Wikipedia:Requests for comment/Level one user warnings

Beginning in late summer last year, a new project to test improvements to user warnings was started as a taskforce of WikiProject user warnings. Its purpose was to try and update select user warnings to improve their ability to dissuade vandals and not bite newbies.

The full report of findings from testing on English, German, and Portuguese Wikipedias is available to read.

Purpose of this request for comment
The A/B testing done during this project had more than two thousand of these user warnings sent, and was the first time we were able to compare the effectiveness of different warnings at diverting vandals and encouraging good faith editors.

The tests showed that if we make the changes proposed below, we can dissuade the worst vandals more effectively, and avoid biting well-intentioned newbies while we're at it.

This open request for comment is to solicit the opinion of editors outside the group that completed the testing, and develop a consensus around how to implement changes based on that work, if any. It is scheduled to run for 30 days (June 27-July 27), and will be closed by 2-3 uninvolved admins.

Templates under discussion
Please see /Proposed changes for the full description.

Yes, let's use the new versions
If you think the versions that have been tested and proposed are acceptable for use, please support this option and explain why.


 * Much, much better (but please expand "congrats" to "congratulations", for the benefit of people with English as a second language). Andy Mabbett ( Pigsonthewing ); Talk to Andy; Andy's edits 18:54, 27 June 2012 (UTC)
 * ✅ Removed congrats altogether; doesn't make sense anyway, given that the edit was reverted. Thanks, Andy! Maryana (WMF) (talk) 19:46, 27 June 2012 (UTC)


 * These do seem like a big improvement. The biography one should say "Wikipedia has a strict policy concerning how we write about living people" though, as Shakespeare is obviously still a person - agree with Andy as well on congrats. Additionally the NPOV one does seem to need a bit more work. -- Eraserhead1 &lt;talk&gt; 19:02, 27 June 2012 (UTC)
 * ✅ Added "living" – good catch :) Maryana (WMF) (talk) 19:41, 27 June 2012 (UTC)


 * Hi. As the proposed ones are much better anyway than the current ones, they should be used even if they need further editing (that's for the people subscribing to the second section here below ;) ). Two little things I want to point out:
 * In uw-delete1 there is a sentence In the future, it would be helpful to others if you described your changes to Wikipedia with an edit summary. This seems a bit weird to me. If I shall describe such a change I would just say "I have removed something." ;) Maybe replace "described" by "explained" or "gave reasons for" or "motivated" or something in that line?
 * In uw-test1 there is that "sandbox" term. Does that word have the same childy connotation in English as it has in German? If so, please better use "test area" or "testing page" or something like that. ;) If not, just ignore me. :) --Thogo 20:13, 27 June 2012 (UTC)
 * We haven't tested calling the sandbox something else, but when we compared just the most common issue-specific warnings (test, spam, unsourced, delete) to their defaults, they all performed significantly better. To me that suggests the issue was not the term that presented a deterrent to newbies. Steven Walling (WMF) &bull; talk   20:48, 27 June 2012 (UTC)
 * This falls outside the scope of the current RfC, but I do think that having something called a "sandbox" is pretty confusing for anybody who's not already a Wikipedian. Whatever the term for it is (I like "test page," for the record), it should be standardized across all links/templates, so we probably shouldn't just change it in these messages until that happens, if it ever does. But yes, thank you for pointing that out, Thogo :) Maryana (WMF) (talk) 21:31, 27 June 2012 (UTC)
 * Well, "sandbox" isn't a Wikipedia-specific term...I actually joined in a conversation about this on WormTT's talk page. Here, check it out.  Not that the term couldn't cause confusion, but there would probably be pushback about proposals to change the name even beyond the Wikipedian's standard resistance to change. Writ Keeper &#9863;&#9812; 22:04, 27 June 2012 (UTC)
 * "Sandbox" is not specific to Wikipedia, but it is still jargon used by the computer geeks who still make up much of our editor corps, for example it is used by my security software to describe a protected environment where potentially harmful programs can be run. We should be using language that the general reader can understand, not just the in-crowd. I would add "orphan" to the list of geeky terms that we should stop using, as the sense in which we use it derives from its use in the context of databases, but that's a separate issue. Phil Bridger (talk) 18:06, 28 June 2012 (UTC)
 * "Sandbox" is a general IT term, and it's a disservice to replace it with a half-cooked neologism such as "testing page" - everyone knows what a sandbox (i.e. a non-VM ring0 pseudovirtual machine in current incarnations, all the way back to plain chroot) is, and, if they don't, it's obvious: unlike some more obscure security and IT jargon, it's called a sandbox because it's self-elucidating. St John Chrysostom ΔόξατωΘεώ 12:16, 16 July 2012 (UTC)
 * You've made Phil's point for him, St.John: this is an IT term, not necessarily comprehensible to anyone who only speaks standard English. If we are to expand beyond our geeky roots, we must abandon some of the legacy tech jargon in favor of human languages. -- Orange Mike &#x007C;  Talk  13:25, 17 July 2012 (UTC)
 * I like the new templates much more than the old ones. They're more personalized and less bitey. This may go a long way to improving what the public believes is Wikipedia's bitey image. It could also help bring in new users that will improve the many articles that need improving. I'd like to see the community switch to the new templates soon. Thanks to all the WMF folks for their efforts working on the new tamplates. 64.40.54.164 (talk) 00:17, 28 June 2012 (UTC)
 * I think they're a massive improvement, and am impressed by what's been done with them. I'd say put them in, as they are so much better and will immediately have an impact, then work on minor tweaks once in. Putting them in front of the editors who'll use them will mean any problems and adjustments are quickly identified and fixed I think.-- JohnBlackburne wordsdeeds 04:29, 28 June 2012 (UTC)
 * Yes, use them now. I normally add to new talk pages and have often previously skipped to UW-2 just to avoid the second "welcome". Mark Hurd (talk) 07:10, 28 June 2012 (UTC)
 * These templates are much better. The usage "less than ", which turns up a few times in the new wording, is rather American, and as an Australian I find it a bit patronising. But that's really an ENGVAR issue, I guess. — This, that, and the other (talk) 07:38, 28 June 2012 (UTC)
 * Yes yes please! These are a million times better. --Conti|✉ 10:56, 28 June 2012 (UTC)
 * I work with new editors quite a bit, and these will be far better warnings then what we have now for good-faith editors who do something wrong, so support. David 1217 13:45, 28 June 2012 (UTC)
 * Support. Tweak per discussion here, revise in future if necessary, but yes, these are definite improvements. More work like this please, well done. Rd232 talk 18:06, 28 June 2012 (UTC)
 * So much better. Nice work! -- phoebe / (talk to me) 23:51, 28 June 2012 (UTC)
 * Support: I'm all in favour of evidence-based decisionmaking. If research shows that the new warnings are more effective, what are we waiting for? bobrayner (talk) 13:03, 29 June 2012 (UTC)
 * Support: I like the idea of user testing, and I love the idea of decreased vandalism. Sure, they might need tweaks, but we'll never get to 100% perfection. Dori ☾Talk ⁘ Contribs☽ 05:58, 30 June 2012 (UTC)
 * Support improvements. → Σ τ  c . 06:02, 30 June 2012 (UTC)
 * Support improvements and more testing of this sort of thing. I also support future tweaking without having an RfC every time. First Light (talk) 15:39, 30 June 2012 (UTC)
 * Support, great improvements and backed by data. Nice work! the wub "?!"  20:59, 30 June 2012 (UTC)
 * Support more personal — Hue Sat  Lum  15:21, 1 July 2012 (UTC)
 * Support with no opposition to further improvements being made. They are much better than the current messages. Ryan Vesey  Review me!  18:45, 2 July 2012 (UTC)
 * Support new ones. I like how they seem noticeably friendlier and open for discussion rather than The Wikipedia Faceless Being coming down on the poor soul, makes more of a distinction between this and level 2 (I'm often ambivalent about using 1(old) vs 2 because they seem so similar in tone. I think delete1 needs some help though. It sounds like I'm questioning the rationale but doesn't mention that I'm undoing it, so it's too close to editsummary rather than the level1 warning set. DMacks (talk) 05:44, 3 July 2012 (UTC)
 * Support using the proposed versions. They might benefit from a few language tweaks here and there but they read much better than the current versions, the principles seem rights and if the evidence says they'll be more effective... Whouk (talk) 12:49, 3 July 2012 (UTC)
 * Support These are sufficiently better that the prior versions that they should be adopted post-haste and further tweaks discusses in due course rather than waiting to perfect them before going live. Eluchil404 (talk) 08:27, 9 July 2012 (UTC)
 * Support They seem OK. If people don't like them they can go to level 2, or just use other templates.  No big deal. Gigs (talk) 13:50, 9 July 2012 (UTC)
 * Yes, I've seen the data behind the changes and this is definitely needed. Steven   Zhang  Get involved in DR! 18:44, 12 July 2012 (UTC)
 * Strong Support - The data shows that the proposed revisions to the templates definitely have a net positive effect, and, upon personal inspection, I find them appropriate, tasteful, and friendly.—Yutsi Talk/  Contributions  ( 偉特 ) 19:45, 12 July 2012 (UTC)
 * Strong support based on both testing data and common sense. I presume that approval of these in their current format won't preclude further editing them as we gain experience, so I don't support the "need more editing" option below.  I also note that it is only the Level 1 templates that are being revised (although I hope someone will look at the other levels), so those who think a sterner tone is needed can still achieve that by using a different level.  --R'n'B (call me Russ) 20:43, 12 July 2012 (UTC)
 * Strongest Possible Support. Retention of new editors is a huge issue and this is a fantastic way to strike at the root of the problem. It may be true that there are lies, damned lies, and then statistics... but the significance of the numbers here are just too hard to ignore. This needs to be implemented with all due speed, as the testing demonstrates an immediate effect on new contributors to the website. -- Lord Roem (talk) 23:23, 12 July 2012 (UTC)
 * Support - Seeing these warnings in comparison to the old ones, definitely. I would support even without the research. The old warnings look ominous and telling users that there is a lot in Wikipedia they should one day follow if they were to act in good faith, and stop them from vandalizing in the short term, but it would also stop them from editing altogether in the long term. The research is good to back it up. Michaelzeng7 (talk) 19:03, 14 July 2012 (UTC)
 * Support as per the unanimous opinion (especially as per Lord Roem and Michael Zeng immediately above me). This is probably my least original and helpful !vote ever cast, "for after the victory is signalled, loyalists and traitors speak the same words". (I normally edit in religion, which gets heated.) St John Chrysostom ΔόξατωΘεώ 12:18, 16 July 2012 (UTC)
 * Support - I read the report, and compared the old vs new wordings, and the new ones are clearly more friendly. --Noleander (talk) 04:20, 17 July 2012 (UTC)
 * Support – The new versions seem more sincere and less bureaucratic. For standard vandalism and spam these templates are a clear improvement. EdJohnston (talk) 05:44, 17 July 2012 (UTC)
 * Support - not flawless (few things are), but good work on reducing the biteyness; and if we need a firmer tone, we still can use Level 2s and above. -- Orange Mike &#x007C;  Talk  13:36, 17 July 2012 (UTC)
 * These are so much better than the old ones this needs to happen ASAP. Maximum possible logged out support. 75.166.200.250 (talk) 18:49, 17 July 2012 (UTC)
 * Support except for uw-biog1 I like the new ones, very nice!  But the phrasing on uw-biog1 needs to be stronger, given the importance of our BLP policy. --Philosopher Let us reason together. 23:01, 17 July 2012 (UTC)
 * Use, but keep refining - incorporating some of the changes suggested below would also be great, but not essential, and we shouldn't feel too bad about using these as-is while we work on improving the text. They're certainly better than the originals. Andrew Gray (talk) 17:36, 18 July 2012 (UTC)

Not yet, they need editing
If you think the proposed templates need some edits before being suitable, please support this option and describe your desired changes in detail.


 * The proposed uw-npov1 template needs to include the phrase "personal comments" or "personal analysis". Far too often I am forced to use uw-npov2 as a first-level warning because that one refers to personal commentary. Also, I can't tell from the examples, but the templates need to retain the capability to refer to the article the user edited. Finally, I'd like to bee able to bypass or replace the invitation to discuss on my talk page, if an article talk page is a more appropriate place to direct the editor being warned. ~Amatulić (talk) 18:50, 27 June 2012 (UTC)
 * I've amended the example to make it clear it will retain the optional clause to refer to the page and diff of the relevant edit. All our testing included that. Steven Walling (WMF) &bull; talk   18:57, 27 June 2012 (UTC)
 * What about the suggestion on my talk page? I think it would make sense to include talk which, if unfilled, would say my talk page, but if filled could link to any talk page.  I'm having difficulty figuring out exactly which template would allow it to place the talk page of the tagging editor for the default. Ryan Vesey  Review me!  19:03, 27 June 2012 (UTC)


 * Much improved. I think there's a little more to shorten without losing either the tone or the meaning: Replace "please feel free to come ask me on my talk page. Thanks, Jimbo" with "please ask me on my talk page. Thanks, Jimbo" or what I use myself, "just ask me on my talk page"  DGG ( talk ) 19:07, 27 June 2012 (UTC)
 * I think the last one comes across as a little blunt to be honest . -- Eraserhead1 &lt;talk&gt; 19:33, 27 June 2012 (UTC)
 * Yeah, it's kind of a delicate balancing act between shortness and politeness, but I agree with Eraserhead1 that "please ask" and "just ask" sound a bit more like marching orders than a friendly invitation, so I'm inclined toward keeping the slightly longer but more informal/colloquial version. It's a tiny difference, but an important one, I think. Maryana (WMF) (talk) 21:25, 27 June 2012 (UTC)
 * There seems to be some sort of a presumably cultural difference here. To me, the longer versions sound smarmy. "Please ask me" is  to me unambiguously friendly and what you would say in the flesh.  "Please feel free to ask me' is the sort of thing people put on boilerplate. Have you every heard anyone say it?  DGG ( talk ) 00:20, 29 June 2012 (UTC)
 * Do they have slightly different phrasing on purpose? I noticed uw-delete1 and uw-npa1 drop the 'please' -- is it because they're less constructive actions? -- phoebe / (talk to me) 23:50, 28 June 2012 (UTC)
 * I prefer the new version also. Yes, I have heard people say "please feel free to ask me," and I've used it on WP and in real life myself because it is less formal and lighter than "please ask." First Light (talk) 15:44, 30 June 2012 (UTC)
 * I believe the templates are great, but the uw-test one seems to say "What you did is fine, if you feel like it next time, use the sandbox". I disagree with the comments above on the closing and the npov template above, just for the record. Anyway, great job making the templates easier on the eyes of the noobs. Keep up the good work! --Nathan2055talk - contribs 19:54, 27 June 2012 (UTC)
 * The new templates look good, but I have a few suggestions. I agree with Nathan about the test one; I feel like it should say something like "If you want to keep practicing, you can use the sandbox as much as you like." instead of "If you want, you can use the sandbox to practice editing."  I also think that the language of the vandalism warning could be a little bit more serious, perhaps by removing the "to me" on the "because it didn’t appear constructive to me" bit.  Finally, I think Amatulic's article-vs-user-talk idea is a good one, although I probably wouldn't use it.  Other than that, though, they look great! Writ Keeper &#9863;&#9812; 20:18, 27 June 2012 (UTC)
 * I have edited the test template based on the feedback from you and Nathan. :) I think it's important to remember here that the majority of true test editors don't actually repeatedly do so, and in this one it's extremely important to assume good faith. I removed the "if you want" part though. Steven Walling (WMF) &bull; talk   20:32, 27 June 2012 (UTC)
 * The new uw-vandalism1 reads oddly to me: "I undid one of your recent edits, such as the one..." I think "one or more" would be better English, and in any case I might use this message after undoing several of a user's edits. -- John of Reading (talk) 20:14, 27 June 2012 (UTC)
 * On this point, two items to note: first, the diff will show all changes reverted, so that's self-explanatory. Second, this change was one of the parts that got tested thousands of times and performed better. While it might seem like "one or more" is precisely accurate, it's actually playing to an edge case rather than the rule, and to a brand new editor it conveys a feeling of, "Well, who knows how many of your edits got reverted. They all kind of suck." Steven Walling (WMF) &bull; talk   20:24, 27 June 2012 (UTC)
 * Several of the messages use the word "article", so could not be used if the reverted edit was in a different namespace. "page" or "Wikipedia page" would be better. -- John of Reading (talk) 20:14, 27 June 2012 (UTC)
 * in uw-delete. The others that use article, uw-unsourced and uw-npov, are about issues that most likely pertain to articles. Steven Walling (WMF) &bull; talk   20:38, 27 June 2012 (UTC)
 * In general they are better, but I don't particularly like uw-delete. I don't like the current one either but the new one doesn't improve upon it much. Basically, that template needs the word sourced in it. We are supposed to remove sketchy unsourced content, not add it back in (particularly in BLPs but it applies to other articles as well). So how about changing the wording to "...you recently removed some sourced content from..." Volunteer Marek  20:21, 27 June 2012 (UTC)
 * Not all reverts associated with uw-delete are about removing sourced content. It's one of the most commonly-used warnings, so playing to that specific case when it may not be applicable is probably not a good idea. Steven Walling (WMF) &bull; talk   20:27, 27 June 2012 (UTC)
 * I agree with Steven here. You should explain why you are removing material even if it is unsourced so this template shouldn't change. Ryan Vesey  Review me!  20:39, 27 June 2012 (UTC)
 * Yes, but we all know that these templates are generally used to intimidate people - i.e. the fact that it's one of the most commonly used templates is probably a manifestation of the problem, not a plus. The new ones go some way towards them being less about intimidation and more about information. But then why are we warning people about edits which may in fact be very much in line with the 5 pillars? If it's not applicable, then it shouldn't be used, which is what adding the word "sourced" would highlight to the person leaving the template. Volunteer Marek 22:43, 27 June 2012 (UTC)
 * I have never used the templates to intimidate, but life is too short not to use them when reverting. -- Eraserhead1 &lt;talk&gt; 06:17, 28 June 2012 (UTC)
 * The template or edit summaries? Either way, there seems to be a self-contradiction in there. Volunteer Marek 00:27, 29 June 2012 (UTC)
 * I agree that the "test1" template reads way too happy. I revert a good many tests (by lurking behind 28bot). Sometimes I template with a standard issue, but at times when the edit seems to have been in good faith, I instead use a personal comment that says I've undone what seems to be a test or mistake:..then I add: "However, it's great that you were bold and edited a page". I prefer that encouragement, rather than "your gleeful splaying junk all over the article was successful"  As well, for all the warnings: the "Hello, I'm Jimbo" seems way too upfront. I don't mind at all having my name and talk location at the end of my paragraph, so I can be located as the enactor / aggressor of the change, but having this first would leave me feeling vulnurable when I don't know who my IPs are, especially.   Fylbecatulous   talk   16:06, 28 June 2012 (UTC)
 * I want to reiterate here how important the "Hello, I'm username" part is. We tested this thousands of times over, and it was one consistent element that stood out as key. The good news here is threefold: speaking up and saying you chose to revert someone discouraged vandals more, it does not increase vandalism directed the reverting editor (we know, we hand checked thousands of diffs in a blind study), and encourages new Wikipedians who have made a few edits. We simply must take responsibility for our reverts and the subsequent warning. It's the right thing to do. Steven Walling (WMF) &bull; talk   16:52, 28 June 2012 (UTC)
 * The new templates are definitely better then the older ones - at times i simply didn't use the warning function in Huggle because the templates would have been overkill if fired at a quality new editor. There are two points i would suggest though; The UW-Spam templates seems to have a somewhat odd "Cause-Effect" relation in the sentence Wikipedia is an encyclopedia written by people like you and me, so we try to make sure articles link to only a few important sites.. It seems to imply that because it is written by you and me, we have to link only a few important sites, which is rather strange. Instead it should explain that not every website is useful to other editors, and that we try to prevent advertising in article's for the sake of neutrality (Or something along that line).
 * The second issue i have is the "Hello i'm Jimbo" first lines in every template. I don't mind it in the first line, but i believe it would be more useful if it stated "Hello i'm Jimbo, another Wikipedia editor". Most editors use some form of made up name for themselves, and i could imagine that not every new editor would understand the more "confusing" names we have. Just a grab from the usernames log for today: "Hello i'm City of Conover" (A talking City?)" - "Hello i'm Howuieurthtgfgghfgs" (You are what?) - "Hello i'm Whip The Alarm" (Whu? What is that odd message i see here?) - "Hello i'm Noble Anger" (What did i do to make you angry?) - "Hello i'm Cheating for a decade" (What? What kind of adult site is this?) and so on and on. Excirial ( Contact me, Contribs ) 20:34, 28 June 2012 (UTC)
 * You make a good point about uw-spam, and the cause-effect confusion. What about a version like, "I wanted to let you know that I removed an external link you added, because it seemed inappropriate for an encyclopedia. If you have any questions, please feel free to ask me on my talk page, or take a look at the guidelines about links."?
 * About the intro... its interesting you brought this up, because some of our original tests said "I edit Wikipedia too, under the username Example". We simply wrote it this way in order cut down length. If folks would prefer a version that includes ", another Wikipedia editor" I would totally be open to it. Steven Walling (WMF) &bull; talk   20:48, 28 June 2012 (UTC)
 * Yes, that uw-spam suggestion would be perfect if you ask me. We don't have a separate template for WP:ELNO links, so i gamble that at least some editors use Uw-spam as a blanket for all Elno's. The suggested text would perfectly fit all these situations. As for the intro, i would definitely suggest the above as long as we don't overdo its length.
 * Re-reading the templates i did notice another point i would ike to raise for consideration, and that is that some templates really jump right into an entirely different subject after the introduction, whereas other are much more "fluent" read. The Uw-Spam is a really fluent example: "Hi Im Jimmy, (another Wikipedia editor). I noticed that you recently added... " (Very logical text progression to the core matter), while others such the Npov version are less fluent to read: "Hello I'm Jimmy. Wikipedia is written by people who have a wide diversity of opinions" (Odd jump from introduction to a apparently generic statement about Wikipedia). Something along the line of "Hello I'm Jimmy, another Wikipedia editor. I noticed you added some content to an article that wasn't written in a neutral point of view so i removed it for now. Since Wikipedia is written by people who have a wide diversity of opinions, we try hard to make sure articles have a neutral point of view. If you have any questions, please feel free to ask me on my talk page. Thanks, Jimmy" would be more of a fluent and "friendly" read in my humble opinion. Note that a few templates seem to be set up this way. Excirial ( Contact me, Contribs ) 21:51, 28 June 2012 (UTC)
 * Okay, uw-spam . Thanks for the replies! Steven Walling (WMF) &bull; talk   22:14, 28 June 2012 (UTC)


 * May I have a second try at improving the grammar in uw-vandalism1? Try reading it without the words "such as". -- John of Reading (talk) 06:18, 29 June 2012 (UTC)
 * Proposals for alt edits are welcome. To be clear though, the clause with the page name and diff is optional. It's only included in the example per requests above and because some automated tools (like Huggle) automatically insert such information. Steven Walling (WMF) &bull; talk   23:12, 29 June 2012 (UTC)
 * Strongly against "please do so!" in the proposed wording of uw-unsourced1. Many times I use this to get rid of unsourced trivia, gossip, etc. The last thing I want is to encourage an editor to re-add the material and add a link to a tabloid, blog, etc. (the proposed change doesn't even link to WP:RS).  The more formal tone of the current message also emphasizes one of our fundamental policies - WP:V. -- Neil N    talk to me  17:38, 1 July 2012 (UTC)
 * Uh... the point of the warning is to warn about lack of a source. If someone can include a source, the problem is fixed. If you want to warn someone about multiple issues, just write them a message about how not only was it unsourced, but unsuitable gossip. (There is also the related BLP template, please note.) But you can't fault the warning for telling new editors how to fix the problem made by their editing, because just throwing a policy at people does not teach them the right way to do things. Steven Walling (WMF) &bull; talk   19:00, 1 July 2012 (UTC)
 * So we waste both the new editor's and the patroller's time by doing another revert and placing another warning when an unsuitable source is added? The unsourced messages are not just for informing about a lack of source, they also inform about a lack of a reliable source. And the new editor has even less of an idea what we accept as a source with the new message. -- Neil N   talk to me  18:38, 2 July 2012 (UTC)
 * No one shares the same concern? -- Neil N   talk to me  19:51, 6 July 2012 (UTC)
 * I do. The template is currently worded for "Addition of unsourced or improperly cited material", this new wording would remove the "improperly cited material" bit for seemingly no reason.  Changing the scope of the template doesn't achieve anything beneficial, imo. - SudoGhost 20:22, 6 July 2012 (UTC)
 * Improperly cited is not something to communicate in this way. Maybe with a specific template "please don't use source x, we've found it problematic for reason y". But if we want to wean people off sources that we don't like, then saying it is effectively unsourced is not an effective tactic.  Ϣere Spiel  Chequers  18:44, 18 July 2012 (UTC)


 * Comment - I do have a concern about the use of first person descriptors in the templates. What if someone else reverts an edit, but doesn't bother to let the editor that was reverted know why?  This effectively means that no other editor can really use that template to let the editor in question know the reason for the edit being reverted.  I know it seems like a minor thing, but it has come up occasionally.  I know that the use of first person is more personal, but maybe there'd be a way to maybe add an optional parameter in the template that lets us change it to third-person if we weren't the ones that actually undid the edit? - SudoGhost 20:57, 3 July 2012 (UTC)
 * For edge cases like warning someone you didn't revert, I would rather suggest that people simply write a few sentences like a normal message, instead of putting very complex conditional phrasing into the templates. <span style="font-family:Palatino, Georgia, serif;">Steven Walling (WMF) &bull; talk   21:03, 3 July 2012 (UTC)
 * Like SudoGhost, I sometimes issue warnings to users whose edits were reverted by someone who didn't provide a warning. I'd like the template to include an option to turn off the part that says "I" am the one who "undid your edit" (or "removed it for now", or the equivalent). --Orlady (talk) 02:10, 8 July 2012 (UTC)
 * Me too. I had forgotten about that, but I sometimes need to add a warning that the reverting user (often an IP) did not provide. The templates are there for a reason (otherwise everyone would be required to write their own warning), and it is not satisfactory to require me to devise my own warning for trivia—an unnecessary imposition on me, with possibly unsatisfactory resulting text. Johnuniq (talk) 02:38, 8 July 2012 (UTC)
 * Comment - The proposed language for template uw-spam1 makes it appear that the warning is based entirely on the opinion of the person who issued the warning; it does not clearly indicate that the offending edit was judged to violate policy. This could be rectified by revising the second sentence to say "I wanted to let you know that I removed an external link you added, because to me it seemed not to meet Wikipedia's guidelines on external links appropriate for an encyclopedia." --Orlady (talk) 02:21, 8 July 2012 (UTC)
 * The uw-spam template does in fact link to the policy, however, one of the clear results from testing is that new or anonymous editors don't respond very well to arguments like, "the policy just says so, and here's a link to our humongous policy". You have give them a reason, even if it's a generic one, if you want them to respond better. Obviously there's a certain number of real spammers who are not going to respond to any admonishments, in which case it is perfectly appropriate to skip the purely good faith level 1 warning. This level of warning is specifically designed for people who are making good faith mistakes. <span style="font-family:Palatino, Georgia, serif;">Steven Walling (WMF) &bull; talk   17:26, 9 July 2012 (UTC)
 * I was aware that the proposed language for uw-spam1 includes a link to the policy. My concern is that the proposed language does not tie that the judgment that the link is inappropriate to policy. It appears from the proposed language that the judgment is totally based on the  personal opinion of the person who reverted the edit, when in fact it's an opinion based on that person's interpretation of policy. I like the idea of saying "it seemed to me", instead of simply saying the link violates policy, but I think that "it seemed to me" needs to be followed by a clear indication that this was an interpretation of an established policy and not merely a capricious personal whim. --Orlady (talk) 03:10, 12 July 2012 (UTC)
 * The proposed language for uw-delete1 is so weak that I believe I would be tempted to escalate to level 2 in almost every case, instead of using this warning. Most of the time when I post a uw-delete1, the user has removed either valid cleanup templates or unflattering reliably sourced information, but I'm "assuming good faith" by issuing the level 1 warning, even though I'm reasonably sure the user is a WP:SPA working on improving the public relations profile of the article's topic. (Also, sometimes the user has removed talk page content that they apparently didn't agree with.) If the level 1 message says only "it would be helpful to others if you described your changes ... with an edit summary", those users are no longer going to get the "benefit of the doubt" that the level 1 message allows, because I will go straight to level 2. I prefer the more explicit language that says "When removing content, please specify a reason in the edit summary and discuss edits that are likely to be controversial on the article's talk page."
 * A second change I'd like to make to uw-delete1 -- for entirely different reasons -- is to keep the part of the current message that says "It might not have been your intention, but your recent edit removed content...". Deletions are the perhaps the only kind of vandalism-like edit that is particularly likely to occur as an accident (I know: I've been warned a couple of times for deletions that I could only explain as "household pet walking on keyboard"), and I think this particular part of the warning makes it friendlier (by avoiding the hard feelings that can result when someone feels they are being bawled out for an innocent mistake). --Orlady (talk) 02:43, 8 July 2012 (UTC)
 * One of the very clear results from the testing is that pushing good faith editors to do things AKA "please use the sandbox" or "please read the policy" does not actually teach them what they did wrong or prevent further action. We compared many templates with such directives to those without them which were even softer than this, and the versions without directives won out every time. The data simply does not support including that kind of language in a level one template designed for very new contributors who have not demonstrated any ill will. As for the "intention" suggestion, I think that's a fine idea. The testing did not suggest it was strictly necessary, but obviously it holds with the spirit of AGF here. I'll make sure it's now. <span style="font-family:Palatino, Georgia, serif;">Steven Walling (WMF) &bull;  talk   17:32, 9 July 2012 (UTC)
 * Thanks for the fix on the "intention: part. As for my other comment, I guess I will no longer pretend to AGF on deletions when I don't think I'm talking to a good-faith editor. --Orlady (talk) 03:10, 12 July 2012 (UTC)


 * As well as DGG's excellent proposal for a shorter and less smarmy wording, I have a concern with the unsourced template. This is one of the most misused templates that we have and one of the most important to get right as it is likely be going mainly to goodfaith newbies. The difficult thing is in concisely explaining why something was sufficiently implausible, contentious or pejorative that it was appropriate to revert it immediately rather than fact tag it. I suspect that this is because that is difficult, and Huggle et al are set up to revert edits rather than add fact tags. Nevertheless the template needs to include something to the effect of "and because this information is pejorative, or controversial information about a living person I have reverted it rather than added [Citation needed] ".  Ϣere Spiel  Chequers  13:11, 11 July 2012 (UTC)
 * Theoretically I don't have objections to your suggestions, but your example shows the weakness of adding more. There is already a different BLP template for the latter issue your example text is referring to, and the number of unsourced reversions that apply as "pejorative or controversial" is unknown. The harm potentially caused by causing someone's edit pejorative when it wasn't outweighs the unknown factor that someone's edit might be so, and we should AGF vigorously anyway. I think it's pretty obvious that the list of issues in addition to the lack of a source is endless, and to include them all doesn't work. <span style="font-family:Palatino, Georgia, serif;">Steven Walling (WMF) &bull; talk   14:33, 11 July 2012 (UTC)
 * In above, "causing" should be "calling". I think the answer is to provide a more pointed level 2 warning, and just skip level 1 for a bad edit (or use a different warning?). If someone, in an article on politician X, says that X is a bigamist (with no source), is a warning appropriate? Which? What about "was arrested for shop lifting" or lots of other possibilities? The proposed unsourced1 is clearly only suitable for what might be a good (DUE) edit, but which is a little too unexpected to stand without a reliable source. Johnuniq (talk) 23:59, 11 July 2012 (UTC)
 * What does uw-vandalism1 mean by "I undid one of your recent edits, such as the one you made to..."? Surely it is the one made to that article that was reverted.  I know that both new and old versions of uw-vandalism1 include the phrase "such as the one" but maybe this is a good opportunity to improve that.  Perhaps it should say "I wanted to let you know that I undid one of your recent edits, with this edit to the page about Science."  Yaris678 (talk) 12:28, 12 July 2012 (UTC)
 * No, for now. Old version seems more formal, which should be the tone when warning users.--Chip123456 (talk) 14:50, 12 July 2012 (UTC)
 * Respectfully disagree with your assessment that formal is necessarily better. Level 1 warnings are supposed to assume good faith, and are ideally suited to greet editors who are unfamiliar with policy — as opposed to those who are flagrantly violating policy. I know it becomes a little annoying on Huggle, where it requires extra work to skip level 1, and I think the wording needs tweaking on a lot of the proposed notifications, but informal is appropriate for introductory notices of policy. -- Jprg1966  (talk)  08:04, 15 July 2012 (UTC)

I prefer the current defaults
If you think the current defaults are superior, please support this option and explain why.


 * N0, I don't think we should weaken them thís much. These are warnings, given out to vandals, spammers, etc.  These are not warnings that should be given out to good faith editors who edit in bad faith.  I do not believe that this will have an effect in changing editors from vandals to good editors, this will only have as an effect that we will have a higher retention of vandals, spammers, POV-pushers etc.  --Dirk Beetstra T  C 04:11, 30 June 2012 (UTC)
 * It is claimed (see comment dated "16:48, 29 June 2012" below) that testing showed the new messages were more effective at removing vandals. Obviously one would have to study the methodology of the test to be convinced, but it is an interesting thought. Johnuniq (talk) 05:13, 30 June 2012 (UTC)
 * Shaming away spammers, school-kids, and POV pushers .. interesting, but I do not believe it, at all - if it is 'true' it must be still within the statistical error-margins. --Dirk Beetstra T  C 05:39, 30 June 2012 (UTC)
 * We have confidence ratings publicly available in the findings, and none of the results we reported on as successful had a statistical confidence of less than 80%. <span style="font-family:Palatino, Georgia, serif;">Steven Walling (WMF) &bull; talk   05:45, 30 June 2012 (UTC)
 * See below - Do I read the text correctly, that the improvement in retention was only for the vandalism warnings and only for registered editors with >5 edits? --Dirk Beetstra T  C 05:51, 30 June 2012 (UTC)
 * That was a result we were most interested in, but it was not the only improvement seen, no. Also, an improvement in retention was not the desirable outcome in some cases (in particular take a look at the first experiments where we separated out different quality levels of editing). <span style="font-family:Palatino, Georgia, serif;">Steven Walling (WMF) &bull; talk   05:59, 30 June 2012 (UTC)
 * Dirk, these are first level warnings. There is no requirement to start out with that every single time in every case for every situation. Any time I put a template on someone's page, I already know if I'm warning a good-faith editor or a bad-faith editor. I am free to use a second-level warning as the first one if it's more appropriate. As I pointed out in my comment above, it is indeed appropriate even for good-faith editors in the case of uw-npov2, because that template mentions personal commentary or personal analysis, which clarifies exactly why the user's first edit was a problem, unlike uw-npov1. And in the case of spammers, the first and second warnings currently are nearly identical, so I start off with uw-spam2 in cases of serial spamming. First level warnings should be friendly, but a second level warning can always be used if the situation warrants. ~Amatulić (talk) 19:05, 2 July 2012 (UTC)
 * One of the advantages of the proposed changes is that I will no longer have to start with a level 2 warning because someone has been around long enough that "Welcome to Wikipedia!" would sound like sarcasm. --Guy Macon (talk) 21:15, 3 July 2012 (UTC)
 * I think Amatulic has it right. Nothing about this proposal precludes anyone from using a higher-level, harsher warning for an editor who really is continuously vandalizing. In that respect, this is the best of both worlds: a note of polite guidance for one-time mistakes and the serious "you're in trouble" paragraphs for those who mean harm. That's a good balance that takes all necessary issues into account. That being said, if this change ends up backfiring-which I see little evidence of happening-we are always able to revert back. -- Lord Roem (talk) 05:01, 16 July 2012 (UTC)
 * I prefer the passive-tone for one clear reason: Instead of "hey there, the community thinks this might be vandalism" we get "Hi, I'm the person who undid your edits because I think they're vandalism". Counter-vandals like "hiding" as a cog in the system. It makes us feel as though we're less likely to be targeted for retaliation. I want to be held responsible for my actions, just not by the vandal I'm warning. Achowat (talk) 13:31, 5 July 2012 (UTC)
 * I understand the instinct to be seen as a face of the system, but we proved that there is no increase in retaliation with the new templates; less than 1% of vandals did so with both the current defaults and the new versions. <span style="font-family:Palatino, Georgia, serif;">Steven Walling (WMF) &bull; talk   17:19, 5 July 2012 (UTC)

General discussion

 * As a preliminary glance, I see this as being a benefit, but want to review this more before I make my decision. I am curious as to whether changes to the deletion messages will be considered.  In addition, it may be useful, for convenience, to display the default messages at Requests for comment/Level one user warnings/Proposed changes. Ryan Vesey  Review me!  18:36, 27 June 2012 (UTC)
 * We got lots of feedback on deletion notifications we tested with Twinkle, but we only tested AFD and PROD, not CSD, so we were thinking that it should be a separate RfC. We could add it here if people are interested though. As for the subpage, that's a good idea. I'll take care of it now. <span style="font-family:Palatino, Georgia, serif;">Steven Walling (WMF) &bull; talk   18:43, 27 June 2012 (UTC)
 * I will be at this point outside the crowds, but my experience is that ANY non-template warning works better than ANY template warning. In this perspective, I think that the best direction would be to eliminate template warnings.--Ymblanter (talk) 06:05, 28 June 2012 (UTC)
 * Agree 100% with you that a personal message is always better than a template. Unfortunately, with thousands of anonymous and registered users making their first edits to English Wikipedia each day, and with many of those edits not being of acceptable quality, the choice isn't really personal message or template. It's template or reverting the edit and not explaining why the edit was reverted at all – which our research shows is actually worse than receiving a template. Maryana (WMF) (talk) 14:46, 28 June 2012 (UTC)
 * I do leave personal messages. Those who revert edits without explanations (with the exception of clear vandalism) should, after warning, have their tools taken away and/or get blocked. Same way as admins who block and do not leave explanations should have their bit removed.--Ymblanter (talk) 18:54, 28 June 2012 (UTC)
 * Just wanted to say that testing a prompt for a custom message instead of a template (whether in Twinkle, rollback/undo, or another method) is something on the agenda eventually for the experimental features team. However, in the meantime, it's probably not realistic to try and force everyone to not use templates, so we should write the best ones we can. :D <span style="font-family:Palatino, Georgia, serif;">Steven Walling (WMF) &bull; talk   23:38, 28 June 2012 (UTC)
 * At the uw-1 & 2 levels, certainly personal messages if well done is better. At higher levels, I will often leave the formal warning, and then explain further. Or sometimes give a personal warning, and then say I am leaving the formal warning to make it unambiguous. these combinations seem especially helpful at level 4im, when there wasn't any previous warning. I have  not usually been adding special wording to the formal warning within the box, because I doubt very much anyone actually reads the text in it--this may of course change, if we get them short enough. But the thing is, Ymblanter, that about half the people around here leaving warnings really should not be encouraged to use anything but the standard until they know their way around a little better. You might reply, then they shouldn't be leaving warnings at all, and actually I think you'd be right, but getting sufficient truly qualified people for NPP and RC patrol is at this point not very likely.   DGG ( talk ) 00:10, 29 June 2012 (UTC)
 * I agree that "Hello, I'm John of Reading..." or "Hello, I'm John of Reading, another Wikipedia editor" is a good start to a friendly message. However, that's going to look pretty silly when the warning is placed by an IP. "Hello, I'm 2001:0db8:85a3:0000:0000:8a2e:0370:7334...". Could the templates be made clever enough to omit that part when placed by an IP? -- John of Reading (talk) 06:59, 29 June 2012 (UTC)
 * The proposals have a lot of merit, but they are not suitable for many cases. For example, changed "list of ship launches" to "list of shi-t launches" and saying "because it didn’t appear constructive to me" just makes me look an idiot (btw: that's a curly apostrophe in "didn’t"!). I use uw-test1 for that kind of thing to give a kind of automaton response: vandal tries to generate excitement, but gets only a bland "not considered beneficial" without any personalities being introduced. I suppose the new uw-vand1 would be ok for a, but I would feel a little silly leaving that message for such a case. Is there a chance that vandals would be encouraged by seeing how dumb the editors at Wikipedia are? Also, what is the "with this edit" stuff? It's too much trouble to get a diff for random vandals, and it might be a bit pointy with a new editor who made a mistake or who didn't understand some core procedure (a pointer to their blunder). Johnuniq (talk) 07:52, 29 June 2012 (UTC)
 * Our tests showed that this kind of construction was more effective at getting blatant vandals to go away than the default versions. You can speculate on the reasons (vandals might actually be embarrassed themselves when they realize a real person is seeing their edits, etc.) but in the end it's a quantitative data-driven finding, pure and simple. As for the diff thing, that's just something that Huggle and other revert/warn tools do automatically, so we have to work with it. Maryana (WMF) (talk) 16:48, 29 June 2012 (UTC)
 * Hmm, yes that's possible—it's growing on me. Is there some explanation of how uw-vand1 and friends would actually work in terms of parameters? Not everyone leaving warnings uses a tool like Huggle, so I'm wondering if "with this edit[diff link]" is mandatory, or whether it will only be shown if some parameter is supplied. Also, I agree with a comment above that many user names would be confusing if presented at the front of a message—perhaps investigate whether there should be a new user preference with an abbreviated user name that could be automatically inserted if nonblank. Johnuniq (talk) 23:54, 29 June 2012 (UTC)
 * Are we sure this is statistically significant? I do not believe that you shy away school-kids, spammers, or POV-pushers.  --Dirk Beetstra T  C 05:40, 30 June 2012 (UTC)
 * Is this relating to the quote: "When we compared the outcome of the control group to the group that received personalized/no directives warnings, the templates had different effects on registered and unregistered users: for registered users who had made at least 5 edits before being warned, the new templates did better at getting people to edit articles. For unregistered editors, the effect was reversed, and the default version was more successful." - this is just for the vandalism warnings, not overall; the first part of that statements needs the caveat of 'at least 5 edits', and the latter part was simply shown to be ineffective - the former caveat also suggests that it is ineffective when looking overall (and that the original warnings actually were better for editor retention). At best, it would suggest that e.g. Huggle would need to detect the number of edits for a registered user, using another warning >5 (following the concept of Don't template the regulars), and otherwise use the regular ones.  Similar should be suggested for anti-vandalism bots (should be trivial) and for 'regular editors' who issue warnings (to who we can't do more than just suggest ...).  --Dirk Beetstra T  C 05:50, 30 June 2012 (UTC)
 * The results that you're quoting from are one aspect of one test. All test results included the editing activity of anonymous editors. I would especially look at the findings from the first two experiments, where we hand-coded thousands of diffs for a qualitative assessment, in addition to the generalized quantitative measurement. <span style="font-family:Palatino, Georgia, serif;">Steven Walling (WMF) &bull; talk   05:57, 30 June 2012 (UTC)
 * But then again, I do not believe that you shy away school-kids, spammers, or POV-pushers. I have been active fighting spammers for a LONG time - editor retention on spammers is high, and unwanted.  Even blocks don't shy away spammers - I can show you a case where a spammed link was blacklisted, and when the link was de-blacklisted by an unknowing and not-understanding admin, the spamming proceeded within days after that.  And I think that the same is true for school-kids (seen that vandalism returns sometimes within days after a block expires shows that they try vandalism during the block, the ones coming through were just lucky that the block expired) and POV-pushers (I can show examples which are similar to the school-kids and spammers there - editors returning hours after a year-long page-protection expires, as if they put it in their agenda).  There must be another reason, otherwise I will believe it is a statistical glitch or a factor that you did not take into account, and that is not filtered out by the testing method.  Maybe a less hard-handed method retains more editors - but do we then also keep more problems?  And seen that for e.g. XLinkBot I see responses sometimes like 'oh, I did not know that, I see your point, you are right, I will adapt my editing', does a less informative message actually help there?  I really doubt it.  --Dirk Beetstra T  C 08:13, 30 June 2012 (UTC)
 * Is the icon necessary? What does it even mean, "Information"? What talk-page message *isn't* information? If the idea is to make the message more organic, why on earth would we include some meaningless symbol that screams "Boilerplate message"? Are we just including it because it's always been there, or what? —Designate (talk) 20:53, 7 July 2012 (UTC)
 * We kept the icon that is present in the current default because when we tested the same templates with and without an icon, there was no statistically significant difference in the editing activity of those who received either. I do think that, when you're scanning a page with more than a few warnings (such as an IP, etc.) it does improve readability. Other than that, I have no strong feelings and would be very much open to removing the icon. <span style="font-family:Palatino, Georgia, serif;">Steven Walling (WMF) &bull; talk   21:02, 7 July 2012 (UTC)
 * If the templates are changed to the new versions, I hope that you will continue to collect and evaluate data? Unless your statistical confidence levels are 99%+, it's essentially a guarantee that some of the conclusions are wrong through random chance. Arc de Ciel (talk) 00:32, 10 July 2012 (UTC)
 * Firstly, I would like to commend the people behind this, both for the quantitative approach taken and the imagination in the specific changes to the template that were tested. However, I would like to sound a note of caution.  The tests were primarily designed to test a hypothesis related to user retention although "discouraging unwanted contributions" gets a mention at the end.  They measured this by how much activity the user showed after receiving the warning.  Some hand checks for quality were done in some cases but I can't see exactly which tests this relates to.  In some cases they looked at who went on to be blocked and found no difference between the cases tested... but this is perhaps unsurprising because by this point users will probably have received further warnings, diluting the effect of the tested warnings.  What would be really interesting would be to see if the wording of a warning changes the number of subsequent warnings received.  A really good warning would reduce number of future warnings while increasing the number of future contributions. Yaris678 (talk) 15:40, 10 July 2012 (UTC)
 * Since I distinctly remember doing the scutwork :) let me clarify – it wasn't "some hand checks," it was 4201 new users, whose contributions before and after the warning we hand-coded (by multiple coders) after two of the Huggle tests, and then assessed on a 4-point quality scale. We felt pretty confident (and exhausted) after that that the proportions of obvious vandals to good faith newbies would be about the same for the other tests, and that we could just filter out the obvious baddies by looking at blocks. You're right, though, that it would be interesting to see what other warnings these users got, if any. But unlike a block/ban, it's actually quite difficult to programmatically identify a warning in the database, since it can take so many forms (e.g., can be delivered by a bot, a tool, substituted, by hand, etc.). I'm not sure the effort of doing this qualitatively is worth it, but I could just be saying that because good lord did that coding take a long time!
 * Another point of clarification: we filtered users by number of edits before/after warning for statistical significance purposes, so we didn't base our findings on the people who gave up after an edit or two. Does that make sense? You can find more details about analysis stuff on our data analyst's journal on Meta if you're curious. Maryana (WMF) (talk) 21:41, 17 July 2012 (UTC)