Frode Hegland: Oh. That was very strange.
Peter Wasilko: Well, I just got dropped.
Frode Hegland: Oh. It’s not what happened? I was wondering if suddenly this was the wrong link. Yeah. I’m very, very, very sorry I’m late. It’s basically Fabian’s fault. I ended up getting a haircut on the way home, so I blame my. Blame the EU. You were talking about scrolling and rendering. Sorry, guys, for the weirdness with the connection. I’m being late. Please, please Please continue. Oh, well, Andrew’s not here, so. Yes, yes he is. Sorry. It seems I inadvertently kicked you all. So the topic you’ve just started is really pertinent for today. I saw Andrew that you’re not ready to show today, and that’s absolutely fine. It’s a bit onerous to expect you to have something every single time, but I’m just logging into something else here. I’ll share a link with you. One second, guys. It’s like. But yeah, if there’s anything else from that conversation, please don’t let me interrupt. I’m happy to wait.
Peter Wasilko: Perfect. Hang on. I’ll be sending a file for you in a second. I just found a link to the prospective wall. All right, let’s see if I can figure out how to get this posted. Okay. Downloads. All right. Zoom, I guess, from inside the chat. Let’s see, how do I drop a file in?
Frode Hegland: Very useful for the transcription, Peter.
Peter Wasilko: Okay, I see it now. Okay. There’s a file icon. Your computer. All right, we’re getting there. We’re getting there. Downloads. All right. And there’s the PowerPoint okay. Let’s see if this works. Are you getting a PowerPoint file?
Mark Anderson: In zoom?
Peter Wasilko: Yeah. In zoom.
Mark Anderson: Yes. Just turned up in the chat.
Peter Wasilko: Okay, pop that up. That’s what I was just talking about. Maybe somebody, like, display that. Okay. And I can’t slide six and seven. They’re visualizations of it.
Mark Anderson: Oh, hold on a second. It’s
Andrew Thompson: Let’s see what you’re talking about. I was visualizing it bent the other direction. This is nice.
Mark Anderson: Is the the concept behind this, Peter, is it just that you can see to allow you some sight of sort of contingent material just outside the frame, so to speak?
Peter Wasilko: The idea is that all of the materials off available on the sides. So if you clicked into a region of it that would cause that whole thing to slide around and become the central focus so that everything on the other side of it would get compressed off on the other wing and it would refocus your display. So through a series of jumps, you can navigate really, really insanely long. Horizontal spans. And unlike the scroll bar, you’re not really guessing. You actually have a sense of where you are, and as you bring something closer, you don’t be able to see a little detail emerge a little bit further down. Then jump to that and then you can see a little bit further down. So sort of like an incremental way of jumping around while also having a visualization of what would be coming next and where you’d see it. So if you had like semi sparse display you could skip over several decades where there weren’t any papers on the topic published and then see where the next cluster was to get to that cluster really quickly.
Mark Anderson: It’s like, so power scroll for papyrus. Yeah. Then.
Peter Wasilko: You’re muted, Leon.
Leon Van Kammen: Is this a horizontal only, or could you adjust the shape into a horizontal n vertical? Geometric shape. Yeah. To be.
Peter Wasilko: Able to do both.
Leon Van Kammen: Ways of scrolling.
Speaker6: Yeah. I don’t see why you like Zig zag.
Andrew Thompson: I feel like if you end up with both, you’re gonna essentially be looking at like a fisheye lens with like the distortions along the outside. Which would almost be like looking at a magnifying glass at a document. Which would be kind of a neat visual effect. This is far beyond the scope of what I know how to do in three.js. But, you know, it’s very neat to think about.
Leon Van Kammen: Yeah, I think it really depends on how you partition a document because, you know, you know, the, the vertical items or vertical cells could, could even be the whole index. And then the horizontal axis could be, let’s say the pages of, of a chapter. I’m just thinking out loud here, but it’s a, it’s a very interesting way to look or think about documents scrolling. It’s it’s a cool example. Peter. Thanks.
Peter Wasilko: My pleasure. Okay. Going back to my eating.
Frode Hegland: So here’s the thing. I’m going to show a slide for today. Lots I can. So that’s something we’ve discussed many times, and I found it very interesting today. I went through use cases over the last few days. Fabian posted about it yesterday and there are many aspects of it and there are many. It’s getting messy, which it should because use cases are messy. But these three things became interesting to me. I’ll read them because it’s very little and it’s for discussion. The point of the left. One of the actual reading a paper. As I was walking home listening to you guys, you were talking about having to frame a document and a rectangle. I do think that for basic reading of texts on a sentence or two, it’s going to always have to be in a rectangle. The rectangle may be small, the context may be huge, but to read things as 3D at an angle or 3D stretched will be fun as gimmicks, I think. And that’s at least the current situation. But if you look at reading a paper and the head academic is now marked in the room, so for an academic to read a document, one of the key things is dealing with the citations and references. Sometimes following to new documents and then choosing to make it findable again, to read or to cite in the future through some means of putting in a folder, reference manager tag, whatever it might be. And here’s the thing I found interesting one of two things. Number one, students generally need to do more than a professional academic because they’re reading the paper not just to learn what this paper says, but also to learn about the field. Oh, for crying out loud. It’s a delivery. I have to run downstairs. I’ll be back in a second, I apologize.
Mark Anderson: Yeah, I haven’t done I haven’t done Fabian’s exercise yet, mainly because I’ve realized that I never seem to have two days that are the same. So, you know, pick something that’s an exemplar. But one of the things I am reminded is that most skim reading is actually an exclusion process. Most of the skim reading I’m doing is is not intentionally, but in a sense as how much more of this do I need to read to know whether I need to do something about it? So basically it’s been keep, been keep is actually what you’re doing, if not with that intention. Because that like is not a deeper read is you’re going to swing back to on a more deliberate basis. Or if you come across a paper you know, someone said something, oh, we’ve had a, you know, had a chat and there’s a window left open at the end of it. And I go back and look at that paper. And you the sort of same thing is occurring. You look at it say, okay, where does this fit? Is it. And one of two things happens is either just clipped a general bin because it’s generally interesting, but I have no idea. So I’ll put it somewhere where I can find it if I can be bothered to come back. You know, if I suddenly say, oh, there was something two weeks ago that, that or it or it’s I generally then if it’s more interesting would move it into the reference manager, which is, I suppose, my equivalent of Freud’s library where it’s innately available as an if and when I need it. But the making it available in that sense is quite distinct from the actual reading.
Mark Anderson: Reading which occurs separately and later. And indeed, I’ve come to realize as I spent some a while doing some academic work, which hasn’t been long, but if I compare what I would, what I do now with what I was doing when I came back to read Masters in 2013, big difference now is that real reading is probably on the third or fourth read. So there’s something you on the first instance, you find something that’s interesting and you think you might need to cite it, or you might need to use a couple of paragraphs from it or something. And, and then over time, for whatever reason, it has more pertinence and you’ll read more of it. But I’ve often found I was just the case for the paper I did with Dave for Rome was I found myself reading papers and picking things out of it that I hadn’t really got before because I wasn’t really looking for that aspect of it. I was reading the paper with a different site, which which goes to show just how tightly packed a lot of academic writing is. Because mainly because it’s being compressed into a very small space and it’s, it’s not quite the same as the compression. It’s not quite the same as the aspect of most things tend to be cited. So if you need to build an argument as to its provenance or something, there is citation. It’s that slightly different thing that you’re just compressing. You’re just compressing, in a sense, the narrative not so much to squeeze out the padding words, but just to try and get more of the story in a limited space.
Frode Hegland: Thank you, Mark, and apologies. And I have to go down again in a few minutes is so bloody stupid anyway, so yes to everything Mark said. So the key with this obvious text on the screen is a little bit that for students, as I was saying, they have to read in a different way than a professional. So they may be the ones using a concept map or a notebook or something, because the professional would know the field and read the paper to see if there’s anything new. On the right is what we’ll be talking about forever. Nothing really new there. What is new here is for me anyway, is reading a proceedings. Proceedings in academia just means all the papers from a conference. So it’s published as kind of a volume. And that is really, really important because if you put something in your library or reference manager, that means that you’ve already decided it has some relevance and meaning to you. When you go through the proceedings, a lot of it is just not going to be interesting. It’s not related to your work. So that’s why I’ve put these very rough kind of bullet points here for us to discuss of go through the proceedings. First of all, is there anybody famous or anybody you know that’s probably going to be one of the first things you look at along the lines of the titles, to have a rough idea of what it’s about.
Frode Hegland: And then you may go on to do the abstract. Once you’re looking at a document, you may want to navigate using a few different means. You may want to annotate for the future, and then finally decide if you want to add to the library along with other things. So I’m not pretending all these points are completely correct or true. I just wanted to highlight that the act of going through proceedings is interesting to us, because our demo in September. We will have the proceedings in the headset. So if we provide a compelling means through which people can go through the proceedings of the very conference they are at. That is a separate thing from a library interaction and paper reading interaction, but it may very well be relevant. What do you all think on that? Well, by the way, last thing on this. It obviously parallels our book as well because our book is collection of many people which are kind of like papers. Please simmer on this. I’m going to have to do the last running down. I’m really embarrassed and sorry. I’ll be back in one single minute.
Leon Van Kammen: I have a kind of understand what he he means by. They’re a bit like papers, but Yeah, to be honest, I don’t really understand what what the difference is between a paper and just a, you know. An article with some research data in it or results like what is. I’m not an academic, of course.
Speaker6: No, I’ll.
Mark Anderson: Try, I’ll try and help.
Speaker6: Yeah. What are the.
Leon Van Kammen: Okay, okay.
Mark Anderson: No, no, you’re right, I mean, and you’re exactly right to be confused because it isn’t clear. Really. The. I think the broad strokes of it are that when people’s paper is used in several ways, paper generically means sort of something published. So it might be it might be something like a technical article where, you know, someone at the university has just written something or someone’s documented an API they’ve written and they put it out there. And and in a sense, if it’s in a document form, it’s by proxy something you can print out. Therefore, in a sense, paper within academic publishing a paper is generally held to it means probably a couple of things. One is it’s been peer reviewed. You don’t know how deeply it’s been peer reviewed, but the the sort of higher up the ladder you go towards something like nature. Suffice it to say, the number of people who’ve peer reviewed it and the expertise of the people who’s peer reviewed it will have stepped up. So, for instance in the hypertext conference where Fred and I published it’s everything’s read by at least two people and commented on and as part of the acceptance process. So, in other words, at least two people, at least two people have to see some, some merit in its inclusion. And there’s a, there’s a committee as well that decides which papers are accepted. And conferences probably run at about a 25% acceptance rate. Which is a sort of unseen part of it. The truth is, a lot of the stuff that isn’t accepted is is is just froth.
Mark Anderson: It’s just people at the at the start of the process, desperately just wanting to publish something without actually thinking about writing something worth publishing. It’s a problem. But, you know, that’s part of the that’s part of why this apparent gatekeeping goes on. The idea is that you don’t want to put something into the effectively citable canon that you then have to take back out, you know, you don’t want you don’t want redactions in in a citation tree that’s supposed to live on that. I think that’s the overarching thing. But you’re right, the term is loose and and actually another thing that if it helps. A the proceedings of a conference is essentially a journal. And in fact, lots of published journals are in fact, they just they’re journals that where a volume happens to be the proceedings of a conference. So the oddity of a conference and it’s peculiar, actually more to the science. Well, probably actually more than anything to the computer world are that that this arose because it used to be that to do it to show computing work at a conference short of turning up with a back of a pack, a pack of sort of punched cards, you had to lug your computer all the way there, sit in a hall and show it to people, you know, because this is sort of pre networks and the web and everything. And so oddly unlike many other areas where conferences are just sort of seen as a place where you go to meet people and you might do a quick, sort of quick fire talk about what you’re you’re doing at the moment.
Mark Anderson: But real work is done in a journal. There are a number of journals in places like ACM and IEEE relating to sort of computer related work where the proceedings are probably the premier journals in the field. So that’s a slight oddity. But in many senses, where we’re talking about proceedings here, you could insert the word journal and you, you it wouldn’t really be any different in, in, in in meaning. So it’s, I guess the, the cut off between a paper capital P and papers in general is probably the being peer reviewed. And they’re in a recognized they’ve come out of some kind of recognized public publication source. In most cases. There’s one other aspect to it is that in. The case of proceedings, there will be there may be items that are published. Sort of. Sections, chapters, if you will, in the book that aren’t actually academic papers. So it might be somebody did a keynote or there’s a note that somebody did a demo, or they ran a workshop or something, which is actually pertinent and interesting to the people, certainly, who attended the conference and possibly those who couldn’t attend the conference. And there as much as anything a pointed to the fact it happened and it gives us a suitable reference as well, that you can refer to the thing that’s happened and someone go say, yes, it did happen. It’s linked here. You might not know what was what actually happened at the workshop, for instance. But it is a thing.
Mark Anderson: So in, in that context, in something like so for instance in the hypertext corpus, the papers or what generally get referred to as papers are the actual research papers. And they fall into two categories of broadly long or short or full or short, but basically long papers are about 10 to 12 pages. In journals they can be much longer. So a journal article could be about 50 pages. And short papers are about about 4 to 6 pages. And then there are also there’s a tradition of posters. So the way you get your, your sort of first foot on the rung in in the academic world is often you go to a, you go to a conference or a, or some such with a poster, which is, you know, you take a big ass poster and have a session where you stand next to your poster and people, people wander around and hopefully ask you sensible questions about it or come around and tell you your idea is rubbish and it’s been done before. Some are lucky than others. And and alongside your poster, there’s normally a 1 or 2 page item may go into the proceedings. It’s a bit variable because from the days of print, whether that sort of stuff made in the proceedings or not often tended to be down to how many spare pages there were in the bind, you know, left over before you flopped over to another cost level on the printing. So it’s inconsistent. But here and in the context of the corpus that you that that I’ve sort of shared with you the hypertext stuff, you’re mainly looking at research papers which will either be short or long, but they’re, they, they, they are of an equal part.
Mark Anderson: Obviously, the longer papers are held to have higher, higher long term value just because there’s more in them. But that doesn’t mean the most interesting ideas necessarily occur in long papers. And one, one last general point, and I’ll pass over to Fred to try and round out sort of all the questions that was asked, which I think are very pertinent, is that journals operate on a much longer cycle, so they may publish quite often, but papers may be knocking around in the journal for years. So you write a long thing, and often the problem is the journal has to find people suitable to actually who know enough about it and agree to peer review it, because, of course, the longer the article, the more effort it is to review properly. And dependent on the nature of the subject and dependent on the nature of the article the article can go backwards and forwards before the reviewers or there are what may happen. Sorry. The reviewers may say it’s fine, but basically it needs it needs some changes. Or, you know, we can test following things or you haven’t proven that and it can go backwards and forwards and then then eventually gets into the light of day, or people give up and then try and publish it somewhere else. So I hope that that rather discursive answer gives you some better feeling of what we mean by papers.
Frode Hegland: Yeah. So I’m not going to ask you, Mark, what you mean by library because you’ve already written a full paper on that. So I’m going to show you something else regarding the rectangular view of work that I put together for you. But I just need to know, as a general thing, based on the previous, the idea of us spending some time on a conference proceeding or book makes sense, right? Because it’s a very specific set of papers that are related that a new person normally in their job would need to interact with. I expect fireworks and all kinds of things.
Speaker6: Yes, Mark.
Mark Anderson: Well, I just say that I mean, in a sense, the future of text is already a. Proceedings are sought in the sense that it has a binding, the parts are separately addressable. I mean, you can put a cigaret paper between a journal, between a conference proceedings, between an edited book, where each chapter is is essentially a discrete.
Speaker6: Yeah, yeah. No.
Mark Anderson: So structurally, they’re they’re actually much of a muchness. I suppose the reflection there for us is what extra in terms of navigational affordances or structure do we wish to put into into the proceedings?
Frode Hegland: Exactly right. So There are a few images. Scream if you can’t see them. What frustrates me, and I’ve spent a lot of time in the vision, reading and writing lately in different contexts. And very often text really does need a rectangle for a lot of it. Not everything. It’s the whole point. So this is the current reader library, my personal library on top. I just added some tabs. This is just an idea that are essentially search terms.
Speaker6: I think we’re seeing.
Mark Anderson: The wrong screen.
Speaker6: Thread.
Frode Hegland: Are you not saying Okay. Hang on.
Mark Anderson: We’re seeing your app library.
Frode Hegland: That’s really boring. Are you now seeing keynote?
Mark Anderson: No. We did for a second, then disappeared. Went back to your app library.
Speaker6: My app library. What?
Mark Anderson: Well, sorry. It’s whatever you. You know, when you mouse over the apps folder and it just shows you a screen.
Speaker6: Oh, really? Okay.
Frode Hegland: That’s weird. Hang on. Let me just close some windows. Then. Maybe the system is confused.
Speaker6: Well, that’s.
Frode Hegland: Okay, I’m going to try now again. Can you now see keynote? Like a slide thing.
Speaker6: Yes. Yep.
Frode Hegland: Okay, so now there’s text on the bottom that says library where the bar on top right.
Speaker6: Okay. Right.
Frode Hegland: So this is the beginning. Simple stuff. Let’s say I want to find everything that has hypertext. So I just type in the search string hypertext. And this is what’s new. This is what I’m thinking about. Each. Found documents. It doesn’t matter why it was found. It could be from a keyword or whatever. It’s not going to highlight the word hypertext. You already know you. After that. What is on here we shouldn’t worry about. I’m currently using. I found keywords and names, but that doesn’t matter. It could be whatever the user wants. The key point here is that if you want to open the document, you click on the title, it opens the document, but all the text here is selectable. So that means that you can be looking at this. You come across Yahoo as in the middle here. Select that, do command F and you now do a search based on that. So it changes the search. In other words, this is a very dynamic way to look up this, that or the other, depending on the kind of criteria you want to have listed here. Makes sense right? The point of this is I may implement a point of sharing it with you guys is visually, it’s really boring, but I found it to be relatively useful. So that’s part of the problem. It isn’t the excitement of XR, right. So. The next thing is We talk about these academic documents, papers on how to make them more readable. Yeah. Peter.
Peter Wasilko: Yeah, I was wondering, do you have some sort of a breadcrumb list you could get so that if I’m going on a command F and I jump to another term and command F and another one could I jump back three searches to get to an earlier search? Easily.
Frode Hegland: That’s not part of the system, but that’s, of course, very Vannevar Bush. So conceptually there’s nothing in the in the way for that.
Speaker6: Good. So.
Frode Hegland: The thing is, the academic papers we’re dealing with in ACM, they’re not that long, as Mark was just saying. So to actually have one in a nice screen and just going through it, it isn’t that horrible. I mean, I don’t like the typography. I think the columns are this and that. But, you know, these are relative details. But this is something that I’ve actually implemented as a little test. So what we have here is the first sentence of after each heading. So if you look here on the right for previous demonstration that’s a heading. The text after that is the first sentence. After that I found just as Doug Engelbart explained, the first sentence after heading is much more useful than just the heading. So in this example here, what the user can do is click on the number after this, and then they get a full not a full, but they get a much longer. So you can read into what is in that section. If you want the whole thing, you click on it. It opens the entire paper. So it’s a way of having a kind of an overview. And it’s kind of okay. But then there’s other things like the question of, you know, should we put images outside like we talked about? Is that useful? By the way, guys, when I’m doing a presentation and it’s just you please just interrupt me. I may not be able to see the hands. So Peter first and then later on. There was an old.
Peter Wasilko: Didn’t lower my hand before.
Speaker6: Oh, okay.
Peter Wasilko: My bad.
Frode Hegland: No problem. Leon.
Speaker6: Oh, yeah. So, yeah.
Leon Van Kammen: Yeah, I was thinking I really like this. I would call it a thumbnail, almost. It’s a generated summary. I really liked it. And I would really love if, you know, I would open my file manager on, on any computer and that it would basically normally you can select tiles or lists and then also the size of the tiles. But I could really imagine this work very well if you could set a bigger tile size and then seeing this and basically to decide if you have many papers, you could you, you could go like boring, boring, interesting, boring, boring. And then sort of like. In the same way, perhaps in XR, like have such a thumbnail as a starting point for a PDF document. I really like this view. I, I’m, I’m a bit frustrated that we don’t have this yet.
Frode Hegland: We’re on the same page there. 100%. Some slides further in. Get closer, I think, to what you were saying. So I just want to highlight when you say make the sign, say something is boring or interesting, that kind of stuff is crucial. It’s not shown here, but that has to be part of it. Mark.
Speaker6: Yeah.
Mark Anderson: I just quickly to pick up on your your question about showing things like tables and figures. I think the answer is yes, but and the but being that and, and and it’s again a problem, given the way we tend to write the documents, that really what would be useful to see is the the image slash table in conjunction with the text that talks about it. So that might be you end up with, you know, two tables, an image and about four paragraphs or whatever. The reason it’s hard to do is I realized looking at some papers when I, I was thinking through this problem is that it’s rather dependent on the author actually saying instead of saying, you know, in the figure below or something. Because when the something like latex gets its hands on it, it then moves that picture six pages away for print packing purposes. So if if the document had the structure within it that we knew what information applies to what image, then those are groupings would be useful. It’s also difficult to because of course you can’t tell in in any quick and easy way how much of the text around the reference that says, see, figure two actually applies to figure two. So probably pragmatically you might be needing to just take, you know, the paragraph you know I mean, I think that’s as close to perfection as you can get at the moment. And the interesting thing I take away from this is it’s really got me thinking about the necessities for proper writing tools for the digital age, which you don’t really have yet. And that’s, that’s sorry. And that’s not a backhanded comment about author at all. But what I’m really thinking about is writing. Writing, put it this way, writing not for paper. So it’s it’s writing. It’s writing something that you could print but is designed to be looked at, something like XLR to do just the sort of decomposition we’re talking about. Yeah.
Frode Hegland: No. Absolutely, absolutely. 100% Mark. And this is some of what we talked about on Monday when unfortunately, you had to be out of town. And that is pardon my French, but, you know, fucking hell, you know, we’re we’re a quarter of the way into the 21st century. But our tools are rubbish. So on Monday we started a little bit on what is an ideal document and I don’t even want to say the word format. Forget HTML or PDF, let’s just forget that for a minute, right? Currently though, I am 100% thinking HTML in this. Of course I’m married to PDF in a different thing, but what I’m going to show you in a few slides is the kind of thing that if we get from ACM proper marked up HTML, then Mark, you are so right. The authoring of this needs to be pardon my French again, fucking joyous. We have to aim for it to be joyous. People can mark up things for social media with hashtags and all kinds of things. It should be nice and fun to mark up semantic values for your own knowledge. So yeah, super important and I’ll stop swearing.
Mark Anderson: I think the challenge in that, for instance, is that whereas now, you know, I might be going and I write something I said and this is, you know, this is discussed in table two and there’s table two has all sorts of tabular contents that might also be then maybe shown in a chart as well, which is in the document. And we relate this. But what I’m thinking about is in the sort of modern writing tool, what you’d actually be doing is you’d almost be setting some kind of a bound on this, this paragraph of writing. That that is basically a semantic bracket that links it to associated materials that, that whether it’s contiguous in the document or elsewhere in, in the manner in which it’s written isn’t the point. You’re doing it entirely for the fact that you are assuming that it’s not going to be consumed in a linear fashion, so that what you’re doing is you’re creating this, this linkage explicitly so that those things can be found and re associated together. So in other words, even if you, you know, you took all the bits out and just threw them on the floor like a jigsaw, that you could reassemble that bit of text, that table and that figure as being something related to one another.
Speaker6: Yeah, that’s super important.
Frode Hegland: Mark, the little thing that I showed you earlier, here it is. And author writes and it’s ridiculous because there’s bullet points. So as we talked about on Monday we should be able to expand and contract bullet points. But also all of these, if there weren’t a proper document would actually link to something else. And just having an end document link isn’t enough. You know, you need to be able to start authoring this document so it appears horizontally, not just vertically. You know, if you want to start writing, check names. So the entire authoring system has to change, but utterly agree with you. And when you have a reference in the text to an image. You know, we’ve talked about link types in the past. Maybe it says this is actually a mural. When you click here, it’ll open massive. You know, and also the data will be contained in there. So you know, the options for how the interaction should happen are absolutely, phenomenally huge, right? So the thing you saw in the beginning is now on the right. It’s exactly the same. It’s the headings plus first line. Yeah. And of course you could also on the side have just the headings. These should be options that the users should have. I don’t necessarily mean we should have white on gray text, but that’s what we have on our webXR experimentation now. So it’s just meant to indicate that. Right. So here’s another way of looking at it. After each one of these headings, plus line, there is an arrow type thing.
Frode Hegland: Click the arrow and it shows the entire section. Right. Really. Now we’re talking about early 90s kind of web stuff of expanding and collapsing text. And I really feel strongly that we need to own that again, because even working in an XR headset with a beautiful display, a lot of this is actually easier. When it is flat, and also two pages. Having experimented a lot with reader, with many pages it can get a bit messy. So to be able to go in and out of what you want to focus on with your knowledge. You know, having a cockpit is probably not a bad idea, but here it’s the same text, but as one column. Sometimes you want that the user should obviously be in charge, right? Here’s a different version of that. And here I’ve collapsed it into a single column. But going back out and this is something I talked to Adam about briefly the other day, because the selecting of text is really, really hard in XR. So if you select something here, there’s a line break after every period. It knows even if you selected something you want to talk about this particular sentence. So maybe you can then make it bigger to focus on it. And maybe once it knows you’re focused on this sentence, it’s easier for you to say the word professionals so it knows what you’re referring to. So how to actually point an XR space is a big issue at the moment. Peter.
Peter Wasilko: Yeah. Do we have any displays for dealing with parallel documents at the same time? Because there are lots of times when I’m working, when I’m writing, I’d really like to have a say. The top two thirds of my display would be two documents that I’m looking at and contrasting with each other. So the top two thirds would be split in half, with one document on the left, one document on the right, and then the bottom third would be whatever I’m doing to integrate the two.
Frode Hegland: Then with reader and author. You can do that now. You can have some PDFs and whatever size you want, you can float things, lift things from them and add that in space. And then you can have author underneath. So at least you have flexibility of of windows. Not as much as we should. I sent the email to the vision people this morning. Let’s see if they even listened to us. Leon.
Leon Van Kammen: Yeah. Regarding that sort of multitasking.
Leon Van Kammen: I think this is a very almost a controversial topic because, like, I really I always like these kind of ideas from, from theoretical point of view, or if I imagine it being able to do that. Super exciting. On the other hand, I also know that there are various reasons concentration wise, to limit an editor to monotasking. And I was curious if Peter Frode and Andrew, you know, have any sort of learned lessons to share to, to share which basically, you know. Point out that certain ideas of multitasking with multiple documents, etc. are cooler as an idea than in reality. I’m curious about that. To to keep it real.
Peter Wasilko: Okay. To give the best example of when I’ll be multitasking. When I’m programing, I tend to create small, very narrowly scoped, single purpose demos of just one little tiny piece of functionality. And I’ll have a whole bunch of those, and I’ll slowly start to proliferate over time. And I sort of call that my scatter mode. Then I move into a gather mode, where I will have a integration document open on one half of the screen of my editor, and this will be the program where I’m trying to combine those little individual working bits together. So then I’ll take one of the little sub fragments at a time, and I’ll then take that code from the left window, which is looking at it in isolation, and I’ll merge that into the right window. And it’ll basically be representing different folders in a mono repo. And one folder will be say the socket server, next folder will be the web server, and those will be working independently. Then I’ll be combining those elements in an integration script in the right window. So. In that one, I’ll be simultaneously firing off those individual demos as parallel processes running on different threads. So I have my integration code on one side. On the right side, and the little example codes that I’m looking at from different tests on the left side. Occasionally I’ll split the left side in half and in the upper left I’ll have that code that I was just describing. And then in the lower left, I’ll bring in a web browser window and pull in documentation related to that piece of code that I’m working with.
Peter Wasilko: Then once I get everything working in the right window, I’ll close out all those little sub projects, timestamp them and throw them in an experiment archives folder. Actually, I won’t even timestamp myself. I have a hazel background process looking for folders being dropped into my experiments folder, and when a folder gets dropped in, it, timestamps it, renaming it with the timestamp so that I can eventually find it later on if I need it. And I’ll close it out, and then I’ll go in to a single focus mode working on the program. Once I know that all the individual working bits have been gathered back together and integrated into the one script then if I want to add something completely new, I’ll start a new sub project in the archive and again, just get that little piece of scaffolding working. Also, as I bring the things together, I don’t like having 10 million different libraries that I’m porting from. So I have a dedicated library script in which I import all of the different packages that I’m using in any of those little sub experiments, and export the pieces that I need, so that everything that I’m importing in any one of my little example scripts is coming from my single library script, which is sort of a nexus for pulling in all the imports reexporting any variables that I’m going to be needing to pull in.
Speaker6: Okay, Mark.
Frode Hegland: This is obviously an import mark. What am I saying? Mark? Peter, this is obviously an important workflow, but it’s more about coding rather than academic reading, right?
Peter Wasilko: The same sort of thing happens in academic reading, though. I’ll be pulling together a couple. In that case, the instead of being programs, it would be individual papers that I know I want to be citing and working into a given reference. And again, I have the few different pieces. And then once I get them pulled together and integrated back into the master document, I’ll close out the individual things that I’m referencing. So. There. I’d be working in Latech and it’s actually still a programing project. Just it’s the programing project where the output, instead of being a program, is going to be the code to generate the academic paper.
Speaker6: Yeah. So. The.
Frode Hegland: I this whole thing about multitasking. Leon, I kind of agree with. However, authoring an academic paper is a multi thing. Still, it’s a stainless single task to go in and out of different papers. I would also like Fantasia, Disney style magician, you know, pull this paper, have a look at it, you know, all without it getting messy because that is looking at these is the single task.
Peter Wasilko: Oh also another major piece of my academic workflow is using the tab groups in Safari. So I’ll have a different tab group on every single subtopic with tabs. So if it’s say info visualization could be a tab group. And then if I discover that I’m getting a lot of tabs in that, I’ll select all of the things related to, say, Perspective Wall and then create a new tab group with just those tabs. So I sort of have like a hierarchical composition of tab groups as I’m developing the paper.
Speaker6: Yeah. Then once I.
Peter Wasilko: Find them, I’ll be pulling the references into Zotero. Then I’ll close out Safari, and then I’ll be pulling the citations in using the zero library to get them in. Or if I have nice dois, I’ll just drop the Dois and author to begin with.
Leon Van Kammen: Okay, so just to summarize, I’m trying to summarize it before Mark is also going to say something. Is it correct, is it maybe correct to assume that grouping is maybe more important than to see everything side by side?
Peter Wasilko: Yes. The grouping is very important.
Frode Hegland: Spatial hypertext, baby.
Mark Anderson: Yeah, absolutely. No, indeed. You’re spot on. I mean, I was I was thinking on Lennon’s point about multitasking and reflecting on the fact that yeah, that whole thing of focus is actually very internal. It’s really hard to discuss because it’s it’s very internal to us. And so for one person, focus is having only one thing on the screen for somebody else is having hundreds of things on the screen, but only the things that really matter. And both those states can be true, but not probably for one person at the same time. So it’s it’s what’s interesting about is it’s really hard to abstract the process from How you how you how basically how you’re seeing it. Which is difficult because I was thinking Yeah. I mean, ideally when writing a paper, in the ideal world, you sort of have all the facts at your hands before you start, and then you write it out and all goes. And the nice thing, in fact, I think what tends to happen is you get somewhere, there are some things you know, you want to use. There are some things that you may know that are background that essentially you either know or can or can easily go and find. I mean, you know, at some point you’ll need to you’ll need to write some preamble for for people, it might not be fully formed in your mind, but it’s going to adjust. It’s in fact, it’s going to adjust slightly depending on what’s in the body of the article. But then you get into something and you think, okay, well, I need to, I need to maybe reference a subject, but what would be the best? What would be the useful reference in terms of not that blog said this, you know, paper X, page 32 because it’s an actual quote.
Mark Anderson: But just if I’m trying to explain to somebody else about this, what would be the best thing to point them to that fleshes this out better for them? And that’s where sort of one’s library comes into play, because there’s, there are the sort of things, I think, and you might get, you might go back to, say, a keyword search of, you know, oh, I want what papers have I noticed having this rough topic. So you might know something, but more often than not, what you might well do is go into there or or it’s the serendipity thing. You think I’ll start looking that. Oh Smith wrote a paper on that two years ago. I’ll take that one. So it’s again, it’s a process. It’s really hard to define because quite a lot of it is looping through here in a way that doesn’t lend itself to, you know, nice little logic flow. I’m sorry if that’s messy, but I’m, I’m trying to help to pull up some of this because, I mean, I do think some of it can be systematized. The hard part actually is trying to work out what is what is reproducible and definable and what is just the very personal interaction with the underlying process. Because at the moment, most of us, I think, tend to to, to to smush the two together unintentionally because that’s how we experience it. And it makes it very difficult for third party, you know, rather as you’re doing here to say absolutely correctly said, yeah, okay. But what’s going on?
Leon Van Kammen: Do you also think fraud that this sort of. That there is an internal map, an external map happening when sort of like having an external external map on the screen of documents, which are connected, but also inside your mind, you have a certain map and they sort of. Overlap or don’t overlap.
Speaker6: Absolutely.
Frode Hegland: There is the layout on your desk. Virtual or physical, may or may not be having meaning. You may have just rough meaning. It may have been very organized. It may be a knowledge map. Most certainly do have a map in your own mind for sure. I wrote here in the in the chat. Mark liked it because I was basically saying what he was saying. Citations. But okay, I just finished my PhD. I have no idea if I’m going to be awarded it or not, but I’ve done my last giving of the paper and a lot of my my thesis was about visual media, was about metadata and metadata primarily to support academics, therefore primarily to support citing. So I wanted to write into my thesis why siting is important. There’s hardly anything I can cite on that. It’s really fascinating. There is something Yale wrote, something really thoughtful a few years ago. It’s gone. It’s not even there. That had a great sentence of citations is how you weave your argument into the academic environment. Something like that. Gone. The university does have something. And Mark told me that, yes, the title of the university is the author. That kind of stuff.
Frode Hegland: Weird. Not really suitable point of it is what Mark said. I have to reemphasize it. Citing is not just to show you didn’t plagiarize. It is not just to show a teacher or somebody you know your points, although that is important if your student. It’s not just to show your academic colleagues that, you know, to cite the important stuff. It’s also a very specific tool because the citation you choose will determine how much weight you give to what you’re citing. Because in a community, people have discussed the same kind of stuff again and again and again. So which version do you cite? Do you cite the original or do you cite the way it was contextualized by someone else? This is a prime part of the academic process, and I’m so glad this has come up now, because when you’re talking about the last part of the little three things we saw at the beginning, the library finding out not just a citation you want to use, but the one that is most appropriate to what you need right now. That is a really important use case, and I think we’ve kind of skipped it till now.
Leon Van Kammen: I think Peter wants to say something about that.
Peter Wasilko: Yeah. I just dropped a little note to a paper describing legal writing methodologies in the sidebar. And that’s a very interesting case, because when I’m working on a legal writing project, it actually, workflow wise, consists of running through checklists in my head. First of all, there’s the Iraq Structural Checklist, which is how you’re supposed to organize the legal paper at very high abstract level, putting the question of law out in front what the rule of law is, how the rule of law is applied. Then when you reach substantive law, you have a whole bunch of checklists for each area of law, there’ll be different elements for each element. You have to establish evidentiary links that they’re supported by. But each one of the evidentiary links has its own little checklist in order to make sure that it’s a valid piece of evidence that will be introduced. So there are evidentiary issues which are sort of like a meta level before you get to the substantive factual issues. And then there are procedural issues that cover how and when you raise the other issues. So when a lawyer is doing legal writing. He has all these checklists in his head, and I have yet to see any kind of a tool that would provide convenient support for flowing through the different checklists. So you sort of like instantiating. Different pieces, there’s going to be a new argument. Here’s going to be the pieces of evidence. They’re going to support that argument. Then, of course, we get when we’re making the actual argument itself, our literature citations into the case law to show why it is that this rule should be applied in this way, in this context.
Peter Wasilko: And that’s where your legal citations come in. And that’s absolutely critical. There was one wonderful people that I met, Stuart Alan Sutton, who was a lawyer turned librarian, and he was actually looking at that stuff when he did his degree. At Berkeley, and he was trying to describe how in the context at the time of. Structured Query Language kinds of things. With a table based representation, you could model the linkages between the checklist and the arguments and the underlying facts. At the time that he was doing that work, we didn’t have open source software tools readily available, though he had absolutely no effort of even touching the implementation side of any of it. Nowadays, everyone is discarding structured approaches and just going to large language models. There’s a group out of Stanford called Codex. We meet on Thursday afternoons online, where people are presenting all of their new companies, developing new legal technologies, and just about every single one of them. I took an LLM, and I have the large language model pulling out stuck contract clauses, and that’s my application. And the next guy will come in and say, well, we’re having an LLM and we’re actually going to use the LLM to look for hallucinations in the output of other LMS. But everybody is a one trick pony. Everything being funded is okay. You’ve got to be using an LLM, I and all of that earlier work on symbolic representations, which actually take time and involve a human entering data into a firm, wide knowledge base to capture that data explicitly has been cast to the wayside, because that’s not where all the money is chasing right now.
Mark Anderson: So. So, Peter, it sounds as though what you’re what you’re suggesting arguing for is, is use of, in essence, the power of AI, but from a different context, which is almost like process completion and monitoring. I say this is interesting because in the work that I’m the fiscal sciences data infrastructure project, I’m the thing I was aware on Monday this is very much what’s been talked about far enough in the context of, well, mainly about chemical lab books, but again, where there’s a lot of process that needs to process, that needs to be captured. Exactly. Having, having so effectively having. So it’s weird. I mean, we’re using we’re using AI, LMS like a magic eight ball when we have something very powerful that could be doing things. That said, it’s sort of saying back to us, you said you were going to do this, and here is a list of things you haven’t done, or here are the list of things that you or here are the things that you have done that you you maybe haven’t, you know, related to what you’re doing. So there’s all sorts of powerful stuff that could be done. And it seems to have its real power deployed at quite small scope, so quite close into what you’re doing. So sort of trying to, you know, write a whole paper for you or something. It’s just, it’s just doing something that it has the attention to concentrate on, which we as humans, by and large, don’t, you know?
Peter Wasilko: Right. Making sure you walk through the checklist, suggesting which checklist you might want to instantiate next, then jumping into that sub checklist. Yeah.
Speaker6: Yeah, sure. Now.
Mark Anderson: Was it fright? Sorry.
Speaker6: No, no.
Frode Hegland: Yeah. Okay. Yeah. I just wanted to say I dare a friend of professor Mark. Les Carr said that I shouldn’t be called artificial intelligence. It should be referred to advanced it. And I think that’s a very clever thing to say, because the term AI doesn’t mean anything anymore. Even LMS is very limited. I mean, when we use speech to text, we’re using a kind of an AI.
Speaker6: Right?
Frode Hegland: I was with someone the other day who was actually quite clever, and we were he was trying on the the Vision Pro, and the Vision Pro was learning the eyeballs. He called it machine learning. Right. And I had to really gently say, this is not machine learning. The machine is learning about you. But anyway, point the point of it is right on Mark, we need to use a lot more AI in our work, but we have to be really careful of which kind and how. For what purpose? And if we use it for something big, that needs to be very, very clear. But for a lot of simple stuff, it brilliant. So. Yeah. Important. Right. Sorry, sorry. Leon. Leon.
Leon Van Kammen: Yeah. I want to I want to play the role of devil’s advocate right now. You said citations are toolset. What about citations are very frustrating teasers. And to give you a example, I was reading in the past, I was reading a self-improvement book, and I noticed that this guy was actually not a genius. He was just basically telling the same story of much older authors in a different way, in a more modern way, perhaps. So I was like, okay, let’s let’s stop this nonsense. Just tell me who this author was, who you’re sort of reiterating. And then I found out it was James Williams. But the problem was that there was so he was I don’t know if that’s a full citation, but it was basically basically a quote. It was very hard for me to find out the book where this quote was written because he had several books. And so I had to do. Yeah, it was a bit like this citation was a beautiful lady who was running away quickly after I got interested. So my question is. Isn’t there a missing piece of the puzzle here where if you cite something, it should be available publicly. For me very quickly to sort of you know, get to the source of things instead of having to hunt down a book or buy a book. All these steps. That’s that’s so frustrating. End of my rant.
Speaker6: I will speak.
Frode Hegland: To this rant. The word cite means to bring something forth. To bring it up. That’s really what it means. That’s why you can have a traffic citation and all of these other citations. What you’re saying is really, really important. But the citation isn’t just one thing. As Mark has been arguing with me, when I talk about library, it is possible to cite the book you don’t own, or a classical text like Hammurabi or whatever, which you don’t have access to the original. So there are cases where it’s absolutely legitimate to be teasing and saying such a thing exists somewhere. But beyond that, when we write, it really comes into the thing Doug didn’t get Doug Engelbart didn’t get. And Ted talks about a lot, and that is the social and political environment of humans working together. Not everybody wants things to be followable. Especially either real politicians or politicians in an organization. Not everybody wants it to be easy for someone to be able to go back and say you know, we have mayor elections in London coming up this week and my favorites. I’m not going to bother discussing who one of the tweets or Instagrams or whatever was screenshots of one of the main other contenders of their tweets. Basically using their own words against them, which I thought was quite clever, but that, you know, it does come into this. So in an academic environment where there is a positive collegiate atmosphere, as we should have. What you’re saying, Leon, is super, ridiculously important. And this is what I’ve been trying to do with author, reader and Visual Meadow, of course, come across the citation.
Frode Hegland: You should have absolute instant access. Now, what is really interesting is the question of what does instant access mean? Does it mean just the source? Does it mean what we talked about earlier today? Some kind of useful summary of the source? Does it mean maybe context of the source, like the example of citing Mein Kampf? Does that mean you’re a Nazi? No. It may very well mean that you are making a point about how horrible it was. Right? So to to bring up a citation is such an amazingly exciting issue for us in the work we’re doing, because of the question of what should be brought forth. If there is that example of Mein Kampf, one thing we talked about a few times is the notion of link types. You cite something that you disagree with, but it’s important for context. Furthermore, how about we provide the means through which when you bring up a citation, you don’t bring one thing only. You don’t bring a truth or a source, so to speak. You bring a entire perspective. That is, of course related to spatial documents. So let’s say you cite Mein Kampf, and the person who made that citation has put all kinds of useful metadata around it. So when you see it, it doesn’t just come up as a book. It comes up with a huge amount of rich context so that you can see how it fits in the field. In other words, hugely interesting topic. Thank you for that.
Mark Anderson: I just quickly add in that another.
Speaker6: Have a lot to say.
Frode Hegland: Don’t pretend it’s quick, please.
Speaker6: No no no.
Mark Anderson: No, I’m I thinking on this thing of sort of citation and trying to answer Leon’s point, which I think is extremely valid, is, is that let’s leave aside bad actors who don’t want you to know where all the cool stuff is, because then they can’t stop you charging you money. Let’s let’s park that side. But I mean, in the genuine thing, there’s also sometimes the sense that close citation may be difficult because What you’re citing doesn’t actually exist anywhere on the page. And somebody might have to go and read 40 pages of a book to. I mean, they’ve either got to take it on trust from you. And again, if we’re talking about good actors, take it on trust that you’ve synthesized something out of it. But if they if you want to understand where that synthesis came from, you’d have to actually have to look at more than just a paragraph. Or, and sometimes it’s almost the case that someone is held to have sort of said something or had an idea that isn’t actually written down somewhere. It’s hinted at throughout what they do, so it’s made manifest by their work, but it isn’t actually stated, which is always problematic in citing terms. And it’s a bit like I was scrolling just as Fred was talking trying to find them. There are a couple of McLuhan references that I keep finding people come back to, and they’re actually rather hard to find in the text. And, and in a, in a world before having fully sort of machine readable text it’s actually quite hard to go to sort of 4 or 500 books in the bookcase and just flick through and try and find find the words.
Mark Anderson: So I think that’s another reason why people don’t it’s probably more the case in the humanities, where people are making much, doing much more close reading and sitting close. You know, the argumentation sits much closer to the actual text that you’re more likely to find very close grain grain references. And another aspect in the way that, that domains differ. So something like the law obviously you, you, you’re going to quote chapter and verse of a, of a statute. In the sciences it’s not really so much. You’re probably more saying that this method, as used in paper by Smith, 2006 or this compound was discovered and described in this paper here. And the general isn’t a need to be more specific. So again, it’s really difficult. It’s hard to generalize. And I say that not not to try and work around the question you’re asking, but it’s, it’s difficult in, in the fact that it does vary quite a bit. As you move across the disciplines between the sort of the empirical sciences, say through, you know, the, the sort of looser sciences and things like computer science through to the, you know, the broad humanities. And once you’re getting into sort of philosophy or social policy or something, well, they will make up all the words anyway. So it’s incredibly hard to follow.
Frode Hegland: Yeah. Thank you. Peter.
Peter Wasilko: Yeah, I think it would really help if we would sort of relax our demand for formality when it comes to citations. Many times I’ve been stymied by only having partial information. I want to allude to it, but I do not have an accurate citation that I can provide. And to avoid any accusations of plagiarism or even self-plagiarism, I choose not to bring that point up for fear that someone will say, oh, he did not have a proper, valid citation. So case example. I remember that at CMU they had a simulation of Albert Einstein that I saw demoed from Carnegie Mellon people at a conference that I attended in Pittsburgh at some time in the 1990s, but I do not have ready access to that paper to be able to say it. So. Rather than go through the torture of trying to identify a paper. Or maybe it wasn’t even a paper, it might have just been a poster, or it might just have been an informal demo, because they did have us visit the CMU conference one day. The CMU campus one day, and they had like a little orientation meeting in one of the auditoriums at CMU before we went off to the reception and visited with some of the lab people. So it would be a nightmare to try to link that memory to a proper vetted literature citation somewhere. So again, I won’t mention it, but if you had the ability to provide a more informal citation, that would help.
Peter Wasilko: And also the target audience is really critical, particularly in law. Part of what makes a lawyer who’s charging a thousand times more than another lawyer to handle the same case, worth paying a thousand times more to handle. That case is that he knows the judge in the case. And beautiful literature example in the Horace Rumpole stories, there was one in which a young solicitor went up to Horace and said she was going to be before judge such and such, and she wanted to know, do you have any advice you can give me on how to present stuff before this? Judge and Horace looked at her and says, oh, he loves the law. The more citations you can provide him with, the more likely it is that he’ll be doing it. When are you going to be before him? And she said, well, it’s going to be Tuesday afternoon. And Rumpole, I just said, yes. Yes. The more citations. Now you provide that judge, he’ll love them. Well, it turns out that he’s the opposing solicitor when she is trying to make that argument. It’s Tuesday afternoon there in front of the judge, and she starts dropping. Academic citation after academic citation to the case law. And the judge starts squirming on the bench. And within a couple of minutes he throws her case out and Rumpole wins and she’s totally confused. And she goes up to Rumpole afterwards to ask him to explain what went wrong. Didn’t I provide enough citations? And he said, well, you see, it’s Tuesday afternoon and the judge has his weekly golf round Tuesday afternoons, so he just wanted to get out of there and you know, be a good sport about it.
Peter Wasilko: Someday it’ll be your opportunity. But, you know, it played out pretty well for my client. And that’s just how the cookie crumbles, basically. So he knew that the judge didn’t want to see a large number of citations, but. Allow the solicitor to go and provide them, naively thinking that the more citations the better. So again. Who’s the audience? There are some judges who hate entire classes of argument. So as a matter of law, strict law, your argument is right. It has all of the backing in the world, but a particular judge doesn’t like that train of legal thought. Or, you know, maybe you’re representing Donald Trump, and no matter what argument you make, it’s going to be utterly irrelevant to the outcome because the judge is fixed on the outcome he wants to arrive at. So the only citations that that judge will be interested in are citations that allow him to justify the decision that he’s already arrived at previously. So you have that rich meta context of who the things targeted for in addition to the official. Formality of how you make the arguments and what in the abstract it should be. So the audience is absolutely every bit as important as the context, and we need to bring that into consideration.
Frode Hegland: Yeah, absolutely. Thank you for bringing in a summary of the state of America as well. As a thing in there of how the sides there seem to not be able to meet anyway. There is along those lines, there is also the notion of explicit and implicit citations. Before citations were formalized, you would try to hide them because it showed how erudite and clever you were. You would refer to the ancient, but in a very convoluted way. So only someone else who was clever would realize what you were doing. You can still do that, of course. It may not be useful, it may be silly, and it may get you in trouble because people may think you’re simply plagiarizing, so. The way we put our together, our ideas together is an interesting challenge.
Mark Anderson: There’s an actually it’s interesting though. I was just as you were saying, that I was thinking that there’s also a thing that sometimes, you know, there should be a citation, but you don’t know you’re unable to make it. This is a subtle difference from I couldn’t be asked to look it up. It’s saying that yeah, I think this has been done, but I don’t really know know where and how and and it is not you know, it is not a five minute quest to go and find it. It could be it could be a lot of time. So in a more in a world of sort of digitally generated documents with good semantic linking that we’re sort of thinking about here is that would be an interesting thing in a sense. So a link type might be a sense, a non-existent link that ought to exist, that maybe that maybe provided a later date by somebody who knows. And it might be either in a sense, you know, that that really ought to be something underlying this. In other words, this thing did not happen without some there being some causal element, but you just don’t know what it is. But but you suspect that somebody else may or just that you recognize that this needs it will be improved by some sort of a citation or some more evidence.
Mark Anderson: I mean, it’s one thing I think that gets very problematic in Wikipedia, where where anyone who just wants to be a pain in the ass can just go around saying, this needs citation. I mean, any fool can do that. It’s like just throwing a brick through a window. It’s easily done and it’s somebody else’s job to clear up, but it’s not helpful. Because not everything is necessarily easily cited in Wikipedia cases, a nightmare because the normally the only people who know the answer are, are deemed to not be allowed to comment on it because they’re connected parties. But that’s a different sort of insanity. And that shouldn’t normally occur in, in paper writing. But I found the McLuhan thing, by the way, because everyone talks about the medium is the message that’s, you know, his, his own thing. And although it’s mentioned as a chapter title in Understanding Media, and normally that’s given as the quote it’s actually said in a book called The Medium is the message from 1967, which is out of print and rather hard to obtain, which I suspect next to nobody is in the current age is read which which just makes this point about the difficulty of, of doing, you know, accurate, accurate citation.
Speaker6: Then the question becomes.
Frode Hegland: Why do citations? And as we’ve heard from Peter Law and from our sciences and humanities is all very different. And of course, in a community like this, different yet again. You know, when you’re. The thing is, science is supposed to be. At least that’s what my examiners keep telling me. Some kind of scaffolding thing going on.
Speaker6: And that is.
Frode Hegland: Fair enough to an extent, but it does to me stifle creativity as well. I think it’s absolutely legitimate to say what I’m saying here now is personal opinion, something like that. As long as you don’t do it a lot, as long as it’s clearly presented as such.
Speaker6: You know
Frode Hegland: My intuition is that my feeling is that. Nothing wrong with it, but it shouldn’t be the meat of the paper that I accept.
Speaker6: But we should.
Frode Hegland: Find ways to allow.
Speaker6: You know.
Frode Hegland: Okay, so we’re talking about different ways of compressing and summarizing and presenting a paper. It probably very useful to have a paper where somebody quite stern can say anything that is just opinion, remove it or show it. Somebody else could say for that paper, if it’s marked appropriately, show me just the new stuff, etc.. The new thinking. So it’s an interesting topic, and it’s very rarely properly considered because it’s mostly thou must cite.
Leon Van Kammen: It’s very interesting because the. Hypertext link in this, you know, in this whole story is a very oversimplified. Thing. Compared to what you’re talking about. Because if the idea is to sort of have this scaffolding. And this knowledge. Fundament. Then a simple hypertext link is to universal to fluid. To to give you what you’re talking about. So yeah, that is an interesting Yeah, it puts the citation in a in a very interesting spot.
Mark Anderson: So I think link types offer. I mean one thing that another sort of go at link types which which got sort of pushed away some years back that would be interesting is I find I find it less compelling. Things like I agree with, I disagree with because that’s why it went wrong the first time. People being overliteral about things. What I think is interesting is, is shading the the less clear things. So in a sense that it’s giving an indication as to why you’re doing that. I mean, it should be broadly be self-evident where you’re saying something for or against the text that the the convention would be the run of the text that you’re writing should indicate how the how the citation is being used. I mean, I don’t mind if somebody does want to do a link type of that type, but I don’t think it’s where it where its value in the, in the immediate sense lies. I think it’s being able to point to just the sort of ambiguity that you’ve you’ve raised in that. So we assume that you can point to the actual causal fact. Turns out it’s sometimes more difficult and a little extra. So effectively a little extra gloss by terms of the link type or some other metadata saying, well, this is you know, this is as sure as a handhold as I can give you towards the thing you’re looking for.
Mark Anderson: And anyway, because you don’t always you don’t always want or need to follow back to source, you’re normally doing it because well, maybe you complete disbelieve, in which case you’re setting out to sort of make a counterargument or you are actually learning in that subject. And what you’re trying to do is sort of you’re trying to sort of, you know, get the overall picture. But often once you have a relative knowledge of, of, of a domain, you’re probably just saying, okay, that I’ll take that on trust from this author. And you’re not even saying that out loud to yourself, but that’s essentially saying, okay, fine, I, I, I get you, I get you sort of given me something to look at. So there’s a scaffolding there if you need it. It’s not a scaffolding without which the whole thing falls down. And I think sometimes, especially in the, in the empirical sciences, people are over literalist in the way they use the term scaffolding. I think it’s a lazy, intellectual excuse to make life difficult for other people. It is important that you help the reader and that you both explain yourself and you help the reader. That doesn’t mean turning it into something. That’s just a process used to make life difficult for the people who don’t know the rules yet. But then we do that so often in life.
Speaker6: Let me ask you this.
Frode Hegland: Question, Mark.
Speaker6: Yeah.
Frode Hegland: The proceedings that we will go through in a few months. How will you read them? What would you try to find out? How will your what will your criticism be as in critical angle to just walk us through that, please, if you don’t mind.
Mark Anderson: But I think thinking of if I use as an analogy, I mean, when do I read the time I sort of read proceedings is for, well, the conference I go to most, I go to. So I sort of know what’s there. The but I still flick back to it because they’re stuff I didn’t, didn’t see. But more interesting is there may be conferences or journals that sit at the edge of what I’m interested in. So with those I sort of go through and it is initially saying, does this, does this mean anything to me? Which isn’t to say. Is this well written or something? But. But you know, should I, should I spend time on looking at this because rather like the old Homer Simpson joke, you know, everything I push and put in my brain pushes something else out. And that is true to a degree. So, so in in choosing not to engage closely with someone, I don’t think it’s it’s actually making a negative assessment. It’s about relevance. So in the ideal world, if someone’s written a really good abstract, which few of us do you get a pretty good indication. And so the first pass is probably to find the, the most interesting stuff and probably engage with that and, and then follow outwards as, as time allows. But I suppose the first flick is really is to look is almost is to look through the titles. The next might be to take a mosey through the abstracts.
Mark Anderson: And given the way you have organized things in your tools, it makes it easier to sort of loop through first pages. That becomes an easier thing to do. So, for instance, on something like DLA it’s remarkably hard to go to the web page for a conference and essentially just look through the the abstracts. Because of the web design. And I won’t go down that rabbit hole. I mean, we can argue web design all day long, but, but, but that library is not designed to make it easy to browse the contents. Which is a shame, because arguably that’s one of the things scholarship should be there for. And and as the ACM is a scholarly organization rather than a commercial publisher, although it is its own publisher. When they rebuilt their library, they could have made it so much better. So titles, abstracts and then and then it’s really just sort of delving into it and sometimes, sometimes you’ll read a part and skip, you know, because the, the narrative doesn’t grab. Or you might think, no, I actually want to come back and sort of read this deliberately. But if I, if you’re in the sort of just skipping reading mode, think, okay, okay, I’ll go and skim some other stuff, but I’ll come back to this one when, when I, when I can concentrate on it. I don’t know if that helps.
Frode Hegland: Right. So do we agree? And obviously, Dean will have to weigh in on this, too. That. The collection of proceedings or papers in the proceedings, rather should be maybe the most, maybe the key of what interaction we’re building. Because it will be new material. Relevant material. Linkable material.
Speaker6: Yes, it’ll.
Mark Anderson: Be useful to strand it. Because because future of text is, is such a sort of open, inclusive pick your term. A subject area. So we have a very wide range of inputs often and regardless of whether we have more in some categories than others, it’s more the point of what it is probably quite useful to be able to do is some more key wording than perhaps we’ve done in the past. Which doesn’t. It’s nothing to do with. It’s not how. It’s less to do with initial organization. It’s more to do with how you might want to consume it or explore it. And this, of course, lends itself to XR or XR use because, for instance, you might just want to pull out all the topics on a particular subject. The thing I’m conscious about with this is key wording is something that nobody wants to do, and most of us do badly, myself included in the sense of certainly when we keyword our own stuff are the mistake we we, I think generically make is we we actually capture our own understanding or misunderstanding of what we’re doing. And that is the last thing that the person who doesn’t know about what we’re doing knows that thinks to ask for. So our challenge is when describing our own work is to think about, well, what would somebody who knew nothing about this think to ask for or look for and that that’s tremendously hard. Which is why I do think it’s almost useful that when all the papers are in that some well-intentioned cells step up to the plate and at least do some key wording because if it’s a couple of people there, there’s a, there’s a chance to actually do some, some general averaging Freud. Sorry. And I see Peter Scott as well.
Speaker6: Just really.
Frode Hegland: Briefly. And my recent experience, this is something I does quite well because there’s two kinds of key wording. One is topic where it does have to do a bit of guessing and you need a human to verify. But then there’s also the thing of entity extraction. So if somebody mentions Bruce Horn for instance, that’s not going to go in the ACM keyword list. But it may be interesting to us. For instance.
Mark Anderson: I would color myself slightly less convinced about entity extraction. The problem is not that it can’t find the entities. The problem is it doesn’t have it doesn’t yet. Or it may be to do with the size of the model. Doesn’t always understand the context of the entities to one another. In other words, it finds all the entities and they’re all given equal weight. And that often is. Not necessarily useful to the human user of that information. No fault there. I mean, I think it’s a thing of the state of the art.
Speaker6: I think we’re.
Frode Hegland: Talking about slightly different things. Let me just bring up the screen here on that particular. Topic, right? Just share her.
Speaker6: So on this.
Frode Hegland: Slide. These are the key words that Gerald. Claude, my new good friend found, right? It’s actually pretty good. And I completely agree that it’s useful to have a human delete or add. But you know which? Semantic web. Semantic web.
Speaker6: You know.
Mark Anderson: Sorry, I.
Speaker6: Yeah, I just want.
Frode Hegland: To. The thing you said about. Wait, they do all have the same weight. I think it’s for the reader to choose how they want the weights dealt with.
Mark Anderson: Yes. And I take your point entirely. And and in the next year I say it’s not by way of pushback. The problem is that’s rather that’s rather saying, oh, well, you know, not my problem. Let the reader do it. But actually what the reader does need is the reader does need some help. Otherwise what they can get is they get too much un structured information, which means that in many cases people just don’t engage and walk away. So in other words, there is the appearance of a rich amount of information to use, but actually it’s too much. It’s not very usable. Nobody’s fault. It’s not done by intent. So we shouldn’t look at it in those terms. But it’s a genuine challenge, which is why I agree. I mean, I think the keyword extraction in is much more useful. I also think some human thematic review, because even if you do first pass, they say, well, we decide, right? What we’ll do is we’ll just get we’ll do some AI extraction of keywords. I have no problem with that. I think it will be useful then to get a couple of people to look through it and just sort of do some grouping because each keyword will emerge in the context of its own paper. Funnily enough, with a system like AI and LM, there may be some optimization. In other words, where where the keywords are are you don’t end up with lots of keywords, which effectively different words for broadly the same thing or slightly versions of them. But I still think some human review is useful after, you know, after the papers have been done, after their set is just to look through and say, right, where are the main thematic groups as opposed to the bucket of all the keywords?
Frode Hegland: Yeah, but then it depends how much they are surfaced, so to speak, because I think a lot of metadata is fine just to have. And then it’s up to the user choosing the relevant interaction or visualization to bring them forward. They shouldn’t be there all the time, that is for sure. But if you’re interested in a specific weird bit, it might be there anyway. This. This is a useful ongoing discussion. I don’t think it’s very black and white. Peter, please. Please.
Peter Wasilko: Okay. I wish we’d move beyond mere atomic keywords, and we would allow people to provide descriptive phrases instead. And there was one tool, and I cannot, unfortunately, for the life of me, remember what tool it was I was using, but it allowed me to add keyword tags to entries in its database, and I was doing them grammatically as phrases, and everything was going along swimmingly until there was a version update, and they got the brilliant idea that will simply alphabetize all of the keywords that have been attached to a document. So suddenly, all these beautiful little descriptive phrases that were allowing me with granularity to pull out the item I was looking for, got wiped out and merged together in the alphabetized list of keywords that were attached to the entries. Okay, Leon, I think you have something on this point. I see him waving or.
Leon Van Kammen: Yeah, I’m leaving. So thanks a lot. And keep up the good work here.
Peter Wasilko: Cheers. Okay. Bye, Leanne.
Frode Hegland: When will we see you again?
Leon Van Kammen: I’m sorry. What was that?
Frode Hegland: Will we see you Monday or Wednesday? When will we see you?
Leon Van Kammen: Probably Monday. Yeah.
Speaker6: Okay. Awesome.
Frode Hegland: All right. Have a good weekend.
Leon Van Kammen: Cheers. Bye bye.
Speaker6: Take care.
Mark Anderson: I have a thought for Andrew listening patiently to us rabbit on about citations. One thing I just thought, you know, thinking on terms of the exile stuff and the design is looping back to something we did touch on earlier, which is this the association of a part of text with images and or tables. I’m wondering whether there’s anything useful to do there. Even if even if we start with a sort of slight cheat. In other words, we we we pretend that the metadata is or we pretend that a suitably structured document is there. But I just don’t know whether that would be something that might be useful, I don’t know. Andrew, do you want to.
Andrew Thompson: Yeah. I’m trying to sort of grasp what you’re you’re asking, and I’m not quite piecing it together.
Mark Anderson: Sure. Okay. Let me. Yeah. No. Exactly. So this is really the reason for asking because. And you and you’re right to to say it is perhaps not clear enough. So let’s let’s pause it. We we’ve we have a paper. We’ve, we’ve we’ve essentially decided there’s a paper we’re going to look at. And the idea comes out from the comments being said sort of before along the way. Well, you know, what, have we showed all the pictures or we showed the figures or we showed all the tables. And I was thinking, well. If we assume that you know, figure one and table two were associated and they were associated with a certain amount of the text, how might we display those together? Because in a sense, they were the author wrote them with the intention of. So the illustration and the data table were provided to provide additional insight or evidence for the writing. The publication process, like as not, has put the the figure and the table on different pages and not next to the text. Now they may be together, but but you know, Murphy’s Law says normally they aren’t, because that’s just how it happens. So one of the useful things that we could potentially do to aid the reader in an XR space where we have control over what gets displayed, where How might we usefully pull those together? And.
Mark Anderson: Okay, I don’t know. I, you know, I’m even thinking about the affordances. Where would you start? Would you. It’s the kind of thing is. Oh, could I could I sort of interact with a with a picture and somehow know that there is information, text table, whatever associated with it and choose to see that foregrounded? I’m mindful that the text is always going to be part of the broader narrative. So what you’re what you’re where this would have most use is where you’ve got maybe a chart of a table. And the meaning of the table is described in some text. Now, seeing those together would be useful because it what you get is you will probably get a table with a caption, and the caption might have a subtitle that gives you some idea. Or in the graph you’ll have a, a sort of very abbreviated subtitle, maybe to the to the legend in the chart telling you what’s there. But being able to put those together to get you the meaning that perhaps the author described could be useful. And so my open ended question is how how we might explore that. With one of the implicit questions being what would you need by way of raw material to start playing with such an idea?
Andrew Thompson: Does that help? The first thing is we have. No real way of rendering the images at the moment. The document is read fully as text, which mostly works. I know that some ACM papers do have images and tables and stuff like that. Tables right now are importing as just like kind of a string of text that’s not formatted. And images don’t import at all.
Speaker6: Okay, which.
Andrew Thompson: I understand is not ideal, but also Unsure how we can approach that with the current direction. We’ve gone with everything. I would say, just like, sort of spitballing here. If we wanted to find a way to import that and we needed the data, it would probably be some kind of maybe we’d grab the image source since this is HTML and we’d have that. And I wonder if I could turn the image source into a. Material and render it that way because I can render. Images as materials and objects so that that might work.
Speaker6: Yeah. I mean.
Andrew Thompson: We’d have freedom to start tinkering with.
Speaker6: Yeah, yeah.
Mark Anderson: No, it’s very useful. You’re surfacing this because it really just shows actually the challenge, the challenge in this. Because in our mind’s eye. So we have pictures. We’ll just put pictures over there and that’s, you know, that seems entirely logical. And you’re rightly pointing out that, yes, we can do. If we had the stuff which we don’t yet have, but could we not if it makes if it makes it easier to explore the problem. I think most of us are flexible enough to say that if there’s basically a box that I have a label saying image, we can kind of live with that, because what we’re exploring is, is less less the fully rendered visual. But the exercise or the problem associated with reassociating these, these insertions within the text, in this case, images and or tables with their associated body text. So to a certain extent it’s not it’s nice, but it’s not absolutely a necessity that we see the beautiful picture or the, you know, the table as a table, but it’s almost a sense of thinking about how the transition. So. Because what’s at the heart of this is, is, is such an affordance actually useful? Because if it isn’t useful, there’s no point of doing all the extra work. You know, even to see the picture for sake of argument. Because whilst it seems exciting to pull this stuff out, I’m just wondering How useful it is. Does that? I mean, does that sort of help shade things any.
Andrew Thompson: Yeah. I mean, it’s certainly a good question and not one that I have a particular answer to. On is it is it interesting or is it useful? I don’t know. Right. It’s one of those things that we could try to implement and then test. It’s also one of those things that would take a lot of time to implement. So it’s really just a choice of Is the time sync worth the test? And the answer might be yes. I try to just prioritize each new feature based on what the group as a whole sees as the most important, because ultimately you all come up with way more ideas, much faster than I can implement them. So we have to choose what’s what’s going to go into the project.
Mark Anderson: Yeah. And just to be clear, before I, you know, I just I hope it comes across. But yeah, I’m absolutely not saying no, no, go do this. I’m deliberately trying to explore the, the the I, the idea so that you can make exactly the observations you have about its impact in terms of you know, time taken and things it needs. So thank you.
Speaker6: Yeah.
Andrew Thompson: For my my first sort of impression on what might be interesting with images if we render them. And that’s a big if would be perhaps as we’re scrolling through the document, instead of having the images appear in line in the document, maybe they sort of pop up to the left of the document in like a larger scale with the, the figure text under it. And then as you scroll down the document, it would switch to the next image. It basically just shows any image that’s within a certain range where it would have rendered it then shows up on the side it’d be almost like a picture book of sorts, where the left page is always the image full sized, and then the right page is just text. I don’t know, might might be interesting since we don’t often see things that way. Thank.
Mark Anderson: Thanks. I’m done. Freud.
Speaker6: Muted. I would have thought that would be a benefit.
Frode Hegland: Right. So, Peter.
Speaker6: I just dropped in.
Peter Wasilko: I just dropped on the sidebar a little example of what I had in mind by having the multiple keyword sequences, as opposed to just alphabetized bags of keywords. That’s a little blurb in the sidebar. Take a look at.
Speaker6: Thank you.
Frode Hegland: Just to make sure. Bach last year’s hypertext conference. We have HTML versions or just PDF.
Mark Anderson: Now suddenly any PDFs because we didn’t use taps.
Speaker6: Okay. But the year.
Frode Hegland: Before we have.
Speaker6: Right.
Mark Anderson: We do 19, 20, 22 and I think it’s either 2021 or 2022 was when ACM basically just suddenly turned this on.
Speaker6: So that’s fine. That’s that’s fine.
Frode Hegland: But the key is this year we will have HTML.
Mark Anderson: Yes. Don’t remind me. That’s my headache to organize. Yes.
Frode Hegland: So what I propose we do and obviously have to run it by Denny. I think we agreed on this on the last call. Call? But I just want to make sure is we tell PDF to go jump in a tree, and we work with HTML from now on. Either, you know? Okay, so I’ve been using Claude and other AIS to get text out of PDF, and it’s pretty damn good, especially when I tell it. Look, be careful, PDFs are awful. All that human stuff. It actually vastly improves the quality, but it’s still not as good as an original HTML, because what Mark was talking about, the relationship with images and text and all that stuff. So, Andrew, for you to work with HTML will be easier than trying to extract stuff and display PDFs, right?
Andrew Thompson: Yeah, for sure. That’s all I’ve been working with so far since we switched over a while back. Right. Of course it’s the HTML renders of the PDFs, but yeah.
Frode Hegland: I don’t think we need to render the PDFs. We may, in some views have literally thumbnails if we have a ton of them. But Dean is concerned about other people especially older academics looking at this freaking out because it doesn’t look like a PDF I’m not worried about, for the reason that we can very easily make the HTML look like PDF, just render it, you know, in roughly that format. And also our prime use case is the ACM Hypertext Conference, not the Humanities conference. So I think.
Andrew Thompson: And I believe, I believe last time we discussed it, Dean was also saying that that was in line as long as we have like a view that kind of looks like a piece of paper with a white background and text. I think that accomplishes it. Yeah. And of course, like the source originally stemming from a PDF, which we also have.
Speaker6: Yes.
Frode Hegland: I think exactly that we’re on the same side same side or whatever in agreement with that. So that means that the thing. Okay. I’ll just Peter. Go ahead. Well, I’ll show a few slides.
Peter Wasilko: I just have a question for Andrew. Do we have the ability to have, say, a headless browser, render something generated for the web, and then get that rendered image imported into 3D space and display it on a object? It’s like a dynamic surface texture.
Andrew Thompson: You can do canvas rendering, which might be what you’re talking about. I don’t know a lot about it. Some of my early tests, back when we were using PDFs themselves, I got the the PDF to render as a solid plane so you could see the PDF as it should look. It was essentially an image texture rendered through canvas. It was useless since you couldn’t do anything to it. But visually you can get them to show up. If you want something that’s like if you’re talking about, like, streaming, almost like a screen that has, like, updates.
Peter Wasilko: Yeah, basically what might.
Andrew Thompson: Be possible with shaders. But I want.
Peter Wasilko: Like a conventional web browser component that can be living in 3D space. So. Oh, you went with the conventional web browser generated user interface, but from inside the 3D world.
Andrew Thompson: Sorry, I misunderstood you. Yeah, I can’t do that at all. In three.js, you would need like an entire proprietary software for that one, unfortunately.
Peter Wasilko: So maybe we could ask Apple to provide that for us. I mean, it seems like it’s such a critical thing to be able to use web tech to generate interactive 2D user interfaces for control surfaces or whatever, and be able to use that in the 3D world. Otherwise, we’re going to have to go basically reinventing all of the existing tech from the 2D web in order to build really nice state of the art user interfaces. To work with in 3D space.
Mark Anderson: That’s interesting. And I can I can see I can see why that would get some pushback, why people said, well, we’re trying to get rid of all that. But I think you’re right. I mean, for some things, I mean, this is an interesting edge case, this whole thing about because it’s about reading in what otherwise many people would just see as a space you experience. But just that it’s a really I think it’s a really interesting use case because the documents are inherently complicated, even if they don’t appear that way on the surface. And it has implications, as we talked earlier in this conversation about in the longer term for better or a different class of authoring tools that, that, that were that would be written with the ability or with the intent of, for instance, Z use in mind. But we don’t have those yet. I mean, it is interesting how how I mean, I was thinking today about the fact that, well, you know, if any PDFs actually had say, say, just had the entire text, not non-rich text of a document, but it still doesn’t help you for figures. It doesn’t help you for tables because, you know, tables are just as as sort of what something like tab columned text or something are still quite hard to read. And and we still use we still use with deliberate intent a minimum of, of emphasis. So we use bulk, we use bold and italics still and occasionally underline broadly bold and italic still have meaningful affordance for most most readers, and which obviously plain text can’t do.
Mark Anderson: What plain text does give us is is a clean text that we can play with digitally. But and of course, the advantage for, for whatever it doesn’t give us one, one thing HTML does give us is it gives us both the semantic structure in, in, in terms of the bits of the document. And, and with that, a form of block addressing, so fine grained addressing and it also gives us the, some of the stylistic layer which we can use or not as we so wish. I think that’s the interesting thing, whether it’s called HTML or something else I find myself still less invested in. But I think that the the the lesson from looking at the sort of the starting materials we have is that. To do what we want to do, given how much work is. So Andrew keeps showing us, you know, turns out the stuff, the easy stuff is maybe hard and the hard stuff is easy. It wouldn’t be the first time, but so the things that we might assume we can do just don’t turn out to be simple and easy. So they’re not going to happen at scale fast unless a really big player decides they’re going to, you know, make a change. But but actually, given where we are in terms of most of the academic corpus, probably being in in PDF, actually, I know that’s not, you know, not going to happen. And they’re not worried. I mean, we’ll deal with that. But going forward I think we can begin to do some exciting stuff.
Andrew Thompson: Mark. That’s actually I like your little comment there. Like, perhaps the easy stuff is what’s hard and vice versa, because that’s very much what it feels like. Anything that’s easy essentially is or like assumed easy is things that we are used to. But to put them into three JS, we have to reinvent them entirely. And then stuff that is hard, which is the more uncommon things. What? We’re making it for the first time. So it’s, it’s much simpler to put together. And it just it’s strange how that works out.
Mark Anderson: The one thing useful reflection on that is if you see anything that just proves to be ridiculously easy, it’s probably worth flagging that up really hard in the sense that we don’t. I mean, we don’t know if it has a purpose yet, but now we know it’s really easy. Maybe we can find we can find more to do with it.
Andrew Thompson: Yeah, yeah, hopefully I’ll be keeping my eyes out.
Speaker6: So in the last.
Frode Hegland: Few minutes of today just wanted to show you a few slides real quick on this topic. We just looked at this, and this is just a notion that we should be able to do the selections that we talked about earlier today to sort of render it this is a thing that I just have an author that highlights only where you are, which is very good for close reading when you’re getting a bit confused. This is having the text in different ways, as we saw. But here’s the thing. I just wanted to end on this.
Frode Hegland: The proceedings are now in many library in a sense.
Speaker6: Thank you.
Frode Hegland: So we really I mean, Mark, particularly you with your academic work, I would really like, based on this kind of stuff, for you to write down a workflow of, you know, you arrive at Potsdam, it’s new. Suddenly the proceedings are available. What do you do? That kind of stuff, you know, that you’ve already done. That would be very useful because we’ve skipped the middle bits. And, Andrew, what I’d like you to not necessarily change any work you’re doing. Just keep in mind that we have this thing called the proceedings slash book. So when we are in your environment, we have currently one document or a library. We need to start thinking more about the intermediate or intermediary thing, which is a collection, which is what the proceedings are.
Speaker6: And that, I.
Frode Hegland: Think, becomes much more. Interesting because finally I go back. These kinds of summary things become useful. Because if you’re only reading one paper, for crying out loud, chill out, dude, read it. But when you present it with a whole pile of new papers, then it’s a bit like, okay, I need a better way in.
Speaker6: So I think that’s.
Frode Hegland: What we already talked about. Just wanted to highlight it and say thank you. Anyone else have any.
Frode Hegland: God latex.
Frode Hegland: Talk to Mark about that. Anyway, anything else? Or will we all just say bye bye until Monday?
Peter Wasilko: Great session. See you Monday. Yeah.
Mark Anderson: My apologies. I it’s national holiday here, so I won’t be here Monday, but I will be here next Wednesday.
Speaker6: Holiday
Frode Hegland: Okay. I’ve heard about these things. See you later, guys. Bye.
Chat log:
16:09:26 From Frode Hegland : https://futuretextlab.info/academic-readings/
16:14:21 From Andrew Thompson : I’ll be working in the background today, so I’ll be sitting with my camera off for the most part. Still present.
16:16:41 From Peter Wasilko : https://dl.acm.org/doi/10.1145/108844.108870
16:20:00 From Peter Wasilko : https://dl.acm.org/doi/epdf/10.1145/108844.108874
16:24:42 From Frode Hegland : Reacted to “I’ll be working in t…” with 👍
16:56:20 From Frode Hegland : Citations are a toolset
16:57:24 From Peter Wasilko : https://www.law.columbia.edu/sites/default/files/2022-06/WC%20Handout%20IRAC%2C%20CRAC%2C%20CREAC.revised%205.22.pdf
16:57:32 From Mark Anderson : Reacted to “Citations are a tool…” with 👍
17:00:02 From Mark Anderson : [sorry for sitting in the darkness here, working through a visual migraine, but don’t want to drop off he call. No sympathy needed!]
17:01:47 From Mark Anderson : Sounds like a task for an AI helper!
17:08:01 From Peter Wasilko : I FEEL THAT PAIN
17:20:07 From Frode Hegland : Explicit and implicit citations and links.
17:24:56 From Frode Hegland : Reacted to “[sorry for sitting i…” with ❤️
17:26:48 From Peter Wasilko : https://web.archive.org/web/20130406022406/http://limingzhu.posterous.com/how-to-reject-a-paper-without-reading-it
17:28:46 From Peter Wasilko : https://sigbed.org/2022/08/22/the-toxic-culture-of-rejection-in-computer-science/
17:30:52 From Peter Wasilko : https://cacm.acm.org/blogcacm/bad-reasons-to-reject-good-papers-and-vice-versa/
17:38:08 From Leon van Kammen : unfortunately I have to go.
17:38:18 From Leon van Kammen : Thanks, very interesting conversation (as always)
17:43:44 From Peter Wasilko : I want to capture concepts like:: Papers translated-from German about User-Interfaces implemented-in Java inspired-by Smalltalk written-in The-80’s
17:48:14 From Peter Wasilko : So we I want a stream of Referents (entity classes or instances) connected by Relations with potential duplication of relation instances and relation name over-loadings that would be distinguished by their referent classes (e.g. in London v. in German being treated as in-location v. in-language)
17:50:22 From Frode Hegland : HTML
17:50:26 From Peter Wasilko : Can we get LaTeX source from anyone authoring in it?
17:51:37 From Mark Anderson : Replying to “Can we get LaTeX sou…”
Possibly
17:59:47 From Andrew Thompson : Take care all!