Wikipedia talk:WikiProject User warnings/Testing/CorenSearchBot
I expect you guys know where you're going with this, but my first impression is that the tone is a bit... off? Maybe it's the fact that I'm an old hand, but I'd feel a little bit patronized by the current wording. I dunno, I guess it sounds a little K12. :-)
Then again, maybe you've got the past experience to know this is what works best. — Coren (talk) 22:01, 13 December 2011 (UTC)
- Thanks for looking over these – obviously still very much a work in progress, and we appreciate the feedback :)
- So, couple of reasons we're writing them this way:
- A lot of people who copy-paste stuff into Wikipedia are coming from cultures where it's totally okay to do that, and they honestly don't understand why it would be a problem.
- Newbies are newbies. Things that seem obvious to someone who's been editing for even a month are not obvious to somebody who's just making their first edits. We get tons of feedback every day from very new editors who can't find their contributions, don't know how to add section headers, and all the basic interface stuff that quickly becomes second nature to Wikipedians. I think we should be as deliberate as possible in any newbie-oriented warning, to helping them clear the initial scary hurdle of editing.
- Nobody likes legalese (except some Wikipedians, heh), and that's what the language of these warnings looks like. Generally speaking, people never read those long disclaimers and fine-print terms of use anyway, and most other websites never use that kind of language for things they actually want people to pay attention to. Instead, they use really clear, simple, precise language... 8th-grade reading level, which is also the standard for most newspapers :)
- Does that make sense? The patronizing tone thing is definitely a problem to watch out for, I agree, especially since there are cases of false positives. But I think most people getting this message are not experienced Wikipedians who've been around for ages, like yourself, and we shouldn't talk to them that way :) Anyway, please do let us know what you think as we're tinkering with these, and feel free to offer more suggestions/advice! Maryana (WMF) (talk) 22:46, 13 December 2011 (UTC)
- Like I said, I wouldn't worry about it. Seriously. I've been on the 'net since the hosts file was still copied by hand, and I've been a tech geek for longer than I care to admit. I'm the first to agree that I'm just about the worst person to consult regarding what is a good user interface, and that what is clear to me is obscure jargon for most of the human race. :-) — Coren (talk) 23:06, 13 December 2011 (UTC)
- No worries. Like I said, you bring up a very good point, and it's always useful to get feedback from the grizzled veterans around here ;)
- Also, it's funny: when you pinged me on IRC, your fluency with current netspeak made me think you're a lot younger than you probably are... way to stay hip and fresh! (Do kids still say "hip" and "fresh"? I don't know!) Maryana (WMF) (talk) 23:55, 13 December 2011 (UTC)
- Constant immersion in memeland, all true geeks remain teenagers well past their old age. And I happen to like lolcats. :-) But I was under the impression we (very briefly) met when Phillippe gave me and Brad a tour of the office last month so you would have seen that my body no longer agrees with my assessment of my own age; perhaps I'm just confused and you were out of the office? — Coren (talk) 00:22, 14 December 2011 (UTC)
- D'oh! I was traveling with Sue in Europe all of last month, so I'm pretty sure I missed you... I'm the lady with the short hair who is currently all over the fundraising banners :-p Do you live in the area? You should come by again so I can get a proper introduction.
- Anyway, back to SRS BZNS: I just had a possibly brilliant thought! So, SDPatrolBot, which we've also been testing on, issues two different kinds of warnings based on the recipient's edit history: one for experienced editors (can't remember the exact parameters, but something like 500 edits or more) and one for beginners. Any chance we could throw that algorithm into your bot and see what happens? We could write really detailed newbie-friendly templates for the n00bs and different ones for experienced editors, reminding them to watch out for copyvio (non-patrionizingly, of course). What do you think? You might ask Kingpin13 (the bot's owner) how he does it, assuming you're working in the same language. He's on IRC a lot – lemme know if you need me to interface :) Maryana (WMF) (talk) 00:38, 14 December 2011 (UTC)
- Well, Sue was there so if you were with her so were you (that was on the 7th). :-)
- As for "per-experience-level" notices, that probably make a great deal of sense; newbies are unlikely to make the same kinds of mistakes, and will need different pointers and explanations. Looking up edit count isn't very expensive if I only do it just before I place a notice. — Coren (talk) 01:10, 14 December 2011 (UTC)
- Okay, I added "experienced" versions – you should definitely take a look at those, because you're their target audience! Feel free to edit as you see fit; they're very preliminary drafts. Maryana (WMF) (talk) 20:48, 14 December 2011 (UTC)
- Constant immersion in memeland, all true geeks remain teenagers well past their old age. And I happen to like lolcats. :-) But I was under the impression we (very briefly) met when Phillippe gave me and Brad a tour of the office last month so you would have seen that my body no longer agrees with my assessment of my own age; perhaps I'm just confused and you were out of the office? — Coren (talk) 00:22, 14 December 2011 (UTC)
- Like I said, I wouldn't worry about it. Seriously. I've been on the 'net since the hosts file was still copied by hand, and I've been a tech geek for longer than I care to admit. I'm the first to agree that I'm just about the worst person to consult regarding what is a good user interface, and that what is clear to me is obscure jargon for most of the human race. :-) — Coren (talk) 23:06, 13 December 2011 (UTC)
(←) The "old hand" versions are very nice; short, to the point reminders. I still find the "Thanks for your contributions!" bit more annoying than friendly, but that's probably just my natural cynicism talking – social politeness from automated processes always give me an aftertaste of bureaucratic-insincere automation.
"Thank you for helping us help you help us all." :-) — Coren (talk) 16:06, 15 December 2011 (UTC)
- Yeah, you're right – the closing line is probably overkill. I was still in newbie welcoming/praising mode :) Maryana (WMF) (talk) 17:46, 15 December 2011 (UTC)
The current versions are fine. I'm a little surprised at the edits to {{welcomelaws}}, though – my impression when I was asked to open new talk pages with it is that it was intended to join the message with pointers to the requirements about copying. I can see why you think that's too heavy for a newbie, but then wouldn't it be simply better to use a generic well-tuned welcome template that's already available? — Coren (talk) 20:40, 17 December 2011 (UTC)
- The foremost reason we rewrote the welcome like we did was because we realized that this particular welcome template is never delivered without an accompanying warning about copyvio. In that context, we'd like to test whether it leads to more action we want if we dramatically shorten the template, make it purely a general welcome, and point to the talk threads below it (few newbies understand how talk pages work). We are just now ending a similar welcome template test with XLinkBot, so between the two of them we should know clearly whether it is better to have short, general welcomes accompany warnings or if the current approach is more effective. Steven Walling (WMF) • talk 17:38, 18 December 2011 (UTC)
- That's a good point, and well made. It looks like you guys are just about ready for me to start using these; are there any reporting actions I could do that would help you get good metrics? Do you have a favourite distribution (alternating, strictly random, Gaussian), or weights you want assigned? — Coren (talk) 19:42, 18 December 2011 (UTC)
- Steven and I spent the day making randomizers and tracking templates for all these guys. These three are for newbies (preloaded with the current and test message):
- And these are the same messages for experienced users:
- So, ideally, the bot would first figure out if the user is new (< 500 edits) or experienced (> 500 edits), then grab a warning from the appropriate randomizer, which will deliver one of two templates, current or test. We also made a randomizer for the welcome message: {{Welcomelaws-rand}}. It'll be interesting to see if there's any difference if a user receives a test welcome and default warning, or vice versa. The randomizers and z templates should take care of randomizing and tracking, so I don't think any extra work on your part is required. Let me know if this all makes sense. If you feel like you're good to go, feel free to flip the switch on your bot at any time – just drop Steven or me a talk page message or email so we can update our documentation. And if you have any questions/concerns, feel free to ping either of us, as well :) Maryana (WMF) (talk) 00:01, 20 December 2011 (UTC)
- I got two bug fixes scheduled to go in and live tomorrow, I'll switch the templates at that time and drop you a note. — Coren (talk) 00:07, 20 December 2011 (UTC)
- Oh – I looked at your templates and there is a problem that's probably going to be significant: the welcomelaws is added to a new talk page in the same edit that the notice template is; and since you use the low order bit of time to pick the template, you'll always get A-A or B-B when I expect you would have wanted both variables to be independent? — Coren (talk) 00:11, 20 December 2011 (UTC)
- Ahh, damn, you're right. Good catch. So, there are two possibilities, I think. One, we nix trying to mix the streams and just stick to current welcome + current warning, test welcome + test warning. Alternately, we could throw something like this into the {{Welcomelaws-rand}} randomizer:
- {{#switch:{{str left|{{{1}}}|1}} |A |B |C |D |E |F |G |H |I |J |K |L |M |N |O |P }}
- That algorithm should put usernames beginning with A-P into one cohort, and (rinse, repeat, but with different parameters) those with usernames Q-Z and numbers into another. A little more hacky, but it can be done. What do you think? Maryana (WMF) (talk) 00:25, 20 December 2011 (UTC)
- On second thought, fiddling with that alphabetical randomizer is probably going to be too complicated. It's fine that welcomes + warnings would be delivered together, as long as they always matched (i.e., default welcome with default warning, test welcome with test warning). Would that be the case? Should we do a test run to find out? Maryana (WMF) (talk) 00:24, 21 December 2011 (UTC)
- It would be, but I might simply have a better way: I can do the randomness on the bot side. This makes it truly random (and independent). — Coren (talk) 05:06, 21 December 2011 (UTC)
- That would be excellent, thank you! Lemme know when you start the test. Maryana (WMF) (talk) 18:42, 21 December 2011 (UTC)
- It would be, but I might simply have a better way: I can do the randomness on the bot side. This makes it truly random (and independent). — Coren (talk) 05:06, 21 December 2011 (UTC)
- On second thought, fiddling with that alphabetical randomizer is probably going to be too complicated. It's fine that welcomes + warnings would be delivered together, as long as they always matched (i.e., default welcome with default warning, test welcome with test warning). Would that be the case? Should we do a test run to find out? Maryana (WMF) (talk) 00:24, 21 December 2011 (UTC)
- Ahh, damn, you're right. Good catch. So, there are two possibilities, I think. One, we nix trying to mix the streams and just stick to current welcome + current warning, test welcome + test warning. Alternately, we could throw something like this into the {{Welcomelaws-rand}} randomizer:
- That's a good point, and well made. It looks like you guys are just about ready for me to start using these; are there any reporting actions I could do that would help you get good metrics? Do you have a favourite distribution (alternating, strictly random, Gaussian), or weights you want assigned? — Coren (talk) 19:42, 18 December 2011 (UTC)
- The foremost reason we rewrote the welcome like we did was because we realized that this particular welcome template is never delivered without an accompanying warning about copyvio. In that context, we'd like to test whether it leads to more action we want if we dramatically shorten the template, make it purely a general welcome, and point to the talk threads below it (few newbies understand how talk pages work). We are just now ending a similar welcome template test with XLinkBot, so between the two of them we should know clearly whether it is better to have short, general welcomes accompany warnings or if the current approach is more effective. Steven Walling (WMF) • talk 17:38, 18 December 2011 (UTC)
Accessibility
[edit]Nice work. A few pointers on accessibility, concerning (as an example):
- Click here to edit your article.
- Click the book icon in the top left-hand corner of your editing window.
- "here" is not adequate as link text, which should be meaningful when read in isolation. Consider using, say, edit your article or Click here to edit your article.
- How will a bind user, or a user with image downloading disabled, see a book icon?
Happy to advise further, but WP:ACCESS can also help. . Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:03, 15 December 2011 (UTC)
- Thank you, Andy, those are excellent points. Will work them in. Let me know if you see anything else that looks off. Maryana (WMF) (talk) 17:48, 15 December 2011 (UTC)
- Hey Andy. The latest revisions should all say things like "edit the article" etc. which are friendlier with people to screen readers etc. Steven Walling (WMF) • talk 01:26, 16 December 2011 (UTC)
Test started
[edit]The SCBot version with the a/b switch (and the experienced/newbie switch) is now live. All we need now is for someone to be naughty. :-) — Coren (talk) 00:52, 13 January 2012 (UTC)
- w00t! Thank you so much, Coren! Maryana (WMF) (talk) 01:01, 13 January 2012 (UTC)
Link to policy in experienced templates?
[edit]I'm not sure it's a good idea to eliminate all links to policies from the experienced templates; Wikipedia:Copy-paste can be useful reading even for editors who've been around a while if they don't understand what they're doing wrong. Experienced or not, if they're copying from other pages, it would seem they might not know yet that they can't do that. :) For instance, evidently this experienced contributor did not know that he can't copy-paste content to a new title. I saw another one like that from a few days ago. A link to Wikipedia:Copy-paste might raise his awareness of that issue without necessarily making the template less friendly or more condescending. It might actually reduce stress more than having to get a follow up explanation from a contributor or having to ask somebody else what they did wrong. :) --Moonriddengirl (talk) 02:53, 5 February 2012 (UTC)
- Thanks for the feedback, Maggie :)
- Looking at the example diff you linked to, it appears to me that the person who received that warning simply moved the content of an existing article to reflect the correct spelling of the organization. He/she didn't actually write the content and probably had no idea it was a copyvio. Unless I'm reading that wrong, it looks like a false positive... which is a major factor to consider with the "experienced" warning. (Case in point: Sue got warned by it once!) That's why we were hesitant to include links to policy pages in these warnings. The new versions follow our assumption (from purely anecdotal evidence) that experienced editors who receive these warnings either accidentally forgot to attribute a source or were a false positive – in which case, giving them a link to a policy page would seem a bit condescending. Of course, we could be wrong, and maybe the people receiving these warnings do look like they need a refresher course on WP:Copy-paste. We'll definitely dig into the data after the test ends next week and see if that's the case. Watch our documentation page for the spreadsheet with all the results (should be ready next Monday), and feel free to poke around and help with qualitative coding if you have the time/inclination :) Maryana (WMF) (talk) 00:37, 7 February 2012 (UTC)
- Actually, a false positive is when there is no duplication. This is a different situation, and not uncommon. CorenSearchBot detected an unattributed split when it matched a new article to a mirror of an existing article. Not all experienced users know that our content here is not public domain any more than content found elsewhere on the web. It's still a copyright infringement if they don't follow the licensing requirements, as this person did not. Wikipedia:Copy-paste not only talks about copying from external sites, but also talks about copying content from one Wikipedia article to another. You'll notice that the person who received that specific caution did nothing to repair the problem. If he had simply forgotten to fix the split, surely he would have done so? --Moonriddengirl (talk) 11:25, 7 February 2012 (UTC)
- Oh, lovely. Now the bot is broken again. :/ --Moonriddengirl (talk) 11:31, 7 February 2012 (UTC)