoken 16:07, 27 April 2015 (UTC)

Traffic Report cited by study

(moved from my talk page for greater visibility): Hey Milowent, you were cited in a study! See page nine, bottom left. cc Serendipodous Ed [talk] [majestic titan] 05:59, 1 May 2015 (UTC)

  • The Traffic report cited is Now introducing ... mobile data (October 15, 2014), which was the first one to include mobile data. Interestingly, the study admits there is uncertainty as to whether mobile views are distributed unevenly across Wikipedia. That first traffic report with mobile data hinted that it might, and it seems possible to me that these differences could effect some of the conclusions in the study. In the six months since we've had mobile data, and Serendipodous and other commenters here please weigh in on your observations, I think we would conclude mobile view percentages do vary by topics, though the extent and significance have not yet been closely scrutinized. Articles made popular by Google and Reddit consistently have lower mobile view counts (in the 20-30% range), as does "Deaths in (current year)". Some pop culture topics seem to have higher mobile view rates than others, but its quite common for articles in the Top 25 to have mobile view counts well above 1/3rd of views (which was estimated to be about the % of views that Wikipedia gets from mobile sources?) -- and closer to 2/3rds. This suggests mobile use is higher for the most popular articles which are often on pop culture and current news events. This sounds like common sense?--if you want to know if a new movie has good reviews or box office income while waiting in line at the theatre, you can easily get this info from Wikipedia on your mobile device. Or if you want to check into a subject while watching a TV show about it? I've noted about three cases this year where a Lifetime| movie put a subject on the top 25, and the mobile views have been close to 80%.--Milowent • hasspoken 13:59, 1 May 2015 (UTC)
If I had the time, which I currently don't, I would "de-mob" the last few Top 25 reports (subtract their mobile percentages) and compare them to three random pages from before the mobile era. From a glance, it appears to me that were this done, the specific articles might change, but the overall trends in the articles would not. Serendipodous 14:16, 1 May 2015 (UTC)

Pause in statistical reporting

Notice that [14] halted at 18:00 UTC yesterday APR-30. This is our data source. I have contacted Dario on the Analytics team directly by email. If others want to raise the notification in other places (technical noticeboards?) then feel free. Unclear at this point if they will be able to back-process this data and what impacts this might have on our reports. Thanks, West.andrew.g (talk) 17:49, 1 May 2015 (UTC)

Fixed by analytics team without impacts. Thanks, West.andrew.g (talk) 19:22, 1 May 2015 (UTC)

"like floral clockwork"

I just love that line. What a great picture it creates in the mind.  SchreiberBike | ⌨  07:22, 6 May 2015 (UTC)

Wow. That's gotta be the quickest response I ever got. I only wrote that nine minutes before you posted :-) Thanks for the compliment, but in truth it was simply a reference to floral clocks. Serendipodous 07:30, 6 May 2015 (UTC)

Inner Core - May 10 to 16 2015

New report up.--Milowent • hasspoken 04:11, 20 May 2015 (UTC)

Retrieved from "https://en.wikipedia.org/w/index.php?title=Wikipedia_talk:Top_25_Report/Archive_2&oldid=1074644358"