Jump to content

Wikipedia:Articles for deletion/15.ai (3rd nomination)

Add links
From Wikipedia, the free encyclopedia

15.ai

[edit]
AfDs for this article:

[Hide this box] New to Articles for deletion (AfD)? Read these primers!

15.ai (edit | talk | history | protect | delete | links | watch | logs | views) – (View log | edits since nomination)
(Find sources: Google (books · news · scholar · free images · WP refs) · FENS · JSTOR · TWL)

Procedural nomination. This article was recently deleted by consensus. A deletion review process was filed; the filer meanwhile created a new fresh draft using different sources which was passed by a reviewer at AfC. I have closed the DRV, so that the new draft may be evaluated on its own merits. By this nomination, I make no judgements on the outcome. BusterD (talk) 17:02, 20 December 2024 (UTC)[reply]

Keep. This AfD was procedurally created because an old version of the article was deleted as a result of a previous AfD, which was largely due to the misbehaving of new Wikipedia editors who participated in sockpuppetry and canvassing, and as such the new research and citations I found after relisting were not taken into consideration. In the previous AfD, it’s worth noting that after my new sources were posted, no new delete votes were submitted. As per advice given to me by User:Liz and other Wikipedia administrators, I was encouraged to create a new draft of the article from scratch and submit it to AfC, which I did, and the newly written article was accepted via AfC within several hours by an impartial third party who was not previously aware of the AfD and DRV. I followed all of the correct processes as was recommended to me by experienced editors. User:Barkeep49, the closer of the previous AfD, can attest to the fact that I did not have a copy of the old version of the article when I wrote this new version from scratch.
According to archived discussions on the Teahouse, a good rule of thumb is three independent and reliable sources that demonstrate significant coverage to establish WP:GNG. Here are the major reliable sources that provide significant coverage for the subject:
  • [NEW] United Daily News [1]. Reliable as one of the largest and oldest-running newspapers in Taiwan. Listed as one of the three major Chinese-language newspapers in List of newspapers in Taiwan. Significant coverage includes an overview of the technology behind 15.ai, particularly noting its ease of use and limited data, and also discusses how 15.ai works, its features, and the viral videos that have spawned using 15.ai. Over 400 (approximate since the article is written in Chinese) words of coverage.
  • Den Fami Nico Gamer [2]. Reliable as listed in WP:VG/RS. Significant coverage includes an overview of the DeepMoji technology used for emotiveness, applications of the voices not restricted to viral videos, and how to use it. Over 400 (approximate since the article is written in Japanese) words of coverage.
  • AUTOMATON [3]. While not listed in WP:VG/RS, AUTOMATON is one of the largest and reputable gaming news outlets in Japan, and has been used in multiple GA's like Only Up!, Visions of Mana, and Sprigatito, Floragato, and Meowscarada. Significant coverage includes DeepMoji, a list of characters available on the application, examples of video content users have created with the platform, an overview of the pronunciation capabilities of the model, as well as a mention of how to use ARPAbet strings. Almost 800 (approximate since the article is written in Japanese) words of coverage.
  • [NEW] Rionaldi Chandraseta [4]. While the article itself is written on Medium (which is not considered reliable), Medium is only being used as a vessel to host the article itself (similar to how Google Docs can be used to host an article), which is part of a very popular newsletter called Towards Data Science, which has almost 800K followers on social media. Following alone means nothing in determining the reliability of a source, but Rionaldi Chandraseta, the author of the article, is an IEEE-published machine learning specialist who has published papers that are listed on Google Scholar [5]. The newsletter has a solid editorial board [6] that consists of multiple masters and PhD's in machine learning and computer science. Over 1,000 words of English-language coverage detailing every facet of 15.ai, from its capabilities to its underlying research.
  • [NEW] Yongqiang Li [7]. Since the article is locked to foreigners without an account, I asked a friend to translate this for me. The article goes into great detail about the technology behind 15.ai and talks about its features, its future, and potential problems. The author is a professor at the Harbin Institute of Technology and has multiple publications listed on Google Scholar [8] and ResearchGate [9].
  • Eurogamer [10]. Reliable as listed in WP:VG/RS. While the main focus of the article isn't 15.ai, it goes into detail the controversy and Twitter exchange that happened when Voiceverse NFT misappropriated 15.ai's work. From However, in now-deleted tweets, Voiceverse was found to have boasted about using its tech for the voice of a cartoon character - which was in fact created using 15.ai, a popular non-commercial text-to-speech service. to "Hey @fifteenai we are extremely sorry about this," Voiceverse NFT wrote. "The voice was indeed taken from your platform, which our marketing team used without giving proper credit. Chubbiverse team has no knowledge of this. We will make sure this never happens again.", this is about 300 words of coverage.
  • Stevivor [11]. After doing more research, I found that Steven Wright, the author of this article, also writes for Inverse, a solid and well-known technology and gaming publication. In addition, Stevivor is reliable and independent, and it is the most-read independent gaming news network in the Oceanic region.
  • Kotaku [12]. While Kotaku is in WP:VG/RS, it also states News posts from Kotaku between 2010 and 2022 are considered reliable, (the article is from 2021, so it meets this criteria) but also states although editors are cautioned of blog/geeky posts that have little news or reporting significance. It's still debated whether an article from the "Odds and Ends" category is considered "News", and the entry in WP:VG/RS says articles should be evaluated on a case-by-case basis. There is no clear consensus to this, but the description for the "Odds and Ends" category is "Odds and Ends - Gaming Reviews, News, Tips and More.". The argument in the previous AfD was that this article did not meet reliability, although it met independence and significant coverage. I personally believe that this article is reliable.
  • Game Informer, PC Gamer, and Rock, Paper, Shotgun [13] [14] [15] All three of these sources are found under WP:VG/RS, but there has been a debate whether these three met significant coverage. While they all pass WP:100WORDS, it is not a Wikipedia policy and their significant coverage can be debated.
  • NME [16]. WP:RS notes NME is reliable in its expertise, and it has been debated whether gaming is one of their areas of expertise. The Wikipedia article for NME states that this is so, and gaming is listed as one of NME's header sections, but there has been debate whether NME's expertise extends outside of music. Similar coverage to Eurogamer, but with fewer words, but still above the threshhold for WP:100WORDS (which, again, is not Wikipedia policy).
  • Andrew Ng [17]. The author, Andrew Ng, is one of the most famous and influential artificial intelligence researchers in the world, with a healthy Google Scholar profile [18] and was included in the Time 100 Most Influential People in AI list in 2023. While 15.ai is mentioned as a blurb and likely does not meet significant coverage, it shows that the subject wasn't a mere curiosity and was under the radar for a large number of prominent figures in AI while the service was active.
However, some of these have been contested in some form, the arguments for which I personally disagreed with. Think of these sources as you will. I still haven't gone over many of the the sources that are used in the newly written page, but I will continue to do my research and update this.
Yes, I'm aware that this is a contentious article that was submitted to AfD. But the AfD was closed largely due to the misbehaving of new Wikipedia editors, who are likely to be children, which is not surprising given the popularity of the application among younger people. I'm committed to doing this subject justice, and I argue that this subject not only meets the bare minimum of notability, but meets it well-within question. GregariousMadness (talk to me!) 17:08, 20 December 2024 (UTC)[reply]
  • Comment I thought the AfC submission was okay but should have been with a different AfC reviewer. Someone who was not involved with these past discussions. This AfD is overkill. – The Grid (talk) 17:28, 20 December 2024 (UTC)[reply]
    @The Grid For clarity, as the AFC reviewer in question, I was entirely uninvolved with the initial 15.ai discussions and didn't know of their existence at all. I was only notified of them after I'd already passed the article. I've been keeping an eye on these newer discussions that followed my review, but for the case of the initial review, I was an entirely uninvolved third party. Magneton Considerer: Pokelego999 (Talk) (Contribs) 18:16, 20 December 2024 (UTC)[reply]
    I have no complaint with User:Pokelego999's review of the article. I've commented at AfC talk that comparing new submissions to recently deleted versions is probably a wise precaution. We should not generally be passing drafts which are currently at deletion review, IMHO. Is your namespace currently the subject of a deletion process? That seems like a quick fail. BusterD (talk) 18:33, 20 December 2024 (UTC)[reply]
  • Keep The references in the article clearly show enough coverage to meet GNG. QuicoleJR (talk) 18:28, 20 December 2024 (UTC)[reply]
  • Keep. Passes WP:NWEB and WP:GNG.—Alalch E. 19:04, 20 December 2024 (UTC)[reply]
  • Comment - the source list above states there are three new sources not discussed in the last, and recently closed deletion discussion. This turns out to be incorrect. Let's look at them:
    1. United Daily News [19] - this was, in fact, considered at the last AfD. I commented on it on 11 December.
    2. Towards data science [20] was also introduced at the last AfD. I commented on that one at the same time.
    3. Yongqiang Li [21] is indeed locked and has the world's most annoying CAPTCHA! I cannot review it.
    So we are here again, but we are, in fact, looking at essentially the same sources as the AfD that just closed as delete, and for which the deletion was endorsed. We still do not have information that establishes the notability of the subject or describes it beyond what the website actually did. Sirfurboy🏄 (talk) 20:31, 20 December 2024 (UTC)[reply]
    but we are, in fact, looking at essentially the same sources as the AfD that just closed as delete, and for which the deletion was endorsed. I'm sorry, but this is false. I posted those new sources on December 9 after the 2nd AfD was relisted for discussion, and all of the delete votes in that AfD were submitted before these new sources were found. You can check the timestamps in the AfD to verify that the deletion was endorsed because they were predicated on the article not having the new sources that I spent all weekend finding. To reiterate, this is what happened: the old version of the article was nominated for deletion (without the new sources), then six delete votes came in assessing that the old version of the article did not sufficiently demonstrate notability, then the discussion was relisted, then I spent an entire weekend researching new sources on different language websites and posted a lengthy explanation of the sources I found, and then not a single person voted delete after that, and then due to the rampant socking and canvassing that had happened before the relisting and before I had posted the new sources, the AfD was closed as a delete. The new sources were never considered by the people who voted delete other than yourself. GregariousMadness (talk to me!) 20:41, 20 December 2024 (UTC)[reply]
    Yes, well that is a different definition of new to the one I was using. Sirfurboy🏄 (talk) 20:44, 20 December 2024 (UTC)[reply]
    What I meant by new was new compared to the reason why the old article was voted to be deleted in the first place. The new sources I found in between the 2nd AfD and DRV was never given a thorough examination, so I included those sources with the "new" sources I found in between the DRV and the 3rd AfD. It demonstrates that this new version of the article is substantially different from the one that was justifiably deleted, and I believe that definition of "new" is the more relevant one in a discussion where editors discuss whether this new article should be kept or not. GregariousMadness (talk to me!) 20:47, 20 December 2024 (UTC)[reply]
    3 is free and not locked, but requires logging in/free registration and a competent translator. It seems to be the same kind of source as the TowardsDataScience article. Aaron Liu (talk) 01:16, 21 December 2024 (UTC)[reply]
    Hmm, the CAPTCHA is gone today. Yesterday it had a CAPTCHA which presented 8 symbols to be placed in order, but the symbols appeared to be totally random. Like a rabbit, a boat, a mountain... Anyway, the article does not provide significant coverage beyond what the website did. As for all the others. Does not demonstrate notability. No idea about reliability. Sirfurboy🏄 (talk) 11:22, 21 December 2024 (UTC)[reply]
  • Delete: Still not notable, coverage was and is limited to "look at this cool app" articles. I haven't found anything substantial, since this was nominated again. Oaktree b (talk) 21:20, 20 December 2024 (UTC)[reply]
  • Keep: I'm against deleting this article, because the site was a big deal when it came out and was groundbreaking at the time. Reading this article is an easy way to learn about the history of the site. I know that's not related to the discussion about sources, but I thought I'd throw my two cents in anyway, just in case. Dogman15 (talk) 21:41, 20 December 2024 (UTC)[reply]
  • Keep per the sources listed above, which establish notability per GNG. Frank Anchor 22:58, 20 December 2024 (UTC)[reply]
  • Leaning Keep, as I agree with the arguments of Madness more than Sirfur's. Aaron Liu (talk) 01:17, 21 December 2024 (UTC)[reply]
    To be clear, I have not expressed any arguments here yet. I have merely pointed out that we are looking at exactly the same set of sources as in the AfD that closed as delete on 17 December. 4 days ago. Sirfurboy🏄 (talk) 09:09, 21 December 2024 (UTC)[reply]
    Sorry, I meant the arguments you argued last time, which seem representative of quite a bit of the opposition from last time. (I could be wrong; I only took a cursory glance at the other opposition.) Aaron Liu (talk) 01:14, 23 December 2024 (UTC)[reply]
  • Strongest delete - not substantially different from the version which was deleted via consensus (a consensus which was well on its way to being confirmed at DRV) 72 hours previously. Consensus needs to be respected and "I submitted new sources at the last AfD but no one looked at them" is not a reason not to. The new sources which aren't on WP:MEDIUM don't add anything new (and I refuse to consider anything which is). I concur with Oaktree b's analysis. ―"Ghost of Dan Gurney" (hihi) 01:59, 21 December 2024 (UTC)[reply]
    WP:MEDIUM says it is considered generally unreliable and should be avoided unless the author is a subject-matter expert, which the author is. GregariousMadness (talk to me!) 02:05, 21 December 2024 (UTC)[reply]
    Four papers on Google Scholar, with only one that has a total of four cites seems low to me, but I'm open to being wrong on this. ―"Ghost of Dan Gurney" (hihi) 02:26, 21 December 2024 (UTC)[reply]
    Well… I'm a 4th year PhD student researching topological deep learning, and I would consider myself a subject matter expert on differential topology and deep learning even though I only have a couple of papers with a total of zero citations… GregariousMadness (talk to me!) 15:24, 21 December 2024 (UTC)[reply]
    Comment. After doing more research, I've found that articles from Towards Data Science are also listed on Google Scholar. This article [22], for example, has 172 cites on Google Scholar, and this one [23], has over 1027(!) cites on Google Scholar, which indicates that the newsletter is a legitimate publication with other papers on Google Scholar frequently citing them. GregariousMadness (talk to me!) 06:06, 22 December 2024 (UTC)[reply]
    I believe the last close was well within the closer's discretion. I also believe that this discretion is only so high due to the amount of disruption, and I believe that we should have some discussion without any disruption.
    In the conversation with Sirfurboy, we agreed that precedent says such reviews should count towards notability, but we also reached a point of ideologically disagreeing whether they should. Note that he did not !vote on those "should" grounds, and that's not what I intend to say.
    I don't see why such "look at this cool app" articles fail non-trivial coverage or don't count towards notability. Now, are we about to argue the "should"? Aaron Liu (talk) 02:24, 21 December 2024 (UTC)[reply]
    Content has primacy over process. Tidy process can't trump valid encyclopedic content. —Alalch E. 09:32, 21 December 2024 (UTC)[reply]
  • Keep Weak DeleteDraftify, as alternative to deletion, per discovering that Business News Ledger appears to be an unreliable source. I think the new version meets WP:GNG and WP:NWEB. It might need improvement, but I think GregariousMadness did a good job. Moreover, I think the strongest source added by far is the [24] which is a publication that talks about the history and impact of 15.ai. Honestly, I don't know why GregariousMadness isn't leading the charge with the Business News Ledger piece instead of the sources they listed above which were mostly about a controversy involving 15.ai(person) and said little about the webapp or otherwise represent trivial coverage/ExpertSPS's. The Business News Ledger article is more substantial and does more than just cover the quirky fun app and its voices. Unless the Business News Ledger is determined to be an unreliable source, it seems like the most substantial source available that pushes the article over the line for me. Anywho, as the person who nominated the article last time I wanted to chime in. I also don't think the article should be punished because GregariousMadness jumped the gun before the Deletion Review finished. Cheers! --Brocade River Poems (She/They) 03:28, 21 December 2024 (UTC)[reply]
    Are we sure the Business News Ledger is reliable though? I have never heard of them and a Google search is bringing up nothing on the source themselves. SportingFlyer T·C 05:33, 21 December 2024 (UTC)[reply]
    I didn't include it in my initial post because I wasn't entirely sure if it was reliable (and I wanted to make the absolute strongest case for the article by only including sources that were 100% reliable), but I did see that the outlet was listed under Google News (that's how I found the article, a search of "15.ai" and "voice"). But judging purely from the contents of the article, everything seems to check out as it essentially retells the story in the developer's Twitter article along with other verifiable details on the subject that weren't present in the Twitter article. GregariousMadness (talk to me!) 05:40, 21 December 2024 (UTC)[reply]
    Business News Ledger are a small outfit registered in Los Angeles, but appear to employ a small number of journalists with editorial oversight from Travon Marner. The piece is written by Esperanza Squire, who has written many articles on the site. It is not exactly the New York Times, but I am not seeing any reason why it is unreliable. Agree that this is the best source we should be focussing on. Sirfurboy🏄 (talk) 09:28, 21 December 2024 (UTC)[reply]
    Wellllllll, I feel like a bit of a silly-billy now. The source did seem substantial, but I didn't do a deep dive into it or anything but at a glance, it seemed legitimate if not small. I checked RSN and there hasn't been any discussion about it before, so that wasn't helpful. Ultimately it seemed to have editors, the person who wrote the article says they're a journalist, there's different articles on the site by different authors and the site has been around since 2018, during which time it said it focused on Canadian business news, but has since changed their mission statement to say global business news. To be honest, I am in the process of actively retiring from Wikipedia so I wasn't planning on diving deep into this but now I'm down a rabbit hole. So, my first problem is that none of the authors seem to exist. Several of them claim to have written for this publication, that publication, or another publication but I cannot find them anywhere on said publications. Moreover, in 2018 we have this [25] article which of the author reads Fay Kadri graduated from Columbia University in 2005, but here [26] same picture, different name, similar story Rachel Knox graduated from Columbia University in 2005. Rachel grew up in Canada but moved to the US and then there's this twitter account [27] same picture, yet another different name? Same story here, [28] fake author using the picture of someone else [29]. This[30] does also say The information is provided by Business News Ledger and while we endeavor to keep the information up to date and correct, we make no representations or warranties of any kind, express or implied, about the completeness, accuracy, reliability, suitability or availability with respect to the website or the information, products, services, or related graphics contained on the website for any purpose. So, my apologies. It would seem that while the article seemed good, the site very likely is an unreliable source. That's on me for not digging deeper into the source. Almost all of the author profiles I mentioned as fake also use the same While studying journalism at [x], with [x] just being an American city as if a university name is supposed to go there. Overall, given that they are definitely stealing pictures and making up journalists I think Business News Ledger is definitely unreliable after doing a deeper dive. Brocade River Poems (She/They) 10:51, 21 December 2024 (UTC)[reply]
    I agree with this assessment. Alalch has now removed content only sourced to BNL and 𝕏. Aaron Liu (talk) 15:07, 21 December 2024 (UTC)[reply]
    I'll keep this brief to explain my weak delete without further editing my own post:
    • Zhihu is WP:USERGEN, the author of the article might qualify for WP:EXPERTSPS, which doesn't confer notability. 知乎，中文互联网高质量的问答社区和创作者聚集的原创内容平台 Roughly translated, Zhihu, a Chinese language online Q&A community and a platform for original content creators to gather.
    • Andrew Ng The Batch, WP:SPS, does not confer notability.
    • Per WP:MEDIUM, Towards data science can be considered WP:EXPERTSPS, does not confer notability.
    • United Daily News cites the GameInformer article at the bottom of the page.
    • Eurogamer, NME, et al. all discuss 15.ai(Person)'s response to Voiceverse and provide no actual content about 15.ai
    • Kotaku is only WP:VG/RS for News, Odds and Ends is the bloggy/geeky content editors are cautioned against. Particularly described as These stories defy categorization, hopefully in a good way.[31] Kotaku as whole reads Gaming Reviews, News, Tips, and More so that isn't a specific notice on Odds and Ends
    Beyond that, my opinion on the other sources remains unchanged from the last AfD in that I do not believe they meet the requirements of WP:NOTINTERNET. Brocade River Poems (She/They) 11:28, 21 December 2024 (UTC)[reply]
    Let's say that the sourcing is a bit lacking according to certain norms. What's the resulting problem with the content, manifesting as an inability to write content which complies with the core content policies? —Alalch E. 12:38, 21 December 2024 (UTC)[reply]
    My problem with the sources provided is, has been, and remains WP:NOTINTERNET which states Wikipedia articles should not exist only to describe the nature, appearance or services a website offers, but should also describe the site in an encyclopedic manner, offering detail on a website's achievements, impact or historical significance. The reliable sources only describe the nature, appearance, or services the website offers. None of the reliable sources provide any detail on the website's achievements, impact, or historical significance. Since Wikipedia:Core_content_policies, despite very confusingly being called "Core content policies", says This page is not one of Wikipedia's policies or guidelines and is listed as an explanatory essay about NPOV, while WP:NOTINTERNET is listed as policy. From my understanding, then, the failure to meet the requirements laid out on WP:NWEB and WP:NOTINTERNET means that the article should be deleted. Likewise, WP:NWEB stipulates Wikipedia should not have a separate article on any web content... for which, despite meeting the rules of thumb described above, editors ultimately cannot locate independent sources that provide in-depth information about the web content [32] emphasis added, which states that even if the webcontent does manage to meet the guidelines for notability, it should not have an article if you cannot locate independent sources that provide in-depth information about the website. The sources in question all provide more or less the same surface level information about how 15.ai was made by some person at MIT and that it uses these specific technologies to allow the creation of [x] voice. That does not, to me, represent in-depth coverage. Brocade River Poems (She/They) 00:11, 22 December 2024 (UTC)[reply]
    Thanks for the reply, I really appreciate it. About achievements and legacy (paraphrasing from the article): In the history of deep learning speech synthesis, this is the the website that offered "voice cloning for the masses". The website is the first example of a website using this technology that reached such popularity. It popularized audio deepfakes in memes and content creation. It interpreted emojis in prompts to generate emotional prosody and "the amount of control that [users could] use to tweak how words and phrases are pronounced [was] pretty deep". As a trending and innovative thing, it was illegitimately exploited and was copied after it became defunct. It had a big reception in the My Little Pony fandom and resulted in many viral videos being created using it. When I mentioned the core content policies I meant the policies Neutral point of view, Verifiability, and No original research (not the page "Wikipedia:Core content policies" as such). NOTINTERNET is also a policy, but because the article, unlike what you say, does in fact establish the website's achievements, impact, and historical significance, NOTINTERNET does not indicate deletion, and the content is suitable encyclopedic content under the core content policies, so there is not need to delete the page. This is a serviceable article, it is not out of place in an encyclopedia and clearly has some educational value. —Alalch E. 11:13, 22 December 2024 (UTC)[reply]
    At the risk of being accused of bludgeoning, I will respond to this one more time, and I will note that unless a substantial source is located, I won't change my mind about Draftify. I will explain my reasoning, though.
    In the history of deep learning speech synthesis, this is the the website that offered "voice cloning for the masses". The website is the first example of a website using this technology that reached such popularity.
    My stance remains until a reliable source says that, it cannot be said on Wikipedia. None of the sources in the entirety of the background section of 15.ai presently, say this about 15.ai Examining the Background section that actually uses sources which mention 15.ai, the most substantial one seems to be Towards Data Science, again, a self-published source. [33] and is ultimately more survey of what 15.ai does, and offers no real information beyond describing what the website does and how it does it.
    • he/she claims that this new technique beats SV2TTS in data efficiency and naturalness
    • The one part I could not agree on is the “faster-than-real-time” claim, because the voice generation is quite slow and not real-time. However, this might be caused by the queue of requests sent by other users.
    • Compared to other TTS technique, 15.ai is able to mimic a character with very little data
    • I could not verify how good is the result with only 15 seconds of data, but the voice of Portal’s Sentry Turret only has ~100 seconds of data and it surprisingly sounds pretty good.My hypothesis is the model can benefit from the training data of other characters too, which explains why a character with very little data can still produce quality result.
    • Another interesting thing from 15.ai is how it uses DeepMoji to predict the emotion of a sentence. Currently, we could not manually set the emotion of the voice, as the only available choice for emotion is “Contextual” which uses DeepMoji.
    • Compared to what I remembered when creating voices with Lyrebird, the quality of voices done by 15.ai is miles ahead of it.
    • Just like DeepFake, this technology has a potential to become a dangerous tool for creating fake speech, but I believe that this would open a lot of possibilities such as creating your own voice assistant to replace Google Assistant’s, Siri’s or Alexa’s voice.
    None of this information differs from what the other sources provide. It also doesn't verify anything about the section that reads A significant challenge... high-quality voice synthesis from limited data, such as transfer learning and few-shot learning techniques which the source is attached to, with the only other source being the creator's twitter account, which is not independent of the subject. Notability is conferred by the sources, and while popularity/fame might enhance the acceptability of a subject, the fact that memes that blew up and went viral were created using 15.ai does not inherently make it notable. Per WP:NRV, the evidence must show the topic has gained significant independent coverage or recognition, and that this was not a mere short-term interest
    As I noted last AfD, the dates:
    • April 1, 2020 - The Batch (Trivial Mention in a long newsletter, SPS, and possibly not independent coverage of 15.ai since Andrew Ng mentions an e-mail with the creator.)
    • January 17, 2021 - Kotaku (Odds & Ends is Bloggy/Geeky Content, cautioned against by VG/RS)
    • January 18, 2021 - Den Fami Nico, GameInformer, RockPaperShotgun, GamerSky (Brief Summaries of site functionality/service)
    • January 19, 2021 - Automaton, PC Gamer, Byteside (Brief Summaries of site functionality/service)
    • January 20, 2021 - United Daily News (Brief Summaries of site functionality/service, sourced from GameInformer)
    • January 21, 2021 - Towards Data Science (SPS)
    • January 22, 2021 - Zhihu (SPS)
    • January 17, 2022 - Eurogamer (Trivial Mention)
    • January 18, 2022 - Stevivor, NME (Trivial Mentions)
    The coverage spans roughly 4 to 5 days, depending on international time differences. The coverage outside of that time frame is either (a) Not independent of the subject, (b) focused more on the product creator's response rather than actually discussing 15.ai significantly, or (c) extremely trivial mentions. This source [34] currently used mentions 15.ai in a single sentence which says it launched in 2020, This source [35] has one sentence that says 15.ai is a notable example of voice deepfakes. The article uses "Equestria Daily" as a source, but Equestria Daily is only considered reliable per RSN in terms of content such as interviews with the production staff of My Little Pony[36], so doesn't count toward notability (and probably shouldn't be used at all). RSN seemed of the opinion that ElevenLabs, Speechify, Play.ht, etc. are likely not reliable sources[37]. Play.ht was noted to be particularly bad at the RSN.
    So from my understanding, NOTINTERNET does indicate deletion because the section of the article you mention fails verification from independent sources. In terms of impact, a video titled "Among Us Struggles" that went viral in 2020 does not demonstrate 15.ai was the cause of that if there isn't a source saying that 15.ai caused these things to be popular. Consider that "Among Us Struggles" was released in September 2020 and accrued 5,000,000+ views, and then look at Google Trends [38]. Among Us itself was extremely popular at the time[39] and the video's popularity likely has nothing to do with 15.ai. Likewise, by February 2021, The RED Bread Bank video mentioned had 150k Views [40] from a channel with over 100k subscribers. The 15.ai "The Heavy is Dead" [41] has at this moment 235k views, the original "Heavy is Dead" has over 15,000,000 [42], so it is likely that traffic simply got driven to the 15.ai version from the large amount of people who saw the original. My point is, there are no sources provided which can be used to verify the significance of 15.ai, and until some can be found, I don't think it should leave draftspace.
    Happily agree to disagree. Brocade River Poems (She/They) 13:08, 22 December 2024 (UTC)[reply]
    I just wanted to quickly chime in on the claim None of the sources in the entirety of the background section of 15.ai presently, say this about 15.ai, the Inverse [43] and Analytics India Magazine [44] that are used in the article do support the claim that In the history of deep learning speech synthesis, this is the the website that offered "voice cloning for the masses". The website is the first example of a website using this technology that reached such popularity.. The subject of those articles aren't 15.ai (rather, voice cloning as a whole), but they both agree that 15.ai came first in achieving mass popularity in the field. GregariousMadness (talk to me!) 16:57, 22 December 2024 (UTC)[reply]
    Going to keep this brief. Neither of those sources are cited in the specific Background section I was referring to, they're used in the lead and "Reception and Legacy". Of those two sources, though, Analytics India Magazine says of 15.ai While AI voice memes have been around in some form since ‘15.ai’ launched in 2020, and Inverse says AI voice tools used to create "audio deepfakes" have existed for years in one form or another, with 15.ai being a notable example. Neither of those sentences from these two sources verifies that claim without WP:OR, neither of them says that it's the first website that reached popularity, they just say they've been around since 2020 with 15.ai's launch and that audio deepfakes have existed for years and that 15.ai is a notable example. At best with those two sources we could, for example, write something like Launched in 2020, 15.ai was an early pioneer of audio deepfakes because that's what the sources can verify without any OR. Brocade River Poems (She/They) 00:57, 23 December 2024 (UTC)[reply]
    That can be fixed editorially, it doesn't require deleting the article: That content now reads: 15.ai was an early pioneer of audio deepfakes, leading to the emergence of AI speech synthesis-based memes. Its influence has been noted in the years after it became defunct, and since then, several commercial alternatives emerged, such as ElevenLabs and Speechify. An additional source is cited, with tighter text-source integrity. —Alalch E. 03:14, 23 December 2024 (UTC)[reply]
    I disagree that Eurogamer and Strevivor are just trivial mentions. An article doesn't have to feature 15.ai as the subject; it just has to address it directly and in detail. While NME did not (hence NME is indeed just trivial mentions), Eurogamer and Strevivor both dedicate over a paragraph to detail 15.ai itself, therefore they count, and the article has Sustained coverage. Aaron Liu (talk) 23:54, 23 December 2024 (UTC)[reply]
    By all reasons Notability exists (to make sure there's enough verifiable information for a non-stub), I don't see why ExpertSPSs would not count towards notability, and I cannot find any Policy/Guideline/Essay that explains why. Aaron Liu (talk) 15:04, 21 December 2024 (UTC)[reply]
    My statement they don't confer notability is based on the discussion about Andrew Ng's The Batch at the RSN where it was said Treat as a SPS, which means that it doesn't count towards notability [45] and also Wikipedia:Identifying_and_using_self-published_works#Self-published_sources_for_notability which states Self-published sources are seldom useful for demonstrating the notability of any subject. Cheers! Brocade River Poems (She/They) 00:17, 22 December 2024 (UTC)[reply]
    That's an interesting and extremely questionable section. I doubt whether it's supported by consensus. Aaron Liu (talk) 15:16, 22 December 2024 (UTC)[reply]
    I would contend that WP:GNG stipulates "Reliable" means that sources need editorial integrity to allow verifiable evaluation of notability, most WP:SPS, including EXPERTSPS, lack editorial oversight. I am also unsure as to whether the author of the Towards Data Science article qualifies as an ExpertSPS which requires it be an established subject-matter expert whose work in the relevant field has been published. Chandraseta has exactly one publication per Google Scholar, cited 4 times, which has nothing to do with voice cloning/audiodeepfakes, and is a 5 page conference paper about scraping articles from the Indonesian Wikipedia[46]. I'm not sure a single conference paper is enough to qualify them as an established subject-matter expert in all things AI, which means per WP:MEDIUM the Towards Data Science might not be usable. Brocade River Poems (She/They) 01:44, 23 December 2024 (UTC)[reply]
    I doubt that's what it means. GNG immediately follows that by ", per the reliable source guideline." I suspect "editorial integrity" to be just another word for "ethos". Notability's purpose is to ensure Wikipedia's articles have enough verifiable information to not be stubs among satisfying other content policies. ExpertSPS articles do not fall short of that, thus they should count towards notability, no matter what the letter of the law says. (The encyclopedic value part of inclusion criteria is already being debated above.)
    I didn't realize TDS was under contention. Gregarious has found that it has has had widely-cited articles, and the publication lists two editors. As for Chandraseta, an AI summarization model is the centerpiece of the paper about "scraping articles". He has a lot of AI stuff on his GitHub profile, including a model to recognize faces and detect their emotions. (He also has another paper analyzing the usage of graphs in video games, but I suspect it to be a dissertation.) Aaron Liu (talk) 23:31, 23 December 2024 (UTC)[reply]
  • Keep: I still believe this website meets WP:GNG. Schützenpanzer (Talk) 03:01, 22 December 2024 (UTC)[reply]
  • Draftify - per Brocade River Poems. I remain unconvinced about the sourcing. The sourcing is the same as was discussed in the AfD that closed last week as delete, and for which the deletion was endorsed at DRV. I concur with the above source assessment, and per Oaktree b. However I do believe there is a distinct possibility that notability for this could be established. The Business News Ledger article, whilst clearly an unreliable source, was the kind of article one might expect will emerge in a reliable source, and if and when that happens, we will have something permanently notable to say about this site. The deleted draft was hurried back into mainspace through a couple of mistaken circumstances, whereas if this had been incubated longer, I expect a return to mainspace down the line would be perfectly okay. Draft space will allow this to develop unhindered by all this process, and I would be happy to assist in the drafting. Sirfurboy🏄 (talk) 07:38, 22 December 2024 (UTC)[reply]
  • Comment. Hi everyone! After another night of research, I found even more sources contributing significant coverage to 15.ai. As discussed in my talk page, finding sources for 15.ai is very very difficult because search engines don't search for punctuation, so search results for "15.ai" get inundated by "top 15 AI" slop articles that I had to sift through. Some of these sources I had to go all the way to page 37 of Google search results. Here they are.
  • [NEW] Yahoo! News Taiwan [47]. Reliable as listed in WP:RSPSS. WP:RSPSS cautions Take care with syndicated content, which varies from highly reliable sources to very unreliable sources. Syndicated content should be evaluated as you would evaluate the original source. Syndicated content will have the original source's name and/or logo at the top., but the article is evidently not syndicated, as a quick translation of the article shows; there is no original source name or logo at the top, and the author is a legitimate author of Yahoo! Games.
  • [NEW] Arkade [48]. One of the largest and most reputable gaming news outlets in Brazil with almost 200K followers on social media. It is a commonly used source on Portguese Wikipedia [49] and has been used for Brazilian subjects on English Wikipedia.
  • [NEW] Anime Superhero [50]. An animation/video games/pop culture news outlet with a fairly popular forum attached to it. Significant coverage, though reliability is uncertain.
  • [NEW] EquestriaCN [51]. A newsletter for Chinese My Little Pony news. Notably, this coverage is in October 2021, which is separate from the other significant coverage of the website listed so far. Significant coverage includes detailed history of the site, its re-emergence, a description of the technology behind 15.ai (including FlowGAN denoising, which is something that previous sources hadn't mentioned, and DeepMoji and ARPAbet strings), and its terms of service.
I was a little hesitant to post this as a standalone comment as it could be considered WP:BLUDGEONING which I want to avoid, so if it would be better edited as part of the original comment, I can do that instead. GregariousMadness (talk to me!) 14:45, 22 December 2024 (UTC)[reply]
No, keep it here, or the history and context of other replies will get mixed up. Simply adding sources won’t be considered bludgeoning. Sirfurboy🏄 (talk) 14:57, 22 December 2024 (UTC)[reply]
Okay, got it! I found even more sources so I'll edit this comment as I keep finding more. GregariousMadness (talk to me!) 15:17, 22 December 2024 (UTC)[reply]
  • Yahoo News: 20 January 2021
  • Arkdale: 19 January 2021
  • Anime Superhero: 18 January 2021
Three more sources that were part of the 5 day flurry of news articles about a new site. See the comments above from BRP, and per WP:NRV. Short term interest - a new site. Tells you what it does. These add nothing, I am afraid.
  • EquestriaCN: discussed above. It is not reliable for this, and in any case it is also just a news report that the site had an update.
Sirfurboy🏄 (talk) 18:35, 22 December 2024 (UTC)[reply]
Thank you for the analysis, but I feel like we're getting too hung up on when these articles came out instead of what they actually say. Yes, there was a burst of coverage in January 2021—but that's because something genuinely groundbreaking happened! The way United Daily News, Den Fami Nico Gamer, and AUTOMATON covered this wasn't your typical "here's a cool new website" reporting. We're talking 400–800 words each diving into the technical details, explaining the innovations in how it handled voice synthesis, breaking down the DeepMoji implementation, explaining how ARPAbet works, what other approaches have been tried before 15.ai, etc. That's serious coverage that demonstrates real significance. AND multiple machine learning professors and specialists came out talking about 15.ai and its technology. It wasn't just some "flash in the pan", it was an application of voice cloning that despite most people using it to make silly memes and videos on the Internet, serious machine learning professors like Andrew Ng and Yongqiang Li took it very seriously.
Also let's not just ignore the fact that 15.ai kept coming up in 2022 with the Voiceverse situation and wasn't just dismissed as just a "cool app". When Eurogamer and other major outlets are covering another company trying to misappropriate your tech, to me, that's a pretty clear sign you've made something significant! And just the fact that a multi-million dollar partnership was still trying to steal the work from a free service that was made by a single person, that's a pretty commendable accomplishment.
Major publications across multiple countries (USA, Canada, Japan, Korea, China, Taiwan, Mexico, Spain, Brazil, and those are just the sources that I've found, I'm sure I'm still missing a lot of them but I've pulled near-all nighters three times now working on sourcing this article) didn't dedicate that kind of coverage for nothing. The timing of the coverage matters way less than what that coverage actually says (I did a whole analysis of the contents of the articles in the 2nd AfD if anyone is interested in that), and I think the sheer amount of coverage does dictate that 15.ai was a notable innovation that changed how people approached voice synthesis in the early stages of AI development.
(Also, I have to point out that EquestriaCN isn't the same as Equestria Daily.) GregariousMadness (talk to me!) 18:53, 22 December 2024 (UTC)[reply]
all-nighters are bad for your health and sanity y'know
I wholeheartedly agree with this. Aaron Liu (talk) 18:56, 22 December 2024 (UTC)[reply]
all-nighters are bad for your health and sanity y'know I know, but when the old article got deleted, I was committed to doing the subject justice. If no one was able to find sources for 15.ai, I told myself I would do it myself, and I want to show the good people of Wikipedia that I'm serious about this. I wholeheartedly agree with this. Thank you. I've been trying to explain my thought process better but I find it much easier to write it in the form of an article than in a discussion, but I'm trying my best! GregariousMadness (talk to me!) 18:58, 22 December 2024 (UTC)[reply]
Trying to be brief here to weigh in on the new sources without retreading Sirfurboy's post:
  • Anime Super Hero used to be known as ToonZone.net, which has been brought to RSN numerous times and mostly found unreliable [52][53] and again as Animesuperhero[54]
  • EquestriaCN identifies itself as a fansite, and to my understanding fansites are considered generally unreliable on Wikipedia with very few exceptions where they qualify as SPS.
I also want to add that the timing of the coverage does matter, per WP:NSUSTAINED. Also please take care of yourself, WP:NODEADLINE. I know it's hard when your brain is fixated on something, but don't hurt yourself via lack of sleep over Wikipedia. This is part of why I feel it would be better to return the article to the draft state, that way you can take your time on improving it without stressing yourself over AfD. Brocade River Poems (She/They) 02:10, 23 December 2024 (UTC)[reply]
After seeing the extensive UseByOthers examples in the first link, I searched the archives. WP:Reliable_sources/Noticeboard/Archive 93#Toonzone.net again features the opposite assessment (brought forth by a prominent ITN participant, but also very old). I would agree that it's sort of 50/50 whether the site is reliable, so maybe we should just assume it isn't. Aaron Liu (talk) 23:51, 23 December 2024 (UTC)[reply]
I don’t understand why you think it’s okay to bold any text other than your vote. What is this silliness with “[NEW]”? Is this a sales pitch? Are you selling me a new product? Totally pompous behavior that does nothing except make it harder for a closing admin to understand what’s going on. HyperAccelerated (talk) 21:16, 23 December 2024 (UTC)[reply]
Ah yes, simply bolding "[NEW]" would surely confuse the closers. While I also don't really see the point, it absolutely does not warrant language like "pompous" and is perfectly fine. Aaron Liu (talk) 22:55, 23 December 2024 (UTC)[reply]
Look, I get that you want to make a joke out of this, but you’re missing the point. The refbombing, the bolding of random text, and the paragraphs upon paragraphs of buzzworded grandstanding makes it harder for people to participate. I personally found it extremely frustrating to parse through the sources provided here, only to realize that they do almost nothing to establish notability. My point is that this behavior is part and parcel of an effort to make it artifically difficult for people to meaningfully participate in this AfD. HyperAccelerated (talk) 00:23, 24 December 2024 (UTC)[reply]
Though my first sentence was sarcasm, I was not joking, and I'm sure you did not mean to assume bad faith. Aaron Liu (talk) 17:43, 24 December 2024 (UTC)[reply]
  • This AFD has only been open for 3 days and already has 57,768 bytes of comments. PLEASE, make your argument and do not respond to every argument you disagree with. Do not BLUDGEON this discussion like the past AFD. This is already a mess for a future closer to parse through. Keep your comments concise and brief. And, even better if you have already had your say and do not need to say any more. Let's hear from new voices. Liz Read! Talk! 08:45, 23 December 2024 (UTC)[reply]
Delete: It is a terrible idea to proclaim that a subject is notable, and then work backwards to find as many sources as we can to support that argument. I've seen this happen far too often, and it appears to be happening in this AfD. Many of the sources brought forth in this AfD are written by authors with questionable reliability or are self-published. For example, one of the sources mentioned in this AfD is a self-published Medium article, and the Taiwan Daily News article is written by someone named "KYLAT" whose profile picture is a picture of Jack Nicholson. Refbombing appears to have convinced people who have merely asserted that the sources are sufficient, but I remain unconvinced. Two good sources are much better than twenty bad ones. If this is what we've found after pulling all-nighters, I feel comfortable saying that there is insufficient sourcing to keep the article. HyperAccelerated (talk) 17:12, 23 December 2024 (UTC)[reply]
I think it'd be much more productive if we directly discussed the sources instead of speculating about editors' approaches.
I just responded about TowardsDataScience here.
It looks like United Daily News allows pseudonymous submissions. They're big and trusted enough that I'm sure they review their articles. Aaron Liu (talk) 00:05, 24 December 2024 (UTC)[reply]
I would need to see some sourcing describing United Daily News's revision process before accepting it as reliable. The same thing goes for the Medium article. I appreciate your response, but my vote remains unchanged. HyperAccelerated (talk) 00:38, 24 December 2024 (UTC)[reply]
Question. Having read through the DRV in its entirety, is this AfD discussion still only open to extended-confirmed users? Madeleine (talk) 20:39, 23 December 2024 (UTC)[reply]
It will be up to the closer to decide that. Best, Barkeep49 (talk) 21:19, 23 December 2024 (UTC)[reply]
Thanks. Weak keep. I find Aaron Liu and GregarioiusMadness's assessment more compelling than HyperAccelerated. A look through the sources evinces that the vast majority of the authors are real with storied edit histories. As funny as it is, I do not buy the Jack-Nicholson-from-The-Shining-avatar-therefore-writings-are-moot argument, particularly when you factor in the fact that using pseudonyms and avatars is somewhat common practice in Asian news media, unlike Western media. It would be wise to invoke WP:NOENG; the Japanese and Taiwanese sources alone appear strong enough to advocate for GNG. Having tried to do a search for 15.ai myself a few minutes ago, I can confirm that it is exceptionally difficult to filter out the useless "Top 15 AI" articles. But the sourcing itself, as is, appears to be sufficient. Madeleine (talk) 21:45, 23 December 2024 (UTC)[reply]
Thanks for voting. Would you mind pointing me to the two or three sources mentioned above that you feel do the most to establish notability? HyperAccelerated (talk) 00:26, 24 December 2024 (UTC)[reply]
Ah yes, simply writing "voting" instead of "!voting" would surely confuse anyone reading this. Stop cluttering this page with your random trivia. HyperAccelerated (talk) 20:13, 24 December 2024 (UTC)[reply]
Keep: With the rise of AI and increasing technological advancements this article would be beneficial to keep. Rager7 (talk) 03:08, 25 December 2024 (UTC)[reply]
Comment. Hi everyone. I found another source that is much better than the Business News Ledger one. [55] Per WP:NG/RS, The Guardian is reliable and demonstrates significant coverage, and the article seems to not be user-generated or sponsored. I'll be adding information from this article later tonight. GregariousMadness (talk to me!) 18:52, 25 December 2024 (UTC)[reply]
This source (published 2 weeks ago) looks excellent to me. A couple of sense checks though: The source is the Guardian Nigeria, which is not owned by the same trust that owns the UK Guardian. I am not sure under what arrangement they use the Guardian style and brand. However, from what I can find, it is one of the most trusted newspapers in Nigeria.[56] so I believe it is a reliable source. The byline on the article is Yusuf Temitope, who seems to exist, but not clear what expertise he has. However he is writing in a properly edited reliable source. This one, from what I can tell, is a Green tickY. It contains in depth significant coverage, allowing an article to be constructed that goes beyond what the website did, places it properly into historic context, and shows why this is notable. My only remaining caution is that citogenesis is possible. If the author of this (extremely recent) article obtained the information from Wikipedia or other self published sources, in whole or in part, the article would not be reliable. That is not a reason to discount this source. It is a reason to keep looking for more sources, and to keep asking questions. But on the basis of this source, I expect we are over the line for a keep. Sirfurboy🏄 (talk) 19:56, 25 December 2024 (UTC)[reply]
Considering that the article describes previous TTS frameworks that arose from 2018 to 2020 that are never mentioned in Wikipedia or other sources, I think the author of this article did their due research. GregariousMadness (talk to me!) 20:45, 25 December 2024 (UTC)[reply]
  • Delete Recreating an article immiedately after an AfD was closed as delete shows a contempt for proper processes that must not be tolerated. * Pppery * it has begun... 00:10, 26 December 2024 (UTC)[reply]
    With all due respect, I recreated the article after I asked other administrators what I should do… I just wasn't aware that it would be good practice to wait a little bit before doing so. But I really did try to follow proper processes, as I went to Liz's talk page to ask questions on what to do and per her advice I submitted the article through AfC. GregariousMadness (talk to me!) 00:18, 26 December 2024 (UTC)[reply]
    The page was recreated as a draft and mainspaced via AfC. Content has primacy over tidy process and a "contempt for proper process" is not a reason to delete under policy. Moreover, valid process has priority over individuals' conduct. If we say that the draft submitter's conduct could have been better, that does not annul a valid AfC pass. Can't just override the AfC reviewer's judgement and role in the process by saying that the draft submitter did something badly conduct-wise. That's because the AfC reviewer made an independent editorial decision when accepting and was fully authorized to do so. Respecting that fact is proper process. —Alalch E. 00:51, 26 December 2024 (UTC)[reply]
    Alalch E., you and I have history. I often disagree with you, but we have grown to trust each other. In retrospect, this submission of a new AFC draft while DRV was ongoing seems naked gaming of the system, IMHO. I greatly regret my actions in this case. I feel I have facilitated bad behavior by not speedy deleting as tagged. Perhaps I should have boldly draftified. I now agree with Pppery this recreation was contemptuous of the process. I nominated this procedurally, but especially given the history of socking surrounding this subject (and the bludgeoning single purpose aspect of the pagecreator), I'd be inclined to delete on the merits. BusterD (talk) 01:07, 26 December 2024 (UTC)[reply]
    I'm sorry but I really am not an SPA. I was editing plenty of deep learning and math articles before this article was nominated for deletion, and ever since most of my time has been spent trying to improve the 15.ai article. I plan to go back to helping Wikipedia on topics like deep learning and math, but I'm traveling for Christmas right now and so I haven't had as much time as before. I'm sorry for any trouble I inadvertently caused, but please don't punish the article for my incompetence. I was also told that making a comment to add a new source wasn't bludgeoning, which is why I added the Guardian source today. I am still learning how to best contribute to Wikipedia and my neurodivergence can make it difficult for me to learn how to appropriately behave. GregariousMadness (talk to me!) 01:14, 26 December 2024 (UTC)[reply]
    I respect your view of things here, and take to heart how you feel about actions that weren't taken but maybe should have been (regardless of what I think), but time can't be unwound, and proper process now is assessing this on the merits of the case regarding content, which AfD is designed for as a process. AfD is not just where articles are deleted, it is also where articles are kept, for example when encyclopedically suitable topics and content are identified. If editors are recognizing encyclopedic value here, it isn't proper process to address the underlying conduct matter by deleting the article. This is "damage done": we get a functioning article on one hand and on the other we get a worry that questionable initiatives might reoccur from an editor with whom you've seen a single purpose aspect. It's that latter thing that needs discussing, but probably somewhere else (for example, I would not feel comfortable expressing my disagreement with the "single purpose aspect" assessment here, knowing that this just isn't the right venue for that) and keeping track of it for the future. —Alalch E. 01:28, 26 December 2024 (UTC)[reply]
    A single AfC reviewer cannot override consensus processes. Those consensus processes were first AfD and then DRV which was made complicated by the AfC actions. DRV can absolutely permit recreation (DRVPURPOSE #3) so the article in draft form could have been considered there. That said I think Buster's decision was a reasonable one given the cluster this turned into following the AfC acceptance. The AfC pieces were done in good faith all around (creation and acceptance) but doesn't change the fact that individual actions need to bow to the will of consensus. Best, Barkeep49 (talk) 03:38, 26 December 2024 (UTC)[reply]
    He can override because there's an inclusionary bias: All it takes to override a consensus to delete is a recreation that is not subject to G4. It takes a group to (hard) delete, it takes an individual or an individual + individual reviewer to create. It almost never takes a group to be able to form a decision to create. That can seemingly happen at DRV but situations when it truly must happen to allow for a recreation are very rare. This systemic favoring of content over lack thereof in the processes is because Wikipedia thrives on content, needs an influx of new content (new topics emerging in the world that need coverage), and is tolerant to retries. —Alalch E. 06:05, 26 December 2024 (UTC)[reply]
    I agree with the general statement that content > process, but in this specific instance the consensus was that the content was not encyclopedic, though. The additional WP:REFBOMBing has done nothing to establish notability of an encyclopedic topic. ―"Ghost of Dan Gurney" (hihi) 04:40, 26 December 2024 (UTC)[reply]
    but in this specific instance the consensus was that the content was not encyclopedic, though But that's why I rewrote the whole article from scratch. I really don't mean to bludgeon here and I'm sorry for that, but I agreed with deleting the old article and starting fresh from a blank slate. Plus I dug up a number of sources, including the most recent Guardian one, which can be used for the article. I thought a process like that to improve the article was supposed to be how AfD's worked, but every time I do anything I seem to get berated for not respecting the process. I'm just so confused. Sorry. I'm just going to cite WP:IAR and just continue to improve the article because that's what I'm good at, not discussions... I'll let someone more knowledgeable than me argue for my case. GregariousMadness (talk to me!) 05:06, 26 December 2024 (UTC)[reply]
Retrieved from "https://en.wikipedia.org/w/index.php?title=Wikipedia:Articles_for_deletion/15.ai_(3rd_nomination)&oldid=1265296207"