Wikiwand AI

Wikipedia:Requests for comment/Visual Editor for new editors

From Wikipedia, the free encyclopedia

Since 2023, when we last had a dedicated discussion about which editor new contributors should get, more research has been done on the choice between Visual Editor (VE) and source editor, and Edit checks have been introduced in VE. The Editing team has written about that research below. The 2023 discussion ended with no consensus, partially due to confusion about options.

a new editor gets a pop-up one their first edit. When they click the default option (start editing, highlighted) they get source, otherwise they get VE
The current journey of a new editor on desktop
a new editor gets the source editor, and can choose to switch if they try out the pencil button. The interface does not give a clue that VE would exist
The current journey of a new editor on mobile web

For those unfamiliar, with edit checks, new editors in VE get a pop-up when they add substantial text without references, asking them to add a reference (Reference check). They also get a pop-up when they copy from an LLM or another website (Paste check), and a pop-up when adding poor-quality sources (Link check). Work is underway to detect some LLMs specifically.

No recent head-to-head A/B testing has been done comparing VE with the source editor in Desktop view (see 2015 data, before automated references were implemented). There is more recent research from the recent A/B testing done with Reference Check, which has effects on revert rates, references added and retention. There is 2019–2022 pre-EditCheck A/B testing on mobile VE vs source editor, which shows more non-reverted edits, but a higher revert rates. In the A/B testing exploring Reference Check shows a lowered revert rate.

RfC summary question

Should Visual Editor become the default on mobile and/or desktop for new editors? Existing preferences are unaffected. In solidarity, —Femke (talk) 🐦 16:52, 6 July 2026 (UTC)

Please discuss in sections below

RfC on the default editor for new Desktop editors

First edit and the VE / source editor choice. Note that editors do not get an explanation of the choice, and that the default choice is to 'start editing' aka the source editor.

At the moment, new Desktop editors are given the source editor when they first click 'Edit'. A pop-up appears immediately, with the choice between "Switch to the visual editor" (grey) and "Start editing" (highlighted in blue; this corresponds to the source editor). Should new editors be given the Visual Editor initially instead, with a pop-up introducing the source code editor at a later edit?

  • Option A: Give editors VE initially, and show a prompt (such as a pop-up or guided tour) once they become autoconfirmed asking if they want to try out the source editor.
    1. Option A1: Afterwards, their interface stays the same: their single edit button will remember the previous editor, and they can switch via the pencil button.
    2. Option A2: Afterwards, they will get a second edit tab 'Edit source' in addition to 'Edit'; they can also switch via the pencil.
  • Option B: Keep the status quo, with a possible follow-up discussion to improve the wording of the pop-up.

Survey (desktop)

  • Option A1. We need less AI slop, more human editors, and I believe VE + Edit check is one key step in that direction. The community has changed the paste check prompt to warn against LLM usage. Now, this prompt only reaches a small subset of editors; those who choose VE despite interface discouragement. We can make these prompts much more prominent, preventing good-faith editors from breaking community norms on AI usage. The editing team is working on getting edit checks into more places (such as draft space), further reducing editor burnout in dealing with this slop.
    On the second point: we know that edit check leads to increased retention, and that when VE was still quite buggy, it did not perform worse on retention than source. I consider it highly likely that VE would lead to increased retention. Not only that, editors using VE with edit check create better edits. If editor numbers decrease, we need to improve edit quality of new editors to keep articles up-to-date. As for option A2: the editing interface is already too busy, even before we got the extra words with the watch star. Editing is overwhelming for new editors, and adding two cryptic buttons to the toolbar does not help new editors get comfortable with the interface.
    In terms of the results from the Edit check A/B tests of Reference Check in the background section: The revert rate on Desktop went down by 9.8%, and people were 2x as likely to add a reference. The additional retention from Reference check (+16% edits a month later) is mostly due to Desktop retention increases. In solidarity, —Femke (talk) 🐦 17:10, 6 July 2026 (UTC)
    What does increased retention mean? Thanks! Dw31415 (talk) 20:59, 6 July 2026 (UTC)
    That who make an edit are more likely to make another one a month later when shown reference check. This is likely because they are reverted less, so have a more welcoming start as an editor. In solidarity, —Femke (talk) 🐦 09:20, 7 July 2026 (UTC)
    A bit of expansion on my choice of A1 over A2: I'm pleasantly surprised to see that editors can find the source editor without having the crowded design with two edit buttons (see Editing Team comment below). We have a 1% difference is ability to find the source editor, and a 18% difference in the usage of VE. Having that extra 18% use source not only makes them less successful, it will also make it more difficult to teach Wikipedia at the Teahouse, help pages, events, as we must first ascertain how editors edit before we can start helping them. Having a clear default would help us help newbies. Like now, experienced editors will eventually find the settings to get both editor buttons. Our settings are a problem: I don't think we have a good way for intermediate editors to highlight which settings experienced editors often enable, but at 10 edits, we're quite far away from the point where you want to start promoting source instead of simply showing editors how to access it in a guided tour or other pop-up. In solidarity, —Femke (talk) 🐦 07:54, 22 July 2026 (UTC)
  • Option A1 per Femke. Between the additions of Edit Check features, the general ease of editing, and my recent experience on French Wikipedia (where VE has already been integrated as the default), this is clearly going to be helpful for new editors learning the ropes. A2 is a second option, although I agree that the interface risks becoming too busy. Chaotic Enby (in solidarity · talk · contribs) 17:53, 6 July 2026 (UTC)
  • Option A1. New editors are the people who benefit the most from being given the visual editor, as well as who need edit check. --PresN 18:04, 6 July 2026 (UTC)
  • Option A1 I think that VE as a default is a better choice for new editors, and clearly the numbers support that. As to A1 vs A2, I think I prefer A1 The pencil icon is more consistent with the mobile editing experience as well. As to whether autoconfirmed is the right threshold, I am uncertain, but it does keep consistent with other thresholds here, which I think is a plus. ASUKITE 18:28, 6 July 2026 (UTC)
  • A1 I've been eagerly awaiting this RfC, as it is high time to make VE the default for new editors. Folks expect a what you see is what you get editor, and the source editor is... discouraging to the non-technically minded. It takes a while to be able to see through the Matrix of source editor, and that just discourages new editors. While VE is not perfect, it is quite robust these days, and has clear advantages in areas like tables, citations, and many kinds of template. We need more newbies, and this is an easy way to retain more new editors. CaptainEek Edits Ho Cap'n! 19:00, 6 July 2026 (UTC)
  • A1 and enthusiastically so. I do not use VE myself but it is clear that it is now mature enough to be the default. We should have only one edit button, as prominent and enticing as possible, and every friction removed on the road between curious reader and beginning editor is a win with massive potential. Choucas 🐦 19:16, 6 July 2026 (UTC)
  • Comment Why is the suggest to only show them the option of sources editing once they become autoconfirmed? I would support VE being the default but why not show them the option up front? -- LCU ActivelyDisinterested «@» °∆t° 19:52, 6 July 2026 (UTC)
    @ActivelyDisinterested, see discussion below. The short version is "any additional friction added before the 'publish' button is probably detrimental". In solidarity, asilvering (talk) 19:55, 6 July 2026 (UTC)
    I see my question was discussed below.
    B VE should be the default but I don't see how explaining the option for source editing is in anyway a friction to editing, and so there's no reason to hide it until editors are autoconfirmed. -- LCU ActivelyDisinterested «@» °∆t° 19:57, 6 July 2026 (UTC)
    I should say "hide or minimise". -- LCU ActivelyDisinterested «@» °∆t° 19:57, 6 July 2026 (UTC)
    @ActivelyDisinterested: Just wanted to make sure you're aware (without trying to convince you of anything) that you've !voted to keep the status quo, which is to have the source editor by default. :) TechnoSquirrel69 (talk) 00:16, 7 July 2026 (UTC)
    Yes unfortunately there is no decent VE as default option, as all option for VE as default given by the RFC include not explaining the source option until autoconfirmed. I would support VE as default without that, but the RFC is setup to exclude it as an option. I'm unsure if everyone has notices that -- LCU ActivelyDisinterested «@» °∆t° 09:31, 7 July 2026 (UTC)
  • A2, per Femke re:A. I strongly prefer the A2 option for desktop editors, and don't think it adds appreciable clutter here. In solidarity, asilvering (talk) 19:58, 6 July 2026 (UTC)
    I should add: I strongly prefer text over icons in general because I find icons mostly inscrutable. I have no idea why a pencil icon is supposed to indicate to someone that they can switch to a wikicode editor; I can understand that "pencil" means "edit" in general, but once I'm editing, what on earth is "pencil" supposed to tell me? We have space on the desktop version. We can add two words for people like me who are icon-illiterate. In solidarity, asilvering (talk) 20:04, 6 July 2026 (UTC)
    The notice when switching to source editor for the first time.
    @Asilvering, we try to have the best of both worlds! This notice appears when switching to the source editor for the first time, explaining how to switch back. Sdkb‑WMFtalk 20:34, 6 July 2026 (UTC)
    But look at all that empty space that could say "edit source"! (Which doesn't rely on someone remembering a popup they probably closed reflexively.) In solidarity, asilvering (talk) 20:59, 6 July 2026 (UTC)
    +1 Ca talk to me! 13:53, 27 July 2026 (UTC)
  • A2. I'm going to say, while VE is helpful, the option to edit using source editor should be more prominent. Like asilvering mentions above, it isn't really intuitive that the pencil icon represents source editing. Epicgenius (talk) 20:21, 6 July 2026 (UTC)
  • A2 > A1 per Epicgenius, Femkie. Thanks to everyone involved for all the work that went into this! Dw31415 (talk) 11:43, 7 July 2026 (UTC)
  • A1, but it really depends on the guide not being shit (ngl if I had to do lots of pop-up guides I'd go do something else, railroading is so demotivating, let people actively learn). Personally I love having two different buttons for the editors, but I highly doubt it'd be as convenient for newbies. Kowal2701 (talk, contribs) 20:40, 6 July 2026 (UTC)
  • A1, as per Femke qcne (talk) 20:50, 6 July 2026 (UTC)
  • Option A1 per Sdkb's quite telling analysis below. I wonder if we should consider renaming the source editor to "code editor" or similar. As long as there's a prominent popup next to the pencil when someone first switches to source editor saying "here is how to switch back", I don't think two edit buttons should be needed. "Hiding" the source editor behind autoconfirmed isn't a problem for me; the users who know enough to benefit from source editor early on probably know how to read the project page that explains how to switch. lp0 on fire () 20:51, 6 July 2026 (UTC)
  • A2: VE is much simpler than the source editor, and with the added Editchecks, it'll make it easier for other editors as well. I slightly lean towards A2 to make it more clear to the editor how to switch, but I'm not opposed to A1 either. ARandomName123 (talk)Ping me! 21:59, 6 July 2026 (UTC)
  • A2, since I think making the option to switch visually obvious is beneficial once the source editor has been introduced, but but I don't feel strongly between A1 and A2 as long as VE is the default. No need to jumpscare the newbies with walls of wikitext, and the visual editor is sufficent for the vast majority of all editing tasks, especially ones suited to new editors. Rusalkii (talk) 22:10, 6 July 2026 (UTC)
  • A1 or A2: Either is fine, just not B. Still not sure how much Paste Check is doing to dissuade poor-quality edits but it's somewhat helpful in flagging them. Gnomingstuff (talk) 23:22, 6 July 2026 (UTC)
  • Options A1 or A2, without a preference between the two. Femke shared this data with me some time ago, and I've believed since then that making VE the default will result in immediate benefit to the project on multiple fronts. TechnoSquirrel69 (talk) 00:14, 7 July 2026 (UTC)
  • A1. As someone who actually has the visual editor disabled, I think this is an excellent idea. Anecdotally – and I'm sure others can corroborate – I've been asked many times if I'm coding while using the source editor with others watching. I think this encapsulates the issue: the source editor simply looks imposing, confusing, and alien to a layperson. It certainly did to me when I started editing at first, years ago. This lack of familiarity is the entire reason the visual editor exists in the first place, so I see no reason not to take the logical step of making it the default for newbies. Yes, some bugs still exist in the visual editor, but the benefits of simplicity and palatability to a general audience outweigh the risk in my eyes.
    Despite the concerns above, I think the first option should fine as long as the pop-up makes it abundantly clear the the pencil can be used to switch editors, say, by using a timer before the user can click off to ensure they digest the information. Unless it can be shown that the first option is harmful, I don't think the added clutter and confusion of the second is worth it. UpTheOctave!  8va? 00:21, 7 July 2026 (UTC)
  • A1 or A2. VE with Edit Check is great, and I believe making VE the default will help more people make constructive contributions. I've opted into the beta Suggestion mode feature, and I'm looking forward to eventual wider deployment of that feature in VE as well. Dreamyshade (talk) 03:00, 7 July 2026 (UTC)
  • Options A, Visual editor should be default, full stop. Don't care beyond there. --📎 JackFromWisconsin (talk | contribs) 03:32, 7 July 2026 (UTC)
  • Option A2 Strong support for the VisualEditor, but I also think there needs to be some clear visual cue about editing Wikitext being an option.--Carwil (talk) 03:37, 7 July 2026 (UTC)
  • Option A, either version. Based on watching "live" as dozens of newcomers make their first edits, both "starting with VE" and "demanding fewer choices" will help enormously with overwhelm. ~ le 🌸 valyn (talk) 04:31, 7 July 2026 (UTC)
  • Option A1; VE is better for the majority of newcomers and this should have been done a while ago. Yes, it still has some issues, but they are a lot more likely to be fixed if we have more people using it. Not hugely concerned about discoverability of source editing as people who desire that option will most likely be technically adept enough to find it. Mir Novov (contribs | talk) 04:36, 7 July 2026 (UTC)
  • Option A1 VE is better as the default for new users. The pop-up is annoying as a new user doesn't have the context to choose between the editors. – SD0001 (talk) 06:26, 7 July 2026 (UTC)
  • Option A (either version). Adding to the above reasons, WP:AFCHD gets too many posts that say "I used an LLM because I don't know how to write wikitext". Helpful Raccoon (talk) 07:43, 7 July 2026 (UTC)
  • Option A1 FaviFake (talk) 11:26, 7 July 2026 (UTC)
  • Strong A2 - Big yes to making it the default. As for why A2 instead of A1: Regardless of how it happens, newbies regularly find themselves lost in the source editor. Once that happens, in a scenario where the interface remembers the last editor used, nobody ever thinks to click the pencil icon (why would a pencil mean "switch editing modes"?). Separately, the pencil icon should be replaced, but until there's a more obvious "switch modes" button, it's A2 for me. Rhododendrites talk \\ 13:59, 7 July 2026 (UTC)
    Yes @Rhododendrites! I train new editors. I have to spend 10 minutes out of an hour session helping people find and change the default setting away from "remember my last editor" because the next thing I will do is have them sign-up on an on-wiki event page so that I can teach them how to sign their name with tildes. It is frustrating to have to immediately change the default new editor settings but what would be worse would be not to, and let them leave the workshop and find themselves at some later date stuck in source mode with no obvious way to get back. I believe this is a significant barrier to editor retention and would love to see it solved. I can't quite work out the A1/A2 options and how these fit with what I'd like to see (brain lag from Wikimania) but something really needs to change. If none of these options involve changing the "remember my last editor" default then maybe this RfC isn't going to help my situation though? DrThneed (talk) 16:37, 26 July 2026 (UTC)
  • Support A2 – I agree with the so far fairly overwhelming support for VE being the default. It's easier and more newbie-accessible. Simple as that. I would prefer the default to be having source editing as a visible alternative though, as the alternative is really hiding it away and as a consequence risks giving it the appearance of an obsolete or legacy way of editing, when in reality it is something that, once you put the effort in to learn it, is in many ways superior to VE and still a necessity for more complicated forms of editing. I sure know that I'm happy to have made the effort to learn it when I began editing seriously. –Maltazarianparleyinvestigate 18:42, 7 July 2026 (UTC)
  • A1 or A2. I agree with previous editors on VE and edit checks, but don't have a clear preference over the sub-options. I understand arguments (e.g., asilvering) that A1 is less intuitive, but also see how A1 would be preferred for simplicity (e.g., UpTheOctave!). Wracking talk! 19:33, 7 July 2026 (UTC)
  • A1 > A2, Oppose B, I think a thing to keep in mind is that for newer editors who are coming in for the first (or even the second time), showing two tabs with "Edit" vs "Edit source" is more likely to confuse them and cause some people to have a paralysis of choice (or click the wrong one). I do want a much better UI to switch editors since I think that would benefit both older editors (like me!) and more technical edits from newer editors who want to use the source mode. I do not see a reason for us to in 2026, default to the wikitext mode for most article edits presenting the new editor with a jumble of markup that they need to learn dramatically increasing the bar for contributions on the wiki. (Disclosure: For transparency, I ended up learning that this RFC had started because someone linked me to this RfC offwiki in response to a question I had. The discussion did not affect my vote.) -- Sohom (talk) 00:33, 8 July 2026 (UTC)
  • Strongly support Option A2 There are simply some things which are excessively complicated to do in the Visual Editor even though it's the best for the vast majority of cases. Both buttons should be visible with VE as the default, with an option to disable the second button in user settings. Option A1 would make these cases needlessly complicated while the user figures out how to switch to the source editor and back again. OmegaAOLtalk? 08:09, 8 July 2026 (UTC)
  • A2 > A1: WYSIWYG interfaces are what people are most familiar with in general. This should be respected by making VE the default for new users. Source editing makes certain tasks easier, but new users already have a lot to learn; Wikitext just makes the initial experience all the more tedious. A2 potentially makes editor switching more clear. Kodning 🌸 (talk) 10:27, 8 July 2026 (UTC)
  • Option A. We really, really need more new editors, and setting VE as the default is an obvious way to make that happen. I don't care which version of Option A we choose. Toadspike [Talk] 15:05, 8 July 2026 (UTC)
  • A2 visual editing is easier, but the option for source editing should be easily visible so a possible later transition doesn't seem a huge barrier. ~~ AirshipJungleman29 (talk) 16:56, 8 July 2026 (UTC)
  • A1 I joined in 2018 when VE was more developed and have thought making Source Editor the default was cruel and unusual punishment to new editors and if Source were the only option, I would have gave up. To post my !vote I am using the Reply function in order to avoid using Source. That's not to say Source is not better at some things but at the same time it is worse at others and what it is better at is not nearly enough to justify keeping it as the default especially for the basic editor. If this passes, which seems it will, there will be some updates needed, for example Help:Referencing for beginners/WP:REFB assumes one is using Source Editor and I'm sure other pages/guides do to same and should be updated before the switch.. S0091 (talk) 20:38, 9 July 2026 (UTC)
    S0091: I'm making a list on the talk page of help pages to update. Feel free to add anything you can think of there! There's quite a lot of dated stuff that can be deleted at the same time at those help pages. In solidarity, —Femke (talk) 🐦 20:56, 9 July 2026 (UTC)
    For transparency, you fixed my indenting because I used the Reply function rather than just leaving it the way I posted it. No biggie but you should have made that clear because now my statement about using Reply rather than Source does not make sense. I will keep an eye out for other pages that might need to be updated. Thanks for starting this RfC @Femke. S0091 (talk) 21:22, 9 July 2026 (UTC)
  • A2... reluctantly – Admittedly, I'm more of an old-school editor personally favoring source editor. Nonetheless, I've yet to hear people off-wiki favoring the wikicode/source editor very much nowadays. Rather something like Microsoft Word or another modern word processor has been what most computer users have been used to... a lot. Really dunno which logged-out editor is willing to register and then become a logged-in editor before becoming autoconfirmed enough to be shown a prompt, i.e. pop-up choice between VisualEditor and good ol' source editor.
    Oh, and why A2? I'd still really like to give newer editors easier access to choose between two editors. Reducing the access to just the pencil and eye icons in one spot is more of a hassle than simply giving newer users "edit" and "edit source" options. George Ho (talk) 21:24, 9 July 2026 (UTC)
  • A1.5. Either way, for new editors, the large blue button showing "Start Editing" will immediately direct users to the VE, and the secondary button should say "Use the Source Editor" instead (or, hide it under some "Advanced..." dropdown menu). After becoming autoconfirmed, or after ten edits, prompt the user with "Would you like to try out source editing?", and some information on basic source editor markup. In the same dropdown, allow users to configure whether they want to show both VE and SE tabs on the top toolbar without going into Special:Preferences (just like what I chose). The asking of whether to "Remember my last editor", "Always use VE/SE when I click "Edit"", or "Show me both editor options" explains why I put a ".5" in my choice. Hason-LEK-SINLet’s chat!My contribs 08:38, 11 July 2026 (UTC)
  • A1 keep the interface simple. – robertsky (talk) 09:46, 11 July 2026 (UTC)
  • A1 — "edit source" just seems utterly confusing for a new editor, and those who are really looking for the source code are probably savvy enough to find it with ease. Prioritizing the visual editor as much as possible is beneficial for everyone. – Aza24 (talk) 22:37, 11 July 2026 (UTC)
  • A1 or A2 - I definitely think VE should be the default, since it's more user-friendly. I use both VE and the source editor, but if I weren't a programmer who's used to plain text, I suspect I would mostly use VE. I haven't noticed serious issues with VE. Cadddr (talk) 19:03, 13 July 2026 (UTC)
  • A2 For essentially the same reasons as others who have voiced this preference. Just-a-can-of-beans (talk) 16:53, 14 July 2026 (UTC)
  • A2 over A1. A2 is already how many other language versions are presenting the options. Rjjiii (talk) 14:41, 15 July 2026 (UTC)
  • Option AEverything should be editable in the Visual editor. Run an experiment for A1 or A2, whatever results in most edits. Rolluik (talk) 00:27, 16 July 2026 (UTC)
  • Option A1 It is shocking that even now the visual editor is not default. It is likely that the source editor drives away countless editors - it certainly did for me for many years until the visual editor was invented. It is half past time for the source editor to be relegated to a power user or emergency tool (even though, frankly, I find the visual editor's features even more powerful than the source editor due to the WYSIWYG interface). I would tend to agree that the interface should be kept simple rather than having multiple edit buttons, something that may be confusing. ᴢxᴄᴠʙɴᴍ () 20:58, 18 July 2026 (UTC)
    You began editing in 2004. VE wasn't introduced until 2012. DS (talk) 20:08, 26 July 2026 (UTC)
  • A1. This would make it much easier for new editors to use VE. I oppose A2; when I was a new editor, I was very confused about the difference between the "edit source" and "edit" button was very confusing, and having it would add unnecessary clutter. Axolitl (talk | contribs) 17:16, 19 July 2026 (UTC)
  • A2. As someone who uses both source and visual editors, having access to both is really convenient and I love having either of them available with a single click without faffing around in menus. JCW555 (talk)♠ 04:05, 22 July 2026 (UTC)
  • Option B first, A2 second choice I’m apparently the only person who finds visual editor clunky and confusing, but that’s my experience of it. Wikitext/source has a bit of a learning curve, but once you get the hang of it, it is significantly easier to use. I don’t think we should be defaulting people to the inferior solution because the better one takes a bit of time to learn. Personally, if VE had been around when I started editing, I would never have become an editor. I understand improvements have been made, but I don’t view the statistics about less reverted edits as being a positive. You lose out on editors who make mistakes that way, and overall, I don’t view any of the positives mentioned as actually being convincing. Well aware I’m going to lose the argument on B, but I think it’s worth stating because hopefully it will get others to reconsider A1 in future comments: that is clearly the worst possible option. TonyBallioni (talk) 12:52, 22 July 2026 (UTC)
    Now that the editing team is here, are there any concrete things you'd like to see improved in VE if it becomes the default? Does clunky mean slow to you? Or clunky in a different way? In solidarity, —Femke (talk) 🐦 15:27, 22 July 2026 (UTC)
    I'm curious about your technical setup though. I do agree the VE can be slow, especially for larger articles and when you have multiple tabs for finding sources. But with a small article and that being the only tab, it loads quite fast.
    I don’t view the statistics about less reverted edits as being a positive Going to rebut this. Since when were edits 100% reverted? It's more like "overwriting" erroneous edits by new editors with an improved version by experienced editors. Hason-LEK-SINLet’s chat!My contribs 23:08, 22 July 2026 (UTC)
  • A1 or A2. I'm about as old-school as you can get. My first word processor was ed on a ASR-33 and roff. But come on, this is 2026. VE is not perfect, but it's miles ahead of editing raw wikitext in that clunky interface that looks like it was designed a quarter century ago. It boggles the mind that anybody would want to inflict that on a new user that we're trying to attract to the project. RoySmith (talk) 19:54, 25 July 2026 (UTC)
  • B VE is still a mess. The learning curve for source editing is not very high. Traumnovelle (talk) 00:04, 26 July 2026 (UTC)
    @Traumnovelle: what about VE would you like to see improved as a priority? In solidarity, —Femke (talk) 🐦 07:44, 26 July 2026 (UTC)
    The inability to handle shortened footnotes, the random white space it often adds to edits, and the inability to name references properly are all issues that effect non-VE users. Traumnovelle (talk) 09:49, 26 July 2026 (UTC)
    ( peanut gallery comment) um, it's actually "affect", not "effect". 🤓
    Also, why did you say "non-VE"rather than "VE"? Hason-LEK-SINLet’s chat!My contribs 09:54, 26 July 2026 (UTC)
    Fully agree about the first one. I submitted a wish that was declined on the topic, as subreferencing is "supposed" to replace sfn. I can't see English Wikipedia making that switch anytime soon despite my enthusiasm for subreferencing. In terms of good news, we might get subreferencing before the end of summer in enwiki.
    My possibly incorrect understanding is that the unification of mw:parsoid (a single parser for both editors) makes it easier to fix some of the weird white spaces or other elements of ugly diffs.
    As for the last one, the wishlist team of WMDE is working on fixing the problem by automatically generating sensible refnames, which should reduce the tension between VE users and non-VE users. Defining refnames is quite unintuitive for VE users, as it's invisible, so I'm glad they're taking the approach of sensible, community-configurable, automatic names instead. In solidarity, —Femke (talk) 🐦 12:04, 26 July 2026 (UTC)
    BTW, these are called slugs. The concept has been around since at least the linotype days and there's slugification libraries available in many languages.. Yes, VE should do this automatically. And, yes, VE should not mess with whitespace. But the primary goal of making VE the default is to improve new editor retention. Neither of these problems are visible to new users, so trying to make them blockers is just gatekeeping. RoySmith (talk) 12:51, 26 July 2026 (UTC)
  • B>A2>A1. You need to know Wikitext anyway in most cases to meaningfully edit templates (like infoboxes), and for talk pages and so on. Fram (talk) 10:20, 27 July 2026 (UTC)

Discussion (desktop)

  • I am open to perhaps being convinced towards VE, although checking now it still can't handle names outside of a last name first name pattern, but for option A I don't understand why the information about having two UIs for editing is left until autoconfirmed. It seems helpful to provide information on this as soon as possible, and perhaps it might entice some exploration for someone wanting to see how the two UIs compare. CMD (talk) 17:18, 6 July 2026 (UTC)
    One thing I hadn't appreciated before asking the Editing Team for research, is just how many folks click edit without publishing an edit (only 17% of edit clicks lead to an edit). Every time you introduce friction (e.g. a pop-up), that reduces the number of edits that are made. We saw this with edit checks for instance, which decreased overall edits (but increased non-reverted edits). I want people to start being hooked before we distract them with an option like this. A further reason is that it's much easier to explain the choice people get when they've experienced one of the two editors first. I see source editor as an advanced tool, and if it were up to me, I'd introduce it at edit 100 or 250. I chose edit 10 because I didn't think there would be consensus for such a high number.
    For patrollers, having the tags from VE is also important. You can see that paste check was shown, so that it's easier to figure out if an edit was AI or copyvio. I know there are quite a few patrollers that are struggling to deal with the deluge of AI slop, and by postponing the introduction of source, we give them more signals to work with.
    The editing team is listening, so hopefully we can put the issue of non-English name patterns higher on their agenda :). In solidarity, —Femke (talk) 🐦 17:35, 6 July 2026 (UTC)
    Good question, @CMD. Assuming it's accurate for us to think you see making newcomers aware of the source editor as a matter of them understanding the full picture, there are a few things I'd add to Femke's reply:
    Unlike most other situations in which we have an autoconfirmed threshold, the goal isn't to block new users from doing anything — they'll still have access to switch to the source editor via the pencil icon if they wish. Rather, the idea is to reduce the sense of overwhelm that newcomers repeatedly tell us they experience by flattening the learning curve.
    It can be easy to forget it when editing has become instinctual for us, but we know from research that newcomers opening the editor for the first time are juggling a lot: the fear of making a mistake, uncertainty about where to get started, and (if using source) confusion about an unfamiliar markup language. At the same time, there is a ton of vital information that we want to communicate to them, both technical (e.g. add an edit summary!) and policy-related (e.g. cite your sources! Don't use LLMs!). However, we've seen that there is an extremely limited amount we can say before users start just ignoring interface messages/pop-ups. By not having the interface prompt newcomers to explore source editing in their very first edit, we free up mental space to draw more attention to the information that they truly need to know from the start. Once they are a few edits in, they are more likely to have familiarized themselves with the basics well enough to have capacity to explore the source editor and its potential to help with more advanced kinds of edits. Cheers, Sdkb‑WMFtalk 19:16, 6 July 2026 (UTC)
    Regarding your question about names, am I correct in understanding that you're referring to how, when adding a citation using VE using the “Automatic” tab, VE may not always understand how to match up names with |first-name=, |last-name=, etc.? To help us explore a fix, it'd be helpful if you could point us to a Phabricator ticket for the issue or create one if it's not documented there.
    One meta point: As I mentioned below, we know that VE still has flaws. We are always looking for input on how we can improve it, and are open to receiving that here, but we wouldn't want a focus on particular issues to distract from the broader picture. We could theoretically wait until we have made VE perfect to deploy it as the default, but every moment we do so is another moment in which a potential future editor clicks edit for the first time, sees a maze of code, and quits. So when considering the options Femke has presented, I encourage everyone to think first and foremost about whether VE in its present state is going to be better for a newcomer than the maze of wikitext. Cheers, Sdkb‑WMFtalk 19:20, 6 July 2026 (UTC)
  • I think I support option A, possibly with a choice between A1 and A2 presented after. As to the autoconfirmed part, I understand the rationale behind locking source editing behind autoconfirmed, but that will likely be controversial. That's a big change, as it will take new editors 10 edits and 4 days before they can access the source option. This would effectively lock anonymous editors out of using source. I think perhaps that bit needs some clarification. I think after "an few edits", the option can be presented, and the ability to change editors in prefs should be retained. (That's just me being nitpicky - I'm reinterpreting this from the perspective of discussing the onboarding process only and now and I realize these might not be sweeping changes, but it's still good to clarify up front) ASUKITE 17:43, 6 July 2026 (UTC)
    The pencil (next to publish in VE) will still be available from the start, so they can access source, and would not even need to go into prefs to gain access to source. In solidarity, —Femke (talk) 🐦 17:48, 6 July 2026 (UTC)
    Funny enough, I've never used that icon. I always click between "Edit" and "Edit source", which I suppose from the perspective of somebody not already familiar with the interface, is itself confusing. I think I'm convinced. ASUKITE 17:53, 6 July 2026 (UTC)
  • For my personal curiosity, what was the reason not to propose as an option what is done by default in most wikis, i.e. showing two editing tabs at the same time with one labelled ‘Edit source’? A2 is the closest to it but still seems to imply that ‘Edit source’ tab would never be available at first. stjn 19:05, 6 July 2026 (UTC)
    I hadn't realised how many languages have those two options. That said, I'm not sure if it was an old WMF default or a conscious choice in these languages. The word 'source' is not really that intuitive for most people, as it's coding jargon. So having these two buttons would introduce a bit of mystique to the interface that could form a barrier to new editors. They might ask themselves if either of the two button is the 'correct' one and decide to click neither because of this friction. I didn't want to introduce an option I thought would lead to a worsening of the user experience (in that sense at least). In solidarity, —Femke (talk) 🐦 19:24, 6 July 2026 (UTC)
    Looking at https://noc.wikimedia.org/conf/highlight.php?file=InitialiseSettings.php (search for wgVisualEditorUseSingleEditTab) it seems the VE default is actually two tabs (edit and edit source). Wikisource, Wiktionary and Wikivoyage wikis as well as ~50 Wikipedia language versions (most of them smaller projects, except arwiki, bnwiki, cawiki, enwiki, kowiki, jawiki, zhwiki and a few others) are using the single tab experience. Johannnes89 (talk) 15:36, 10 July 2026 (UTC)
  • Hi all! The ongoing discussion about options A1 and A2 made us curious about:
    1. How newcomers interacted with the edit button(s) on wikis where buttons for VE ("Edit") and source editing ("Edit source") are both visible compared to wikis with a single VE button
    2. How likely newcomers are to visit the source editor on single- and two-button wikis where VE is already the default
So, we analyzed some data that we'd like to share. Overall, we are prepared to support either outcome, but we hope this is helpful in informing your decision.
Among first-time editors who opened the editing window at Wikipedias where VE is the default and two edit buttons are offered, ~28% entered the source editor. (For comparison, on Wikipedias where a single edit button leading by default to VE is offered, ~10% entered the source editor, likely because they were logged-out editors who had a cookie saved with their editor preference.) Although the situation would be a little different at the 10th edit, overall we think this shows that the presence of a second button leads many newcomers to initiate an edit session in an interface where we know they are less likely to be successful.
On single-button VE wikis, 66.6% of new users who made at least 10 edits within the 90-day analysis window visited the source editor at least once by their 10th edit, compared to 67.2% on two-button wikis. We think this shows that the presence of a second, "edit source" button does not meaningfully affect the likelihood that newcomers will discover the source editor.
Based on this data and the recommendations of our design team, we think A1 is preferable as it keeps English Wikipedia a single-button wiki. Having both buttons visible is a great choice for experienced editors who want easy access to them (personally, I changed my settings to enable it), but editors with just 10 edits are not yet experienced. They are still learning to navigate Wikipedia's interface, and the simpler we can keep it for them the better. A single "Edit" button is less cluttered and clearer than "Edit" and "Edit source" for someone who may not yet understand well how those options differ. As they become more experienced, they'll naturally open their settings at some point and be able to opt for two buttons then if that's what they want. And for those that enter the source editor and wish to switch back to VE, the notice I shared above prevents confusion.
Cheers, Sdkb‑WMFtalk 19:20, 21 July 2026 (UTC)

RfC on the default editor for new editors on mobile

The following discussion is an archived record of a request for comment. Please do not modify it. No further edits should be made to this discussion. A summary of the conclusions reached follows.
Option A has unanimous support for mobile, closing early per WP:SNOW. Desktop discussion can continue above. Levivich (talk) 15:32, 10 July 2026 (UTC)

On mobile, new editors are now given the source editor. Should the initial editor for new editors be the Visual Editor instead? People will still be able to switch.

  1. Option A: Make VE the default for new mobile editors.
  2. Option B: Keep the source editor as the default for new mobile editors.

Survey (mobile)

  • Option A: The data is even more clear for mobile. Mobile editors really benefit from the extra support they get in VE. The key numbers from the research linked in the introduction: In 2019–2022 (pre-EditCheck) A/B testing on mobile VE vs source editor, new editors shown VE made more non-reverted edits (+44%), but had more reverts (+26%). In the A/B testing done with Reference Check there was a −24% revert rate and mobile editors were 17.5x more likely to add a source. That 17.5x times is not a typo. In solidarity, —Femke (talk) 🐦 16:55, 6 July 2026 (UTC)
    Option A per Femke (Summoned by bot) Dw31415 (talk) 14:10, 7 July 2026 (UTC)
  • Option A, again per Femke. Chaotic Enby (in solidarity · talk · contribs) 17:56, 6 July 2026 (UTC)
  • Option A. The data is clear here. --PresN 18:05, 6 July 2026 (UTC)
  • A Same as my above reasoning, plus, VE on mobile has really improved over the years, and is actually quite pleasant to use these days. CaptainEek Edits Ho Cap'n! 19:02, 6 July 2026 (UTC)
  • A for the same reasons as above. Choucas 🐦 19:21, 6 July 2026 (UTC)
  • A VE should be the default for all editors, and we should be moving towards eliminating any differences between desktop and mobile editing. -- LCU ActivelyDisinterested «@» °∆t° 19:59, 6 July 2026 (UTC)
  • A, per Femke. Benefits obvious. In solidarity, asilvering (talk) 20:01, 6 July 2026 (UTC)
  • A, per the above. Epicgenius (talk) 20:21, 6 July 2026 (UTC)
  • Option A. I support having 17.5× as many references. lp0 on fire () 21:07, 6 July 2026 (UTC)
  • A, per Femke's data (17.5???), plus my own experience using VE on mobile. There are very few cases where you want to switch to the source editor, and new editors are unlikely to encounter them. Rusalkii (talk) 22:05, 6 July 2026 (UTC)
  • A Per Femke. ARandomName123 (talk)Ping me! 22:10, 6 July 2026 (UTC)
  • A, same reasoning as above. Gnomingstuff (talk) 23:23, 6 July 2026 (UTC)
  • Option A, as above. TechnoSquirrel69 (talk) 00:17, 7 July 2026 (UTC)
  • A, per my comments above and the remarkable data provided by Femke. UpTheOctave!  8va? 00:23, 7 July 2026 (UTC)
  • A, as I said above. --📎 JackFromWisconsin (talk | contribs) 03:33, 7 July 2026 (UTC)
  • A Clearer choice on mobile than on desktop.--Carwil (talk) 03:35, 7 July 2026 (UTC)
  • A per Femke and my reasoning above. Mir Novov (contribs | talk) 04:39, 7 July 2026 (UTC)
  • Option A per above. FaviFake (talk) 11:26, 7 July 2026 (UTC)
  • A, the mobile mode wikitext editor is unusable. There are zero tools available and the editor expects you to write perfect citations from scratch in wikitext, which I'm pretty sure would styme even the most Wikiholic editor. We definitely should not be throwing new editors into this mess. -- Sohom (talk) 00:37, 8 July 2026 (UTC)
  • Option A per Femke and Sohom Datta. I can't believe the source editor is still the default on mobile. OmegaAOLtalk? 06:13, 8 July 2026 (UTC)
  • Option A per Sohom. Mobile wikitext editor doesn't have syntax highlighting or RefToolbar. VE is a clear improvement. – SD0001 (talk) 06:59, 8 July 2026 (UTC)
  • A per Femke and Sohom. VE is a more familiar editing experience for newcomers, and editing wikitext on mobile is annoying even for experienced editors. Kodning 🌸 (talk) 10:33, 8 July 2026 (UTC)
  • A, per my desktop rationale. This is a no-brainer. Toadspike [Talk] 15:06, 8 July 2026 (UTC)

Discussion (mobile)

  • While the survey for desktop has a wider array of thoughts and opinions that merit extensive discussion, I think this survey on mobile default might be heading shortly towards a WP:SNOW close. I wonder if the proposal will wait for both questions to be resolved before moving forward, but if not perhaps thought should be given to how early this question should be closed to allow the volunteer team to make the necessary planning and changes for the mobile experience while we still hash out the community's opinion on the desktop edit messaging. — ♠ Ixtal ( T / C ) Non nobis solum ♠ 23:32, 8 July 2026 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Comment from the Editing team

Thank you for launching this proposal, Femke! To support you all in discussing it, we wanted to share some background information/data that we hope will be helpful as framing.

The core impetus behind our work on VisualEditor is that, as we all know from having seen backlogs grow, we need more editors. To help you recruit them (and patrol the edits they will make), we're looking at how to make every stage of the newcomer journey friendlier and less bitey. But just as interactions can bite newcomers, so too can the software. VE was designed to address this problem by providing a "what you see is what you get"-style interface familiar to anyone who has ever used a word processor.

A sankey diagram showing where newcomers stop editing
The editing journey of newcomers[a]

One way to see VE's effect is by looking at data about the journey editors take making their first edit.[a] As the chart above shows, currently three out of every four choose the source editor (likely because the interface suggests it as the primary option), but those that take that path are far less likely to succeed. More than two thirds (67.5%) of them, on seeing the often-complex wikitext code, close the editing window before typing any changes, compared to only a little more than a quarter (28.3%) on VE. On both editors, about half of those that make changes never publish them (51.0% on the source editor and 51.2% on VE). And of those that publish, 41.2% who used the source editor are reverted within 48 hours, compared to 30.4% on VE. Ultimately, only about 1 in 10 editors who choose the source editor make it to publication, compared to nearly half of editors who choose VE. The abandoned edits from editors thwarted by wikitext ultimately add up not only to a staggering loss of contributions to the encyclopedia but also an end to the journey of newcomers who might have gone on to become the experienced editors of the future.

We want to acknowledge that when VE was first introduced back in 2013, it had numerous bugs that led the community to decide not to enable it by default. 13 years later, it's not perfect, but it's vastly improved,[b] and we're continuing to work on it.[c] In particular, it now includes edit check, a VE feature that provides an expanding set of in-edit-window guidance when an editor does something like adding content without a reference or pasting text from a potentially copyrighted source,[d] which we've seen be highly effective.[e] VE is the default on mobile in all 346 other language editions, and on desktop in 344/346.[f]

We believe VE is now clearly a better environment for newcomers than wikitext and increasingly effective at helping experienced volunteers detect and deter problematic edits. We think that steering first-time editors to it will lead to more edits and more future editors. We are happy to see this discussion, and will be available to answer questions and to support the direction the community decides to pursue.

Cheers, Sdkb‑WMFtalk 17:04, 6 July 2026 (UTC)Updated 18:16, 6 July 2026 (UTC) to add data on other languages.

This conclusion ("abandoned edits") presupposes that everyone who clicks "edit" intends to make an edit. Someone may click "edit" not because they intend to make an edit but just because they want to see what happens, or they want to see what the editor looks like, or they want to see or copy the wikitext (which would explain why source editor has a lower publish rate). I don't think there is a logical basis for concluding that everyone who clicks edit but doesn't click publish has "abandoned" an edit. I don't think you can distinguish between those who intend to make an edit and those who don't without asking the users. There are other good reasons for making visual the default for new users over source, but I don't think publication rate is necessarily one of them. Levivich (talk) 14:04, 8 July 2026 (UTC)
Regardless of the term being used to describe the action "user selecting the edit function without proceeding to submit an edit", personally I don't feel that the choice of editing interface would significantly alter the percentage of users who select edit intending to make an edit, out of all users who select edit. So while the absolute values of the completion rates may not fully reflect difficulty in using the editing interfaces, I do think there is validity in comparing the values for the two editing interfaces. isaacl (talk) 18:33, 8 July 2026 (UTC)
I don't think you can distinguish between those who intend to make an edit and those who don't without asking the users. @Levivich, we were curious about this same question, so recently we actually did ask the users (via a survey presented when you close out the editor)! We haven't had a chance to publish the full results yet, but to give you a preview, we found that, among users who closed the editor within 2 seconds, most (86.7%) just clicked "edit" by accident. Among those who spent at least 2 seconds in the editor, the chart below shows the breakdown of responses.
More information Editor exit survey results ...
Close
If you're interested, I can let you know when the full results are published, and I'm sure the folks who ran it will be happy to dive deeper into the data with you. Cheers, Sdkb‑WMFtalk 19:58, 10 July 2026 (UTC)
Thanks, Sdkb, that's great! Yes please let us know when the results are ready (I'm sure I'm not the only one who's curious). Looks like about 1/3 abandoned an edit, I wonder if that 1/3 changes depending on whether Visual or Source is the default. Levivich (talk) 22:39, 10 July 2026 (UTC)
as we all know from having seen backlogs grow, we need more editors – could you please clarify: in this wording, what is perceived as the source of the growth of the backlogs? I'm asking because the backlogs were juxtaposed with needing more editors. And having more editors doesn't necessarily guarantee a reduction in the backlog burden. What if the efforts of the Editing team in this scope only make the backlogs worse? —⁠andrybak (talk) 19:27, 15 July 2026 (UTC)
To some extent, the expectation is that some backlogs will grow when we get more retention from new editors (like the NPP backlog). They're new, and they'll make mistakes. But a few key ones might be going down: I hope we'll be able to teach editors about LLM and copyvio much earlier, and get fewer sections tagged with unreferences when we teach editors that they do need citations. In solidarity, —Femke (talk) 🐦 20:58, 15 July 2026 (UTC)
Think I’m gonna propose a watchlist notice asking for help w backlogs, CCI, AFC, NPP, AINB are all overwhelmed and far too important for that to be the case Kowal2701 (talk, contribs) 21:36, 15 July 2026 (UTC)
Good question, @Andrybak. We think there are at least three factors contributing to growing backlogs:
  1. The number of experienced editors motivated and knowledgeable enough to work on the issues backlogs are filled with
  2. The number of new edits to review
  3. The toil involved with discovering an issue and addressing it
We know experienced editors start off as newcomers, so we're trying to improve the editing experience for newcomers to widen the pipeline of future experienced editors. This does mean that there will be more edits to review, and that newcomers, being newcomers, will sometimes make mistakes that can add to backlogs. To help you all with this, we're including features in VE that can make it easier to deter newcomers from publishing problematic edits and detect problematic edits when they do get published. For instance, paste check, in addition to directly deterring many copyvios, also tags edits with pasted content to help patrollers decide whether to revert them. The same is true of reference check (which, as Femke noted above, has been effective at causing newcomers to cite the new content they're adding) and tone check.
Other aspects of Edit Check, and many of the Growth team's features, are also focused on helping newcomers to become more successful long-term editors. Further, the Moderator Tools team is investing in ways to help newcomers develop the know-how needed to help patrol edits and for existing moderators to more easily discover edits that might benefit from their attention. This work is happening as part of the broader Home initiative. If you have specific ideas for software features to help with backlogs, we'd always value hearing them!
Cheers, Sdkb‑WMFtalk 05:43, 17 July 2026 (UTC)


Notes

  1. This chart has data for editors making their first edit at English Wikipedia on desktop in November 2024. Data for other groups can be found in this presentation. We don't have data on mobile but believe the situation there is similar.
  2. Major changes since 2015 that came to mind for our engineers include visual diffs, Citoid with the "automatic" panel, automatic save recovery, the 2017 wikitext editor, rich text paste handling (including table paste), table editing improvements, interwiki link improvements, command search (ctrl+/) find and replace (including regexes), improved template editing interface with TemplateData, and fixing numerous performance and IME bugs.
  3. In the coming fiscal year, the Editing team will, at a minimum, introduce additional edit checks and suggestions to help mitigate and detect AI-generated content (e.g. LLM-Paste Check and Reference Survival Suggestion), explore the feasibility of using LLMs to detect hallucinated references, and introduce additional tools to mobile VE. WMDE is also investigating improvements to how VE handles references, including how references are named and adding support for PDFs within Citoid.
  4. Edit check also tags edits it is activated within to help patrollers spot edits that may need attention.
  5. For instance, we found that reference check more than doubled the likelihood of newcomers including a reference with an edit introducing new content, and paste check reduced by 18% the likelihood of an applicable edit being reverted.
  6. Chinese Wikipedia and Spanish Wikipedia are the other exceptions.

Related Articles

Timelines

Top Qs

Fact Checks