[00:02:23] https://jwmeeting.miraheze.org/wiki/Help:Wiki_editing/Recommendations_for_mobile_friendly_articles_on_Wikimedia_wikis [00:03:18] [1/2] how do i remove the default mode thing (i already removed DarkMode) [00:03:18] [2/2] https://cdn.discordapp.com/attachments/407537962553966603/1213637790600925184/1BRgJgW.png?ex=65f63345&is=65e3be45&hm=c7714d4ee7083013ecc6916645065c5b92b77c2eabec96394c6584d7899dcbf9& [00:06:38] [1/2] or you uncheck the extension: [00:06:38] [2/2] https://jwmeeting.miraheze.org/wiki/Special:ManageWiki/extensions#mw-section-other [00:06:54] i already did???? [00:07:31] It shouldn't be there. But It can be the cache [00:07:45] nope, still there [00:07:50] The server needs to refresh, just like Commons.css [00:08:17] i already refreshed the web page [00:08:17] It's teh server cache, not your wikipage cache [00:08:40] how do i clear it :3 (i already did ctrl + r and it did nothing) [00:09:07] Remember when you uploaded your logo? [00:09:25] It took a while, because it is the SERVER cache [00:09:37] Not your wikipage cache [00:09:58] It doesn't help ctrl+r [00:10:13] It will go away soon enough [00:11:50] a day has already passed and its still there [00:11:50] :/ [03:37:09] Go to chrome dev tools by tapping F12 the right click the reload button and choose “Empty Cache and Hard Reload” to be certain it isn’t a local cache artifact. [04:31:14] i only speak Firefox :moonch: [04:31:45] how do i use the pop up extension? [04:31:55] i enabled popups by default yet it did nothing [05:11:39] Popup? Sometimes works, sometimes don't [05:12:13] Like, it require some certain things for it to fully works as you pleased. [06:12:54] I would also double-check to ensure that it didn't get hardcoded into your MediaWiki:Sidebar page, as that also controls what displays there. [06:14:48] RE: Popups, that's one that can take a LONG while to start having the intended effect, much like Visual Editor. [06:17:28] Also worth noting: it'll only work in content (main) namespace by default [06:48:42] just for updating, can I replace Phabricator with Phorge, at this moment? [10:10:10] is there a template in meta Miraheze to show on the user page if a user is dead? [10:35:05] No, not that I am aware of [10:47:33] there is a setup for it, not sure if it's a template per se [10:47:58] [1/2] yeah not a template, but this [10:47:58] [2/2] https://meta.miraheze.org/wiki/Deceased_Mirahezians [11:05:50] There should be a userbox for this [13:00:01] [1/7] Does anyone know whether the automatically generated layout of file pages is stored in a system message? I suppose it is like this: [13:00:02] [2/7] ``` [13:00:02] [3/7] ==Summary== [13:00:02] [4/7] $1 [13:00:03] [5/7] ==Licensing== [13:00:03] [6/7] $2 [13:00:03] [7/7] ``` [13:06:22] It is, for the summary you can customize it by editing a MediaWiki interface page [13:11:34] Special:PrefixIndex?prefix=&namespace=8 [13:11:43] Here you get all system messages [13:30:48] I want to customize those two headings (`==Summary==` and `==Licensing==`), but there are so many system messages, and I don't know of any system messages that store them😵‍💫. [13:32:39] In Upload wizzard? [13:36:00] Or just uploading a file? [13:36:02] MediaWiki:Upload-default-description [13:36:59] MediaWiki:Uploadtext [13:46:10] MediaWiki:Summary [13:55:44] wait [13:55:53] I think I know where to set it up [14:07:42] [1/8] It is indeed possible that they were customized through UploadWizard🤔... My supposition comes from seeing the following file pages of layout on Wikimedia Commons: [14:07:42] [2/8] ``` [14:07:42] [3/8] =={{int:filedesc}}== [14:07:42] https://meta.miraheze.org/wiki/int:Template:filedesc [14:07:42] [4/8] ...... [14:07:43] [5/8] =={{int:license-header}}== [14:07:43] https://meta.miraheze.org/wiki/int:Template:license-header [14:07:43] [6/8] ...... [14:07:43] [7/8] ``` [14:07:44] [8/8] Because normally, those two headings should come from the corresponding strings that are in the same language as the site content language. [14:09:52] Oh... interesting. I didn't use that in my wiki so idk [14:10:39] That is only for wiki's that use many languages [14:11:04] It will pick the language of the user [14:11:44] My wiki for example only uses EN-gb [14:12:00] So I don't need to have it translated. [14:13:48] [1/3] MediaWiki:Upload-default-description [14:13:48] [2/3] ```{{Information |description = |date = |source = |author = |permission = CC-by-sa-4.0 |other versions = }}``` [14:13:48] https://meta.miraheze.org/wiki/Template:Information_ [14:13:49] [3/3] The system will put ==Summary == above it, and == License == underneath it. [14:14:33] (If you choose a license from the dropdown menu, that is) [14:14:58] I just use my own pre defined templates instead [14:30:11] [1/2] But the structure itself is hardcoded in the software, so you can change the word Summary to File information, but not the ==File Information == [14:30:11] [2/2] You can remove it afterwards though [14:32:44] So summary would be at `MediaWiki:Filedesc` and the license is at `MediaWiki:License-header` [14:33:04] indeed [14:36:51] https://cdn.discordapp.com/attachments/407537962553966603/1213857626564206653/1080933379215736862.png?ex=65f70002&is=65e48b02&hm=1f5972d2bf3aa9f89829f826bf70d3cd9b98df920b97af380e583376c6d29b51& [14:37:22] What did you have in mind? [14:43:30] `File Information` CAN be changed using `MediaWiki:Metadata` [14:44:28] It's all been listed at `Special:AllMessages` [14:44:44] Metadata shows the stuff delivered in the image itself, like the camera [14:45:14] yea, I know, but I'm talking about the File Information header [14:45:31] All of the actual messages from the software has been listed at `Special:AllMessages` [14:45:45] I know, I mentioned a few above [14:46:02] at least now you know- [14:48:58] https://jwmeeting.miraheze.org/wiki/File:Gage_Fleegle.jpg [15:06:45] [1/3] Chinese is divided into simplified and traditional forms, each with multiple variants. I want to edit two headings to accommodate the diverse content of these variants, because when the site content language is `zh`, those two headings default to outputting simplified Chinese (`zh-Hans`) content. [15:06:45] [2/3] If I can't edit those two headings, I can only use `MediaWiki:Conversiontable` to achieve certain needs. [15:06:45] [3/3] https://cdn.discordapp.com/attachments/407537962553966603/1213865150608838676/400911745767571456.png?ex=65f70704&is=65e49204&hm=adee80e0dc2097585cd647d7303ffc214eb9e135f7f7328e3157f7cccfac8fd7& [15:17:09] Ah. Okay [15:19:09] [1/3] Each message has different languages. [15:19:09] [2/3] Like MediaWiki:Metadata/zh and [15:19:09] [3/3] MediaWiki:Metadata/zh-hans [15:19:21] So each user will see it in his/her own language [15:20:00] Just change your language, and check after upload. You'll see that the language will have been changed [15:22:28] [1/3] So in: [15:22:29] [2/3] MediaWiki:Filedesc add: ```{{int:filedesc}}``` [15:22:29] https://meta.miraheze.org/wiki/int:Template:filedesc [15:22:29] [3/3] MediaWiki:License-header add:```{{int:license-header}}``` [15:22:29] https://meta.miraheze.org/wiki/int:Template:license-header [15:29:36] ok [15:38:05] I am in a great mode, but wondering, should banner images for a page have alt text? [15:38:50] I know, but those two outputted headings are not system messages. I just tried to add the corresponding i18n magic words in `MediaWiki:Filedesc` and `MediaWiki:License-header`, but an error message "Expansion depth limit exceeded" appears. [15:40:41] hmmm, strange. [15:41:25] MediaWiki:Filedesc/zh [15:41:41] MediaWiki:Filedesc/zh-hans [15:42:02] MediaWiki:License-header/zh [15:42:22] MediaWiki:License-header/zh-hans [15:42:42] They do have the system message in these languages [15:43:03] It automatically puts that between == == [15:45:10] [1/6] Commons.Wikimedia uses a different system. They translate through [15:45:10] [2/6] ```=={{int:filedesc}}== [15:45:10] https://meta.miraheze.org/wiki/int:Template:filedesc [15:45:11] [3/6] ...... [15:45:11] [4/6] =={{int:license-header}}== [15:45:11] https://meta.miraheze.org/wiki/int:Template:license-header [15:45:11] [5/6] ......``` [15:45:11] [6/6] But we have all messages translated as far as I know, in the most spoken languages [15:45:23] I am in a great mood, but wondering, should banner images for a page have alt text? [15:45:52] I always think you shoold, especially in case someone uses a screenreader [15:46:04] Alt messages are more descriptive [15:49:04] [1/2] People who are visually impaired, rely on these alt= messages. So they know what is shown. So not: "House on hill" but "A brick house on a hill with tall trees and fence. A red tile roof and a wooden green shed is beside it. People are waving." [15:49:04] [2/2] Just as example [15:49:41] Also those who are colour blind for example can use that [15:51:48] Just having to work out what to add for 300+ AI images [16:17:44] alt desc has to be as short as the context allows, so "house on hill" may be sufficient if that is all the image is meant to convey [16:20:28] For visually impaired is that to little info. [16:21:38] it's still interesting that alt image info would actually increase your scores in SEO [16:22:56] Because visually impaired people depend on it [16:23:24] Their screen readers can read what is described [16:23:55] Therefore you need to be more explanatory in your alt [16:25:18] @HandtheGene probably means what you write in the thumb under the image? [16:25:27] [1/2] https://en.wikipedia.org/wiki/Wikipedia:Manual_of_Style/Accessibility/Alternative_text_for_images [16:25:27] [2/2] alt text should be short and convey the same essential information as the image, so in some contexts, "house on hill" or "a house on a hill" will be sufficient., whereas other contexts may warrant a lengthier description like "A brick house on a hill with tall trees and fence. A red tile roof and a wooden green shed is beside it. People are waving." [16:25:49] LOL [16:25:53] Wikipedia [16:26:21] Wikipedia is never been friendly for visually impaired or Mobile users [16:26:31] Don't trust to much on their MOS [16:28:06] what about the ALT function that social media started adopting recently (twitter, tumvlr) [16:28:27] I rarely use any social media anymore [16:28:34] So I don't know [16:28:44] sometime it was "abused" to add more captions to the post [16:28:53] people put very detailed descriptions in ther [16:29:00] yep [16:29:24] I stay clear of twitter or Facebook or whatever, I just use discord [16:29:41] maybe implement something like that as an alternative to default alt [16:30:04] [1/3] there are other guides that say the same thing about alt text having to be short and serving the same function as the image, which depends on the context [16:30:04] [2/3] https://accessibility.psu.edu/images/alttext/ [16:30:05] [3/3] https://www.w3.org/WAI/tutorials/images/tips/ [16:31:13] point is just that you provide all of the essential information without giving superfluous details or highlighting the wrong aspects of the image, in the specific context where the image is used [16:31:35] Yeah. "House on the Hill" Try to imagine what kind I mean. [16:31:41] It’s important to not only provide ALT text, but also provide one that is useful in the context of the document. You may not need to include every detail about the image, and you may be able to skip over purely decorative images, but those that contain critical information need to have it to spelled out in the ALT text. [16:32:17] You don't need to spell out that it has a rain drain, and a chair, and grass around it [16:34:10] [1/3] But you need to describe, what kind of house it is. What the scene is. [16:34:10] [2/3] You need to be able to picture that house on the hill. [16:34:10] [3/3] Is it a wooden house, or a red brick house, does it have tiles or straw. Is et a two under one roof house, or separate [16:34:29] The blind person needs to be able to picture it in their head. [16:35:27] You do not use long sentences. But you have to share Key info to help the impaired to picture what it is. [16:36:26] If I write "House on the hill" you don't know anything else, then a house on a hill. [16:38:51] i feel like it's going in circles so i'm just gonna drop it [17:30:12] can someone remind me if col/rowspan can be placed in styling? [17:31:42] I don't think so, at least it was a HTML thing [17:31:43] I don't think so [17:32:00] But you can use grid to do pretty much the same thing [17:33:36] Maybe flexbox also [17:40:29] I just joined and am trying to make a wiki, im logged in but I dont see an edit option any ideas why ((on mobile)) [17:40:54] make a wiki, you mean requesting one right? [17:41:02] Are you on your wiki ? [17:41:12] Already made and verified [17:41:34] the wiki [17:41:36] not account [17:41:46] Yes, and im confirming that [17:42:03] Did you create a page ? [17:42:27] Yep, page is all made, it has a single edit by one other, just trying to figure out how to do it myself [17:42:57] can you link? [17:43:00] your wiki here [17:43:22] wait, on mobile [17:43:37] try to switch to desktop mode, at the very bottom [17:43:47] default mobile mode nerfs a lot of things [17:44:02] You can edit on mobile display normally [17:44:03] Ah, I see, thank you ill try [17:44:11] It's ugly and unpractical but still [17:44:41] If you don’t enjoy sanity [17:44:55] I can deal with ugly, it works thank you! [17:44:57] Ye [17:45:07] we advise to change skins, either all together to some mobile adaptive and disable MobileFrontend, or pick another for mobile mode skin [17:45:19] instead of default Minerva [17:45:30] I did some template with CSS and js on mobile [17:45:39] Why ? [17:45:57] well, as I said, Minerva disables a lot of things [17:46:02] especially for admins [17:46:26] What skin do you recommend? [17:48:45] tbh, I uses Citizen- [17:48:55] it's the only skin I use- [17:49:12] [1/3] Timeless or Cosmos, and people also seem to like Citizen [17:49:12] [2/3] but I personally would make these default skin and disable MobileFrontend altogether [17:49:13] [3/3] if you really want "traditional" skin like old Vector or Monobook, then mobile mode is needed (if you want to please mobile users) [17:52:18] Well last time we tried disabling mobile frontend it broke the wiki for mobile sooo [17:53:00] it broke? then use an responsive skin [17:54:14] yeah, the point is mobile responsivness [17:54:33] what I mean is either you pick such skin as the main one disable frontend [17:54:48] or put this skin in frontend setting, instead of Minerva [17:56:55] I mean we juste kept the default vector-2022 [17:57:22] is possible to add a footer icon? [17:57:23] Is there a way to like test the skin without actually changing the parameters? [17:57:35] Bc I don't have this power [17:57:45] ?useskin=nameoftheskin in the url [17:58:12] ?useskin=oasis is my favourite [17:58:16] For example, here's Meta's main page in Monobook: https://meta.miraheze.org/wiki/Miraheze?useskin=monobook [17:59:07] Thx [17:59:48] w/ JS [18:00:17] and Miraheze allows it? [18:00:22] yes [18:00:23] yes [18:00:29] this is not fandom lol [18:00:33] there's no js moderation [18:00:41] There is a lot of Wiki request open and a lot of people are waiting https://meta.miraheze.org/wiki/Special:RequestWikiQueue [18:01:13] well, if you try to like mine crypto on your visitor's browsers, Stewards _might_ get involved tho [18:01:27] ofc lol [18:01:48] but comparing to fandom [18:02:04] we are cooler w/ js [18:02:58] obviously I will not mine crypto [18:03:25] I'm happy with 3000 MXN [18:05:43] But FANDOM lets you open a wiki in a instant ||and get your wiki deleted by a little inactivity in a instant|| [18:11:06] I'm doing a few of them rn [18:36:58] can i use non installed skin (i guess not) [18:37:14] it has to be enabled on ManageWiki [18:46:11] The fact that 5 wiki creator RfPs had to be opened to deal with the huge backlog of wiki requests may be a sign that Miraheze has outgrown the manual review system and that it should be replaced [18:52:29] Or just contract more Wiki creators [18:52:54] That can only work so much. [18:52:55] That’s not how it works? [18:52:57] ^ [18:53:03] We don’t contract them lmao [18:53:13] They get promoted via community consensus. [18:53:21] So yeah. [18:53:28] WCs are community members who are nice enough to take their out of their lives to volunteer [18:53:30] I was once one, so I know. [18:53:31] How else would we handle them? [18:53:33] AI? [18:53:46] I mean, our turnaround time isn't terrible [18:53:55] I rather have reviews than see stewards getting backloged w/ actually troublesome wikis [18:54:05] Yeah that too [18:54:18] How can we apply, what are the conditions, and is there anything else in the job ? [18:54:23] cough fandom experience cough [18:54:27] If people are actually capable and competent to become a WC, more power to them. [18:54:36] I was once able to, so there we go. [18:54:58] So if I can do it, pretty much anyone else can if they work hard enough to achieve it. [18:55:11] plus let's be honest - not all people don't read all policies before jumping into request [18:55:12] General life lessons moment. [18:55:32] sadly [18:55:37] I seem to be of more worth than even I give myself credit for. [18:55:50] lol [18:57:18] I mean that's just a given. [18:57:39] Not everyone is going to take the time to read them, honestly even though they should. [18:59:56] Wasn’t that already part of the plan [19:00:58] [1/3] I'm still figuring out the exact details, but the gist is that an AI is trained to determine the likelihood that a wiki will violate the Content Policy before and after it's created (0.00-1.00). If the likelihood is above a certain amount, the wiki is flagged on an IRC/Discord feed (something like `#flagged-wikis`) in the Miraheze volunteer channel; all wiki cre [19:00:59] [2/3] ations will also be logged in a public `#wiki-creations` channel in case the AI misses something an alert user notices. Once the AI reaches a satisfactory level of accuracy and the logs are set up, we allow anyone to create wikis with the peace of mind that any CP-violating wikis will most likely be shut down before they can do any real damage; to avoid putting [19:00:59] [3/3] more work onto Stewards, the wiki creator role will be converted into a "wiki reviewer" role that has the power to delete/undelete wikis (which also helps wiki reopen requests get handled faster) [19:01:17] [[Wiki Creators]]. Tldr apply on [[RfP]]. Not much and not really [19:01:17] https://meta.miraheze.org/wiki/Wiki_Creators https://meta.miraheze.org/wiki/RfP [19:01:18] [1/2] [19:01:18] [2/2] [19:01:59] [[WC]] [19:01:59] https://meta.miraheze.org/wiki/WC [19:02:00] [19:02:07] Okay I'll take a look [19:02:16] You need to follow the volunteers conduct policy [19:02:23] As an extra condition [19:05:01] My Wiki request was on hold and can you please check it out [19:24:05] The reviewer who put it on hold, has certainly asked fro more details? [19:25:44] That has been the idea since an early stage but it would have to be a very good model for that to work [19:26:12] but the idea would be that clear cut +1s and -1s could be handled by the AI and human reviewers would process the ones that it's unsure about [19:26:25] but that's the issue with AI GPT models, they rarely admit they're unsure, they make stuff up [19:28:51] I asked ChatGPT to review some wiki requests and it got them all right but it might not always work [19:37:08] I added more details about my Wiki request [19:55:14] A fly just flew into my ear [19:55:33] I think it left thi [20:00:53] uuh [20:01:01] I tried the citizen skin [20:01:14] [1/2] and things looks like css was killed in the process [20:01:14] [2/2] https://cdn.discordapp.com/attachments/407537962553966603/1213939260550553752/image.png?ex=65f74c09&is=65e4d709&hm=a6a6a1415fdd15d338ea78562e638b645febe431e6b24a64ed4ac8d34fdf193a& [20:01:28] [1/6] with this comment [20:01:28] [2/6] ```css [20:01:29] [3/6] /* [20:01:29] [4/6] Problematic modules: {"skins.citizen.icons":"missing","skins.citizen.styles":"missing","skins.citizen.icons.wmui":"missing"} [20:01:29] [5/6] */ [20:01:30] [6/6] ``` [20:01:46] does that come from us or is the skin just broken ? [20:04:02] and this does not seems to work when i use firefox dev tools to simulate a mobile environment is there a way ? [20:50:01] Works as it should on my wiki [20:50:42] This looks more like a css problem [20:56:18] Just a note that there a lot of wiki creator request open, three without comments. In your spare time, consider leaving a comment for some of our candidates! [20:59:01] three without comments???? [20:59:09] You mean !votes? [20:59:19] only one i saw waa raiddar [20:59:59] Yes, but comments incompasses !votes ({{s}}) and comments ({{comment}}) [20:59:59] https://meta.miraheze.org/wiki/Template:s [21:00:00] [21:00:00] [21:00:48] yeah, 2 in the past few hours [21:02:08] Raidarr applied 24 hours ago, And I gave him some questions [21:02:34] Okay, great [21:02:44] oof [21:03:06] I have faith in raiddar [21:03:41] me too [21:04:54] I love the influx of requests. [21:05:11] It is encouraging yes [21:05:17] ...though I know the possibility of me getting support is next to zero. [21:05:21] So I won't. [21:05:23] But I will vote. [21:05:42] Great, always good to have more people involved in discussions [21:06:00] I know right. [21:08:50] I'm actually quite comfortable helping in other areas where advanced permissions aren't as necessary. [21:15:58] resolved it by purging 🤷 [21:17:06] https://tenor.com/view/purge-button-press-fast-gif-17107922 [21:17:27] lol [21:30:36] automatic purging when [21:37:29] I'm new to MH bc I didn't have time to work on any projects, and I heard they're doing approval for wikis now. whats up with that? [21:37:37] Like you have to message someone [21:38:15] to start a new wiki on Miraheze you should make a wiki request [21:38:56] a wiki creator, volunteer, will review your request and accept it if it's fine w/ policies etc [21:42:34] [[meta:]] [21:42:34] https://meta.wikimedia.org/wiki/ [21:43:08] http://meta.miraheze.org [21:43:17] Why'd they change it from that to the earlier method of just requesting? My friend made one for us and he had to message a few people regarding the topic of the wiki and continously had to elaborate the topic [21:44:32] well yeah [21:44:40] creator can ask question [21:44:44] in request [21:45:11] wow that's a lot of requests up [21:45:15] idk what you mean by messages, I guess they tried to ask here? [21:45:36] what's w/ influx tho? what new fandom controversy I've missed lol [21:46:58] Thing is, that sometimes the scope and managing isn't clear. So that's why a reviewer asks questions [21:48:14] the process is about the same though, it's just that the expectation for filling the form more completely + known problem areas needing to be addressed has increased since earlier times [21:49:31] Ahhhh okay that makes sense [22:01:11] who??? [22:02:11] hi [22:02:17] when making an infobox template [22:02:37] it says "template loop detected" [22:10:56] show your template code [22:17:55] https://meta.miraheze.org/wiki/Meta:Requests_for_permissions#Collei_(Wiki_creator) [22:18:20] https://meta.miraheze.org/wiki/Meta:Requests_for_permissions#Blad_(Wiki_creator) [22:18:33] https://meta.miraheze.org/wiki/Meta:Requests_for_permissions#AirmanKitten203_(Wiki_creator) [22:18:39] 3 now!! [22:19:26] JESUS [22:20:20] https://cdn.discordapp.com/attachments/407537962553966603/1213974269692870776/message.txt?ex=65f76ca4&is=65e4f7a4&hm=9a7a5c6d0653bc525694ea6112db8109f7e7328f33478548ed089aab2f640de5& [22:20:29] [1/2] It's probably because you copied the documentation page on the template page. [22:20:30] [2/2] Click the [edit] link on top of the page and copy that text. don't click the [edit] link IN the page [22:21:02] I don’t even feel confident to be a WC and now I’m feeling pressured [22:21:19] You don't have too [22:21:40] But the list of new questions becomes LOOOOONG [22:21:42] 😄 [22:23:46] from wikipedia right? [22:23:58] mon dieu [22:25:04] i'm kind of new to this stuff [22:25:27] Soon, very soon 😏 [22:26:50] [1/3] Correct, but remember, that copying from Wikipedia means that you have to copy a LOAD of other templates, modules and styles over. [22:26:50] [2/3] It's not as simple. [22:26:50] [3/3] Read [[Infoboxes]] [22:26:50] https://meta.miraheze.org/wiki/Infoboxes [22:26:51] [22:27:44] ok thanks [22:29:37] yeah LMAO [22:29:44] peer pressure goes craaaaazy [22:31:55] In all fairness, it's also proof that the system works. Growing our volunteer base is always a good thing. [22:32:55] Yeah. [22:33:19] I think it’s also a snowball effect definitely [22:33:26] Miraheze has traditionally had a narrow and struggling volunteer base [22:33:36] Strengthening and prospering is good [22:33:42] People seeing others run for WC, deciding to give it a shot [22:33:47] And we seem to have energy at the moment [22:33:54] That's my take of the day before i sleep [22:34:17] It's absolutely a snowball effect haha - we start with UO and OrangeStar, progress to Pisces/Rodejong, then raidarr, and now 3 others [22:34:44] Let’s recap real quick [22:34:50] In the span of ONE WEEK [22:35:15] 7 new ones [22:35:16] We have 8 Wiki CreatornRequests [22:35:23] 8??? [22:35:26] wait [22:35:30] UO [22:35:30] 8 indeed [22:35:32] OS [22:35:38] oh wait [22:35:39] 8 [22:35:39] Ro, Pic [22:35:41] raid [22:35:43] 3 [22:35:50] that's ridiculous tbh [22:35:53] Jesus [22:35:57] LOL [22:36:11] nah not jesus here [22:36:17] [1/5] That would still be true for the system I proposed earlier (posted below for convenience): [22:36:17] [2/5] ```... the gist is that an AI is trained to determine the likelihood that a wiki will violate the Content Policy before and after it's created (0.00-1.00). If the likelihood is above a certain amount, the wiki is flagged on an IRC/Discord feed (something like #flagged-wikis) in the Miraheze volunteer channel; all wiki creations will also be logged in a public #wiki-c [22:36:17] [3/5] reations channel in case the AI misses something an alert user notices. Once the AI reaches a satisfactory level of accuracy and the logs are set up, we allow anyone to create wikis with the peace of mind that any CP-violating wikis will most likely be shut down before they can do any real damage; to avoid putting more work onto Stewards, the wiki creator role will b [22:36:18] [4/5] e converted into a "wiki reviewer" role that has the power to delete/undelete wikis (which also helps wiki reopen requests get handled faster)``` [22:36:18] [5/5] It will take a lot of work to accomplish and I don't expect it to be completed within a week or even a month; once it does, however, it'll be worth it for the influx of new wikis and reduced workload [22:36:18] JIMBO CHRIST [22:36:21] I'm inventing questions full time at the moment 😄 [22:36:44] AI is a very long term product [22:36:47] Ask GPR to write wiki requests lmao [22:36:51] As in years in the future [22:36:53] LOL [22:36:57] Hey [22:37:06] You know at least some people are gonna do that [22:37:20] snecking a few in is good practice [22:38:03] It could most likely be done by the end of the year if SRE were willing to work on it every day [22:38:06] I think the system fandom and wikimedia uses to detect vandalism could be used as an idea foe a base maybe idk ai [22:38:15] x to doubt [22:38:19] That would leave no time for just about anything else [22:38:48] And also I'd add that the most talented engineers on the planet are currently trying to break the problem that is AI and the current product is GPT 4 [22:39:05] Which is a great product - but really doesn't meet standards as it currently stands for WCing [22:39:20] Encouraging [22:39:56] Then again we are fundamentally talking about two different technologied [22:40:20] Generating text isn’t the same as understanding text [22:40:32] Yes [22:41:16] Eventually, AI would be great [22:41:34] Short-to-medium term, it's likely manual [22:41:43] Has anyone tried feeding wiki requests to gpt? Im fairly sure someone did i just dont recall who [22:42:46] Surely before that you should have like a minimum character limit on the Wiki description [22:42:57] We do [22:43:02] Collei and I both did; Collei reported that GPT got all of the answers rights, while GPT got 3/5 correct for me [22:43:12] What it get wrong [22:43:28] 🤷 bigger minimum character limit [22:43:53] One for Content Policy #6, another for insufficient information [22:44:18] [1/2] Wake up babe new wiki request just dropped: [22:44:18] [2/2] `This wiki will be on just how cool mailboxes are cause they baded [22:44:41] Should be 250 [22:45:18] sent early mb [22:47:03] ..which one was 6 again [22:47:10] anarchy [22:47:18] I didn't understand when you said "riaici" but "……" has me intrigued [22:47:54] my point is char limits are kinda useless they just shove it full of spam [22:48:24] Brandon saw a request that was smt like “give me a wiki…………………………………………….” [22:48:25] iirc [22:48:31] @bwm0 was dat joo? [22:49:00] Does anyone have a working or non working prototype of the ai system? [22:49:11] yes [22:49:14] multiple [22:49:15] That's #4; #6 is the prohibition on sexual content involving minors (the one that ChatGPT missed, which is somewhat concerning but expected as the internet search plugin for GPT is restricted to GPT-4 and it couldn't have known about the topic of the wiki as it wasn't specified) [22:49:20] Ah [22:49:39] Feels concerning [22:49:48] more so than anarchy, cuz, ya know [22:49:50] yeah [22:50:02] What was the request prompt for that one [22:50:26] It was actually taken from a real request for a wiki named "Showtime X Monsters Wiki" [22:50:35] I don't have the link on me [22:50:54] cause that seems like more a fundamental issue with any AI system that can’t access external information [22:52:41] What if you install a system where every other word has to be fish or relating to Cod, and if people can't describe their wiki using that then they can't have a wiki :nffsUp: its a solution no one else will propose I'm sure [22:53:52] Then you'd have people spamming "wiki fish" to try and get a wiki [22:54:57] You got a point, if the Call of Duty wiki ever decides to migrate, mods will have a pretty difficult time discerning whether they are talking relating to fish or Call of Duty [22:56:43] [1/2] wot [22:56:43] [2/2] https://cdn.discordapp.com/attachments/407537962553966603/1213983423283335198/image.gif?ex=65f7752a&is=65e5002a&hm=23b498df826f27e7c9e5657c41023f03c21274f1b244101ce5564c8dbcaf889b& [22:58:06] Cod of Duty [23:00:10] Hello! [23:00:18] olleh! [23:00:58] "I want cod a movie wiki ocean for in-depth cod analysis of box office cod, this includes fish and it includes a full fish of cod on actors and behind the planketon" - Mods "can you not use planketon to describe your wiki" [23:01:22] I am the creator of the Worlds Chat Wiki. [23:02:22] Hi [23:05:23] welcome. Do you have a question? Or can we help you with something? [23:06:09] Not at the moment. Figured to join the Miraheze discord. 🙂 [23:10:05] :mirahezelogo: [23:12:21] Welcome. [23:12:26] Glad to have you. [23:14:45] I disagree thoroughly that AI is an ideal horse to go with right now [23:15:15] re, as the cornerstone for wiki approvals [23:20:59] [1/2] AI has no judgement. [23:20:59] [2/2] You have to be very specific to get the right answer. [23:21:00] not right now [23:21:38] In the long term future it could be a useful assistant [23:23:02] For simple repetitive tasks, it is ideal, like math. But where it comes to judgement.... no way [23:24:05] Its not really good at maths from what I recall [23:24:51] I have no issues with math. But I don't use very complex math either so Who am I to judge it 😄 [23:25:39] Thing is.. AI can only spout out what has been put in there by humans in the first place [23:25:50] welcome! [23:26:15] It can recite, It can extract, it can calculate the data that we have put in there in the first place [23:27:48] [1/6] I use it daily, to extract database data, and to put it in wiki templates. [23:27:49] [2/6] But even then I have to be specific in how to ask it. [23:27:49] [3/6] For example: [23:27:49] [4/6] Replace daynumbers with daynames [23:27:49] [5/6] Type the time as hh:mm not as hh:mm:ss [23:27:50] [6/6] Write meetingtime syntax as: dayname, hh:mm [23:28:38] Otherwise it will add i.e. Meeting on Day 2 at 19:30:00 [23:29:46] But as such, I can easily write a couple of articles in an hour, with the same structure, by using substitutable templates [23:30:58] Last time I asked about the superlative in Latin he couldn't give me the right answer [23:31:09] 😒 [23:31:21] Then it's not put in there yet [23:31:57] Same as asking about a current seoson in Formula 1, or football, or Tennis. [23:32:11] It's data goes back to september 2022 I believe [23:32:35] Latin hasn't been updated since Century 😂 [23:32:41] LOL [23:33:52] It's funny though, I tried once to ask several questions in different languages. Each answer to that question came in the same language it was asked 😄 [23:34:17] Oh I haven't tried asking it something in Latin [23:40:02] [1/4] ```J'ai besoin de te ask a question, comment translate-on" ich möchte dich essen" en english``` [23:40:02] [2/4] Answers me in french but [23:40:02] [3/4] ```J'ai besoin de te ask a question, comment translate-on" ich möchte dich essen" in english``` [23:40:03] [4/4] Answers me in English [23:41:47] lol