 So you should see a browser on the right and on the left. You should see what turned out to be the major categories of the, I guess it's basically the J-Wiki server file system. So this is the default set of major categories. And the way to read this, hang on. The way to read this is there's a history, excuse me, not a history, I have a bar chart running down the left side of the table of contents. And where you see a little bar on the left, a little red bar, that indicates that there are more than one document in this major category. So in personal taste, for example, there's only one, it's called personal taste. In P Invoke, there's only one, it's called P Invoke. And as I mouse on it, it loads up on the browser on the right. So in that sense, it's very similar to the other demo. New York City J Users Group has a bunch of documents, so it's got a larger bar. I'm gonna start at the bottom, just because there are some points I wanna make. So by default, we can fit as many as three columns of documents in the detail area on the left. So these are all J-Wiki meeting reports going back to December of 2021. We've got two columns worth of material. I am familiar with these. I would imagine. Get to something a little richer, like vocabulary. There are actually six columns, and we can't accommodate those at full width. So what we do is the first column is expanded to full width and there's a gray ribbon running down the left, running down the column. And as long as your mouse stays inside that ribbon, this is the column that will remain expanded. But if you leave the ribbon to the right, the next column expands next, and the next, and the next, and the next. And the idea here is to try to accommodate larger numbers of documents and larger numbers in subcategories. Like in this case, the absolutely essential term subcategory, the about new book subcategory, or it's plus and whatever else. If we move up a little, there is a category called system. I guess that's currently making the point I wanted to make. The basic notion here, the basic notion in general is to try to take all of the subcategories of a major category and all of the document titles of the major category and flatten them out in the detail area. And we can do that as long as there are too many documents and too many subcategories. A point that Raoul made last week that was extremely well taken was that the original implementation of the original prototype wouldn't scale very well. If we suddenly double the document count, for example, or if the subcategory structures underneath the major categories got richer, it wouldn't really handle it very well. What I'm trying to do here is show that there are ways to handle that well and still preserve a lot of the interactivity that characterized the original demo. We come up here to help. Help has a lot of documents. And if we open it up, it has so many that we can't actually accommodate them all in columns. And so the approach we take there is simply to take the subcategories and arrange them in an outline form. So we indent as we get to subcategories. And as you touch each subcategory, we arrange the documents in the remaining two columns. Now, a fair question is what happens when you've got more than two columns worth of documents on the right? Let's look at that. If we go to all of these categories have been the default server file system categories. If we move up to the top, we get to Bob's categories. So community developers, J Playground, newcomers, references and a Wiki. And Bob, I realized just today, I may have a brawler glitch. I may not have gotten everything. I apologize for that, if that's not to be true. So if we open up references, it's huge. There are a lot of subcategories, but we have no trouble accommodating them all. And what happens is we squeeze the columns. So release notes has five columns worth of document titles, but that's okay. They'll expand as we move to the right. And as we touch them, we'll load up the pages. By the way, part of the reason the pages are loading as fast as they are, is that I've got a local HTML cache that I'm using. So I'm not actually touching the network when I load up these documents. All right, but there is one incredible category, subcategory, and that is orphans. Orphans is, as far as I can tell, all the otherwise uncategorized pages. And orphans has well over a dozen columns worth. There are several hundred pages. And it still works mechanically. The columns expand appropriately as you move them out to the right, but it's not a great experience. You really can't read enough of the compressed columns to be able to see anything, to guess at what they might actually contain. So I think probably, there are ways to deal with that. I think probably one way to do it would be to take orphans and create a few synthetic categories underneath that category is called page one, page two, and page three, for example, each of which would have just a couple of hundred pages in them. I think that would take care of the problem, but I'm sure there are other approaches. Yeah, we could always add real categories. I'm sorry? We could always come up with new categories and start sticking them in these new categories or existing categories. Sure, I think it's important to distinguish between curation on the one hand and experience delivery on the other. And if your experience is failing, there are two approaches. I think the honest approach is to say, I'm gonna fix the experience. I'm gonna make the experience more sophisticated so it can handle it. And I think the forgive me slightly less honest approach is to say, well, it's Bob's problem. Let him fix it. Okay, so... I'm not as enamored of that approach. I'll just stop you there because in the case of orphans, it is definitely Bob's problem. And I'll tell you why. Because what orphans are are just any page that doesn't have a page that refers to above it. So if you think about all your navigation bars, nothing goes to a navigation bar as a parent. It's its own thing. So it's an orphan. Now what you see listed is every page that navigation bar touches. And we're looking at all the navigation bars, including NuVoc. So there's no way, like orphans was useful when Eric first went into the web crawl because it allows us to identify pages that might be off on themselves and they're not hooked to anything. But orphans also applied to navigation stuff, which is there's no way you wanna do references into navigation, especially when navigation itself, you basically are going to the thing, you're not just going to the top level, which is a navigation bar, you're going to everything that that's touching. It's just not a useful category. Yeah, all right. So don't worry about it. Okay, I won't, we'll do a category nav, hidden category nav or something like that. Something like that. Yeah, we could do that. We could split them out that way. Yeah. Now I wanna contrast this style of navigation. And when I say this, I mean being able to touch both super categories, major categories and within them subcategories and immediately see all the kids and then immediately load up those kids with the approach that's taken by media wiki. And I don't wanna single out media wiki, but it's because it's typical of the web. It's just the way things are. So you've got here a hierarchy and you can click to open up individual sub topic, subcategories. And if you click on a subcategory, you get not pages, but a list of pages, one of which you could click on. And then it's incredible. There's only one page here. They could have shown that on the hierarchy page. I can dive into here. And the feeling I have is that I've just lowered myself into a cave system. And if this were a working browser instead of just a WD web view, I would use the back button to laboriously crawl my way out back to the listed subcategory pages, back to the category outline as well, which it's actually easier to get there like this. The contrast between that web style of navigation on the one hand and this style on the other were even a very small mouse movement. And by the way, I'm not clicking. There's almost no clicking involved in this interface. A very small mouse movement opens up whole good vistas of information for you to explore. The contrast between the two approaches increasingly strikes me as very stark. And I think that a big part of the reason that video games, certain kinds of video games are so engaging is that they're just incredibly responsive, incredibly fast. And there's something intrinsically engaging about the mere speed with which they respond to your input. And I think the same sort of benefit of engagement could be recruited by an interface like this one. I'm definitely conscious and this is a bias and individual data point. I've been exposed to more content in the J website in the last week than I have in the last couple of years. It's just so easy to move around. Now I'm just satisfied with one thing, which is I'm moving up and down here selecting pages and they're not loading in the browser. I would like to figure out why that is and fix it. The left side of the application responds like a video game. The right side responds more like the web and that's annoying. That's not as engaging as it probably could be. And I would like to fix that. Okay, real briefly, let's talk about search. So search works, it works more or less the way. It works more or less the way it did in the last version. So you type rank and now the bar graph on the left shows where the hits are. So there are hits down in vocabulary for obvious reasons, quite a few of them. There are hits up in reference, quite a few of them for obvious reasons. And so Orphans has a lot of hits because Orphans includes Nouveau, release notes and so on. I kind of like the idea of putting hits in the context of a map which should become familiar but I'm not married to the idea. So there is a search pseudo category up at the top and that actually shows all the hits that have happened and they're categorized. So it effectively works out to the same thing except rather than poking around in the category map you get to see them all flat right in front of you. So there's somebody to that. And then last point, there's a pseudo category called super category called Nouveau and that behaves exactly the same way as it did last week. So you mouse on a glyph to get a list of valence links and you mouse on a valence link to load it up. And that is the story. And I will just as a code mention that I've done this kind of work before and the best possible outcome of a demo like this one is a feedback that leads you to pitch everything you've done and start over again through scratch. So with that in mind, if there are any comments or questions I'd be very happy to take them now. Yeah, I have a question. Yeah. How do you scroll the browser for age? Okay, good question. If you click on whatever you're currently touching so I just clicked on sort down and I go and I load other pages. The page I clicked on becomes the default. So if I leave, if I'm not mousing on any page it comes back and then I can scroll. It's not the best interaction but it's the best I've been able to come up with so far. So if I clicked on raise or raise in, excuse me and then I go and bring up sparse but come over here, raise in comes back and I can scroll it and I can follow things and so on. That's pretty good. The only thing I would want is just a little more feedback that it happened when I clicked like a slight background change for everything else that highlights the one that I clicked on or something maybe. Okay, that's a good point. Yeah, it's entirely hidden that that would need to be made more transparent for sure. Thank you. And just to follow up on that to make it even a bit clear is there a way to do almost a film over the right side so that when you click on that that locked page now looks different than the other pages you're scrolling around. I see, possibly, I'm not really familiar enough with WD to know whether I can do that, but it's entirely possible. I wonder whether it might also make sense when you click on a page to put a button up above the browser, that's just that page title. So the last half dozen pages you clicked on become bookmarks basically. Yeah, kind of like, well, not breadcrumbs because you're not following your way back out but bookmarks is a good example. Yeah, yeah. Yeah, so I think it could be made more transparent. I certainly agree that it needs to be or would need to be. Well, I think that what you gain by making it more transparent and seeing multiple things change at one click you're reinforcing that behavior much clearer, more clearly. Yeah, like a little tab system of a tab queue across the old style browser tabs. Yes. Now, when we have this display here, if you click on the green heading, what happens? Nothing right now. Can you think of something that you'd like to have happen? No, no, I was just wondering, I kept seeing the headings, I thought, well, that's interesting. I mean, I can see what happens whenever you're over the black. I wonder whether it's more, whether it's possible when you look at my category trees that might be more useful to open a subtree. I'm not sure. I'm not sure I understood that, I'm sorry. Okay, well, the page you're on right now, reference. You've got to say J602 is highlighted. If you were to click on J602, well, say you clicked on anything, it didn't have to be highlighted. Anything you clicked on would open up the subtree, the rest of the subtree to that. So I have miscommunicated something. All the subtrees are open right now. If it looks like a subtree isn't open, that's a crawler glitch. So the whole part of, I was about to say the whole point, but it's really only part of the whole point is to show a flat representation of the entire sub hierarchy underneath the currently selected major category. So maybe it's a little clearer if I get rid of this. This is everything under references. Right, okay, okay. So what we were seeing was an artifact of the search. It locks you into the different spots. Yeah, yeah, one of my goals is to avoid drill down, avoid having to drop yourself into that dark cave complex. Yeah. I could really use some criticism. I could use some, this doesn't work. Otherwise, I'm pretty much out of ideas. But I mean, it's okay. If there's not, I'll take feedback. I'm just, I'll come up with something, you know? I've got lots of ideas that are just not good ideas. Well, no, you don't. No, that's absolutely the wrong way to think about it. Even bad ideas can be the start of good ideas. Yeah, that's exactly the way I work too. Well, before you had a three level hierarchy, and when you get a lot of columns like this, it might be worth introducing like pages that are like letters, you know, the initial letter as intermediate, and then on the right, it scrolls to that point in the list where that... So you're thinking of maybe page headings across the top, but as you select them... Yeah, like on the orphan... ...for sets of columns or... When you had orphans, go back to orphans here. It'll be in reference, I think. Yeah, right, so orphans. So here, if you had like a two... You had a... You have your orphans here. I guess it would be actually a... If you could split this... Instead of having one, two, three, four, five, six, seven, eight, nine, 10, 11, 12, 13 columns, split the whole thing into two and you have... One's a fairly narrow thing that's got two A, B, C, D, E, F, and then when you... Or two, it scrolls. It does it kind of like on the far left where you have the magnifying glass effect. Have you that on the far right for... Right, so you're doing it horizontally... You're doing it horizontally rather than vertically. Right, with a thin narrow column, it's a quick nav within that. All right, what would you think of the idea? I'll take that and I'll keep that in mind. What another notion that I've been kicking around was what if we introduced several synthetic categories under orphans that would basically be the same as what you just described? That way you keep up with the vertical navigation. So there'd be a page one, page two, and page three underneath orphans and invented. They don't really exist in the hierarchy on the server. They would be, as I say, synthetic as a way of splitting up the documents into a more tractable set of columns. Well, they're as synthetic as any other category in that whole tree. Absolutely, you could make them. The thing that happens there is over time, as things get added and subtract, your orphans migrate into different pages. So you have a little bit of... Maybe what we're really saying is we don't care a lot about them and we just wanna have a way of accessing them. In that case, it works. But if we also care about content in orphans, then all of a sudden it's gonna be hard to read to go back to where you were before after a few months. Oh, the synthetic categories wouldn't occur on the server. Bob would have nothing to do with them. They would be generated dynamically by the client. No, but that's why they would be, it's because they're dynamic that you lose track because if stuff goes out of orphans, it goes into orphans, it's gonna move stuff around. You're absolutely right, that's a good point. Although that's true in general of all of the categories that things move around, but orphans is gonna be particularly volatile, that's true. Pages in particular is gonna shove stuff will move to a different page as opposed to in the category, it will just move a different position within the category. You're right, but it'll still be visible. Interesting. Okay, all right, I'll think about that. Okay, this was probably, oh, go ahead. I was gonna say, I think this, I'm not sure where this is what Raul was touching on, but if you go back over to your left so that we see the layout for the orphans, what about if rather than doing a table, essentially, of columns, if you took your reduced screen, the horizontal lines and everything, and just put those instead of even worrying about seeing the content, except for as, I think we're talking about the same thing, but I think what I heard was Raul was talking about using the magnification to highlight a section of all those horizontal lines, but you don't need to see all of them at one time. No, that's true. Like right now we're seeing all the pages all at one time. That's not actually necessary. No, it works for smaller numbers of columns because you actually have a hope of figuring out what they're about, even when they're compressed, but you're right that in the case of the orphans that's really no point. And it's not like your muscle memory is going to kick in and tell you or remind you where something was because, again, orphans are gonna be pretty volatile. They're gonna move around over time. But if you go right over to your far left and you scroll up and down with your magnification there, it's going so quickly that if you think of that same thing happening off the pages linked to one of your green categories, you get through them very quickly. That's true. That's true. I don't think you feel like you weren't seeing some. Yeah, that's possible. The same is true of just moving horizontally. Horizontally, yes, you can all see them, but you're only, given this is a really large group, but you're only being able to see part of them with the large group. Smaller groups, it's not as important. It becomes more useful to see them side by side because you can see everything at once. But having such fast response when you go up and down, I'm not sure you lose that much. That's fair. And even more so if you're doing a search because the ones you wanna get to are gonna be highlighted. Right. You're gonna know exactly where you wanna go to. That's true too. Bob, I have a question for you and maybe I'm fishing here for something to do. What's the biggest problem you face right now? Right now, my biggest challenge, and I think actually in some ways, this may solve that problem because I think it's going to go away, it's gonna get around the issues I have with creating these category trees, which was my solution to it. But what those require is an awful lot of content put into an established context. And with having access to it. I'm not in my head, but I'm not sure what you meant by that, content to established context. Okay, so if I go to one of my category pages, and I think you touched on it once with one of those, what was it, an Android page? It only had one page on it. Right. At the top of that page, there's a little gray blurb that I've put in basically saying, this is a category in the future, this will be expanded to link to other categories for more information. That area can be expanded to provide context for everything that would be linked below. I see. So when I'm building my category tree, my biggest challenge is gonna be going to each of these categories. And in the case of J on Android, I'm sure there'll be other pages come in. But then I have to go and establish context for why we're going to see those pages. And I can even put the links for those pages up above. That's a ton of work. But I'm having to do that work because I'm trying to build an experience as you descend down your pipeline, right? You are moving away from that experience because it's so darn fast. Yeah, it's interesting that the pages actually, the actual pages come up about as fast as tooltips do in a conventional system. Yeah. Which is nice. I mean, we're habituated, there's a couple of things that I think we've been habituated to that are unfortunate in a way. We don't like to click on things at some subconscious level, I think, because there's a huge commitment associated with that. We know it's gonna load a page and that's gonna take time. We know that by loading the page, we're gonna lose our orientation, our context, whatever map we're looking at is gonna disappear. And we know we're gonna have to incur the cost of going back. So that original map page is gonna have to be reloaded, reparsed and re-rendered and that's gonna take time too. And then, of course, we have to incur the cost of reorienting ourselves because the map disappeared on us and then reappeared. I think we can get away from all of that. If we can, what we are comfortable doing is hovering and this is all about hovering. Again, there's no clicking. And I think that that recruits part of our interaction repertoire that we're a lot more comfortable with. We're very comfortable with hovering. And because hovering gives you the whole page, it doesn't just give you a little description of a page in the form of a tooltip, I think that'll dramatically increase people's tendency or willingness to explore. Maybe without their even noticing what's happening. I think you're right. Thank you. Yeah, I'd agree with that. And then the only, and I think you addressed it pretty well with the way you're using the clicking to lock down a page so that you can return to it. But what you do have to watch is to make sure that your click doesn't establish a new mode, right? So it kind of locks things in and now you're hovering but you don't know that you've locked something somewhere. Yeah. And then your next click will lock something else. So the language of clicking becomes more important and you have to maybe signal a little bit more strongly when hovering is your main mode, right? Yeah, that's true. Yeah, I need to think about the click. That's so far one of my two main takeaways and I appreciate it. And the click has tremendous power. It's like the deep click and what the Apple abandoned it, but it was a very powerful thing if people had started to use it that when you press harder, you get a different thing. It gives you another level of interaction. Of control, yeah, yeah. All right, well, this is longer than I expected to take and I apologize. Don't apologize very much. We're making it longer.