Community Wishlist Survey 2019/Editing/Flag edits by new editors (editor retention tool)

Flag edits by new editors (editor retention tool)

  • Problem: Wikipedia is steadily losing editors. If a new editor has all their edits reverted, they are much less likely to become a long-term editor (survival drops from three-in-five to one-in-five). Even one revert discourages newbies. Revising or tagging new editors' edits does not have the same discouraging effect. It can even be taken as praise[1]; personalized constructive criticism is especially helpful.[2]

    In other words, every time I help two to three new editors make their first retainable, productive edits, I win Wikipedia a long-term editor and multiply my contribution; and to do this I need to treat the new editor with additional care. Problem is, I don't know who they are, and the user interface makes it difficult for me to not auto-bite newbies.

Experienced editors can deal with a bold revert and a line of jargon, and it's efficient, but it scares new editors off. I want to know when I am interacting with a new editor, so it's easier for me, as an editor, to behave in ways that promote editor retention. For instance:
  1. leaving edit summaries which are educational and comprehensible to a newbie (e.g. link all jargon), so the newbie can learn community norms
  2. tagging edits with Inline cleanup tags, so the newbie can learn what is wrong with their edits
  3. fixing edits (rephrasing copyvio or bias, sourcing, etc.), so the newbie can learn how to make good edits
  • Who would benefit: Increasing retention is critical to the long-term survival of our community.
  • Proposed solution: An icon-style flag on edits, saying this edit was made by a new editor, would be nice. It would also let me rescue edits others have reverted. A list of edits by new editors can already be generated by using filters in Recent changes, but I'd like to see the information in the article history, so I see it in my regular editing practice (that is, without going to a dedicated page, like Recent changes, or using specialized tools such as Snuggle or STiki). Others may prefer a similar flag in watchlists.
While I hope it would prompt help, such a newbie flag might also stigmatize new editors, and thus hurt their integration into the community. This should be tested. One alternative might be to flag only reverted edits by good-faith new editors, and have an edit notice prompting anyone reverting a new editor (especially using a tool) to be aware that this is a new editor, so they can react appropriately.
As I need to mention in the edit summary if I am fixing an edit of a declared-COI editor, a (different, obviously) COI flag for COI edits would also be useful. Flagging edits by vandals reverted with "rvv" with yet another flag might also be an easy extension.
Since helpful advice when reverting good-faith newbies almost always includes a referral to the Teahouse, it might be nice to have that added to the revert notice automatically for the first 2 months/100 edits (or empirically-determined thresholds).
I'm very much open to suggestions here, as I am aware that many others have more knowledge and experience than I in this area.
  • More comments:
  • Phabricator tickets:


Discussion of rescoped proposal

The proposal has been changed as above; comments on the new proposal are welcome here. MMiller (WMF), do you have views on how the modified proposal might interact with the Growth Team's work? As DannyH said, the proposal had morphed into a bit of a "have a Growth Team" proposal, but it's smaller now. HLHJ (talk) 01:40, 15 November 2018 (UTC)Reply[reply]


  •   Support Stussll (talk) 00:51, 17 November 2018 (UTC)Reply[reply]
  •   Support Liuxinyu970226 (talk) 04:25, 17 November 2018 (UTC)Reply[reply]
  •   Support Gnangarra (talk) 09:37, 17 November 2018 (UTC)Reply[reply]
  •   Support Barcelona (talk) 18:56, 17 November 2018 (UTC)Reply[reply]
  •   Oppose Old serious editors now dislike Wiki because any newcomers can wreck well done pages, which requested hours of effort. Once the time all stubs were welcome, now this phase is ended: all main subjects are covered, quality of articles is now needed. Too much indulgence with newcomers has the only effect that old serious editors will leave Wiki. A ntv (talk) 08:23, 18 November 2018 (UTC)Reply[reply]
    • This is an interesting hypothesis, A ntv, that wiki maturity caused the abrupt transition from exponential growth in editor numbers to a slow decline. However, the same transition occurred at the the same time on many, but not all, other wikis, which mostly have far fewer articles than the English Wikipedia.[7] It seems that old editors are leaving at the same rate as they did before the transition, but we are getting fewer new editors, leading to a steady net loss of editors.[8] Possible causes are discussed on Research:The Rise and Decline. HLHJ (talk) 23:24, 18 November 2018 (UTC)Reply[reply]
      • IMHO the abrupt transition occurs, in any Wikis, when the easy-to-find-sources (i.e. mainly online other encyclopedia or similar) have been fully used to create new articles (fun and rewarding job). After that the number of contributions depends from two items: a) the easiness to find sources (google book / libraries), and b) the need to defend the articles from troll/newcomers. In the future, to improve an article or create an interesting new article of true encyclopedic interest, it will be more and more difficult to find sources (libraries or scientific papers) and off-line work it will take more and more time. Editors who use now most of the time in selecting sources, are less and less interested in taking care of the newcomers who write trivial articles on not-encyclopedic people or make edits without even having read all the article they modify. Please protect the old editors.A ntv (talk) 07:52, 20 November 2018 (UTC)Reply[reply]
        • Research:The Rise and Decline and its associated paper discuss this theory, citing it as having been proposed by Suh et al. in 2009. There is limited evidence to support it; in a sample of human-rated newbie edits, good-faith newbies dropped moderately from 92.2% to 79.8% of all new editors during 2005 (total editor numbers were rising sharply and Wikipedia was much in the news, so this probably represents more vandals), and newbie edits to longer articles are more likely to be rejected, for instance.
          The transition occurred at the same time on de-wiki, tho, and they not only have fewer articles, they have much less well-developed articles (and a lower requirement for sourcing). A substantial proportion of the editors can read English at a near-native level, too. If writing useful new work had become more difficult because de-wiki editors had exhausted the easy sources and easy article topics, one would expect that the monoglot German-speakers might be less productive, while the people with good English would still be just as productive (as comparison with en-wiki clearly shows that there are plenty more usable English-language sources, and article topics). I have not seen any evidence of this.
          You are, however, now very likely to get an edit on either wiki rejected if it isn't initially perfect. If it has grammatical errors, formatting errors, or is unclear to the reviewer, it will probably get reverted, and it probably won't get fixed by someone else. The learning curve for new editors has become precipitously steep, and anyone who does not scale this learning cliff does not become a regular editor. The effects of learning-curve steepness on retention are known from video games. Most people won't climb cliffs. To keep going, they need a hike, not too steep, not too flat, continually challenging. As you say, the old editors need care; if we lose old editors faster than we train new ones, our community will dwindle and die. But we will lose old editors, if only to death, and we must train new editors fast enough to replace them. Currently we don't; editor number are falling, and areas of the wiki are falling silent. It's starting to feel like a ghost town. HLHJ (talk) 04:52, 21 November 2018 (UTC)Reply[reply]
    • I think that the best prophylactic against wrecking answer is the edit-approval (Sichten) system used on German Wikipedia PJTraill (talk) 23:14, 26 November 2018 (UTC)Reply[reply]
      • The edit approval system of the German wikipedia marks edits by newbies (less than 30 edits). It is positive feedback to get an edit approved. Minoo (talk) 21:36, 27 November 2018 (UTC)Reply[reply]
        • I think it's 50 confirmed edits or 150 edits, and 30 days of registration (policy). I don't know what proportion of edits by good-faith new editors are rejected under this system, and I'm not sure if anyone has tested what effect this has on recruitment and retention. HLHJ (talk) 03:46, 28 November 2018 (UTC)Reply[reply]
  •   Support NMaia (talk) 10:29, 18 November 2018 (UTC)Reply[reply]
  •   Support Timeshifter (talk) 15:15, 18 November 2018 (UTC)Reply[reply]
  •   Oppose reminds me of the StackExchange New Contributor Indicator --Frozen Hippopotamus (talk) 11:22, 19 November 2018 (UTC)Reply[reply]
    • That sounds like a useful experience, Frozen Hippopotamus. Thank you for the link. Do you know how it worked out long-term? Are there stats? The page you linked to has some comments which seem pertinent; one I liked was "Build the new user experience so that it explains how to use the site instead of relying on users to "be nice" by explaining it to them"; taking this view, this tool suggestion (which I frankly think is better than mine here) might be preferred to the help desk focus, though in practice they are probably complementary. The StackExchange comments also ~contain the ones I made here: testing the effects and making the new-user notice invisible unless the interaction is going to be negative. HLHJ (talk) 05:31, 20 November 2018 (UTC)Reply[reply]
      • I've since quit SO/SE, not because of this feature but because of the general low quality of the user contributions, especially the questions on SO (different story). Having been a simple user of these sites, I have no insight into how the feature was received once implemented. SE (the company) is not very keen to share raw data or stats about the success of such features. It may be difficult to measure, especially if there is no A/B test (half with, half without a feature) done. --Frozen Hippopotamus (talk) 08:01, 20 November 2018 (UTC)Reply[reply]
        • That's unfortunate, to say the least, but thank you for the info. I would strongly oppose making any changes that might significantly affect the editor community without good A/B testing. HLHJ (talk) 04:52, 21 November 2018 (UTC)Reply[reply]
    • It sounds as though you dislike the SE New Contributor Indicator, but I think it a quite reasonable (in spite of the flak in the answers on the linked page), though more important there than here. PJTraill (talk) 23:14, 26 November 2018 (UTC)Reply[reply]
  •   Support Zeromonk (talk) 08:25, 19 November 2018 (UTC)Reply[reply]
  •   Support Benjamin (talk) 10:29, 19 November 2018 (UTC)Reply[reply]
  •   Support BugWarp (talk) 01:15, 20 November 2018 (UTC)Reply[reply]
  •   Support Novak Watchmen (talk) 01:13, 21 November 2018 (UTC)Reply[reply]
  •   Support RIT RAJARSHI (talk) 19:38, 21 November 2018 (UTC)Reply[reply]
  •   Support I often look how many edits a contributor has when reviewing changes, but this would help too. PJTraill (talk) 23:01, 26 November 2018 (UTC)Reply[reply]
  •   Support WeegaweeK ❀  t  c  08:46, 24 November 2018 (UTC)Reply[reply]
  •   Support It will be of benefit for newbies. I know that some experienced Wikipedians dislike newbies' inexperience, but it's not the reason why we choose to discourage them. It is a way to balance newbies and the experienced, although it may not be the best. Mariogoods (talk) 13:35, 25 November 2018 (UTC)Reply[reply]
  •   Support — AfroThundr (u · t · c) 02:10, 26 November 2018 (UTC)Reply[reply]
  •   Support Dvorapa (talk) 13:17, 27 November 2018 (UTC)Reply[reply]
  •   Support Reminds me of the way rookie drivers on motorsports circuits have special identifiers on their cars (and how rookie firefighters also have markings on their helmets) Though I think new contributors should be informed of this and allowed to opt out ... some may not wish to be so singled out. Daniel Case (talk) 04:20, 28 November 2018 (UTC)Reply[reply]
  •   Support Tgr (talk) 08:12, 30 November 2018 (UTC)Reply[reply]