User talk:Jimbo Wales

(Manual archive list)

Graph of editor activity 10 years

The graph below, 2003-2013, shows the counts of article-edits (not talk-page edits), to easily see the stabilized pattern.

During 2012, the numbers declined by about a slight 2% per year, like saying a hot day was only Template:Convert/2. Meanwhile, support for old MSIE browsers remained poor, and if support for the world's IE7 and IE8 browsers improved, then the pattern might reverse as a slight growth when supporting the world's most-popular browsers (although, more likely the world will move to IE9 faster than WP moves to support IE8). The middle (purple) line shows the typical "semi-busy" editors (now ~9,200 people), with 25+ edits per month, which includes most everyone we find talking, because many veteran editors just cannot handle 100+ edits per month, but instead make perhaps "19 changes" in each of fewer edit-saves each month. The top line of 3,400 busy editors (100+ edits) includes the frantic small changes, such as putting a category link into each of 45 articles, and many veteran editors rarely perform mass edits, and so, they rarely exceed 99 edits per month. In fact, many of the 3,400 busy editors might be too busy to even talk very much about their editing. Anyway, that quick graph of editor activity shows "They are here to stay" after the past 10 years. -Wikid77 (talk) 13:52, 13 March 2013 (UTC)

Your conclusions might be valid (my guess is not, though), but one cannot tell from the graph. I think the edit counts of busy editors are down materially, in contrast to the claim of "stabilized" but it is difficult to see with that scale. Can you repost the numbers of the busy only, with a Y axis recalled to 0-5000? I think it will show a material drop, but one either needs the numbers, or a better scale.--SPhilbrick(Talk) 14:52, 13 March 2013 (UTC)

See monthly numbers in: http://stats.wikimedia.org/EN/TablesWikipediaEN.htm, including counts of editors who made over 2500 edits per month. -Wikid77 12:59, 14 March 2013 (UTC)

Thanks for the link. From a high of almost 4800 editors in 2007, it has dropped to 3400 in January of this year, an annualized drop of almost 6% per year. That is significant. The rate of drop has ameliorated in the last two years, but I would not yet call it stabilized.--SPhilbrick(Talk) 16:05, 16 March 2013 (UTC)

"support for old MSIE browsers remained poor" - how do you mean exactly? MediaWiki is one of a shrinking number of web platforms which still attempts to support the ancient horror that is IE6 (see mw:Compatibility#Browser). If you have examples of specific failings in IE, I suggest you bring them up at bugzilla. the wub "?!" 15:10, 13 March 2013 (UTC)

I think the unformatted text and lockups have been reported in Bugzilla, where after a few edits, the text just scrolls down the page, such as a navbox listing "95" items down a page (not a rectangular box of 95 entries), or right-side images appear scrolled down the left-side margin, or wikitables have no border lines. I cannot confirm IE6, but multiple sites of IE7 and IE8 showed Vector or Monobook skin merely scrolling down the screen, with few boxed areas except Search[_____]. -Wikid77 12:59, 14 March 2013 (UTC)

The stereotyping of "busy editors (100+ edits)" here is highly dubious - after all just over 3 edits per day, or 25 per week, will get you into this group, while hardly leaving most people "too busy to even talk very much about their editing". Johnbod (talk) 15:38, 13 March 2013 (UTC)

Well, the numbers are there, as 9,200 make 25+ article-edits per month, but only 3,100 people log over 100 monthly edits. -Wikid77 12:59, 14 March 2013 (UTC)

I know that; what I am objecting to is your subjective characterization of those "frantic" people. Johnbod (talk) 15:45, 15 March 2013 (UTC)

@Johnbod. Editors making 100+ edits a month is one of the main metrics tracked by WMF. The official term is "very active Wikipedians," I believe. See: THIS. Carrite (talk) 17:39, 13 March 2013 (UTC)

Ah, now I see your objection to the characterization of these very active Wikipedians. I agree that 100 edits a month is not at all a "frantic" pace. One could hit that number just by stopping by AfD every morning and adding a few opinions each day. Carrite (talk) 17:44, 13 March 2013 (UTC)

That's over 100 article-edits not talk-edits: A common misread is thinking the 100+ article-edit counts include talk-pages, or template edits, and stop around 100, but instead, the count at 100+ includes higher, such as 750 article-edits but no "wp:" edits nor talk-pages nor templates nor images. -Wikid77 12:59, 14 March 2013 (UTC)

This is hard to measure - I cant tell you by experience that older editors do fewer edits not because they are adding less, but due to the fact they edits smarter. Thus resulting in less edits for the same amount of information. Moxy (talk) 17:18, 13 March 2013 (UTC)

I have also noticed those multi-change edits, where some long-term editors might make 37 changes during 1 edit-save, rather than 9 edits. -Wikid77 12:59, 14 March 2013 (UTC)

I both agree and disagree with this. I can write entire sections in one sweeping edit, then go off and make a half dozen one-byte changes because I realized I screwed up repeatedly. (My recent performance at 1988 Winter Olympics demonstrates both quite nicely.) Though I would lean to the general idea that veteran editors can edit smarter. Familiarity with the tools helps. Resolute 17:57, 14 March 2013 (UTC)

While there is still plenty of work to do, we do now have over 4 million articles, so there are fewer new articles to write as well. I would assume that as Wikipedia matures, the number of total edits will go down. Most of the "big" articles are already done, although I can easily picture us having over 20 million article topics that are worth writing about, they are just minor things, like red slaw or 1950s American automobile culture (two of my more recent starts, which are a bit obscure.). It took 10 years to get those articles written simply because so few people have even heard of it or thought to write them. The big topics, like Ford, Romania, and Freddie Mercury were covered early on and few big topics are left unstarted and are already past the "stub" stage because many were interested in writing them, and so many reliable sources exist to source them. The most obscure (or minor, if you will) topics take longer and build slower. When I started in 2006, I found more errors as the main articles were rougher than they are now, thus more chances to make minor edits. If anything, that chart shows very strong continuing interest, not a lack of interest. Dennis Brown - 2¢ © Join WER 21:53, 13 March 2013 (UTC)

Perhaps 1/3 of minor topics have articles: There are so many missing articles, for common notable topics, that I find them redlinked everyday, such as recent "prison blues" or "creative incompetence" or "salad fork" (but "dessert spoon") or "slave brick" (but "mudbrick") or "commercial loan" or minor TV character actors, or minor footballers, where WP has over 74,000 footballer articles but among 242,000 known players. Many other terms, not redlinked, are couched as redirects to related articles ("consumer loan" redirects), but the terms could be in separate articles. -Wikid77 12:59, 14 March 2013 (UTC)

More to the point is the huge number of articles, many with high views, that have existed in a poor state for years, but have hardly been changed since an early spurt of editing. But statistics won't tell you that - only looking at some of them will. Johnbod (talk) 15:45, 15 March 2013 (UTC)

Whenever I see this graph I always wonder what the impact of the growing legion of bots is, especially ClueBot NG. With many millions of quick and mindless edits being done by machines it may mean that human resources are freed-up, making fewer edits but overall more substantive ones. We may see an acceleration of this as Wikidata comes into force. Has this ever been investigated? --LukeSurl ^{t c} 23:55, 13 March 2013 (UTC)

I find it rather hard to believe this graph is accurate. Even if it doesn't include bots or talk pages a lot of high output editors have punched out in the last few months to a year and I find it extremely difficult to believe that we have recruited enough new people to cover that. Rich F did over 10k a month and often more than 25K. I (Yes its Kumioko again) usually did over 10K edits a month and there were several months last year I didn't do any. That's just 2 of several. So although I find this graph interesting, its hardly believable. 108.18.194.128 (talk) 00:11, 14 March 2013 (UTC)

Other people have replaced the editors on break, even 5 people making 10,000 edits per month, or 50-60 who made over 2500 article-edits monthly. -Wikid77 12:59, 14 March 2013 (UTC)

• That is actually a very good question. I also wonder what the number of vandalism/reverts is now, compared to 5 years ago, as each incident creates at least two edits that are essentially null, whether they are reverted by a human or a bot. You could ask User:Okeyes (WMF) who works for the Foundation and does data work like that (his regular admin/editor account is User:Ironholds). He has the access and skills to provide some useful answers to that question, and from my experience, he has a general interest in those kinds of questions. The chart above is interesting and informative, but it is only one piece of the puzzle, so it is difficult to draw any definitive conclusions without more information. Dennis Brown - 2¢ © Join WER 00:16, 14 March 2013 (UTC)

I don't really feel like anyone cares or wants to hear my comments these days so I won't be asking about those questions. I just felt compelled to leave a note about my skepticism in the data (although I have no knowledge of how it was generated or how old the source data might be). If you want to ask though you certainly have my permission to do so (not that its needed mind you). As an extra note I think the point you made about the Vandalism/Reversion cycle is a good point too. I wonder how hard it would be to build an algorithm to create some categorization. I also think that using the revision history of cluebot as a measure of vandalism or even maybe a combination of that and edit summaries (it'll never be perfect mind you) that would be a good vetting process for some of the zero sum gains.108.18.194.128 (talk) 00:26, 14 March 2013 (UTC)

Graph shows live data as of January 2013: The graph above reflects the recent recounting of live Wikipedia article edit-history data, as dumped into data files for analysis. See data counts:

• http://stats.wikimedia.org/EN/TablesWikipediaEN.htm

The edit-activity data is not from "old reports" to combine guesses from years ago, but rather regenerated from the live contents of all of Wikipedia's current 4.2 million articles (as of January 2013). The data also includes counts of editors who made over 1000, 2500, 10,000 article-edits (or more) per month, which has also stabilized for years. For talk-page edit counts, see numbers in that "stats" data file. -Wikid77 12:59, 14 March 2013 (UTC)

• Compare stability of other-language Wikipedias: There are numerous similar graphs for the other languages of Wikipedia. However, looking at just the German, French, Italian, Arabic, Greek, Russian or Japanese, all show a similar stability of editor activity levels in recent years. See others:

• German: http://stats.wikimedia.org/EN/PlotEditorsDE.png
• French: http://stats.wikimedia.org/EN/PlotEditorsFR.png
• Spanish: http://stats.wikimedia.org/EN/PlotEditorsES.png
• Italian: http://stats.wikimedia.org/EN/PlotEditorsIT.png
• Arabic: http://stats.wikimedia.org/EN/PlotEditorsAR.png
• Greek: http://stats.wikimedia.org/EN/PlotEditorsEL.png
• Russian: http://stats.wikimedia.org/EN/PlotEditorsRU.png
• Polish: http://stats.wikimedia.org/EN/PlotEditorsPL.png
• Japanese: http://stats.wikimedia.org/EN/PlotEditorsJA.png
• Esperanto: http://stats.wikimedia.org/EN/PlotEditorsEO.png
• Latvian: http://stats.wikimedia.org/EN/PlotEditorsLV.png
• Romanian: http://stats.wikimedia.org/EN/PlotEditorsRO.png
• Swahili: http://stats.wikimedia.org/EN/PlotEditorsSW.png
• Chinese (growing): http://stats.wikimedia.org/EN/PlotEditorsZH.png

Numerous languages seem to have stabilized in their editor-activity levels, although some still vary widely. -Wikid77 (talk) 15:13, 15 March 2013 (UTC)

• I would bet that some of the growth on those other Wikis represents people who formerly edited in English Wikipedia, in spite of their having little ability to communicate in English, and have now started contributing to the Wikipedia in their own languages instead. --Orlady (talk) 15:28, 15 March 2013 (UTC)

To me, one of the reasons why it appears we hit a ceiling is because of the edit conflict, which is a technological barrier to collaboration. Jimbo, shouldn't the WMF put goal of having real-time collaborative editing (like google documents) down on paper, even if it is for year 2020? Biosthmors (talk) 16:43, 15 March 2013 (UTC)

I'm not sure why edit conflicts would have got worse over time. But of course edit conflicts are one of the major reasons why this site seems so snarky to newbies. It isn't that obvious how to resolve an edit conflict without losing your work, and I suspect that we only retain the editors who are sufficiently IT savvy to suss that out. There are several Bugzilla requests for various minor tweaks to our Mediawiki software that would reduce the number of edit conflicts (just treating the hash sign as a new paragraph and therefore a separately editable item would go a long way) Unfortunately the devs consistently rate such requests as less important than white elephants like the AFT and the Moodbar. This is something where Jimmy could intervene to great effect - a board level priority to the devs to find ways to resolve half the current edit conflicts without losing edits would make a real difference to this site and should be fairly easily achieved. ϢereSpielChequers 07:49, 16 March 2013 (UTC)

It seems to me that the developers are by and large just kids, with no real IT experience or understanding of the real world. But they bought in to the Wikipedia way, which is all that matters. Malleus Fatuorum 08:09, 16 March 2013 (UTC)

• Consider major innovations to increase editor activity: For my editing, the biggest increase in edit-article levels comes when assisting the wp:GOCE backlog-elimination drives, every 2 months. However, I wonder if some better tools could make editing more enjoyable, to encourage more article-edits per typical editor. The idea is to improve more articles, rather than just repeatedly edit the same articles, more times, but I suppose either increase in edit-activity could be considered an improvement. Definitely, the upcoming Lua-based wp:CS1 cite templates, to reformat major article cites within 3 seconds rather than 11-35 seconds, should re-attract the users who often avoided editing when they realized every edit-preview had become intolerably slow, slow, S-L-O-W. What other improvements could encourage editors to edit more? -Wikid77 17:26, 16 March 2013 (UTC)

Why editor activity stabilized in many languages

16 March 2013: After seeing clear evidence of a steady-state balance in the editor-activity levels, of many other-language wikipedias (but not Persian), then people question why editing continues, at a steady pace, regardless of either major problems or recent innovations. It seems there is some gigantic, over-arching force which attracts editors to contribute along an even keel, rather than have alternate periods of, perhaps, 50% higher offset by 50% lower swings in editor activity. However, some of the weekly variations have been smoothed, as disguised, by the small-scale chart shrinking the up/down curves into a narrow band. -Wikid77 06:18, 16 March 2013 (UTC)

There are probably several factors at work here. The big contrast is between our rapidly growing readership and our broadly stable editorship. One part of this is the rise of the edit filter, both in the sense that the more sophisticated the filters the more vandalism and vandal reversion we lose from these charts of editing, and because nowadays vandalism is rarely up long enough to be seen and reverted by our readers. So we need new ways to recruit editors to make up for those we no longer recruit as vandalfighters. We also need to improve the editing experience for mobile users. The Devs have done a great job at making Wikipedia readable on mobile phones, and our readership is now growing faster than the Internet. But we haven't yet cracked editing on a mobile, hence editing levels are not increasing in line with readership, and we are beginning to see the greying of the pedia as we become more dependant on the generation who are less mobile phone oriented. I'm documenting some of this at User:WereSpielChequers/Going off the boil ϢereSpielChequers 08:09, 16 March 2013 (UTC)

If you had to do it all over from the beginning...

If you had to go back and rebuild Wikipedia from the ground up, what- if anything, would you do differently?--Amadscientist (talk) 05:15, 15 March 2013 (UTC)

Were there other ideas you thought should have been used sooner, in the earlier days of Wikipedia? -Wikid77 17:26, 16 March 2013 (UTC)

Not really. Just curious.--Amadscientist (talk) 19:34, 16 March 2013 (UTC)

Some issues Jimbo has noted: I am not sure what wording could be searched in the archives, to find things to have done differently years ago, but some issues I recall from Jimbo's comments:

• Lower numerical consensus, perhaps as 63%: There have been several cases where a 2/3rds majority was needed, and the percentage was extremely close, such as a 64% majority, but not 66.7% or such. In some cases, Jimbo explained that the consensus was abundantly clear (unlikely to reverse), so that perhaps 63% in some decisions (among numerous editors) should be considered sufficient to decide an issue, but this would not apply to a small set of people making a decision, but rather when dozens of people respond with support/oppose. Such a redefinition for a two-thirds majority might have allowed major decisions to occur as months, or years, sooner. In the U.S. some obvious "landslide" votes are split only 55%-43%-2%, and it is often clear how the 55% are a sizable majority. -Wikid77 10:00, 15 March 2013 (UTC)

• Have a WYSIWYG interface sooner: Jimbo has commented extensively on the VisualEditor feature, as needed years ago, in possibly helping more people to edit articles. I am not convinced, because longer edits can be tedious in a point-and-click environment, where I think some people would tire of the too-many-keystrokes navigation needed to make multiple changes, and when edit-conflict occurs, then people would not have been using copy/paste to salvage their portions of the total page, so an edit-conflict would tend to require re-entering all those point-and-click keystrokes manually (rather than just cut/paste from the wikimarkup buffer we edit today). However, there has been talk of "forced-save" mode to ensure part of a WYSIWYG edit gets saved before an edit-conflict occurs, as less new text rejected by the edit-conflict. So, in that case, Jimbo is probably right about ease of use, with little loss from edit-conflict during longer edits. -Wikid77 10:00/14:53, 15 March 2013 (UTC)

• Some prior method to limit size of major articles: Jimbo has noted when many major articles were smaller, then they seemed easier to read, or more effective at covering topics (as well as faster to load or edit). If there could have been a smaller size limit, earlier, to insist on moving tangent text into subarticles for the myriad, rambing details, then perhaps major articles would still read like a comfortable overview of a topic, as perhaps 2-3 book pages, rather than 17-24 screens of text, tables, and lists. Now any talk of smaller pages might incur claims of "unfairness" against new topics, relative to older long pages. -Wikid77 10:00/14:53, 15 March 2013 (UTC)

• Higher thresholds for notability: Another issue has been too many articles about minor participants, such as villages of 23 people, or new schools of 35 senior students. So, there should have been more-stringent standards to require a school to have higher attendance, or longer history, before creating an article for every corporation who hires a few teachers and declares themselves a "school" rather than focus on larger public schools, or private schools with longer histories. -Wikid77 10:00/14:53, 15 March 2013 (UTC)

• Auto-assign names - If I could do it all over again, I would auto-assign names to new editors, then let the ones that stick around choose their own name. This would mean that almost all of the 18 million names used up would still be available. It isn't too late to try the approach see User naming convention proposal, but it would have been better had it been implemented at the start.--SPhilbrick(Talk) 16:16, 16 March 2013 (UTC)