This is an archive of past discussions about Wikipedia:Top 25 Report. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.
Sorry about that. This list really plays tricks with your eyes, though I usually catch them before anyone else does. Good job with the research! Serendipodous08:09, 28 May 2015 (UTC)
The raw copy has been posted to [1] for this week while I continue troubleshooting efforts with posting per the WP:VPT thread that was referenced last week. West.andrew.g (talk) 18:47, 1 June 2015 (UTC)
Done -- Posting issue is believed to be fixed. Old reports have been re-generated and posted and I expect the automated posting to succeed at its usual time over the weekend. West.andrew.g (talk) 15:19, 3 June 2015 (UTC)
Report has not updated
I think the current report is stale, the present version was scheduled to end on June 6, 2015 and then apparently refresh. Nothing happened so far. Brandmeistertalk11:04, 23 June 2015 (UTC)
I'll admit I was also surprised by the low totals for Charleston church shooting, but the numbers check out against stats.grok.se after one accounts for the mobile ratio. I'll note that the cause of missing reports over the past weeks lies purely in the ability to generate the reports (specifically queries against the API to obtain quality markers) and then post the reports' content over API. The nigthly raw data ingestion to the database never stopped and there should be no suspicion about the quality or accuracy of the underlying data (at least not anymore than usual). Moreover, the report cut-off times have been reproduced exactly as they would have been if the reports generated on schedule, and these are noted in the reports/versions that Milowent lists below. Thanks, West.andrew.g (talk) 20:52, 30 June 2015 (UTC)
I think I'm reasonably intelligent, West.andrew.g, but I am mystified by your explanation. Can we just say that the gods on Olympus delivered the recent Top5000 reports because they noticed us wailing and gnashing our teeth? Some day I will educate myself about APIs, databases, queries and the like. LizRead!Talk!21:11, 30 June 2015 (UTC)
Liz, look at the history, it looks like he posted multiple reports to history (and I added who should prepare Top 25 report if we follow recent order -- though I'm gonna be sparse ...)
Well, it would be nice to have all three charts just for completeness sake. And it would be great to have an up-to-date, June 21-27 chart for this week's Signpost. Have fun! LizRead!Talk!21:08, 30 June 2015 (UTC)
Histmerged back and redirect deleted to allow you to create the report as you wish. Although, having said that, why do you cut and paste instead of move when archiving? What are you doing about attribution of the edits? C&P gives the impression that the page creator is solely responsible but moving the page shows who wrote what before the archive was created. Compare WP:GO (page archived weekly by move, talk page stays in place, everyone happy). BencherliteTalk00:32, 8 July 2015 (UTC)
OK, in future I'll provide a link from the subpages. And while we're on the subject, there's another histmerge that probably should be done if possible. Milowent moved his first archive rather than copy/paste, so the first year or so of attribution got moved to Wikipedia:Top 25 Report/October 12 to 19, 2014, which you then deleted. Is it possible to get that history back? Serendipodous10:29, 8 July 2015 (UTC)
Done Actually, the history had been moved to Wikipedia:Top 25 Report/October 12 to 18, 2014 but I've moved it back. A suggestion: why not create a new subpage straight away each time you create a new report, and transclude that onto the main Top 25 report page? That way, all the history stays together for the main page and for the weekly subpages, and you don't have to worry about moving/cut-and-paste every week. All you'd need to do to show the new report on the main page would be either to update the dates of the transcluded subpage or find some template coding to update it automatically. BencherliteTalk08:43, 9 July 2015 (UTC)
Hi, just to fill in the blank - Cilla Black's link to the Smiths is that she indirectly destroyed the band. Morrissey wanted to record a cover of Cilla's song "Work Is a Four-Letter Word" (theme song to the maligned movie of the same name, also starring Cilla), and Cilla's inoffensive "girl-next-door" shtick was about as far from the rock-and-roll image that Johnny Marr wanted as it was possible to get. They eventually did cover the song (as a B-side to "Girlfriend in a Coma"), but that recording session was the Smiths' last one ever. Smurrayinchester12:13, 12 August 2015 (UTC)
Well, I did briefly waste some time wondering it was related to the Mandela Effect, which made some waves last week, but has fairly small views, so I must have come across a very weird internet niche when I read about it. One of the Mandela effect's erroneous memories is people believing the guy who stood up to the tank was run over and killed.[5].--Milowent • hasspoken13:26, 17 August 2015 (UTC)
I'm sad to see that week skipped over. I wanted to see how high Yogi Berra would rank.[6] Does anyone mind creating a report for that week? Totally understand if noone does. – Muboshgu (talk) 22:55, 7 October 2015 (UTC)
Sorry; I fixed the link. It's not a full report but you should be able to read it now. Yogi Berra was no 4; something I, as a lover of words if not baseball, lament. Serendipodous22:57, 7 October 2015 (UTC)
Thanks for the reddit link! I just got far enough down the list to realize you had to deal with black hole last week as well! I'll ping Oliver but leave it on the list for one more week unless I hear from him in time.--Milowent • hasspoken21:35, 26 October 2015 (UTC)
Black hole might have to do with the new images that NASA released about what it looks like when a black hole consumes a star. _dk (talk) 22:38, 27 October 2015 (UTC)
Samburu people
Given that traffic spiked on 19 October, it was almost certainly this reddit thread that did it (TIL in a 1980s Nike commercial, a Kenyan tribesman’s words were translated as “Just Do It”; however, American anthropologist Lee Cronk alerted Nike that the Kenyan really said, “I don’t want these. Give me big shoes.” Nike responded back saying “We thought nobody in America would know what he said.”) Smurrayinchester16:09, 27 October 2015 (UTC)
It doesn't help that they didn't actually use the word "Samburu" either! I looked at the page history and saw all of the edits were to the "In Western popular culture" section, so I tried searching reddit for the phrase "I don’t want these. Give me big shoes", which sounded like the most reddit-friendly bit. Smurrayinchester10:20, 28 October 2015 (UTC)
AyanP, thanks for weighing in and making a swap. (Though your substitute works in Canada now, haha!) I assure you it was not intentional, not sure how we did that. We are not experts in Indian cinema, so I encourage you to review our coverage of that subject regularly and comment to make sure we get it right.--Milowent • hasspoken14:07, 6 November 2015 (UTC)
New Report: Doodles of popularity
New report is up. I decided to finally exclude Black hole. In addition to finding no source, the editing frequency has remained constant; if it was really so popular we would see more editing activity.--Milowent • hasspoken14:46, 10 November 2015 (UTC)
Excluded for non-human views; If the mobile view % of an article is very near 100% or 0%, the views have proven not to be legitimate, so we exclude those entries. There are usually a few every week. Dulce Marie has .07% mobile views out of 2.5 million, and Anahi had .33% out of 536K. Web scraping was also excluded (1,312,744 views; .17% mobile), and well as an odd "link" for "compensation" (564,480 views, 0% mobile); the redlink Computer virus192.168.14.195/192.168.14.195-GET (395,327 views, 100% mobile), and a long script call starting "[[-webkit..." (388,446, 0% mobile).--Milowent • hasspoken16:41, 10 November 2015 (UTC)
box on talk page for notable entries?
Thinking about Lamar Odom at #1 last week, what do people think about creating a template like Template:Press that could be optionally added to article talk pages with a title like "This article was the most viewed article for the week of _________ according to the WP:TOP25"? The word "most" could have a variable allowing "2nd most" or other appropriate number put there. I think this information could be of interest to editors.--Milowent • hasspoken14:44, 27 October 2015 (UTC)
I like the idea: (1) it's good press for our statistical and editorial efforts, (2) it should make more people aware of the report, and (3) might spurn more user contributions to the articles themselves if authors are aware of "their" content's popularity. I don't think we want to automate the process, as we wouldn't want to repost the same/similar template week after week to very popular articles; but it seems non-intrusive enough for articles experiencing emergent popularity. Thanks, West.andrew.g (talk) 18:42, 27 October 2015 (UTC)
West.andrew.g,Muboshgu,Serendipodous, I created a simple template at Template:Top25 to try this out, please take a look. The only parameters are "place", where you can put "2nd" etc, if needed, for placement in the Top 25. If left blank, it is for the "most viewed" article instead of "2nd most", etc. The other parameter is "week" where you will ideally put a link to the archived report where the entry appeared, see example below which I just placed on Talk:Justin Trudeau:
I agree with Andrew that we don't want to automate this. "Template:Press" is not used for every press mention of a wikipedia article. We don't want to template Deaths in 2015 again and again. I would suggest that the first time an article hits #1 may generally be notable, such as a movie article, or a google doodle. Hopefully we'll pick up an editor that has an interest in posting the template from time to time on talk pages. I welcome any comments on the template (feel free to mess with it), I have no expertise in this area.--Milowent • hasspoken14:50, 4 November 2015 (UTC)
I appreciate its simplicity. Two comments:
I wonder if a link to WP:5000 might be warranted with a text description of "(raw stats)" or something even more brief. The wittiness and curation of the top-25 is certainly something to be appreciated, but being able to track topics more broadly on a week-to-week basis outside that format might interest some users and influence repeat views to both of our efforts. A tiny bit of vanity might also come into play here :-).
A simple template like this is able to have top-of-page prominence and will affect watchlists like any other edit. However, this isn't going to spur discussions the same way an explicit section might. Is one better than the other? Could we get away with both? i.e., use your template atop, then add a section that says "hey, we added this template to the top of the page... congrats on the exposure for your hard work... generated from the raw stats at [link] ... not a breaking news event you are aware of? an article discussing the weird reasons popularity sometimes occurs is [link]... want to discuss? come over to [link]."
I'd be willing to take on the routine task of posting or updating the templates on each page each week. Also I could add a note on the talk page with links as decided. A few times a year, I'm away from a computer for a week, but I could do it when I get back or ask for someone to do it for me. SchreiberBike | ⌨ 00:53, 5 November 2015 (UTC)
I was just thinking this through and it seems that the way it is set up, we wouldn't be able to place the template on a talk page until after the current week is over. A link to Wikipedia:Top 25 Report would be out of date in a week, and a page like Wikipedia:Top 25 Report/October 18 to 24, 2015 doesn't exist until a week after the report is posted, which might be two weeks after the article was highly viewed. One option would be to put a note on the talk page as soon as the new report goes up - that would have a link to the current report, then put up the template after the archived version is created. I'm also wondering how far down the list we should go - could start with top 10 and see how that goes, then go up or down as seems appropriate. SchreiberBike | ⌨ 03:35, 9 November 2015 (UTC)
SchreiberBike , I agree the redirect idea would work perfectly. As for how the list we should go, I think starting with the Top 10 would be good, I didn't have a particular number in mind, but the Top 10 is also all that we put in the Signpost.--Milowent • hasspoken14:04, 9 November 2015 (UTC)
In case anyone's interested about the other contestants. The mobile percentage looks legit (matches the other celebs), but Lady Colin Campbell is a massive outlier here. Of course, she's also one of the few people in the show who isn't already a TV star (Eubank in 2nd place and Dyer in 4th are both sportsmen; Shelley at 7th and Hadley at 11th are singers) and one of the few with an interesting life outside celebdom (she wrote some controversial books about the royals, especially Di, and she was raised as a boy because she was born with ambiguous genitalia – she claimed that when she divorced Lord Campbell, he spread the rumour that she was trans). Smurrayinchester14:22, 25 November 2015 (UTC)
(One note - I'm a Celeb draws 10-11 million viewers. If these stats are right, 1 in 4 viewers looked up Lady Campbell's Wikipedia page - or probably about 1 in 2 households. That's ridiculously huge number.) Smurrayinchester21:27, 25 November 2015 (UTC)
Smurrayinchester, awesome chart! And thanks for verifying that the mobile percentage issue is similar among contestants, that's a great cross-check. I wonder how many of these views came during the show, that is something that could be checked by Andrew, but it could take some effort and I don't have the technical expertise myself. --Milowent • hasspoken05:11, 26 November 2015 (UTC)
I say yes, exclude them -- they were somewhat suspicious last week and now its continued. Also if we go back to the WP:5000 posted on Nov 29[7], they were both about 50% mobile views. Change in percentages in red flag, I think, though not one we've had to focus on much before.--Milowent • hasspoken05:41, 15 December 2015 (UTC)
I'll agree with this judgement while noting that if Google were to manipulate its search results, its possible for such an effect to be completely legitimate. Imagine a mobile device search that returns the Wikipedia article first, whereas the desktop version does something completely different. I can't remember the paper, internal WP research, or bit of anecdotal evidence -- but it was shown Wikipedia's SEO power and people absent mindedly clicking of the first search result drive a humongous amount of WP traffic. Thanks, West.andrew.g (talk) 06:37, 15 December 2015 (UTC)
thanks for the link! i love the blogpost even though Solomon Northrup's #1 position on December 12 was caused by bot views, not actual human interest. We can see this because the hatnote report for that day shows it got 1.1 million views, and stats.grok.se says the same thing, which means it had no mobile views. Also the article has had no edits since December 8. This is a dilemma I don't know how hatnote and other automated uses of the API can easily fix. But this is a huge improvement over prior tools, I hope someone can replace stats.grok.se using this new API, having the mobile data there would be awesome.--Milowent • hasspoken03:20, 18 December 2015 (UTC)
well, We can look at the per-article endpoint, which has the breakdowns! Ok, so all agent types are 1.05 million, spiders are only 55 hits and users are the rest. So maybe stats.grok.se got silently updated? Not sure what's up, maybe we're still miss-identifying some type of automata. The low count on mobile-web and mobile-app access types does seem to prove your theory correct. I've filed a task to look into it. But I wanted to also write this analysis here so you could potentially use it to vet the top articles endpoint. By the way, we only use "user" agent types when we compute the top 1000, so spiders shouldn't get in there, but it does seem like we're not detecting all the spiders as you say. — Preceding unsigned comment added by Milimetric (talk • contribs) 23:50, 18 December 2015 (UTC)
See the force awakening in daily views.Milimetric, thanks for the update! I doubt anyone external is intentionally trying to hype Solomon Northrup. When the WP:5000 is updated in about 24 hours, I'll take another look. BTW, the new pageview api tool is completely awesome. Using the Hatnote 100 daily reports, I know Star Wars: The Force Awakens is almost certainly going to be the #1 article in the next weekly Top 25. Using the page view tool, I got the exact counts for Dec 13-18 (6 days out of the 7 day week for the Top 25) and see it has 4,137,490 views already. That is among the best weekly performances of an article for the year. These tools will also allow the Top 25 report to include more granularity on when an article was most popular in our commentary when warranted.--Milowent • hasspoken15:25, 19 December 2015 (UTC)
Hey gang! I am on top of my script writing this year, so I expect being able to produce two year-end reports in the hours immediately following the new year: (1) A report similar in format to the WP:5000 will appear at User:West.andrew.g/2015_Popular_pages, and (2) I will produce/link a text-based report showing which articles had the best "single day(s)" in the entirety of 2015. I will notify everyone below when these actually generate. We are unlikely to be the only ones producing such reports, so I think it is in our best interest to expeditiously produce and promote our work. These reports tend to generate interest both internal and external to Wikipedia; let's see how much traction we can get. Happy new year everyone. West.andrew.g (talk) 08:23, 31 December 2015 (UTC)
I look forward to this as well! On a related note, I am slated to prepare next weekly report (Dec 27 - Jan 2). The Signpost apparently skipped a week, so I think we'll need to combine two weeks into one Traffic Report - I am thinking of just putting both top 10 charts into one article listing each week, with a global summary at the top adapted from the two Top 25 reports. This would be separate from a 2015-full-year report, which I am happy to let Seren do if he wants to. cc: Ed.--Milowent • hasspoken14:02, 4 January 2016 (UTC)
Happy to help with this wherever needed, as I'd like to cross-publish the results on the Wikimedia Blog and (maybe?) pitch to a few reporters. :-) @West.andrew.g: do you know when the information can be released? Ed Erhart (WMF) (talk) 10:22, 5 January 2016 (UTC)
Done -- The year-end report has finally generated. Can someone help sanity check these results to make sure there hasn't been an egregious error in my code? I assume the 5000 report didn't generate because the CPU was so busy. I'll look into that now. Thanks, West.andrew.g (talk) 00:20, 6 January 2016 (UTC)
Legitimate events (not bot-driven) become more dense as one moves down the list. Note that the Main_Page has been removed from this list (it would have the top 365 spots). Remember that we measure days by UTC time. Feel free to beautify and distribute in any way the team sees fit. Thanks, West.andrew.g (talk) 17:19, 6 January 2016 (UTC)
@Serendipodous: Much of the last paragraph of your draft can be explained by the introduction of mobile statistics. This wasn't our fault, as the WMF wasn't giving them to us, but they play a major factor here. In 2013 and 2014 we saw no mobile traffic, yet mobile was likely becoming a more popular medium of viewership. Numbers on the whole probably weren't decreasing as much as indicated; readers were just shifting to the mobile format where we couldn't count them. The massive jump in 2015 is because we had mobile stats this year. Let's be careful about this in this write-up and that for the Signpost. West.andrew.g (talk) 17:38, 7 January 2016 (UTC)
I agree with Andrew about the mobile traffic (added since Oct 5 2014 in the weekly data), it makes it hard to compare totals. I think the drop WMF speculated about could well be the move to mobile. Also, I wonder if the Top 25 is a big enough sample even if had apples-to-apples data. A year with a few big events could get more views in the Top 25, but the Top 5000 might be similar.--Milowent • hasspoken18:00, 7 January 2016 (UTC)
@Serendipodous: You wrote "One colour I am very happy to see the end of is pink" referring to the Other category. However, the TV show category looks like a darker shade of pink to me. Any chance of having a more drastic color difference? Thanks! GoingBatty (talk) 01:25, 8 January 2016 (UTC)
Schedule
Whoa, the Signpost published on time already. We'll want this to be front and center in the next Signpost. Somewhere in there we can include a link to the Top 25 weeks which didn't make the signpost as a catch up and bring us back up to schedule. I'll take a look at adding titles to the tables.--Milowent • hasspoken18:00, 7 January 2016 (UTC)
Steven Avery Was Framed By The Police For Murder
New Top 25 for Dec 27 - Jan 2 is up, though its very late. But gotta have the historical record! User:Serendipodous, since you are preparing the year end report, I will offer to do next Top 25 (Jan 3-9) now if you are OK with that.--Milowent • hasspoken14:48, 11 January 2016 (UTC)
Negative feedback on talk page notice
@Boilingorangejuice: There was some negative feedback on the talk page notice at Talk:Martin Shkreli#This article made the Top 25 Report and I'd like to widen the discussion. I agree that the notice does not strictly follow the instructions on talk pages, but I think it does help make the encyclopedia better by drawing attention to the impact of the article. I'd note that I have been placing those notices on the appropriate articles in the top ten of the Top 25 Report since mid-November and most have had no response but there have been maybe 10 or so positive reactions and no other negative ones. The original discussion that lead to this practice is at Wikipedia talk:Top 25 Report/Archive 3#box on talk page for notable entries? Thanks, SchreiberBike | ⌨ 20:43, 24 December 2015 (UTC)
I don't really understand the issue here. An article's popularity has absolutely nothing to do with its quality; it has to do with relevance. Drawing attention to attention isn't about "jerking ourselves off"; it's about alerting people to articles that are generating views and so may need to be expanded to accommodate their audience. Serendipodous02:21, 25 December 2015 (UTC)
Yep, I wouldn't let one dissenting voice discourage our efforts. The postings are at minimum interesting, possibly spur editing efforts and retention, and are minimum in their space consumption. We are already applying human discretion to avoid annoying/duplicate postings. Proceed as usual. West.andrew.g (talk) 02:37, 25 December 2015 (UTC)
I respectfully disagree with your position. Any non-article discussion on the talk page consumes resources and attention from improving the article. Also we are not a for profit business so i don't understand " expanded to accommodate their audience". Our job is to make high quality article on even the most esoteric subjects. In should not be the case that a Justin Bieber article is much higher quality than an article on the metabolism of Proline.Boilingorangejuice (talk) 06:47, 28 December 2015 (UTC)
Are you going to put an RfC on this, because the original discussion seemed to be just people who were involved in the report page, rather than the subject pages? I agree to some extent with Boilingorangejuice: if someone wants to find out the most popular pages each week, then that's fine, they can look at the report, but I don't think a reverse lookup is relevant or appropriate. There is an implication (maybe unintentional) that these pages should receive more attention from editors whereas, actually, these are exactly the pages which already get the most attention (correlation between views and edits). There is already a system of ranking page importance within its subject area which is designed not to pander to what the latest media trend may be and it is a distraction to see pages on a watch list being flagged because they are flavour of the week. It's not about space on the talk page, but rather the jostling for attention on the watch list which bothers me. Btljs (talk) 07:53, 28 December 2015 (UTC)
I stayed out of the original discussion because for me the weekly surges weren't as important as longterm trends. That said, I do feel there is value in alerting Wikipedians to the popularity of articles, particularly over long periods. If an article, like, say, United States or World War II, has consistently received tens of millions of views a year without becoming a featured article, then we have to ask ourselves what are we doing? Is Wikipedia providing a service for its users, or is it simply a hobby site for its editors? Serendipodous09:47, 28 December 2015 (UTC)
Wikipedia's only customer is knowledge. It is not here to pander to trends, popular topics or memes. If War War II is not a featured article it is for more reasons than not enough superficial, poorly sourced, non-vetted, rushed edits. Featured articles happen over time and are not products of acclaim designed please the current demand of the populous. I'm sure Jimbo will back up this position. Boilingorangejuice (talk) 11:17, 28 December 2015 (UTC)
"Wikipedia's only customer is knowledge". What? That is the most meaningless statement I have ever read. Of course Wikipedia has customers, just like any other service provider, for profit or not. Wikipedia exists in the real world and is intended for human use, not for the alien archaeologists who will pick through the remains of our society after we're gone. If Wikipedia's sole purpose is for the masturbatory self-gratification of its editors, then those editors should just admit it, not make some contentless self-aggrandising statement about being a servant to knowledge. Serendipodous11:24, 28 December 2015 (UTC)
It sounds like I hurt your feelings for using that abrasive sexual metaphor. I apologize for that. No one is stopping you from personally editing on popular articles. That is your prerogative and you should edit what articles bring you enjoyment. But to argue that it should be a formulated policy that wikipedians should disproportionately focus on popular tends, I disagree.Boilingorangejuice (talk) 11:31, 28 December 2015 (UTC)
How do you know what will be important? Slender Man was locked from editing for months until I took the trouble to expand it, and lucky I did because a few weeks later two girls stabbed another in his name, causing the internet-illiterate boomer generation to rail aimlessly at threats they didn't understand. Most ended up quoting the article I wrote because they had nowhere else to look. The idea that Wikipedia should not "pander" to popularity is an abrogation of its responsibility as an information provider. Does that mean that I think List of Game of Thrones episodes should be a featured list just because it's popular? No. Does it mean that more attention should be paid to the environment in which Wikipedia exists? Yes. Serendipodous11:40, 28 December 2015 (UTC)
The slender man article looks really good, I'm glad you were able to add a lot of value there. I know quite a few individuals from the "boomer generation" that are light years more literate that I am when it comes to using the internet. Let's let some other users weigh in on this narrative. Boilingorangejuice (talk) 11:45, 28 December 2015 (UTC)
There is certainly some segment of the editing population which is concerned with "impact". If it would take 5 minutes of editing effort to improve article A and article B by X%, and one is disproportionately more popular the other, it would not be unreasonable to choose the more popular article. We are about to implement something similar in the anti-vandalism space. When doing recent changes patrol in editing assistants, one could prioritize edits for inspection based on page popularity. An edit with a 1% chance of being vandalism on say, a breaking news event getting several views a second, should be inspected before an edit with a 99% vandalism probability on a page that gets 1 page view a year. This would minimize the number of "(expected) vandalized article views". One could construct similar metrics surrounding article quality and impact. I am just making the point that these statistics are useful in practice. This isn't a claim that the talk-page use-case is the best way to make this information available to the editors/consumers for which it might have the most influence. West.andrew.g (talk) 02:03, 30 December 2015 (UTC)
Each editor can use whatever prioritisation system they see fit to decide what to edit, obviously. I would pay absolutely no attention to whether a page was popular or not before deciding to edit it, therefore I have no reason to see popular pages flagged in my watch list. Surely, someone who was interested in editing popular pages would actively seek them out themselves? This discussion is about inclusion on talk pages, not editing practice. If Wikipedia's future is under threat due to poorly edited popular pages (e.g. donations falling or bad press), then clearly that needs to shape policy, otherwise, I can't see any reason to promote pages which are, in any case, more likely to get edited due to greater visibility. Btljs (talk) 12:51, 30 December 2015 (UTC)
I'm not too worried about a comment that refers to statistics as "j*rking ourselves off." Yes, the stats show that inane topics get a huge amount of traffic and that is worth knowing. Now that we've done these notifications a few weeks, inevitably we would get some feedback, and that is good. Maybe an RfC would be a good thing to pursue for feedback. I am away for a few more days but will chime in further when I return! Happy New Years to all.--Milowent • hasspoken13:47, 30 December 2015 (UTC)
Given that objections have been stated to posting notices of articles being on the Top 25 Report, I feel that it would be in bad faith of me to continue to post the notices. Do we need an RfC to resolve the issue? SchreiberBike | ⌨ 15:35, 11 January 2016 (UTC)
The page for the 3-9 January report is showing just as a plain text, tab-separated list of page and image titles and numbers. The link at the top to the previous report (spanning the new year) redirects to the live Top 25. I'm not awake enough to try and sort this out. Thryduulf (talk) 23:46, 24 January 2016 (UTC)
I'm interested in the overall trend of mobile vs desktop pages views (for all articles, but the Top 25 would be a start). If you have easy access to such numbers, would you mind including the percentage in The Signpost? WhatamIdoing (talk) 01:30, 11 February 2016 (UTC)
The end of that page reports the sum of desktop and mobile views for pages in that list, but (a) most people won't look at that, and (b) it's only for pages on the list. Editors, especially experienced editors and admins, are overwhelmingly desktop users, and I think it would be appropriate for The Signpost's traffic report to help editors understand how different "what it looks like on my screen" is from "what it looks like on a typical reader's screen". WhatamIdoing (talk) 19:00, 11 February 2016 (UTC)
New report up
NB: While I don't have to be neutral when writing this, I try not to offend anyone, but it was really hard to be objective this week. Serendipodous12:40, 12 February 2016 (UTC)
It is difficult to be objective when the topic is "most hated men on Earth" :p I think you did alright in that regard. Perhaps you could describe some criticism of Bernie Saunders? He seems to be the only politician that isn't criticized this week. Regardless: very nice job! :3 ~Mable (chat) 14:41, 12 February 2016 (UTC)
I see the page view statistics link from every page's history page now leads to the nifty new wmflabs tool (e.g., [10]). Anyone know when this happened, or was reported anywhere? Seems worth mentioning the next report summary unless its been broadly covered while i've been absent. cheers.--Milowent • hasspoken22:30, 15 February 2016 (UTC)
JMinor (WMF), hello, this sounds like an exciting project you are working on. The talk archives of this project have more discussion on this topic from time to time, but generally speaking any article that has almost all mobile views (~95% or more), or almost no mobile views (~2% or less), is likely not being driven by human views. So if you look at the current WP:5000 version [11] I used to create the latest WP:TOP25, I discarded Web scraping (0.01% mobile); XHamster (94.49% mobile); Superintelligence: Paths, Dangers, Strategies (99.9% mobile), Java (programming language) (1.46% mobile, a long-term removal); and Images/upload/bel.jpg (0% mobile, not even a page). I am not a programmer, but checking the mobile percentages along those lines would probably filter out many of the current problem articles. A second thing we often look to is daily views to see if an article has a huge spike on one or two days, with very very minimal views on either side of that spike -- extreme spikes are not consistent with real human viewership, which always has an obvious curve in daily views whether it be over just a few days or a longer period. That might be a second parameter that could be monitored. Last thing I would note is that certain articles are inherently used more on desktops than smartphones - thus the main page was 31% mobile last week, and Deaths in 2016 was at 26.92%. Articles fueled by Reddit popularity usually have less than 50% mobile views (this is creeping up over the past 18 months), and most popular articles are today well over 50% mobile views. Hope this helps.--Milowent • hasspoken17:35, 1 March 2016 (UTC)
Thanks for the details. Agreed this seems like a very promising heuristic. For those following along we're going to put a "dumb" version of this in the first version of the iOS, and suggest we upstream a more sophisticated version as an optional parameter on the API. JMinor (WMF) (talk) 19:39, 2 March 2016 (UTC)
I can see why you got frustrated at the end, haha. I never knew wrestling was this big... Or maybe Wikipedia is just really big among wrestling fans. ~Mable (chat) 18:15, 1 March 2016 (UTC)
Yes, we seem to get a lot of wrestling traffic on the days of major "events", etc. Maybe there's not a good Wikia for WWE so they turn to us? JMinor (WMF) (talk) 19:41, 2 March 2016 (UTC)
Fixed both issues, but would appreciate someone double checking my edits. Thanks! GoingBatty (talk) 03:44, 28 May 2015 (UTC)
Done -- Posting issue is believed to be fixed. Old reports have been re-generated and posted and I expect the automated posting to succeed at its usual time over the weekend. West.andrew.g (talk) 15:19, 3 June 2015 (UTC)
Done -- The year-end report has finally generated. Can someone help sanity check these results to make sure there hasn't been an egregious error in my code? I assume the 5000 report didn't generate because the CPU was so busy. I'll look into that now. Thanks, West.andrew.g (talk) 00:20, 6 January 2016 (UTC)
