A Discussion with Wikipedians Sam Klein & Annie Rauwerda

All right. Hello, everyone. Well, we have something slightly different for today. I'm here with Sam Klein and Annie Roarda, and I consider this sort of an 18 year follow up on a conversation that I had with Sam here on, let's see, the last week of May 2005, and we had met at some event at MIT and Sam was showing me Wikipedia and in particular was showing me, look, there are all these great things on Wikipedia. And he showed me my own page on Wikipedia, which I'm not sure I had even seen at that point. And that moment the Wikipedia was still young enough that things had been hooked up so that there was a little chirping sound every time there was an edit to Wikipedia. And I think I considered maybe we should do that for our source code updates in our office, but don't think it ever actually happened. We might have had a, you know, in, in birds Inside, so to speak, with, with lots of chirps going on. But anyway, so that was 2005, and then Sam got me to come to a, a Wiki Mania. Sam, you were saying it was, it was an art project. Okay, how long did it last? How long did the birds last? How long. When did they, when did they go extinct?

Daniel Wunsch set up the project and I, I just set up a display of it at the Cyber Arts Festival in Boston. I don't remember exactly when it stopped being deeply integrated. It was, it was reading recent changes off of irc because back then there was an IRC channel that got a ping for every edit.

Huh. Okay, well, fair enough. It's any case. So since in the last 18 years Wikipedia has hopefully presumably grown up in some way or another, and I think its life and times might be. Well, and since that time we built WolframAlpha, which came out in 2009, and now I haven't looked at my Wikipedia page in perhaps even since you showed it to me. Maybe I've looked at it once or twice since then, but not, not for many years. I, I kind of have this principle of not. I, I always feel bad when I don't read stuff that people write about me because I know they put effort into writing it and if I'm actually going to meet them, I will usually read it, but otherwise I don't because I figure it's, it's a, it's a bad sort of self. Self dynamic, so to speak. Anyway, I thought, you know, one of the things I wanted to do was to talk about, you know, I put, we've put lots of material out into the world that some of which I think has been picked up by Wikipedia. Much of it probably hasn't. Some of it maybe should be picked up by Wikipedia, but can't be because of some, something we've done in, you know, putting some, some copyright thing that makes it poisonous or whatever else. And I thought it might be interesting to. And useful for the world and so on to, to see what things we have that might actually be useful to Wikipedia. And maybe we'll have to ease into this. Maybe we get to look at my Wikipedia page, but that's. We'll have to ease into that. That doesn't come so easily. That's the. You guys, you guys presumably both have Wikipedia pages. How often do you look at them?

A lot.

Okay.

I worry about it a little bit, but I can't. I, I don't, I don't edit my own article, so I suppose it doesn't help me at all to be aware of it.

Right. Well, that's, you know, I think one of the features of, of doing, you know, I, I do lots of live streaming and one feature of that is, you know, you do what you do and you're kind of in control of what you say and if you say stupid things, well, you say stupid things. And it's kind of, It's a, it's a, it's kind of a, A well defined system, so to speak. I, I think. Well, let's see. Where should we start? We could start by. Let me, Let me. We could just start by like, what. Think of one.

One grounding context. I'm listening in the background to a. The modern successor to RCBirds, which is called Hat Note. And it is a very satisfying way to listen again to all the edits made on Wikipedia.

Okay, can we, can we hear it? Can you, can you put the audio into the, into the stream?

Yes, I think screen share. Let me screen share. Zoom tries to filter out audio coming from my computer, but usually there's a.

There's a thing you can check where it says send the audio as well.

Oh, yeah, I'll see if I can do that, but I'll screenshot while I'm doing.

Sounds kind of like wind chimes.

Yeah.

And the size of the circle corresponds to how big the edit was. So some of the edits coming in are huge, some are tiny.

So what is a typical. So these are lots of edits to people's pages. What, what typically are those?

If someone's like fixing a typo, maybe that's the difference is just a few bytes Maybe someone's adding a photo. That would be like a lot of bites. Maybe they're rephrasing stuff like that.

But so when there's a fairly. Okay, Tom Jackson Jr. Whoever that might be, it's interesting that most of the things we're seeing are people here. So is that. Is that if you look at the volume of. None of the people that I've seen so far have people I've heard of. Well, Sears Roebuck. Okay, there's a company. I've heard of that one. But when there are big changes, what like for a people page? I mean, fixing a typo, presumably, is not showing up as a giant circle on this plot.

Right. A big change might be pasting a whole section. Or sometimes if you're moving things around or adding a lot of references, it ends up looking like quite a large diff in terms of character count.

And so what are the colors here?

Oh, I know, it's. So the green ones, I believe are from people who haven't made a user account.

Okay, look, finally. Okay, finally. I saw Ilya Prigozhin. That's a person I knew. Okay, so there we finally hit one. That. Finally hit one. That I know.

And there's a setting that you can do to add different languages because I'm pretty sure we're just watching English Wikipedia. When you scroll down. Yeah.

You pick a language.

So that's the frequency of edits that we're seeing there in the.

Right. If you want something really fast, we can switch to Wikidata.

Does that have automatic feeds, automated feeds?

Yeah. And so it's even more dominated by bot edits than the wikis, than the encyclopedias.

Well, so how many of those. Of the things we were seeing there, how many of the things will have been bot edits of just the regular Wikipedia editing? How much of that was bots is that?

It's easier to see when you look at the recent changes. So you can see the ones called. It looks like the. Yeah, the bots are not filtered out here, so maybe one in three.

Right.

Right now this particular bot is making a lot of edits. JJMC89.

And what do you think that bot is doing? What? What are the typical bots doing?

I don't know. Let's. So when you click on it, it will show you the user page. It's active. It's only partially exclusion compliant. It does not have a disable button. What is it doing? It processes categories for deletion. It's processing categories for discussion. So anytime someone thinks a category should be deleted. It applies the various templates. Maybe right now it's going through a backlog of queued up requests.

Okay, so is it useful? I mean, here, let me. Maybe this is cool. And that was an interesting view of modern Wikipedia. So.

I met the guy that made it one time. We did go karts in Santa Cruz.

What? Wait a minute. Maybe, maybe you should mute that audio, otherwise we're going to be listening to the, the, the music track. Well, let's see. So, so maybe where should we start? We can talk about some of the stuff that we have. So one thing I, you know, I just don't know what Wikipedia has, what it doesn't have. I mean, we could start. Let's start with, I mean, here's a really simple thing is WolframAlpha. I have no idea whether there are links from Wikipedia to things you can do on WolframAlpha, which for our sins is a free thing that everybody can use. And I have no idea whether, if you go to a page about, I don't know, let's take a look at the, The, let's see. I'm, I'm, I mean, if, if we go. I don't know, what is this? This is some computation thing that should probably be linked to the page Wikipedia about transfer functions. Because look, you can compute it here. I have no idea if it is. What's the right way to make that happen? Or should we not bother? I mean, it's fine. It's like, so tell me what should be done with something like that.

Typically, articles don't they avoid linking to a particular source. If there could be many sources and it's not somehow.

Let's see what, let's see what the article. I have no idea. I mean, we just picked this one.

Yeah. So this is, it's a great example of what I think of as the ISBN problem.

What it links to.

So right now there aren't any external links to a place where you could render transfer functions, which seems like a shame, even though it would be useful. And similarly with ISBNs, people who see an ISBN, they often want to visit the record page somewhere where they can get a book or, or they can find the standard, the standard metadata about how you can get it, either in a library or to buy a copy.

And presumably, presumably if I go to this. Okay, I'm right there, whatever that is.

Right.

And presumably if I go to a book here, well, there's an ISBN that wasn't very easy.

If you click on the actual ISBN number, it takes you to this meta page, because otherwise you'd have to non.

So it's not taking you to the Amazon page to go buy the book.

Right. And if you're non neutrally choosing one source, why are you supporting Amazon or AbeBooks or some French.

But I mean, in this case here, this is a thing we're paying for the servers to provide this for the world. It seems like the world should take advantage of it. That seems like a mistake. And I mean, I don't know. You know, that seems like one of those inefficiencies in the world that, you know, given that we've done the philanthropy of providing this for free to the world, seems like the world might as well take advantage of it. Or is that not the right way to think about it?

I think it would be great. I also. I'd love to hear your take, Annie. So I think an example of the kind of landing page that everyone could get excited about would be a site that was not owned by a company, but it was owned by a library somewhere that generated things like this. And maybe Wolfram Alpha was the only instance of a good visualizer, and it could feature whichever visualizers it wanted. And the context of the page would be, here are ways you can visualize transfer functions online. And the page gets.

So far as I know, this is the. I mean, I don't know about this particular thing. We literally picked this completely at random. So, you know, and there will be, you know, 5,000 other things like this in, you know, Wolfram Alpha. So it's, It's. I mean, like this one. I doubt there's another place. It's a fairly specialized thing. I, I doubt there's another place where you can do what we're doing here. So, you know, and, and obviously, you know, the, you know, for our sins, you know, we're running the servers that make this possible, just like, you know, Brewster is running the servers that make archive.org possible, where, you know, he's. He's got a bot that's linking lots of things to archive.org?

Which sometimes, even though it is a nonprofit library, sometimes runs into arguments as well, where people say, why are we automatically linking everything to the Internet Archive? There are other archives out there.

Are there?

There was a question about whether this was spam that was just adding extra crafts to every. To every web citation. They've mostly been resolved.

Seems useful. I mean, it seems useful to me. I mean, it seems like. So, I mean. Yeah, you know, one question here is, okay, so we've got something I mean, this is, this is something where, you know, this seems useful. What's the issue?

Well, I'm. Okay, so I'm not a transfer function expert and nor am I on the block for longer than I have. But if it's helpful, I think that if I were reading that Wikipedia article on transfer functions and I had, you know, a little thing embedded from Wolfram Alpha where you can visualize things as a reader, I would enjoy that. The logistics and arguments of why you can and cannot put that in there. That's where I think SJ might know more because I. This seems pretty great to me.

The fact is that what would be. I mean, so look, the things. There's WolframAlpha, which has tons and tons and tons of material that's all freely available. There's, I mean, another example. Well, so, you know, another example that. Another example is Math World, which obviously predated Wikipedia and was, according to Jimmy Wales, at least a, you know, one of the things that he looked at as a kind of a model for how to do Wikipedia. And you know, Math World is much more textual. And in the early years of Wikipedia, basically, you know, Math World was steadily mined by Wikipedia, I think often with citation, sometimes not. I don't really know in detail. But, you know, in, you know, a Math World is. Oh yeah, MathWorld obviously has Wikipedia links on it. I mean, it has Wolf Malpha links on it. But, you know, so this is a, this is a textual thing that we put out in the world that I think, you know, that hopefully with citation, I would hope. With citation, I don't know, but I would hope, you know, was, you know, was successfully mined by Wikipedia. I would consider that a. I mean, I think, you know, from our point of view, we put lots of effort into building MathWorld and insofar as Wikipedia just took the content and, you know, took the users away from it, that's kind of a lose for us. But I'm philanthropic enough that I consider that, okay, you know, it's good that the information is out there.

I think we had to be very careful. I mean, I think even in our very early discussions when I would meet people who had worked on MathWorld, all I wanted was for them to license them under a free license. Then it would have been included very explicitly with attribution and people had to rewrite these things in some cases from scratch or finding other sources. To the extent that it wasn't under a compatible license, I don't think we ever. People weren't able to reach an agreement Maybe with Eric. How have it worked out?

Well, what would happen? Okay, so let's say, let's take another example. Okay, let's take a textual example. We've got both textual and non textual examples. Let's take, okay, let's take a textual example that, whose licensing is at least well understood, which is, you know, my big book, which has certainly. I mean it's, it's full of pictures that people might like and I don't know where's a good picture. Those are a lot of trivial pictures. Those are easy to generate. But the, you know, there's a picture people might like now I have no idea whether. Okay, so you tell me, could you take this picture? What, what would it. What will be involved in taking that picture and using Wikipedia?

You or the publisher would have to, would have to release it under a compatible license. So CC share like.

So let's find out what the license is for this. So I don't actually know. Good question. Where is it? Let's see.

Copyright page.

The copyright page is long and complicated, but I think. So this is probably a regular copyright page and probably not. Let's see. Yeah, okay, so it's a regular copyright page. And see, what is this? This is. So if we take. I see somebody is putting this in there. Where is that? Where is the footer? What should I be looking for here? See, I don't know any of these things. Oh, there we go.

Right there.

Look at that. What's that? Oh, how nice. Okay, so is that, is that what you need?

Let's see. Yes, that looks good. Looks like. Sure.

Exactly what's needed. I mean, I have no idea what, what. Should we click on it? Let me look.

Attribution, Share alike, Share like is exactly what's needed. That's perfect.

Okay, so. Okay, so that means in principle people can take stuff from NKS and put it on Wikipedia. So I don't know whether they've done that, but they could.

Is that fantastic?

And so have at it. You're welcome.

Flood Wikipedia with cool diagrams.

Right? I mean there's, there's, there's lots of stuff here. So now maybe nobody ever noticed that. I mean, I, I'm sure that footer has been there for a long time. I mean this is maybe. Actually, you know what? One thing that's probably a bug on our part is that footer is not on every page, but it probably should be. And that would be a. You know, that's an easy thing to fix. But yeah, that, that's so that, that's so that, okay, so that, that means there's no structural issue there. And this is just content. I mean, we're not, we're not talking about. Now another thing that I would say is in here, which is really quite. See, why do we have this like. I don't know, let's look at this section here, for example. So this is basically huge amounts of kind of. This is very Wikipedia friendly stuff because basically summaries of things that I wrote for, you know, that I wrote to as essentially footnotes to the things I was doing. So they're mostly not. They're much less about what's in NKs as they are my attempt at being an explainer of things in general. And I have no idea whether these have been mined. I mean, I do occasionally run into, you know, I use Wikipedia and I look for things and I have actually looked at stuff, now that I think about it, about mathematical logic and it's largely incomprehensible compared to what I've written. I mean it's much, it's, it's, you know, it has, you know, this, this is at least nice and clean. Some of, maybe some of what's on Wikipedia is more technically sophisticated. I don't know. But you know, this would be a good example of something where somebody could just lift this and quote it. Because there are pieces here that are pretty much, you know, if you try and if you try and rewrite it, it's a mess because it's actually quite, quite. You know, it's quite a complicated thing to describe. But you know, so I mean, I assume with that thing which should be on every page with this, whatever they are, these, these little icons that it would be allowed for somebody to just pick up a piece of text from here and plop it on the appropriate Wikipedia page. Is that, is that a true statement?

Absolutely. Yes.

Okay.

Yes.

Well, fine. So people should have at it with my blessing because it's kind of like I wrote this for people to read it and it's, you know, I noticed. So that, so that's one big source, I guess is nks then I suppose another big source is all the things that I write in. Let's see. Ah, well, that was definitely not. If I misspell my name, I won't get there.

It's your Massachusetts doppelganger.

Right. The. So, you know, this is a ton of stuff. I'm sure some of these are linked on Wikipedia. But I think, for example, if you go to historical perspectives, that's pretty much the, I mean, you know, There are lots and lots of things here. There are lots of. Unfortunately, there are lots of obituaries here, but they seem very popular among, well, people who knew the person as well as other people. And they are prime linking stuff. So let's look. Does this have a footer that. So this does not have a. Okay, it says terms here. Okay, so what's this? What does this say? So I'm learning here. What. Okay. We encourage the use of aforementioned items and educational resources around the web. For example, on Wikipedia and university websites.

Amazing.

Okay, so.

Well, and it looks like the writings.stephenwolfram.com has been cited on five different English Wikipedia articles.

That's pretty lame.

I know. It's not very many. And one of them is, is Wolfram Mathematica. And then there's Julia, the programming language. Erica, Jen Moses, Sean Finkel. And then binary combinatory logic. And in those cases, it was all used as it. There was a statement that was made that probably paraphrased you or, or like wrote in different words. And then.

So that's really lame. I mean, honestly, I've written, you know, 10,000 pages or something of stuff that, you know, and some of these posts get, I think. Well, you know, some of them are in the millions of views on, you know, on my actual page. So, so it's, so what's, so what's going on? Why are. I mean, like, like this one. I, I'd be amazed, I, you know, if this one isn't cited by Wikipedia. Wikipedia is, is stupid because this is this, I mean, I wrote this mostly because I wanted to understand this myself, but turns out it's been, you know, it's turned out to be really popular. And I, I. And it, it even. In fact, Paul Graham of Y Combinator fame, who I've known since before Y Combinator, actually sent me mail saying, you know, you should publish this as a book, which I was a bit skeptical about. But we did it, and it's been quite successful as a book. But it's also just free on the web here.

I'm sure it's because people have not gone to the terms page to try to parse it since the, since the footer on most of the pages just says copyright. Stephen Wolfram with a link to the terms on a blog. 99 times out of 100, that just means full copyright.

Well, fair enough. So we should add, I mean, it does say terms there.

Yeah, yeah. So firstly, I think if you added the little MIT license icon there, that.

Would be a useful thing to do. That sounds like that's a good thing to do. So that's a takeaway here. Add that if we want people to actually read. I mean, it's not clear people read ever. Although I suppose the conceit of all of the stuff, of the things I write and all those kinds of things is, you know, people read it otherwise. Well, actually, for me, a large part of the point of writing is to understand things. For myself, it's. But it's a pleasant side effect that other people read them and hopefully get benefit from them. But, yeah, I mean, I, you know, it's. I mean, I. I think people have been finding this page, so it doesn't really matter that. That, you know, that it isn't linked to from Wikipedia. But I mean, that list of. Of things that link to writings is kind of. Is kind of depressingly small. I mean, it's kind of crazy because there are pretty much definitive, you know, biographies, you know, obituaries, you know, explainer pieces and so on about all kinds of different things here. And it's kind of. It's kind of weird. I mean, like. Like, this is. This is a thing that I did. Small thing, but it has things like, you know, the game graphs of Tic Tac Toe, which surely somebody wants to have on Wikipedia somewhere. And they're, you know, hopefully nice pictures of those. And it might have. I'm sure it has many other kind of. Now, what's this? This is some other type of game and so on, and there are probably lots of things here. Okay, this, I'm sure this is an obscure. Is a rather obscure thing, which probably I'm linking to Math World for that.

But I see the problem. So the Terms of Use releases the images associated with the writings under the MIT license, but it does not license the writings themselves. And the writings are covered by the Wolfram Terms of Use, which says basically full copyright.

Okay. Okay. Okay. So maybe we should change that. Should we change that or will nobody care? I mean, because obviously to change that is, you know, I mean, you know, I don't think. I don't particularly want actually. Do I care if people sort of mirror these pages? Not really. It's fine, actually. I mean, I wrote them for people to read them, but is that. I mean, you know, will anybody care?

I would care a lot.

Okay, fine. So you're a good. You're a good. You're a good sample. So, you know, that's. So far as I know that something, you know, this is. These are my pages. So I think so far As I know that's the thing that we're perfectly in a position to do, I think. Let's. Let me give you an example actually of a recent thing. So I wrote this is a history of the second law of thermodynamics, which so far as I know exists nowhere else. I mean, I wouldn't have written it if I thought it existed somewhere else because it was a huge pain in the neck to write. And my main, my main interest in this whole thing was writing about the things that I figured out about this, which. But I wanted to kind of anchor that in the history. So, you know, now here there are images that are probably. I mean, that's a book from 1680s, I think, which is presumably firmly out of copyright. And my guess is that, that. Okay, so that says source there and that went to Brewster. Okay. And so this will be an example of, I mean, this particular document. There's probably. Well, this, this is a very citable document, so to speak, because it has. Well, it's a long, you know, it's 100 page long or more actually story. But so, so what should we know? I mean, what, what, how can we make this. You know, I think people have been reading this. It's got lots of views, which is nice. And it will also come out in a book about the second law that, that should be out in another country, couple of months. But what. You know, I'm sure there's an article on Wikipedia about the history of the second law. I am completely certain that this provides a lot of additional information.

Yes.

So two challenges.

One is Wikipedia is not a good place for. It's not a great place for original synthesis and it's often. It's a decent place for incorporating synthesis from an existing reliable source. Your site is one of the exceptions to personally hosted sites that are obviously reliable sources. I would say the default. The average blog on the Internet is not considered a reliable source. And so there will be.

Usually these blogs get posted also to archive. To, you know, Paul Ginsburg's archive. Archive.

Yes. Great. Yes. So.

So I'm sure that's.

That's a nice way to think about it. Anything. So anything that's on that has a DOI that's on. Paul Gainsborg's archive.org is a good source for things. And these blog posts are also a good source for things. They're obviously a good source. Whole scale reproduction, as the text of an article might be debatable. Depends on the topic. Depends on whether other people have opinions about what they want to say. But long, full quotes or just filling out sections that didn't exist make great sense. It's really nice. I mean, it's so much easier to fill out an existing article when you don't have to worry about copyright and you can take as much as is appropriate from the source.

Right. Well, I mean, so that's fine. It's super disappointing that only, you know, only five, you know, that's kind of a negative, you know, Pavlovian negative. Why the hell should I care about this type thing? You know, given that, that I've got, you know, many, many years of, of endless, you know, I mean, not all of these are, are, you know, some of these are Wikipedia fair. Some of them are not. I mean, you know, some of them are. I mean, you know, a piece like this is surely interesting, but probably not. It's not, you know, fact, fact, fact. It's got. It's more something where if somebody's writing something about the philosophy of numbers, this would be a relevant thing to cite. But it's not a thing where it's got, you know, it's not a fact pack, so to speak. But let me ask you another question here. So let's take a picture, for example. I don't know that random picture there. Okay, Could. What. What is. So somebody wants to take that picture and stick it in Wikipedia with, with attribution. What can they do that? What, what do they need to, what do they check to do that?

All of your images could be uploaded en masse or at any time to Wikimedia Commons, which is its own site. And that would be very welcome. And there could be a, there would be a category that was there, probably a family of categories. You could have a category for NKs, a category for each of the sections on your site, and then anyone who wanted to could embed them in any.

Article that makes sense. Sounds cool. What's that?

I said that would be cool. I would be excited.

I don't see why that. I mean, I don't see any reason we wouldn't want to do that. And that's, I mean, you know, we, it would be some, some, you know, is there anything else we should know about that? I mean, this is an image. It can be. It can be. I mean, by the way, this of course has something different that is not part of that story. But this has the code that made that image as well. Right. So, you know, that's the code. You can just rerun that image. There it is. That's a big piece of modernity that I don't think is quite, you know, Wikipedia isn't quite at that point yet. I mean, that's something that requires a whole different technology stack to support that. Something that would be, you know, it's, it's eventually, it will be the future of sort of all scholarly publishing and so on that people say, there's a picture in your paper. You know, I want to be able to reproduce it. Let me just click it and get the code and reproduce it. That's that, that's the kind of thing that, you know, talk about sort of the reproducibility of science. That's the, that's the ultimate version of that. But people haven't the. But I guess the question here is. So, I mean, you know, we can do the homework of sort of uploading lots of images to Wikimedia. What is it? Wikimedia Commons or something? Is that the thing? I mean, it's just a program that would do that. Is there anything we should know about the. But I'm not going to add them. I mean, I don't know, should I add. I'm not personally going to add them to Wikipedia pages because I think that's a, it's just, I don't, you know, I, My, my service to the world consists of writing, you know, many thousands of pages per year of stuff and doing, you know, hundreds of hours of live streams. It does not extend to, you know, editing Wikipedia and having people tell me that, you know, tell me how horrible I am, and so on and so on and so on. Not, not, not my, not my idea of fun. So I think it's, it's kind of like I'm, I'm, I'm a primary producer, so to speak. I'm not the, not the person to do that homework, so to speak, the. Sam, you're suggesting. So you should you say what you're suggesting here because.

Yes. So data points, claims with sources that can go into Wikidata or new Wikidata entities and media that's already under a free license, you can post yourself. That's the best, that's the best option. And then the person who is adding it clearly has some part of the provenance of.

So how would somebody know. So let's say we upload. You know, I suspect we have a hundred thousand images we could upload. So is anybody ever going to know that.

That'S enough that you might want to do a bulk upload and you can cue those up there?

No, no, but I mean, I'm Asking the question, you know, let's say we do that. It's going to be a certain amount of work on our part.

Yes.

If, if I find out in another year that only two people ever used those images, I'm going to be upset.

Well, usually when, when stuff is on Wikimedia comments, it's tagged. So then when someone's writing their article about, I don't know, whatever, whatever the photo is, they'll hopefully find the tag. It's just a lot of work to tag stuff.

What do you mean by tagged?

Well, just like a metadata tag, maybe. See, I don't really.

Presumably these have alt tags. I'm sure these have alt tags. Is that what you mean? Or is that.

Well, I don't think that's what I mean.

Like.

Well, let's take an example. Let's take a less silly example because. Let's take an example from. Because that particular one was just not. So let's take this picture here. Okay, so that is a picture of, you know, so that could say it comes from. I mean its tags could be. Okay, it's part of the tic tac toe section of my thing about games and puzzles as multi computational systems. And then one could, you know, essentially all of the graphics here has the form of, you know, statement colon graphics.

Yeah. So you would, that would be like, you know, the tic tac toe multi computational systems. Maybe you would even tag Stephen Wolfram work puzzles.

So. Okay, but so, so then it's a collection of tags that go from, you know, that in this case would be the title. This. It probably has an alt tag. How do I see the alt tag? I don't even know how to. What do I do? Do I say inspect expect accessibility properties? Is that right? How do I do this? I should know this, but I don't. What do I do there? How do I have it? Show me. It's his name. Surely. How do you see an alt tag? What do I do if I switch off images in this browser? I would certainly see the alt tag.

Can you right click to see it?

Well, I thought I could spectax properties.

I'm trying to visit the page to see if I see any alt text. But in general, a nice idea for a product this size would be to make a small project page about it on Commons or on Wikipedia and then invite the people who work on related projects subject areas.

Well, I mean, I think that the practical thing that. Okay, so if, if we, you know, we can certainly have somebody go and it'll mostly be A bot go through and take. I mean, we have versions of all these pictures. Obviously we have this picture. You know, you just take that picture, right? It's a gif of some kind. And I'm sure I could just take this and. Whoops, just take that. Ah. You know, when I first saw Windows like this, somebody showed me. I visited Xerox PARC in probably 1979, and somebody showed me, you know, windowed operating system. And they were like, oh, you can just, you know, drag this to there. And I was like, well, what happens if the window is covered up? And it's like, this seems crazy. And it's like, well, it was one of those cases where I was wrong. That. Okay, that. That just is giving me the. That is. Is. Okay, so that's. That's the PNG for this. Okay.

Are you looking for the alt? I was, I, I was looking through the HTML, trying to find the alt tank too.

Is there no alt tag? There must be an alt T tag.

Maybe. Maybe I'm crazy, but I didn't see one SJ.

There.

Maybe it got stripped out in publishing.

It's possible. Yeah, it's possible. I'm a little surprised. I. I thought these things had alt tags. Maybe. It also is messed up by the fact that there's. That it has its, you know, computational version as well. But that's okay.

That's because often you can hover over and see the alt.

Right? Usually you can, except that it just says copy input to clipboard here.

Which is. Which is also cool. But not an alt.

But not, not an alt tag. It's a different, different thing. But so, I mean. Okay, so let's just go through. Let's say we were tagging this. Assume it had an alt tag. Maybe it doesn't have an alt tag. You know, maybe somebody's written an alt tag for it. Maybe they haven't. But let's say we're tagging this. This. We would give some number of tags. We would give, you know, the title of the piece, probably, and, you know, this tic tac toe thing and. Then. And then this. I mean, yeah, it'd be good to actually see this process. But let's. First of all, let's just get a scope for what's out there. So there's, there's, you know, there's MathWorld, there's NKS, there's all my writing stuff. Then another big area, which, again, God knows whether anybody's bothered to make use of, is our demonstrations project, which is. So this is about how many is it 12,000 plus? It says, you know, I'm sure you've seen this. I'm not going to pick control theory again because I randomly picked that and it's an area I know very little about, so it would be a bad one for me to pick.

I would love to include these in articles. These are great. Now my recollection is these are all under an NCSA license, so non commercial only, which makes sense.

So maybe we can fix that. So one of the issues is these were often contributed to us by other people and now, and so we have certain obligations to them. But. So let's see, where is the license information? Well, here, CC by ncsa. What does that mean?

Non commercial use only.

Okay.

That's good for Wikipedia, right, Sam? No, never.

Unfortunately.

Why do I see it sometimes?

I don't know. Some Wikipedias will use things that are. They'll basically use them as fair use.

But, well, so what should that say to make this usable?

CC by essay.

And what does that mean? So, I mean, let me give you an example.

Yeah.

Of what? Okay, so then, then on Wikipedia, I mean, okay, first step is people should surely just link to these pages, which requires no incorporation. So if they haven't done that, it's bad Pavlovian conditioning for us to say, let's do something great for Wikipedia with these pages. If nobody cared enough to link to them, that's a bad sign. So I don't know whether people did. I'm sure you can look it up. But you know, so that would be, the first step is linking to these pages, right? Now the second step would be using, I mean, you know, use a, use an image like this, right? Use a static image from the page. Right? Because, because to do more than that, you have to, you have to embed this page, right? So, so here there's, there's some kind of embedding mechanism. Okay, there it is. There's some embedding mechanism.

Yeah.

So what, what would be the preferred way to, to do this? So that, that seems to be a fancy embedding mechanism, I think. Yeah, go ahead.

It would be, it would be the best user experience if people had an embed while they were reading something and didn't have to shift pages. That would require someone probably on the Wikipedia side to write, to write a gadget for this that did the embed in a CSS friendly way that embedded seamlessly into a media wiki page. But starting with a screenshot is certainly a lovely, a lovely first step.

And maybe, presumably, presumably what could be Done. I mean, for lots of these pages, just having one static case of, you know, if I say copy the clipboard of this thing and I go and I am just paste it in here, I will get that thing. Isn't this amazing that you can just do that and I can be, you know, just using it right there.

Right.

Isn't that cool? And it also, it will probably also work directly in the cloud here, but it will be slower. Yeah, I mean, I'm sure this, I'm sure this. Yeah, there we go. This, this is just in my web browser, but my web browser is kind of slow and, and it's having to do server transactions every time I move something here. But so, I mean, you know, one possibility which might, by the way, have a different license for us and for the original creators is, is, you know, take adjust, you know, it to the configuration you want and then take that screenshot and use that screenshot. Yeah, I mean, so, you know, that's a. But, but. So for example, you know, we could probably, you know, expand this. See, I don't know whether. Well, again, I don't know what the upstream commitments that we've made to our contributors are. My guess is that they'd all be fine with this, you know, if it links back to this page and it links to their, their, them as the source of it and so on, I think everybody will be fine. I mean, they, they know that they're signing up for something which is intended to be a public resource, I think So, I mean, in terms of the embedding. Well, yeah, somebody can absolutely take this embed code and figure out how to make that work in MediaWiki. I don't know whether that's. It may be straightforward. I mean, I'm sure if I take that piece of code, I'm not an expert in. Okay, so this is. Okay. I don't know whether you can, I don't know whether you can decode this, but cross origin script. I don't know exactly what this is doing, but that's. Yeah, I don't know what this is doing, so I can't, can't opine maybe. Do you know this stuff to the point where you can.

It looks like it's just calling other JavaScript that you have.

Okay.

Tell you how you're interacting with it.

Okay, but I mean, so, so the question is that will be the thing that somebody would have to make a media wiki version of. Is that right? Yeah, and. Well, somebody should do that. And so you're suggesting here a button that takes a snapshot. Well, I think it's a button that takes, I mean, that, that's, you know, that software engineering work. You guys have got to show me that enough people are going to care that we should do it. Because if I find that, you know, if I go look on Wikipedia and find that only three people have ever cited 12,000 demonstrations that we have out there, it's a big, It's a big negative of a. Like no way do I get to allocate software engineering resources to support.

Right. There's a. I guess the legal version, the legal. Only version would be to add a clause to the licenses that says you, the viewer can take a screenshot of, of, you know, this image in this configuration and use it under, under.

I think, I think the point is maybe what we should do is say the images.

Right.

See, I think the problem with this is, you know what the problem is? The problem is there are educational publishing companies that both license some of this stuff from us and want this stuff from us.

Right.

And you know, by the time somebody's charging 200 for a chemistry textbook. Yeah, I don't really think it's, it makes a lot of sense to say, you know, we're going to, you know, give away all that work, so to speak. So.

And they want the interactive.

What's that?

They want the interactive that the students.

I don't know, I assume they want the interactive part. And I, you know, and I think Wikipedia, honestly, the interactive part seems like a bunch of heavy lifting. And it seems like the screenshot is the part that's probably of highest immediate interest. And the screenshot, it's for sure these screenshots here would be, you know, we could absolutely put a different license for those screenshots, which. That's an easy thing to do. And I think that that will be a fine thing. I mean, and that, that's, you know, and we can, you know, I don't know whether people, people might want to take this part off, which would be okay. But. Yeah, I mean, to put a separate line that says underneath here, this content is, is, you know, whatever it is, you know, whatever that doesn't have that NC part to it. That, that seems to me to make perfect sense and to be, be fairly, you know, a straightforward thing to do now. Okay, so then there's demonstrations. That's 12,000 of these demonstrations. And I'm, I'm not even going to. Maybe I'd be too horrified to go look at how many people have, have used them already. I mean, they've been out for, what, since 2005, I think maybe even earlier than that. Maybe maybe 2006. But. So it's. So if they haven't been used, it's kind of like, why. So, okay, so let's.

And you have data on the most popular.

Me? Are you asking me?

I was asking Stephen, but YouTube.

No, I mean, of. Which ones of these are popular? Oh, sure. I mean, we could find that data, but, I mean, it's not hard. I. You know, what I would do is the popularity of these on our website will not necessarily be related to their utility for Wikipedia. I mean, you know, I have no idea which of these, for one reason or another, you know, have somebody linked to them because, you know, who knows what you know it. And quite a lot of these are. So a pretty common thing is for people to. Let's see, what is this? Okay, we're even mentioned on. Oh, for God's frigging sake. See, this is the kind of thing. This is why I shouldn't open these pages.

My page was deleted at some point, and it has definitely improved my mental health.

What's that?

My page was deleted at some point, and it has definitely improved my mental health. Seeing people arguing over pages.

Well, so, I mean, you shouldn't show me this because this is like no software engineering resources. Sorry.

Right. Well, this is a classic Wikipedia problem with all partners, actually.

Right. I mean, so and so often the.

Challenge is the easiest to find text about an organization was produced at some point by a PR group. And so when people summarize it for an article, they're summarizing pr. And there's just. There's an immune response because there are also paid PR companies that are going and sort of subversively trying to edit things in favor of clients. So.

Yeah, yeah. I mean, look, when I look at people's Wikipedia pages, for a slightly obscure academic, I would eat my hat if the page wasn't written by the person themselves. Right. Because it contains information nobody else could have, nobody else would care about. It's, you know, and it seems to me it's a perfectly useful thing. You know, I want to know about this person. It's a page which maybe some other people have looked at. And so it doesn't contain pure nonsense, but it's perfectly, you know, it's like. And nobody else cares enough to do it. And so, you know, seems now, you know, people may say, oh, but it's shocking, but, I mean, it seems useful to me. I mean, it's not. I'm. I'M not signing up to do that, but. And I'm less obscure than the people whose pages have that characteristic, so to speak. For better or worse. I don't know whether it's a good thing to be less obscure, but I am less obscure. But anyway, I mean, coming back to, okay, let's see what's on the foraging list, so to speak, but can I.

Let me highlight the strengths of the wiki network. People often think about the article on a topic as the way the topic gets represented, but often the category of things from that topic is more interesting. And so the category of media, especially where something really is a primary source, it's less that, like the, the interesting thing is to find all the instances of things from that source. Right. So media, media from an educational project or topics, you know, topics covered by Khan Academy. I think the article at Khan Academy is probably bad, but somewhere there's a really interesting category which is. Drawn from their ontology of all the things. All the things that you can learn.

Yeah, I mean, obviously, look, the new gorilla on the block is LLMs. And LLMs are pretty good at pulling together disparate information from across the web. You know, I find myself actually, I had a use case just yesterday where I actually got sort of primary information from LLM. I was, I wanted to know what the correct description of words like styled and some other past, you know, past participle, ish type words were. And I didn't know whether it's past tense, past participle, you know, what is it? So I asked GPT4, I think, and it did a really nice job. And it was, you know, it was a really nice job summarizing kind of things that it effectively had foraged from very disparate sources. And I kind of think that that's, you know, maybe the future is. You know, everybody gets a page created, you know, on the fly, so to speak. And it, you know, and the whole, you know, somebody with their pitchfork is saying this and somebody else with their pitchfork is saying that, and it doesn't really matter because some AI is going to make some totally arbitrary and often completely wrong decision. And, and every time you press the regenerate button, it's going to give a different one. But, but I mean, just to. Okay, so look, here's just in terms of other sources of things, the. Yeah, no, I. Sam, I wasn't looking for that specific word. I was looking for the category of words that were past tense, like words. And let's see, I'm, I think I, I recorded what I had there, so I can. I could remember what. Where did I have it? Where did I put it? Where would I put this? You see, I record everything, so it's certainly recorded in. In screenshots that my machine automatically takes, but I'm not sure whether I can immediately pull it up. But it was sort of an interesting case because it was one of these things where there wasn't a head word that I could go look at and find the information because that's the. You know, and. But so, you know, another source for me, if somebody cared, is, you know, there are transcripts from, for example, science and technology, Q and A. So there are explanations of God knows what. I mean, well, you can see the word clouds here. Of things. I don't know what I was discussing there. Why is it important that the periodic table is structured as a table rather than a list of elements? Okay, so there'll be some long transcript of me answering that question. Some of this stuff is probably quite useful because it's probably questions that there isn't really a great other source for. And who knows whether I did a good job in real time. I think I often do a decent job in real time with these. With these kinds of things. So this is another. I mean, this, I think is on the order of 50 million words. Maybe something like that. Some. Some immense amount of material. Material here that, that somebody could mine. It's more work to mine than I'm sure we've got. Oh, this is actually going to go to the YouTube. No, that's not what I want help. Not what I want.

Q and A material is. Doesn't really have a home right now in the. In Wikipedia sibling projects. But there's a. There's a. A live demo of what a wiki answers could look like.

It's an interesting idea. It's been. Oh, wow. From AI systems. Well, at least in my current configuration, I'm not an AI system. I'm just a. Just a poor human. And this is. Oh, yeah, let's see. This was the. See, this was the thing that. This was my actual example here of maybe. How do I get this? That's a good.

So any. There's a lot of spontaneous beauty that comes up in these visualizations, makes them very satisfying to.

Yeah, so here, this is the thing I asked. I asked yesterday what grammatical format and this was a really sloppy asking. I'm embarrassed. It's so sloppy. What grammatical form are translated and styled. So it says past participle of the verb whatever and this. And then I was saying cluelessly because I should know this stuff is a past participle, an example of something in past tense. Yes, a past participle. Blah, blah, blah, blah, blah. So that was useful. I was writing something and I needed this. This fact. And I'm not sure I could have easily found that by, you know, I could have perhaps by reading headwords of, you know, past tense and this and that and the other. But this was a, you know, was an example where kind of the, the aggregate wisdom of the AI was useful. Let's see. So. So I mean, I'm looking at. I'm trying to think about other places. Oh yeah, here's another. Well, there's another thing that is just. Now let's see what other things are there? There's. There's also. I mean there's a data repository which would be useful for. Again, it's. It's the kind of thing where these are just cleaned data sets about all kinds of random stuff. And I think many of these look quite special purpose, but they're probably a page. What's that?

That's great. Now I want to browse Danish fire insurance claims.

Yeah, you see but the thing that's neat about this will be if I pick this up, right. I copy it. I mean this is now our tech. If I pick this up, I can paste it in here and I should get, you know, there's the actual time series and I can just say, you know, yeah, we're, we're alive and you know, we can do things with it and we can compute. I'm sure if I did, you know, moving average of that. Well, let's just take moving average of that time series with. Ten steps or something. That should give us a more. A smoother version of that. But anyway, I mean those are just. But you know, that seems a bit specialized. Let's see what things am I forgetting that are really big chunks. Oh yeah, here's an example of something. This is an interesting case. This is a project that we did.

For.

Museum of Math that I've been involved in since it started. And this was an attempt. This was sort of the beginnings of getting sort of an inventory of things that one could get for a physical gallery. So this is just Now I wonder what the. Okay, so this is various images and so on and Good question. What licenses? Oh, okay, so then there's a whole explanation of this. So this is again, this is good material. I mean this, this would be, I'm sure if there isn't an article about Turing Zeta Function Machine There should be one. If there is one, it would surely be useful to have these kinds of things in it. Now let's see, what, what does it say? Okay, so it's got a whole description about. This thing. Wow, this is a long one. Wow, it's a really extensive story here. And. Okay, so that just says copyright, blah, blah, blah. See legal and privacy policy. What does that say? That's not very interesting. So this is an example of one which is currently not Wikipedia friendly from what I can see, but it could be made. So I mean, I, I think what is about the site, I don't think, I mean, I, I don't know. Looks like, looks like this is simply our copyright now. Okay, so here there are image credits. So some of these images are probably. Some of them may be freely uploadable. The ones that we've made are presumably freely uploadable, but others are not. But so that's a. So I mean, you know, this looks like a thing that has a lot of content that should be Wikipedia friendly content. Now, you know, I, I would assume that in your, you know, don't put original research. Well, any piece of history that anybody writes is an original representation of that history. So it's kind of like, like that can't be a correct. You know, so it's like unless you exclude all historical material that doesn't make sense.

You include. Wikipedia tries to hone in on being a tertiary source that summarizes the perspectives of a range of secondary sources.

Fair enough. But. Okay, so I mean, I would say that, you know, if you look at. Let's, let's, let's look for example, at. So I mean this, for example, this I think would be considered a secondary source. That. Fair, Fair assessment, Yes. I mean this is not. We didn't. I was really hoping that we would get to do some kind of Indiana Jones style, you know, finding mathematics artifacts in the, in, in, you know, in jungles and things like this. But although there were a few small cases of that even one in South America, but not in the jungle. It was mostly. This is very much tamer kinds of kinds of sources. And this is, This I suppose is the. So I mean, that's kind of secondary source type type material. So then. And again, I don't know with these, with these pictures, I have no idea what our image, you know, what the. I'm sure we give somewhere at the bottom here. I'm sure we give. Yeah, we got a lot of material here.

Wow.

Yeah, it's interesting. So probably anything where we do not specifically cite a source for the image. It's a deeply public domain thing because these are mostly from books that were published in the 1700s type Thing. And I wonder what happens. It doesn't look like we have any. It doesn't. It doesn't click to anything. So I'm assuming that this is a. Ah, this. This one does have an alt tag. Cataldi's divisor table. Well, so let's see what you're suggesting. You're suggesting things.

Sam, Annie, what do you think of this potential workflow?

Okay.

What the workflow is. I copy your workflow out so that I can put it somewhere. Let's. Let's copy it. And for people.

Working from the easiest things to add to the potentially most complicated, you have all of these neatly organized documents that identify some, some concepts. If you extract the concepts, you could see whether there's a wikidata entity for it. If not, you can make one with.

With these articles we have, we use. Okay, so we have a wikidata data.

Function. Yeah, Library.

Right. Well, it's not a library. It's just a built in function. I mean, so if I say, if I go and look at our entity for Alan Turing. Okay. Oh, that's interesting. What does that. Why does that not do what I. What did it do? Why does it not do that? Oh, it needs to have a property. Yes, it needs to have a property. So it needs to say something like. I don't know whether this is that something that's going to work. No. Well, let's take an example. Let's take an example here. So that's using wikidata entities. God knows whether it's correct. Let's see. Mass of the moon. I mean we've. Okay. And that divided by line 14. Well, we're close. Maybe some part of the moon fell off between. But so I mean, I think this is okay, right? This is. I'm not an expert on how this works. This is what we call an external identifier. So this is using the. Let's see, how do we get this? I think we say. Okay, so if I say for example, have an example here of take how in fact we should be able to do this here. This should give us the. Here we go. So this is now showing the wikidata information about all these things. So now presumably if I take that wikidata. Oops, what did it just do? Oh, if I take that wikidata item, I say copy that. I would guess that we can get that we have a way to get the entity that is that in our system. How do I do that? I think. Oh, yeah. Actually, that's an important point. Right. So we. We have. Yeah, so we have a wikidata search mechanism. So if I say wikidata search for, let's say, Alan Turing. Oh, my. Who's the real one challenging. Oh, that's a sculpture. Okay, let's try that one. Let's see what we have. Let's try that. Yeah. So, I mean, we have some knowledge of what's here. And I would say that our. We've not been particularly enthralled with our ability to use this data because it's mostly really messy compared to stuff we already have. But it's a nice way to cross check lots of kinds of things.

Please clean it up. I mean, that's another easy way to.

Yeah, but that's not. Why would we do that? Why we do that. We spend huge amounts of money in software engineering to build our own sort of curated source of data. And that's not. That's not a reasonable ask.

You know, slice off the power law. Slice off the power law of the things you care about. And if you run into something outside the power law, that's in. You know, it's a lot.

What we have done, I think we have put some of our identifiers into wikidata. Right. Because that's what. That's what you need to be able to link up, you know, so, for example, the fact that there is an Alan Turing here, that is. That is, you know, our. That is the actual person. That is our, you know, Alan Turing person. And I, I'm really confused by the fact that I'm not able to. Well, I wonder what happens if we just say image here. I wonder whether that's going to. No, it just says missing, not available. There's a way to get our corresponding entity from this, because what. What's there right now? I mean, if I just say, what. What's there now? I mean, look here. Here's the basic story, you know, from a philanthropic point of view. You know, we're putting a huge amount of information out there in the world, pictures, et cetera, et cetera, et cetera, which, you know, we and I are doing. Well, maybe as a side effect of me writing things, so I actually understand them myself or whatever else. In some cases, they're really just for kind of knowledge, philanthropy, so to speak, and anything where we can kind of with small, you know, if it's just a question of flip this switch and it makes things easier for Wikipedia. Great. If Wikipedia throws shit in our face as a result of doing that, then obviously we say, well, we shouldn't be doing this. And that's kind of the. But you know, things which are easy to do, like changing footers on pages and things that's completely fair game. And it may. And I, you know, those, those are really useful suggestions. And as I say, you know, from a philanthropy point of view, it's a great thing for us to, for us to do things that require real serious effort and software engineering and so on. Pretty hard to justify. And I mean, in a. And I think, you know, we're sort of at a. So I mean, I think that that's. But, but let's see. Did. Did we. Did Annie. Was Annie about to make some great suggestion which got cut off here?

No, I wasn't. I wish I were.

Okay, okay, fair enough. So, I mean, it sounds like there are, you know, uploading a bunch of pictures to Wiki Wikimedia Commons, changing some footers. I think those are things that make perfect sense and we'll do them. That sounds like a good thing.

And for everything that's out there for one of these educational portals, those summaries and updates should make their way back into articles eventually. But a nice way to start, and this doesn't have to be done by you, once it's under a suitable license, it can be done by. By anyone with a topic extractor is to create entities for things that don't have articles yet. And then there's an entity. Then when you upload an image about it, the image commons likes to have links to related wikidata entities. So you've already made the entity with a thing that's depicted in or being visualized by the image. And then when someone goes to start an article, they already have all of this context.

Well, that's something. I don't know how easy that is to do. Maybe our wikidata search, if Wikidata has good. I mean, I believe we have the technology to easily create a sparkle wikidata thing because so insofar as we can, you know, to get the things up to our standards of canonical entities and so on, that's a whole different story. To do something up to wikidata standards of canonical entities is probably much easier. I don't know whether that's a. I mean, you know, in modern times we might just ask an LLM pick out, you know, pick out the, the keywords in fact, I think we even have. I can Let me find this in our new prompt repository that just went live. There is a thing in fact we should. Maybe we should have. Somebody should put in some Wikipedia related prompts here, I'm pretty sure. So like there's a glossary generate prompt which takes a piece of text and asks for a glossary. And this is all right, it's programmatically accessible so you can just use LLM resource function of glossary generate of whatever text it is. And it will. It will generate the glossary now I think also so we could easily ask it for. In fact, I'm pretty sure there is one here for. Oh, maybe it doesn't. Oh, index terms suggest. There we go. So okay, so here we go, right? This is suggesting index terms for. I bet that's actually something I wrote about Ada Lovelace. Hey, that's pretty good. Well decent at least. I mean that's probably good enough for a. For a. Here are some tags type. Type thing. In other words, if one feeds the LLM a paragraph of text. Yeah, actually this is a good idea. So feed the LLM a paragraph of text that describes the image which obviously we have in the case of any of those images I'm showing you because they are images that were part of a piece of narrative text. So we can easily put those tags into the Wikimedia Commons kind of. You know, there's presumably a place to put tags there that a true statement. The. So what what happens?

I mean there's a folksonomy of categories, so you can. You can search for matching categories or create new ones.

Okay, but is that. Is that. That's presumably what our Wikidata data search. What our Wikidata search is doing in Wikidata at least.

Possibly.

Now another source, by the way, another whole different source of this stuff is our documentation which increasingly has. I don't know, I'm going to pick a random page and I have no idea whether it's going to have anything interesting on it. But I bet if I pick try and pick that that looks like a lose. That one doesn't have anything interesting. But quite a few of these pages have. Have big kind of narrative stories about. About these things. And again, you know, I have no idea which of these will have that. I actually, I suspect the Graph Theory team has been bad at putting these in. Let's try another one. Let's try. Let's try something like this. Let's try some. Let's try that guy, for example. Wow. That's lame. The. Yeah, let's see. There's lots of stuff. Okay, this is. Yeah, this is all stuff about this. Actual data about people. That's a different. That's a slightly different story. What I'm. What I'm looking for is in. I'm surprised here that I'm not seeing more in this. But. But many of these stories about some particular obscure mathematical function have a whole section that is basically background information which somebody could lift. And I mean, these ones don't seem to be very big on that. I think these functions are not particularly ornate. But again, I mean, you know, I think. I don't know, I wonder what the. What the. Yeah, this is. Okay, so again, these pages are, you know, that's something that. I guess we do like that. Yeah. So, I mean, these, These pages seem to be. Seem to be not. Not directly usable. Maybe they should be. I'm not sure. And I mean, I think. Yeah, so there's a whole other. Let's see, what's another. Another thing. Another big area that I know very little about is our community website. And this has. I'm not the best navigator of this, But this is. How do I find. I have no idea what this is. This is. This is a. A thing contributed by some user of ours that. I don't know whether that. I mean, what does it. What does. What does it take for somebody. For somebody who contributes this. Let's see, where does it say what. What somebody can do here it says, cite this notebook this way. What does it have? Doesn't seem to have any. So the. The whole page has a copyright. But what does. What. What should this have? So this is people contributing material. They're contributing sort of essays and so on. What should this have so that somebody can pick it up and use it in Wikipedia?

If that had a clear license, a CCSA license, that would be easy. And also if the initial contributor is part of the contribution process, that also makes it obvious that there's no license obstacle.

Right. But so can this person, whoever this is, you know, Ms. Az, whoever she is, you know, if this person goes to Wikipedia and says, hey, look, I got this cool picture I made here it is. It's this exoplanet discoveries per year. Seems like a relevant picture. And they sort of say, I have rights to this picture. They upload it to Wikipedia, they link back to this article. Is that considered fair game? Or is that not considered fair game? Or do they have to get a friend of theirs to do it? How does it work.

Annie? You Want to field this one?

I'm so sorry. I was talking to somebody in the chat about. About a meta project.

Oh, great. So I think.

What I'm asking is.

They have to make sure that they.

Have the right person who. Okay, apparently a high school student posted two days ago, you know, this particular picture, and maybe she wants to just put this on Wikipedia because it fits in some article about exoplanets. Okay, and what should be done? Should. Should she just go and do that? Assume she has the rights to this picture, which she does. You know, it's like she can go log in and just do that. Or is somebody. And then. And then she'll link back to this piece she wrote here. Now, is somebody going to say, oh, you can't do that because you're linking to your own piece? Or how does that work? I mean, this is clearly a person.

The image to Commons, as long as it's under the right license, that's very welcome.

Right, but then what should the person do? I mean, so let's say this fits on a Wikipedia page. What should they do?

They could insert it into the page themselves. I'm pretty sure that's fine. Right. They could also, if they're not sure where it fits, they could just do a bunch of tags.

Yeah, but then nobody's ever going to find it, right? I mean, that's just a waste of time. Right? So, I mean, if one says to somebody like this, you know, you've done something nicely, you've made a nice picture, go put it on Wikipedia, you know, that say. And, you know, and then it, you know, that seems like a thing that should make sense. I mean, in principle, one could get me to do that. I mean, in principle, when I generate hundreds of images for some blog I wrote, in principle, the world could get me to go to Wikipedia and go and add those images in relevant places. In practice, with me in particular, that's very unlikely to work because it's just like my personal calculation is I'm going to do things where I just put them out in the world and people can do with them what they want, rather than, you know, I'm going to fight tooth and nail for this particular thing. Not my, you know, it doesn't. That's not a sort of a personal calculation, so to speak.

But so a small change that would make a big difference to encouraging people to post on Wikipedia could be identifying articles that are. That are relevant and inviting them to suggest adding an image that they made on the top page of that article. Wikipedia just operates on its own timescale. So they should be ready for what could be a relatively slow timescale of talking to people who visit that article on and actually read the talk. But that's probably more satisfying than trying to add it into the article when they don't quite know how and then getting some pushback that feels like it wasn't wanted when the safe option is.

To put it in the talk page. But I feel like you can just put it in the article and if someone gets mad, then you can discuss with them.

That's what. So, so that's what I do too. If there's something to make it satisfying, we should have a way for curators want to help people who are doing this to find the family of, of edits that were made that way. Because often people will make a change, it'll last for a while, and when they come back, maybe a couple months later, it's not there anymore and they're not quite sure why. Maybe someone else had a different opinion, maybe someone thought they had a better image that could set up, that could fit that goal and the image is still there, categorized appropriately on commons. But that might be a little bit confusing to them. So anything that would make that process.

Feel, yeah, I think you can probably assume that this high school student, whatever, might be an enthusiastic, yes, I'm going to put my image here. I made a nice image. It fits. It's a good thing. And, or it might be, you know, we have, I mean, I, I happen to click on a page that's a high school student. We also have, you know, the, the acknowledged world experts on this and that also, you know, contribute to our community. So could be an acknowledged world expert on this or that rather than a high school student. But that's, that doesn't, doesn't really matter. But any of those people, if they go and add this picture to Wikipedia and people kind of, you know, you know, call them names and, and say, go away, you'll never have them again. I mean, that, that's, you know, and that's, that's kind of the, that's sort of the, the calculation. And so if there's a, I mean, you know, if there's a thing, I mean, I don't know whether we're. The, you know, if, if, You know, if there's a thing to tell people that, oh, go do this. I mean, you're basically saying, go, put it on the page. But then, then you have to be prepared to go, I don't know, you know, fight in the Talk trenches or something thing. And, you know, is there. Is there a way. I mean, if, you know, is it the case that the way Wikipedia now is, is a place where everything is fought? If that's the case, then. Then that's okay. You're. You're shaking your head. That's a good sign that that's only.

On controversial pages, really. I think Annie's right. For most pages.

Yeah, just clean it up.

Just. Yeah, just.

It might get removed, but often the person that removes it will have a reason that's at least understandable. And maybe the person that's removing it is right. And usually people are, like, really reasonable. I don't know. I think that you should tell these people to put them on and then see what happens.

Fair enough.

And don't freak out if there's feedback. If there's feedback from someone else who cares about the presentation of the page, that's nice. It's someone who you can collaborate with too.

Right. But I mean, you know, for us, like, we can say, we could put a post on our community saying, hey, we talked to these Wikipedia people and they said, you know, we've changed the footer here so that it's Wikipedia friendly. And we could say, hey, if you're posting things here, you know, think about putting interesting pictures onto Wikipedia. Right. Is that a sensible thing to do? Or are people going to say pitchfork? Pitchfork? You can't tell people to post things on Wikipedia.

I would say tell people to post things on Commons, which welcomes it and exists for this, and then to start conversations on the appropriate articles to see where the image would make a good illustration. Because they're separate. The image is always going to be useful in the context of this original post. And then there is a more nuanced discussion about which articles it might illustrate well, and where it goes in the article and that actually.

Yeah, right. But realistically, the people who are writing about something are the people who know, you know, they're going to know where this fits, and they're either going to put it there and, you know, maybe somebody will say, you know, maybe this high school student doesn't know this or that thing, and they might put it in the wrong place. And that could happen. It's much less likely to happen. I mean, you know, if I were editing Wikipedia, I think it is much less likely. I'm much less likely to make a mistake, so to speak. In other words, I'm, you know, I. I kind of know what I'm talking about. Usually when I. When I think I Know what I'm talking about? I typically do know what I'm talking about. So I'm much less likely to make a mistake. But you know, other people and it's, it's not. Not. I don't know this person. So I mean, I have no idea this person. I'm, I sound like I'm, I'm, I'm down ranking this person, but I'm not, I'm just whatever. I think the. But I think, you know, anything that's too long, you know, too long a path to gratification. I don't think we're going to be able to get anybody to do. I mean, I think, you know, some auto upload to Wikimedia Commons that's possible. Something that involves a human in the loop. I think they've got to, you know, the human has got to see some payback of gratification, of philanthropic warm fuzzy feeling or whatever else it is in finite time. Others not going to do it. And well, let's see, we've gone a long time here and we've got lots of interesting. Sam, you're saying one maximum don't edit in your area of expertise because the sense of somewhat hyper focused on an area about where and how prominently a fact should go may be quite different in the sense of a general editor. Okay. Is that really a formula for getting good information that sounds like you're getting bad information?

If you're a newcomer who's tentative about editing and you might only make a handful of edits and you have a very particular opinion, like you very much want a particular statement to be in the article, or you very much want this image to stay in the article. That is a recipe for unhappiness.

Right. Well, I, you know, speak to what, what would happen. I mean, you know, I would say on, you know, the truth is, you know, for somebody like me, Wikipedia lost me as a. I mean I never was a contributor but, but you know, the stories I hear make me say no way, I'm not doing that. I've got enough things to do, thank you very much. This is not a good thing for me to be doing.

The stories that take all the, like people spend all their time in like their, their memory of their experience tends to focus on areas of drama. You get drama where there's contention. You get contention where multiple people with different perspectives all care about how things. Something is presented and everyone thinks they're right and you know, it inevitably ends up on a wrong version that perhaps no one is happy with. But most Topics don't fall into that category. The long tail of human knowledge.

Well, I think the thing to start with is the fact that there are only five links to all the stuff I've written is very disappointing to me. It's very personally disappointing to me. I mean, it kind of is a. Is a. You know, even though I know, you know, these things have been read millions and millions of times, it seems wrong, and I kind of, you know, and it seems like a big, you know, I shouldn't care about this type thing. I mean, I'm. I'm talking to you guys because I wonder. And I didn't know that number, so you told me that number, so I, I was not aware of that number. And so I guess my question is, you know, if. If I mean, my only ax to grind with respect to most of this material is I bothered to write it. You know, the more people that read it, the better because, you know, I bothered to write it type thing. But beyond that, it doesn't really, you know, my, you know, I don't think I have a. A deeply, you know, politically polarized view of the history of, of the second world of thermodynamics, for example. You know, this is just, you know, a bunch of historical work that I did was a fair amount of work to try and untangle these things. And I doubt, you know. Well, I know because lots of people have read this and people have commented on it. I don't think anybody has yet said, oh, you got that horribly wrong. In fact, rather to the contrary in this particular case. So, you know, it's all right. Do I have the nerve to click my own Wikipedia page? I think I probably don't. Should I do it or should I not do it?

I certainly avoid it for people I know.

Okay, I'll avoid it. Well, it will be good, you know, the more accurate it is, probably the better. But I'll. I won't do it. For now, the. Sandra on our live stream is commenting for very, very, very specific things. He said he's had good experiences editing Wikipedia, but for more popular pages, it always got reverted by some. He says, quotes super Wiki editors because it is their personal thing, or so I won't necessarily interpret that, but.

The more popular a topic is, the more likely there are other people who care about it and how it's presented. The wiki as a whole is not paper. Go ahead.

Oh, if you have a lot of Wikipedia edits, like, if you've done your time, people trust you a lot. When you have like 20 Wikipedia edits. People assume you're going to vandalize. So that's like, you know, I'm. That's. That's people's own fault. They say they're open to everyone, but in reality, it's. It's really hard to make meaningful changes on controversial topics. You don't have, like, a really powerful argument and the willingness to defend it.

Yeah, well, fair enough. I mean, that sounds perfectly sensible. I mean, that sounds rational. You know, it well. Okay, what else should we cover here? We've been yakking for a while and a number of good suggestions and a number of things where I'm not convinced, but. What else should we do to make the world a better place, so to speak? That's a rumbling sound. I don't know what that might be. The. That might be the answer. I think maybe we need to cue the birds again.

Do you have any favorite articles?

Say it again.

Do you have any favorite articles? Any goofy ones that you like?

I don't know. I'm thinking about that.

There's a list of sexually active popes. There's. I just. I'm pretty known.

There's a lot of, you know, the things that I like tend to be these sort of collective folk information things like the thing you just mentioned. Right. I mean, not that that particular one would be my. My particular speed, but. But the, The. These things where somebody contributed or a collection of people have kind of contributed folk information in a collective way. That, that to me seems like a good. A good use of things. I mean, I think also to some extent, I like the places where there's sort of a basic summary of something I have to say in more technical articles on more technical topics. I think Wikipedia very often defeats itself because somebody came in and, you know, there's some math topic or something, and they. I. I can imagine somebody saying, oh, but that's not quite right. We should say it more precisely in this way and that way. And the result is that even somebody like me can't really understand it quickly. I mean, you know, it's. And I'm like, yeah, I'm gonna go somewhere else. But I think these. I have to say that, You know, in. In this kind of a. A pithy summary of things. You know, I. I write lots of pithy summaries of things for different reasons myself, and there are a lot in the notes to the NKS book and things like that. So I'm a complex consumer of pithy summaries of things. And I would say that Wikipedia, I don't know, a quarter of the time I'm like, hey, that's a pretty good pithy summary. Three quarters of the time I'm like, it's not so great. Usually the problem is that there's kind of a, A, a core point and people didn't quite hit the nail on the head. They kind of, maybe because they were dancing around the nail, so to speak, because, you know, the nail was something that, which required a little bit more intellectual heavy lifting which the people who were, you know, non experts in that thing weren't going to be able to do. And the experts were too, too kind of focused in on some micro detail and so it didn't happen. But I think this kind of the folk information thing that's good. The thing that we find because we're always looking for data that we can kind of make computable is the thing we tend to find is there's a certain amount of, for example, linguistic folk information. Like what is the name by which this is known? That one gets very nicely from Wikipedia. The, the thing that, you know, things. Systematic data. No, it's, it's usually a very bad source for that because what's happened I think typically is somebody, you know, somebody said Great, we found 500 of these things, let's upload all of them. And then over the course of the next five years or something, people said, oh, that one isn't quite right. You know, that one's just in my backyard, I can change that one one. Then they change another one, they change another one and pretty soon the consistency that might have existed in the original source is all gone. And that's, you know, that's a, which is, you know, it's just like different things are good for different purposes. And I would say that, you know, in now that that kind of foraged folk information, I have to say I think that the LLMs are going to be really pretty good at that. So I don't know what the, you know, and I don't know what actually here's a question. So let's say I ask an LLM for like here's an example. What colors were popular in fashion in such and such a year? Okay, now maybe there's a page that lists that maybe there's a single source or maybe that's a kind of information that is best kind of foraged collectively from the, the web. And so the question would be, let's say you ask an LLM that question and it gives you an answer. I wonder. I'm just a curiosity Question, kind of the, you know, what's the kind of. Okay, so can somebody cite an LLM and say, this is my source, or is that considered? Or has that not yet come up?

Well, there have been a lot of discussions. No, you can't do that.

Okay.

Do you want to.

Humans only. Okay, so wait a minute, wait a minute, wait a minute. I have to get this straight. So a human who is the actual primary source of something is not okay? An LLM who is not a human is not okay?

Correct.

So what's okay is a human who is what? So it can't be a primary source. It has to be somebody who read the primary source and then is. Is writing the thing as a secondary. As a. I'm. I'm confused here. Help me out.

For a lot of science, like, I don't know, I edit a lot of like, biology articles. I don't edit a lot of math articles. But yeah, they like, they prefer review articles sometimes. If you're talking about personal statements like, hey, I got divorced, if you tweet that, you can cite the tweet, but if you tweet, hey, I don't know, the sky is blue, the sky is like pink, you obviously can't cite a tweet for something like that. I will say this is. So this is a little bit different, but I was at the Wikimedia Hackathon this weekend and my friend Tony, he did something that you're gonna like. He made a. A thing, a section summarization tool. I'll put it in the little. Here, I'll put it in our zoom chat here. But you can add this to your Wikipedia customizations page. And what it does is it takes the sections of Wikipedia and then it will use an LLM to make a. A little summary of the section.

Right? Yeah, right. No, that's what I mean. Summarization prompt in our prompt repository will do the same thing. Yeah, right. This is. It's a good thing. And it works surprisingly well.

And there's a. Funding from.

Okay.

And just another AI thing if you're interested, but there was a talk like about a year ago, Meta announced this project, I forgot the name of it, where they were gonna like check the Wikipedia sources to see if the source is saying what Wikipedia says it says, if that makes sense, to make sure that we could.

Did they do it?

I don't think it's in use. I haven't heard about it in about a year.

Yeah, that would be a good use of LLMs. That would work nicely.

Yeah, I Never saw it. Enterproduction. That'd be amazing, right?

That, that, that seems highly useful.

It's about 100 million calls. So what's that? It's. It would be about 100 million calls to their function.

Yeah, right. But, but I mean, Let's see. In. In. Yeah, I mean, so, so here's, Here's a meta question. Talking of meta, but not that matter. The. I mean, you know, clearly I personally, and we as a company have spent lots of effort kind of putting out knowledge into the world and so on, and I think the Wikipedia community is kind of interested in the same things. So my question would be, you know, how, you know, what's the best way to be friends, so to speak? I mean, we have a certain, you know, we have certain resources, we have certain unique capabilities. We have. And, you know, it seems like Wikipedia has not made good use of what we have, except maybe for math world, and, but only. And so the question would be, you know, I think our goals for the world are fairly aligned. How can we, you know, what can we do? I mean, you know, from our side of this, there's things we can't do because that's, you know, on the Wikipedia side. But, you know, what, what, what can we do to be kind of to be better friends? So I don't think we're friends at all, actually. But I mean, what can we do to be better friends?

You know, Wikipedia is not a single organism. It's a. It's a colony of organisms, lots of organelles and probably the, the spikiest. There's a. There's a narrow subset of the colony that has, that has spiky reactions. Right? The sort of antibody reactions the projects really get along well with. This is just my. I was reflecting, thinking about the same thing. The product really love lists and feeds and categories and some way of generating things that synthesize the category of things on which we could be friends. That allows for the time factor of the wikis, like the time factor in which things happen might take a few months. It takes people who are interested in the thing to meet people who can automate the thing to develop an idea to bring people together. There's some. There's something organic about it, However, that's possible. A little piece of it. We were talking about. We talked about some very specific pieces for a known body of images or a known body of entities that don't really have a name that, that deserve a name that can be referenced, but there are a lot of other little facets of that that you've been talking about. And just if we can incubate a community that loves the stream of these things that's coming out of the Wolframverse, and then they can help curate how that communicates with the Wikiverse. I think there are so many people in both communities that want good.

So how do we do this? That seems like a good idea. So we've got probably. I don't know, you know, we've got, you know, we went through a whole bunch of material that we have, right. And so the question would be, I mean, whether it's a blog of mine, whether it's demonstrations, whether it's posts on our community site, you know, whether it's the history of math project, all these kinds of things. I mean, we could certainly make something where there is a, you know, you tell me. I mean, is there some, you know, project that somebody could set up where we say, fine, we'll upload a bunch of images to Wikimedia Commons and somebody there. Somebody else, not us. Because, you know, we'll say, great. We'll, you know, we'll send it along with tags and things like this, and somebody will then say, okay, let's have a project to see where these images can be usefully placed in Wikipedia. I mean, that would be a. So, you know, how. How does one get such a thing to happen? I mean, you'd be an example of a person who would probably be in the. In the. In the. Kind of. In the. This is something you could do. So, I mean, if. If we uploaded. I mean, God knows how many it'll be. I. It's. I would guess it'll be a hundred thousand images. So we wouldn't feed that. We. We could not feed that to you all in one day because that would. That's a bit of overkill. But, I mean, let's say that we did it, you know, and I don't know how automated it would be. You know, we did it progressively, and we, you know, we. We notify something somewhere, or we just uploaded a bunch of other images related to this topic, et cetera, et cetera, et cetera. What. You know, and. And then, you know, okay, anybody who cares, have at it. Go put these in relevant pages or whatever else. Or we say, I mean, let me ask this question. If there's text, one thing we could do is say, okay, here's a piece of text, and this is our moment where we're telling people, look, it's a new thing that exists, and you can have at it and put it on Wikipedia, you know, this is like, you know, like my history of the second world of thermodynamics or something would be an example of that. What is the right way? You know, I mean, that page has been read by a large number of people. So the fact that nobody, you know, presumably, since you said there were only five references to my. My whole blog verse, so to speak, writings verse. I don't think it's really. I don't think it would. It's quite. I refer to it as a blog from time to time, but I'm not really sure it's quite sure it's fair to call it a blog. It's a combination blog archive, source, book source, etc. But anyway, occasionally there are articles that are more blog like, but those are pretty rare at this point. But okay, so I mean, you know, the thing exists. What does one do to say, okay, here's a thing that's a useful source, do something with it.

Yeah, if there's any. You have run edit a thons more recently than I have. But if there's a tag for things that are wiki friendly, then people who are excited about that would know. And there's a lot of work that goes into one of those. And there are probably four or five processes that could be batched to figure out how to also.

Okay, there's a thing called glam Galleries. Galleries, libraries, archives, museum. They do similar things where they have massive, massive repositories of images of like artifacts that they'll upload in batches. And it's an extraordinary amount of work. But they have. There, there are a lot of people out there that have experience doing such massive things.

The software engineering is not the big problem. The big problem is if we do the software engineering, will anybody care?

Yeah, once it's up, people know that it's about.

How do we tell people that it's there?

Okay, well, I mean the thing that Sam put in our little chat I think is like putting a tag on the blog. Like this is Wikipedia friendly. Tiny reminder.

What kind of tag should that be? How should we do that? Give me a practical suggestion.

But you already have the little thing that says like CC, MIT license, blah blah blah. You could be really explicit and just write at the bottom. You can put this on Wikipedia, which isn't necessary. But if people are kind of lazy and don't go in the terms, then maybe that's just a little nudge like hey, share this.

Fair enough. Well, maybe that maybe there needs to be a little icon that means Wikipedia friendly. Maybe you guys.

Maybe Wikipedia should do that.

Yeah, or just a category on, like if there were a category in for the writings that said that was wiki friendly.

What do you mean by a category? You mean on our side or on the Wikipedia side?

I mean, I mean, on your side, you have. You have a bunch of categories that you attach to each post.

Yeah, but they're all. They're all Wikipedia friendly. I do not care about. I mean, they're all equivalent in that regard. So I mean, they're all things. I mean, some of them would not be of interest to Wikipedia. That's a different issue. The. But I mean, in terms of being kind of, you know. So I mean, that's the question is how do we. How do we tell anybody who would care? You know, assuming that we, you know, we don't have time, inclination or. And people probably wouldn't like it if we said, here are all these pictures. You know, let me. I mean, I was thinking that what I would do in this actual live stream is actually create a login on Wikipedia and see what it takes to, you know, and find out front lines, you know, add something and see whether somebody reverts it during the course of the live stream. But we've. We've now used our time, so we got sidetracked. Yeah, right.

For a lot of these things, I think you. Right, you want either a bot, either a script to do the upload, or you want people who are one step removed to do some of the edits, but maybe including people from the Wolfram community. So it feels like there should be a glam partnership where Wolfram. The Wolfram site is the archive. And that includes, like, that usually includes handing. Passing on a catalog with some. There's often some sort of prioritization information like how. How relevant or how popular are the different topics. So if you, if you have limited time, start here and.

Explain this to me. I mean, this glam thing. So let's say. Let's say I go tell somebody, take 100,000 images and upload them. Right?

Yes.

This is a. That's a perfectly doable thing. Right. So then what do I tell them to do to, you know, and, you know, and put all the relevant stuff. I don't know exactly what it is, but, you know, we figure out the relevant stuff to say, yes, this is an image. You can use it. Here's what you have to cite, et cetera, et cetera, et cetera. Right, That's. So then. Okay, then how do we, you know, is this glam thing, a mechanism for telling people about that or how does that work?

Yeah, they often hire a Wikipedian in residence, like at the Smithsonian, and that person will put everything in Wikidata, make sure it's interoperable or sometimes organized edit a thon. Sometimes they'll put them into articles, things like the Kansas City Library, things like the British Museum.

Okay. People will be horrified. There'd be pitchforks. Right.

Well, there's a large approval process, so if you get approved, then it would be great.

So tell me, so what are you suggesting there? I don't quite understand that. So the, the. You're suggesting we pay for somebody to do what?

Well, okay, you were asking. So the thing that I've seen being done before, not telling you to spend all your money on this, but I've seen things like the Smithsonian pay a person to be the Wikipedian in residence. If you're an organization like the Smithsonian, who cares about opening up access to knowledge, etc, then.

Well, obviously I've spent God knows how much money and time, you know, opening up access to knowledge, so I, I count. But we're not, I'm a very. It's not, it's not a. You know. But we also have a company that makes money selling software, but that's a separate branch from the, you know, from the world of, of, of creating knowledge. And like, you know, my own writings effort is, you know, is just me, so to speak. The. But so what. Oh boy, is that cat a Wikipedian at this point?

This cat has made a lot of errors that got reverted very quickly.

I thought it was like at the critical moment, the cat is the one that presses return. So anyway, but okay, second thing.

What Annie said you could have a Wikipedia in residence yourself. The minimal thing would be for us to define the project together. Like for you to. Either we could talk through it, or there could be someone who understands the structure of what the 100,000 articles in their metadata look like. We could have a little conversation and then we summarize that in one of these requests for batch uploading. And I posted a link in the shared chat. And that request usually just adds a link to a source website. So the link could say, here are the categories on the Wolfram sites from which these things should be extracted. And then as people get to them, they would write a little script that would do the appropriate scraping, or better yet, maybe you could package up the entire thing as a download and it would just be a question of getting it to someone who could run the batch uploader.

Yeah, I mean, look, in terms of the engineering, I'm not concerned about the, you know, we're a, I believe, competent software operation. And, you know, if it's like, take 100,000 pictures and upload them here, you know, that's a piece of Wolfram language code that is probably on the order of two lines long. That's easy.

But this is more of a barn raising. And the. The barn raising is not about the engineering of building a barn. It's inviting the people who want the barn to exist to come together.

That's the part I don't know anything about. So that's the. That's the part that, you know, that's the part that relies on you, so to speak. I mean, because that. That's. That's just not, you know, this is the part I don't understand at all. And. But that's, you know, what I'm. I mean, we have a lot of material. Happy to do the engineering part of it. I don't understand the barn raising. And if there's a, you know, if it's a. If it's a question, I mean, you know, I don't know if. If it's like, oh, can we hire an intern to help do this? Probably the answer to that would be, sure, fine. You know, if it. If it's like, oh, let's get a, you know, expensive, you know, software engineer to do it, that's a different story. But, you know, in the, you know, let's hire an intern to, you know, to help with this. Absolutely. And if you've got suggestions for people, I mean, you know, let's say. Here's the thing. My point of view is we went to an awful lot of trouble to create all this material. It's a shame if the world doesn't see it. And Wikipedia is, you know, a fine venue for it, and I'm happy to see it there. But on the other hand, it is. Life is too short to have kind of the, you know, trench warfare, so to speak. So it's kind of. Yeah, I mean, I think the thing of greatest interest to us, because it has, you know, it's the place where there's real, you know, real. It really makes a difference is Wikipedia itself. I mean, Wikidata. Most of the consumers of Wikidata are people like us, you know, doing. You know, mostly it's computers. And I think we. I think we're. We're. We have. We have better ways to contribute to the computer ecosystem. You know, I think we're the the way for us to contribute what we're interested in more is, You know, is. Yeah, so. So, I mean, I think. Yeah, I mean, so look, it seems like the, the things for us which are very useful that you guys have pointed out are, you know, our, our terms of service are kind of muddy and sometimes not Wikipedia friendly. That's a thing which, you know, it's kind of like we're prepared to sign away those. That. That all those things to make that Wikipedia friendly. That's easy. That's, frankly, insofar as that's not already the case, that's kind of just a mistake. It's, you know, we can do that. We'll tell you we did it then, you know, we can find ways to put tags on that link also to Wikidata, probably to put tags on images that we have, possibly. Would it also be useful to put tags on the articles that we have? Is that useful? Let's say we could put a Wikidata collection of Wikidata tags onto some section of some. One of my writings. Would that be a useful things to have or is there no way to consume that?

You can do that. Give one very specific example.

Okay, so let's say we've got the section of my piece about the history of thermodynamics, and it's a piece where it mentions the following 10 people and it mentions the following 8 concepts or something. And we can find the Wikidata tags. We could also probably find the Wikipedia pages. Actually, that's a good point. We would know the Wikipedia pages that were relevant to that. So we could say, here's this block of text, here are the Wikipedia pages. And the same will be true of images, by the way, we would know which Wikipedia pages are relevant to those images. We could put those that put that in the tags.

So that would be insanely useful. And then anyone. So something that does happen in other contexts for glam projects is someone gets the archive to update and tag all of their media or all of their collection entries. Someone else writes a script that anyone can log into and say, show me 100 entries from this corpus. And then they go through and say, yes, yes, yes, no. Sometimes they give feedback back to the source and say, this tag's probably wrong.

Right, that sounds good. I mean, look, so then the idea would be, okay, we got two categories of things. We've got images, we upload those to comments to the Wikimedia Commons, confusingly named Commons. What's that?

The confusingly named commons.

Yeah. Why Is that confusing?

Sorry? Because within the wikiverse, people refer to it as just the commons, but there are many commonses, of course.

Oh, I see. Okay. Okay. Yeah. Right. So that's not a confusion that I had, but. Because I had the prefix word, but. No. So then, okay, so we upload there, we tag it with the Wikipedia pages that are relevant, the wikidata entities that are relevant, and some human readable tags. Is that a reasonable approach? And then, so that would be. And then we tell you we did that. And then there's some mechanism. So now tell us the mechanism. So, okay, so we got this thing and we say, let's say we do 100 pages to begin with. We just, we just upload these, we put in some tags, we can ask people, are these tags useful? Are these the right tags? We can change the tags. You know, the, you know, we do that. Then it's like, okay, here's the thing. So does that, does that sound about right for the images? Yeah.

Yes.

Okay. And then, okay, for lumps of text. Is it. I mean, maybe there isn't a mechanism for those. But, but if, if we have a lump of text that's, you know, this section of, you know, thermodynamics in the 1880s or something, which is, you know, will be. Not quite, but roughly some or the, the, you know, ergodicity would be the history of ergodicity. This will be a section in my, in my piece, piece about history of thermodynamics. So what could we do with that text? How can we make that text more Wikipedia friendly?

The same process of identifying related articles and wikidata entities would be quite useful.

And where would we put that data?

It could just be a bit of JSON attached to each article. However, nobody will find that.

I mean, that seems like it could.

Be a version of the article that even has hyperlinks from the specific terms that's probably more densely linked than you might want from the normal.

Yeah, we don't want that because that makes it. The page will be blue in that case. Plus, we have our own links for many, many of those terms will have our own links.

Right? So it could just be a JSON file that one of these scripts could read. And then again, there's an interim step which takes a moment, waiting for someone to realize there's a beautifully annotated corpus and make a script that relies on that and then shows people places that they could, they could engage.

But I mean, so, but is there, I mean, for the images, there's already this Wikimedia Commons where people upload lots of images. But what this really is, is a, is a, an area of tagging information for text. It's kind of like, is there, is there a place for that? So this is basically making it easy. You know, we can get a system that goes through and figures out these are the Wikipedia pages and these are the Wikidata entities and so on that are relevant to this particular note in the NKS book or some such other thing. Right. So then the question is, you know, where, where on earth do we put that? I mean, we can do that. That's a. From a software engineering point of view, it can be done. You know, is there a place in Wikimedia Commons, perhaps where one could just upload a bunch of those JSON files, for example? Would that make sense? I mean, if one does it. If it's like the. Oh, that's an interesting idea that Daniel is suggesting here. If we just did an entry in our data repository. Yeah, that's a good idea.

And then we'd make an entry for that in Wikidata and people could go and mine it and use it for their scripts.

Yeah, that's a good idea. That's a good idea. That seems like a. I mean, then once again, it's like people, people have to know that it's there. But that seems like, from a software engineering point of view, that seems like a good solution. Yeah, I like that. That's a, that's interesting, actually. And that. Yeah.

Danielle, now you have to come on screen.

What's that?

Now Danielle has to come on camera.

I don't know.

That's great.

That's great. No, no, she doesn't. She's waiting. Okay. I think she just got back from being away for a while on some thing. But, but anyway. All right, well, look, this is, you know, this is, this is, you know, we may have done some good for the world. I don't know, maybe not. Maybe. Maybe we, Maybe we. Maybe it's all reverted, so to speak, to not to not use some. But.

Hopefully I'm very excited about that. About the. What to me feels like a change in, in licensing. You can't revert my happiness.

Right. Well, it's, It's. That's good. That's. Yeah, I like to, I, I like to think of that for myself as well, that you can't revert happiness once you've once been happy. It's. Anyway, but, but. All right, well, we should, we should probably wrap up here. And anything else that we should make sure to cover. And I, you know, appreciate it's been. This has been informative. Nice to see you, Sam. Nice to meet you. Annie And Sam, are you still in the Boston area?

Yeah, I am.

Wow. I might actually run into you somewhere. Who's to know? I think I used to run into you, like, a decade and a half ago. We seem to. Maybe you've ascended or something to different circles.

No, maybe just less time on campus. But if you ever are, Annie's in New York, I believe.

Sometimes I know less about wikidata than Sam does, but I know about English Wikipedia. Also, I do comedy shows about Wikipedia, so if you want to be in one, just let me know.

Oh, wow. Wait a minute. How on earth does that work?

Andy is the best communicator about Wikipedia that I've ever met. And. And it is so delightful. Talk about unrevertible joy.

I'll be in Boston in the fall.

Oh, let me know. Yeah, that sounds like.

Maybe I'll make you.

What's that?

Maybe I'll make you do trivia on stage, and maybe I'll make it be about really obscure stuff. So I stump you.

That's, you know, in the category of doing things that I'm very much into doing things that I've never done before. So being part of a comedy show about Wikipedia would definitely be in that category. So let me know when you're in Boston. That sounds like.

I will let you know.

All right, well, very nice to chat with you both and be in touch. All right, and thanks to folks on the live stream. See you all another time.

Thanks, everyone.