[16:50:27] On a New York Times podcast today I heard about a network of local news websites in the U.S., which practiced "pay to play"—publishing news articles because businesspeople or politicians paid the network without disclosing it. That podcast mentioned there are some FTC rules that require (at least in the U.S.) that advertisements are not supposed to look like news articles. I easily found a page on the FTC website: [16:50:27] https://www.ftc.gov/tips-advice/business-center/guidance/native-advertising-guide-businesses [16:51:59] The reason I'm bringing this up here is that, as I mentioned a couple of times, articles created by AW should also be labeled as such, and not look totally like Wikipedia articles, and it should be universal and default, and not customized per wiki. [16:53:37] AW is not advertising, but of course. But it does remind of this. Advertising is not the same as news, and shouldn't look the same. Machine-written articles are not the same as human-written articles, and shouldn't look the same. [16:54:34] Even if it's not a law, AW should adopt this as an ethical rule. [17:13:05] Should lsjbot or ranbir be required to make that explicitly visible? [17:13:16] *rambot [17:14:57] Lsjbot already does make it explicitly visible that the article is created by a bot (re @vrandecic: Should lsjbot or ranbir be required to make that explicitly visible?) [17:18:47] Lsjbot already does by itself. Other bots should be required, too. [17:20:50] It would be good if the knock-on effects of lsjbot and the others were somehow visibly tagged as such in other wikiprojects like wikidata. [17:21:31] Afaik they're only visible in the wikipedia they were created in. [17:26:03] We should ask the English Wikipedia community if they prefer to be in charge of deciding whether and how these articles should be marked, or whether they would prefer this to be centrally decided by... By whom actually? The Foundation? [17:30:31] As it can be a legal issue in some countries, I think the foundation needs to decide, not the enwiki community. [17:31:53] When it is a legal requirement, yes, agreed. In general, we should comply with laws, agreed. [17:35:25] I'm not sure how we would go about keeping up-to-date on all the jurisdictions touched by the enwiki and agree with Amir that it should be universal and by default. [17:36:07] When you add in the rest of the wikis, it's going to be a huge mess unless it's universal and by default. [17:39:00] But who should make that decision? [17:52:07] I think the foundation should. [17:52:33] And I'm normally very anti-let-the-foundation decide :) [18:05:57] A good question, and I have a good answer, I'll write in an hour or two (re @vrandecic: We should ask the English Wikipedia community if they prefer to be in charge of deciding whether and how these articles should be marked, or whether they would prefer this to be centrally decided by... By whom actually? The Foundation?) [18:07:35] I'm playing jazz on Wednesday [18:23:51] I'm playing jazz on Wednesdays :) [20:56:07] OK, I'm done with my jazz practice, the first after many weeks of lockdown. [20:56:08] Why the English Wikipedia community specifically? [20:56:09] If the English Wikipedia community wants to use AW in the first place, it's already a win, isn't it? :) (re @vrandecic: We should ask the English Wikipedia community if they prefer to be in charge of deciding whether and how these articles should be marked, or whether they would prefer this to be centrally decided by... By whom actually? The Foundation?) [20:56:17] More seriously, if you scroll up, you'll see that I cann this marking "universal and default". Here's what I mean by these two words: [20:56:18] * Universal: the same way across all wikis. For example, if the French-speaking AW community writes rendering functions that add the words "created by Abstract Wikipedia" to every paragraph, this is not universal. I'd be happy not to see this happen. It's a feature that can be useful to all languages, so there must be a universal way to do it. [20:56:20] * Default: Enabled by default for all AW output. [20:56:21] What I am not talking about at this point is the form that it will take. By default it should be both machine-readable and human-readable.By "machine-readable" I mean something that is easily readable by machines, but not necessarily even seen by humans. For example, the function that output the first sentence of the Beverly Hills article can output something like this: [20:56:23] * en: Beverly Hills is a city in Los Angeles County, California. [20:56:24] * ro: Beverly Hills este un oraș fiind situat în vestul comitatului Los Angeles County, statul California. [20:56:26] * ca: Beverly Hills és una ciutat localitzada al Comtat de Los Angeles, Califòrnia. [20:56:27] Note that the "class" is the class and the data-abstract-constructor are the same in all languages, and that's part of being "universal". For being machine-readable, it's the best thing. By itself, this markup doesn't affect rendering for end-user readers or page editors. [20:56:29] (I'm giving an example in HTML because that's the most ubiquitous format. It's conceivable that it will be in JSON or something else. The format would be different, but the metadata idea would be similar.) [20:56:30] For being human-readable, there should be some styling or extra text that is automatically added to elements with this class. For example, a different font, a different background color, etc. There should be some default definitions for it, and they should be overridable by the local wikis' communities to the point of being invisible. [20:56:32] More seriously, if you scroll up, you'll see that I cann this marking "universal and default". Here's what I mean by these two words: [20:56:33] * Universal: the same way across all wikis. For example, if the French-speaking AW community writes rendering functions that add the words "created by Abstract Wikipedia" to every paragraph, this is not universal. I'd be happy not to see this happen. It's a feature that can be useful to all languages, so there must be a universal way to do it. [20:56:35] * Default: Enabled by default for all AW output. [20:56:36] What I am not talking about at this point is the form that it will take. By default it should be both machine-readable and human-readable. [20:56:38] By "machine-readable" I mean something that is easily readable by machines, but not necessarily even seen by humans. For example, the function that output the first sentence of the Beverly Hills article can output something like this: [20:56:39] * en: Beverly Hills is a city in Los Angeles County, California. [20:56:41] * ro: Beverly Hills este un oraș fiind situat în vestul comitatului Los Angeles County, statul California. [20:56:42] * ca: Beverly Hills és una ciutat localitzada al Comtat de Los Angeles, Califòrnia. [20:56:44] Note that the "class" is the class and the data-abstract-constructor are the same in all languages, and that's part of being "universal". For being machine-readable, it's the best thing. By itself, this markup doesn't affect rendering for end-user readers or page editors. [20:56:45] (I'm giving an example in HTML because that's the most ubiquitous format. It's conceivable that it will be in JSON or something else. The format would be different, but the metadata idea would be similar.) [20:56:47] For being human-readable, there should be some styling or extra text that is automatically added to elements with this class. For example, a different font, a different background color, etc. There should be some default definitions for it, and they should be overridable by the local wikis' communities to the point of being invisible. [20:56:48] More seriously, if you scroll up, you'll see that I cann this marking "universal and default". Here's what I mean by these two words: [20:56:50] * Universal: the same way across all wikis. For example, if the French-speaking AW community writes rendering functions that add the words "created by Abstract Wikipedia" to every paragraph, this is not universal. I'd be happy not to see this happen. It's a feature that can be useful to all languages, so there must be a universal way to do it. [20:56:51] * Default: Enabled by default for all AW output. [20:56:53] What I am not talking about at this point is the form that it will take. By default it should be both machine-readable and human-readable. [20:56:54] By "machine-readable" I mean something that is easily readable by machines, but not necessarily even seen by humans. For example, the function that output the first sentence of the Beverly Hills article can output something like this: [20:56:56] * en: Beverly Hills is a city in Los Angeles County, California. [20:56:57] * ro: Beverly Hills este un oraș fiind situat în vestul comitatului Los Angeles County, statul California. [20:56:59] * ca: Beverly Hills és una ciutat localitzada al Comtat de Los Angeles, Califòrnia. [20:57:00] Note that the class and the data-abstract-constructor are the same in all languages, and that's part of being "universal". For being machine-readable, it's the best thing. By itself, this markup doesn't affect rendering for end-user readers or page editors. [20:57:02] (I'm giving an example in HTML because that's the most ubiquitous format. It's conceivable that it will be in JSON or something else. The format would be different, but the metadata idea would be similar.) [20:57:03] For being human-readable, there should be some styling or extra text that is automatically added to elements with this class. For example, a different font, a different background color, etc. There should be some default definitions for it, and they should be overridable by the local wikis' communities to the point of being invisible.