>> Susan Dumais: I'm Susan Dumais, a researcher in Microsoft Research doing a lot of work in information retrieval. It's my pleasure today to introduce William Jones, who's a professor in the Information School at the University of Washington.
I've known William for almost 30 years. He was a -- did his graduate work at
Carnegie-Mellon in cognitive science with John Anderson, did a post-doc with some of us at Bell Labs. And our paths have crossed many ways over those years. William's a large part of the reason that I'm at Microsoft from Bellcore. We've collaborated on several things since he's been at University of Washington.
And so today he's going to be here telling us about a body of work that he's been involved in for the last decade or so, I'd say, on helping people basically keep found things found, to improve the way in which they manage, organize and use information. What he's going to talk about today I think is some of the newer developments in that, something having to do with a project manager, and mostly just try to frame this work on personal information retrieval in a broader picture.
So, William, it's my pleasure.
>> Williams Jones: Thank you, Sue.
Now, Sue was one of the original members of the Keeping Found Things Found group back when we got our first grant. And so I'm writing an article -- I'm almost done with it, but I've been saying that now for a couple of months -- with the same title. And I--
>> Susan Dumais: And you have a book.
>> Williams Jones: Pardon me?
>> Susan Dumais: And you have a book by the first title.
>> Williams Jones: Keeping Found Things Found. I have a book called Keeping Found
Things Found: The Study and Practice of PIM . I don't know how many people have heard of personal information management. I know some of you have. Anybody else?
Ed, you certainly have. But I don't see a lot of people. So a lot of people are sort of like, well, what the heck is that? And that was kind of my reaction the first time I -- I -- let's see. Now, how come I'm not -- oh, there we go. Okay.
The first time I -- actually, Mike Eisenberg, the former dean of the Information School, suggested when I first started working with the Information School back when I was just an adjunct professor -- he said, Well, it sounds like you're doing work on personal information management. I described what I'd been doing since my days at
Carnegie-Mellon doing memory research. And I thought, what the heck is personal information management? That's sort of everything, isn't it, and nothing, and does it have anything meaningful behind it? And I've concluded since then that it does. And a lot of us have concluded that too, including Jaime, who just came in, who is -- she and I co-edited a volume on personal information management.
So basically I like the title because the world is at our doorsteps in a lot of ways, certainly digitally we've got all sorts of ways to get at digital information, high bandwidth, everywhere, not just at home, but when we travel. And we can also change our physical
world too. Most of us make reservations online these days. And I did a thought experiment with a class at UW where I said, Suppose you could -- if somebody gave you
$500,000, if you would stay in your apartment for the next year and not go outside. And they started thinking about it and said, yeah, I guess I could live with that. All right. I'd order things, I'd order things through Amazon, you know, and you can get groceries delivered through Amazon, so it's doable. I think one person stipulated an exercise machine. But beyond that, we can sort of reach out and we can do lots of things. And, in fact, that's probably the way most of us effect change in the world around us is through the information that we send and receive.
But our houses are -- our virtual houses are a mess. And so certainly our physical houses are -- have -- mine has lots of paper documents everywhere and e-mail inboxes are overflowing. I have lots of old memory sticks and -- with all sorts of -- there's something somewhere in our house with a priceless video recording of our son when he was two and a half years old saying that he was going to marry so-and-so, a friend of his. And I don't know. We don't know where that is. We recorded it. It's somewhere. For all we know, you know, is it still -- is it still around? We don't know. It's sort of there but not there.
So that's sort of part of this theme of keeping found things found. And then Cathy
Marshall with Microsoft has done some work showing that people have old laptops and desktops with hard drives; they may not even be able to boot them anymore, but they don't dare get rid of them. First of all, there might be good information on there. Second of all, they certainly don't want that information to be falling into someone else's hands.
And I see lots of nods from the audience. We all have our laptops around, lying around.
And then there's another phenomenon that we've probably experienced that we can sort of be in the push and pull of a day and responding to e-mail messages, going to meetings and be totally exhausted at the end of the day. And then if our spouse or a friend says,
So, what'd you do? Huh. I don't know what I actually did, but I know I'm just exhausted.
I was working every step of the way. But maybe not in ways that added up to anything.
So that's what I think PIM is about as much as anything is ordering our houses of information. And I think that there's a real desire that PIM should just go away; it's kind of like the garage that we can't park our car in anymore or the messy closet that we open up and then quickly close again because we just don't want to think about it. It's kind of like eating your broccoli, it's like your parents telling you to straighten up your room.
You know, why can't it just go away.
But I think there's another side of that to say that we have tremendous opportunity here, especially in the domain of digital information, to do things that really make PIM just sort of the work for us, make the things that we need to do just happen in the course of our interactions with our information in ways that wouldn't be possible with paper information.
I came across this recently, and I thought it was an interesting definition. So personal information management, what's the management part? Well, managed -- manipulate. I like that. You're interacting with your information. That's an essential part of it. And I really like this definition, the art of getting things -- the classic definition of management in corporate is the art of getting things done through people.
So what about PIM is the art of getting things done through information? What do you think? What do you think, Jaime? Put you on the spot. If you don't like it, I don't -- we don't have to talk about it now, but I think it's kind of an interesting --
>>: (Inaudible) person.
>>: Yeah, you left personal out.
>>: You left the person out of that one.
>> Williams Jones: Pardon me?
>>: I think --
>> Williams Jones: Oh, the art of getting things done of the person -- yeah, the art of one of us getting things done. And, yeah, it probably needs to be.
>>: It served our information.
>> Williams Jones: Yeah, yeah. Exactly. Yeah.
But I just sort of wanted to substitute people with information and try it out. But obviously you want to be more specific about, it's at a personal level.
So, anyway, to get to the point of today's talk, first of all, I want to provide an overview of what the Keeping Found Things group is doing these days, and also give a little slant to PIM.
So I'll say that "P" is for a mapping, a very personal mapping that weaves together information and need, and I'll kind of come back to that later. And I'll say that "I" is for fragmentation. We talk a lot about information overload, but I think in a lot of cases what we think of as information overload is more aptly described as information fragmentation, that -- and there a lots of examples of this. Sue and I are familiar from cognitive psychology you can do chunking or you can do patterns. And something that's of blazing and buzzing confusion could all of a sudden coalesce into something that's meaningful to you. And I think that a real challenge is to do that with our information, to have it sort of coalesce in ways that are meaningful to us so that we're no longer overwhelmed by it.
There's that effect that -- we're looking for some research on that, but we have it probably when you hear people talking a foreign language, that it sounds like they're talking very fast sometimes. And the explanation for it that I've always thought was very interesting was that it's because we're processing at the phonetic level; if we're hearing English or a language we understand, we're processing at a higher level, so we're chunking. And if we're sort of measuring speed by the number of events, distinct events, there are fewer events. So in a real sense -- what's -- so you apply that to PIM and you say how should your information be organized.
And then we've got "M" is for everyday living and managing our information along the way. This sort of gets back to the plasticity of digital information and how we can
leverage the things that we're going to do anyway. So to be a lot smarter about how the information should be organized. So those are themes that drive the Keeping Found
Things Found group.
And so I'll review three things. I'm going to sort of reorder the letters here. The first is the -- something called The Devil is in the Details . And this is a study comparing the use of tags and folders. The second is -- and I'm just going to give you -- we don't -- I only have about -- I think I only have about 40 minutes left or 35 minutes, so I'm going to try to go through these fairly quickly.
And then the second one is sort of very preliminary results from another study we're doing under a National Science Foundation grant. The first one was done under a Google award. And then I hope to spend a little bit more time on sort of a demo of the Personal
Project Planner. And if the demo fails -- and we're pushing to get Web release 2 out on that, and we have it available for download, so the upshot of that is I took a gamble and downloaded the current version from my developer, and that's always dangerous. So if that fails, I'll give you a quick demo of it.
All right. So does that seem like a good plan for today? So here's the project, or the -- and we're calling it a group now because we have several projects, and these are the current people in the project. Dawei and Deen are the developers and Elisabeth and Julia are on the fieldwork side. And so let's see.
A brief history of KFTF is that it began back in 2001 under an NSF-funded project, and we were delighted to have Sue, one of our original team members. And I remember you gave some very valuable feedback on the grant proposal that we first got funding through. And so we've -- that three-year grant-funded studies look at how people keep information, how people refine information and how people organize information.
We were focused on the Web, but we quickly realized that you can't sort of bound things that way, that the things people need to do just quickly go beyond the Web. So when we're looking at the Web, for example, we would see one of the most common ways people keep information is to e-mail their URL to themselves. And so we did a functional analysis looking at why people would do things like that, which was quite revealing, and it turned out the bookmarks were not quite as popular for a number of reasons. One thing about e-mail, for example, is that if you're in the habit of checking your inbox, there's a reminder there; whereas if you create a bookmark, you can easily forget it later. And we saw people repeatedly do that: out of sight, out of mind; they'd forget about it. We did a refining study where for various reasons, not just forgetfulness, but half of the time they could use a bookmark, they didn't when they were prompted to get back to information.
>>: (Inaudible.)
>> Williams Jones: Pardon me?
>>: Where do you get the low, medium, high from?
>> Williams Jones: That's just our -- we're sort of doing a functional analysis. That's us deciding where the high -- so reminding -- yeah, and I don't want to spend a lot of time
on this because this is published work going back, way back. But, yeah, so, for example, we've got reminding is high for e-mail to self, but relatively low for bookmarks. And even now I'm not aware of any systems where you can set a reminder on a bookmark, even though people forget about them, people forget about the bookmarks.
Another thing about e-mail is it encourages sort of a -- kind of a very casual expression in which you want so you can sort of type in a few words, whereas if you -- you have to sort of trick-game the system when it comes to an old-fashioned bookmark. Nowadays you can do some of that with tags using something like del.icio.us. When we originally did the study, del.icio.us wasn't widely used. I'm not even sure it was -- I don't think it was available until the year after.
Anyway, and then we also did -- in the final year we did something and published something, a CHI called Don't Take My Folders Away . And this is sort of the canticle example. We've seen this in several instances. But I'll have some disclaimers here as well. So this is an example from somebody planning her wedding. And just a folder structure in Windows, and what's really striking about it is in a real sense it's a problem decomposition.
And that got us to thinking, well, what are the folders about? And we did this sort of -- we called it the "Google Question": I suppose you can get back to your information just by typing in a couple of words; can we take away your folders. And people said absolutely not, because the folders turned out to represent a lot of things beyond just a way to get back to information. They were a summary; they were a way of managing tasks; they were a reminder of things to be done. Even though none of that was really designed into the use of folders to begin with, that was how people were using folders.
So that got us thinking about a lot of things. And I'll sort of come back to that later when we talk about the Planner.
So "I" is for fragmentation. So this is an illustration from the book. I think it really aptly describes a lot of us. We start -- sometimes we don't dare touch an e-mail message because we know that can easily mushroom into, oh, you know, I better check this. And
I had one recently where our personnel person said, So how long do you want this job application to be up for? And I said, well, I thought I'd already said something about that, but I better go back and check what I'd said before to be sure I'm consistent, and then I thought, well, maybe I should e-mail somebody. And I started to e-mail, and then I said, no, I really should be -- I should figure this out on my own, because I was the one that posted the job application. So then I'm back and so on and so on and so forth.
Sometimes they can easily mushroom into -- so an e-mail message that looked like it would only take a few minutes can end up taking a half an hour or more. And Victoria
Bellotti has done some good work on that, showing that that can happen.
The other thing about fragmentation is that even simple decisions sometimes require that you go in many different places to look at distinct forms of information. So can you take a job candidate to dinner next Thursday? Well, it's kind of your turn to do it, isn't it, so you'd like to be able to say yes, but -- or think of another scenario. Anyway, you'd like to be able to say yes, but where do you check? Well, you check an electronic calendar, but, you know, that's not enough, you might check your e-mail messages because you might have made a commitment in e-mail. And if you're sharing a household with other people, still one of the most common ways to communicate is through not an electronic calendar
but a paper calendar. And you've got Web pages, maybe you wanted to go to some conference next week and you want to check and make sure that's there, and you may have your child's soccer schedule. So those are all distinct forms of information, and that's part of the fragmentation we deal with, even to make fairly simple decisions. And the interesting thing about that is that, you know, sometimes four out of five isn't good enough. Four out of five, you can say, yes, I can make that, and then the fifth one is what presents you with a conflict and the double-booking.
Okay. So that's a segue into the first study I'm going to talk about today. And this is done under a Google award. And Andrea Civan was the lead on that, and she did a really nice job. Basically people have been talking for a long time about folders versus tags, but if you -- it turns out there's just not that much empirical information about it. So we decided to do something and to sort of investigate that. And the tag line is Devil is in the
Details . So what we did is we originally said, well, heck, we can just hack up folders versus tags, so that's no problem. And then we started thinking, you know what? No, we can't. There's a lot of little things that are just going to be an enormous drain of our resources, and at the end anything we find out may be more an artifact of our poor implementations than of anything else. So what we decided to do, we said we'd have a nice, convenient comparison here, sort of the Web -- the simplified Web versions of both
Gmail and Hotmail, provided us with a comparison of two metaphors: placing into folders versus tagging with labels.
And I think some of us -- how many of us would think that tagging would be a clear winner? Okay. How many people think that folders might be a winner? Okay. So it's not -- some places you'd probably get tagging, obviously, that's that -- folders are so -- why do you say folders?
>>: Because tagging is a lot more work for me, have to think of the tags and make sure they're not (inaudible) --
>> Williams Jones: Yeah.
>>: (Inaudible.)
>> Williams Jones: Yeah.
>>: (Inaudible) they call it personal information management or PIM or whatever they --
>> Williams Jones: Yeah.
>>: So for me folders are -- the concept is easier.
>> Williams Jones: Okay. Anybody else want to make some comments about that?
>>: Sometimes something -- you'll sometimes have pieces of information that you really kind of want to put into two folders, so --
>> Williams Jones: So you're a tagging advocate. Yeah. You're a tagger. All right.
Jonathan, and then -- yeah, this is a good discussion. Jonathan and then --
>>: And individual differences (inaudible) how they deal with e-mail, I would think that --
>> Williams Jones: There's certainly individual differences, yeah. But what about your own individual differences? Where would you come down on that, tagging versus folders, for yourself?
>>: (Inaudible.)
>> Williams Jones: All right. We'll come back. Yeah.
>>: Well, e-mail, I think I prefer folders than --
>> Williams Jones: Yeah.
>>: For other types of things, like photos, tagging I think is --
>> Williams Jones: Tagging. Yeah. So individual differences in it also -- these are all really good points. And so we're probably going to make you very dissatisfied because we only scratched the surface of that.
>>: I think that there's a lot of history people using folders, you're more familiar with it --
>> Williams Jones: Yeah.
>>: -- and all the interfaces with tagging is not as advanced as it really --
>> Williams Jones: Right. That's right.
>>: -- (inaudible) tagging --
>> Williams Jones: Yeah.
>>: -- folders (inaudible).
>> Williams Jones: One more comment and then we'll move on.
>>: I -- I use rules to tag my incoming e-mail quite a bit, but then I definitely want to be able to uncollect it by tag and jam it in a folder, so I would definitely like to have both.
>> Williams Jones: You would like to have both. Well, yeah, that's the right answer, I suppose, is why not have both. But the question is how should they combine together.
Yeah.
>>: For me, search is I think a form of tagging. Every word is a tag.
>> Williams Jones: Every word is a tag. Yeah.
>>: So if I have folders and search, then I'm happy then.
>> Williams Jones: Yeah. Okay.
>>: You have to know what to look for.
>> Williams Jones: Yeah. I -- yeah.
>>: But what if I don't know what the content is.
>> Williams Jones: Yeah. Let's --
>>: (Inaudible) tags do.
>>: Well, but if I organize my folders, then I have explicitly said --
>> Williams Jones: Yeah.
>>: (Inaudible) individual differences.
>> Williams Jones: I'm going to -- yeah, that's right.
(Laughter.)
>>: Normally help.
>> Williams Jones: This is a -- well, you know, this is -- I think to quote -- was it Bork who said that -- the Supreme Court-nominated -- this is a thing about which reasonable people can differ, so...
But let's go on. I will say one thing in favor --
>>: I think there's a really important difference between the functionality that current tag and folders support, which I think are actually fairly different, and that the way in which they're implemented.
>> Williams Jones: Oh, yeah.
>>: I think if they're implemented well --
>> Williams Jones: Yeah. Well, no, and then -- but that's also sort of the -- it gets back to what Jonathan was saying, too, that it's not a matter of either/or; it's a matter of both and maybe more. Okay. And what should that be and how should the two combine with each other. You don't just sort of -- you don't necessarily just slap tags onto folders and say, okay, we're done. And -- yeah.
>>: Just a very, very quick thing, which is I think related to what Marty mentioned as well, which is just sort of the ways in which they're implemented now and the logical ways in which they could be implemented are very different.
>> Williams Jones: Um-hmm.
>>: And so, I mean, the idea, for instance -- you know, somebody mentioned, well, is it
PIM or is it personal information management or whatever. Well, the same could be true of folders or anything else. It's just sort of the way in which they're exposed and how you apply them.
>> Williams Jones: Well, yes, there is. So -- and that is -- that's a real challenge with this. There was a -- we -- there's a -- we do cite in the papers -- gonna be presented at
ASIST in October. We cite work by Pausch and others. And they get into some fairly quantitative differences between tagging and folders. But the problem is that so much of it is so easily dismissed by differences in the way -- I mean, they had people -- I mean, differences in the implementation. They sort of had this paper-and-pencil way of dealing with your folders and with your tags. And it's like, well, you know, how much of this has really just not effected the implementation.
So it's tough. It's tough. And we wanted to get as close as possible to the underlying metaphor, which is, you know, if it's -- the metaphor in a folder is put that there. And if it's there, it's not over here. You know, this is not quantum mechanics. It's not -- it's right there. Okay. So that's -- I'm probably doing havoc to the sound system. So that's -- that's -- that is -- whereas tagging obviously has lots of -- that's more like label this. If you go around with a label maker in a physical space, you can sort of give things as many labels as you want.
So that's what we try to stick as close to as possible. And we also -- we had this sort of method that actively involved people. And a -- within -- so there's comparison of the two. So we gave them both and we had them keep a log of their experiences of both over a period of five days with -- where each day they got five articles in each condition: the
"label this" condition and the "folder" condition on projects of their choice. So, basically, ten participants -- not many participants, but then this is an expensive study to set up, and we were after qualitative comparisons. It wasn't like we're going to at the end of the day say we have statistical significance here. We weren't after that. We were after qualitative differences that we would have overlooked if we just sort of thought about it. Okay.
And so we're kind of letting people be smart in ways that we couldn't and get a reasonable sampling of the ways in which they could do.
So basically we had people select two projects from a list of ten. And we kind of based them on things under SoYouWanna, if you've ever seen that Web site. SoYouWanna run a marathon, they have a lot of things that they -- they have SoYouWanna do a sex change operation. But we didn't have that as one of the projects. And then for each project, people received five how-to articles on each of five successive days. Okay. And so e-mail really -- the point wasn't to look at e-mail; the point was to use e-mail as a delivery vehicle for this information, as a way to control that and as a way to very roughly sort of simulate what happens to a lot of us. We don't get our information all neatly packaged at one time; it's sort of distributed over time. So that's the -- and here's some of -- here are the list of things that we let them choose from, just things that you might -- and basically we ask them to pick things that they didn't know about or didn't know that much about but were interested in. So they picked two of those, and then we randomly assign, or semi-randomly. We tried over -- over people to sort of balance so that the same project was represented roughly an equal number of times in each of the two conditions,
Hotmail and -- or label and folder.
Some things to say about this. The study was exploratory. We've already talked about that. So we did some quantitative comparisons too. We had -- at the end we had them recall three facts about each of the topics that they'd picked. And then we also gave them a cued-recall test. But we didn't get that many huge differences between them. Yeah.
>>: Could the information they were set to be applied to multiple --
>> Williams Jones: No. I mean, the projects were meant to be -- no, the projects were meant to be -- first of all, they only had two projects or topics, and -- we call them projects because they were sort of structured that way, you know, run a marathon. And, no, they were clearly distinct from each other. Okay. Yeah.
>>: (Inaudible) like two months need to --
>> Williams Jones: Yeah, sure. And we just didn't have time. That's a really interesting -- I mean, there are all sorts of interesting things. We just sort of scratched the surface on it. So you're absolutely right. And clearly 25, that's not very much, is it? So, you know, we wish we could have had the time to sort of -- but it was a very small study.
Google didn't give us that much money. So, you know, it was enough to keep us going and that's what we did.
So I'm going to move on to some of the results. Within-subjects, I mean, I'm a cognitive psychologist by trade, and within-subjects has maybe a more specific meaning here than it does here, so I'm putting it in quotes; I'm not sure if I'm violating some sense of what it's defined to be. But basically participants are experiencing both conditions. And we're asking them to actively engage in thinking about the two conditions and writing down their impressions of each.
And then two actual systems, and then people worked over time on their own computers in their own workspaces, so we wanted -- and that's sort of a general problem methodologically I think with PIM is that you're not going to get much about PIM I think from bringing people into a laboratory, where they're away from their information, working on someone else's computer working in an environment different from theirs.
And certainly not in a one-shot deal. It's very much you had to be thinking about the lifecycle of information.
So sort of results. Well, there's some similarities between placing and tagging with respect to retrieval performance. Not much difference. It seemed like people were able to recall a little bit more on the tagging condition, but it certainly was not significant by any means. We weren't expecting it to be. But it wasn't -- it didn't -- it was minor. And
I'm not even going to go into those numbers.
There's kind of an interesting similarity in the organizational scheme over time. So where the interesting difference is, there with regards to sort of essential activities of
PIM, keeping and organizing information and refining information, and I'll drill down into those differences now.
So on the keeping and organizing side, we saw some of -- things that people sort of nicely expressed, better than we could have, maybe, you know, I don't know if -- better than we
could have, were trade-offs between cognitive and the physical efforts. So, you know, one thing about tags versus folders, well a folder is if it's apples or oranges you have to decide. It's got to be one or the other. You can't have both. Where with tags you can, sure, it's both, I'll tag with both. But then once you start tagging with both, you have to be consistent, you have to consistently do that. So we sort of expressed that as a difference between cognitive versus physical effort. It really kind of focuses the mind if you have to decide what one folder that it goes into. But, on the other hand, if you say both, then that's more manual effort over time.
And we have quotes for each of those. With tags it's easier to work through and pick more tags, but with folders I kind of had to think: does this really fit in this category?
On the other hand, they said: I felt like I spent more time on tags just because I had to apply multiple tags so I had to apply one and then the next and then the next. And so, you know, we got both of those expressions. We also got trade-offs between hiding information and visibility. And that was a really interesting one. With folders I don't like that things get hidden away and then I might miss something important.
On the other hand, one problem with Gmail, I think they've got a new feature now that let's you sort of filter on things that aren't tagged. But one thing people expressed as a frustration with Gmail is that they couldn't move things out of the inbox. And so it was always there. They could filter away things or they could highlight things or they could explicitly tag. But what people really like doing was -- or some people, at least, is to say, okay, once I move something out of the inbox into a folder, I processed it at some level so it's done. So what's left of the inbox is stuff that I still have to get to. Okay. And so we see quotes there to both -- expressing both of those. Okay. On the refining side, there are trade-offs between -- well, we called it between flexible and systematic search.
So apples and oranges, you're not sure which. Well, with tags you can look under both and have a reasonable chance of getting at it. But, on the other hand, if you want to say, I know this is in there somewhere, I notice -- I remember seeing this article, it's sitting there somewhere, you want to do an exhaustive search, it's harder to do with tags than with folders because you keep on seeing the same articles again and again that you've already eliminated. And so in a sense folders -- well, not in a sense -- the folders are our partition. Okay. Me see. So mutually exclusive, collectively exhaustive. And so, anyway, trade-offs and refining cues also, folders are visual.
Now, it's interesting that they said that because I don't think there's much visually different between folders in Hotmail and tags in Gmail. I really don't. So I think they might have been extrapolating a little bit on that. But it was interesting that there was a visual component to folders that they -- and I don't know whether that's just because it's so deeply engrained in us to think of folders as putting something there so that it really engages a location metaphor. I don't know what. But that's -- that a quote. And then -- but, on the other hand, in Gmail, at least, with tags, they would be at the beginning of an e-mail message, so you could sort of -- it was describing and it was a way of enhancing the description.
So we were -- this, by the way, is one -- we also had this kind of nice way of representing what's going on, which we thought was pretty useful to sort of see the evolution, a mapping over time, either for folders -- and this is for one person, for folders, running a marathon, and for labels, publishing a book. Okay. And what you have here is
everything happening on day one is coded with red and day two with orange; three, a yellow; four -- you get the idea. And what that sort of -- it's kind of -- it's a nice -- we had this, like, how do we figure out what's going on?
And this gave us a nice way of visualizing what was going on. So you can sort of see that somebody originally had -- well, these are -- the dash ones are things that they eliminated. They got rid of those links. And then they got -- well, they got rid of marathon there, so they deleted that. And then what they did instead was for folders they kind of built something into the names to sort of in a sense represent this part whole relationship. It's all about marathon, but I want to specialize marathon events, training, and diet. They start out with just marathon and realize that wasn't going to give them the distinction they wanted.
>>: This is really an evolution of their --
>> Williams Jones: Yeah, it's an evolution. So you can sort of -- at a glance you can sort of see the evolution, and you can also -- for any -- these are items over here. I should have made that clear. These are items and these are folders, the names of folders. So you can sort of see for -- obviously for items you'll see what you would expect, that there's only one line going to that because it only goes to one folder, whereas in the tagging condition you can see that people did use more than one label, although not a whole lot more than one.
>>: (Inaudible) the evolution, so like in the one case you see a structure getting set up earlier, meaning more or less malleable?
>> Williams Jones: Yeah.
>>: I mean, that's only a week --
>> Williams Jones: Yeah. We sort of had this inclination that the structure seemed to be more stable in the label condition, but we're not sure of that. Which was sort of counterintuitive for us. But the other truth is that we still have to go back and look at that data in more detail. We haven't had a chance to analyze that.
>>: Did anyone make hierarchical clusters?
>> Williams Jones: Hierarchical clusters? Well, we -- you know, of course, neither system supports the hierarchical arrangement of either folders or tags. But what you do is you get some effect of that in things like this, sort of -- they're sort of building it into the name itself.
And then what we also asked the people to do, which we thought was the most fun part of this, was we asked them to sketch their understanding at the end of that time. And so you get sketches here or here that show a lot more representational richness than you could achieve with either folders or tags. And you know what? I guess I could have just given a talk about this.
>>: You could go for another 20 or 25 minutes.
>> Williams Jones: Yeah. So this is another example from someone else, they're understanding of writing a book proposal. So they got prepublishing, query, letter, book proposal, nonfiction, author, agents, agent, publisher. I mean, they've basically mapped this out. It's a little bit of a -- it's crude, but it's a flow of how things are going to happen.
And that -- that intrigued us. It was thinking, well, gee, maybe rather than have a drop-down list of tags or folders, you could give people the ability to arrange things. I mean, it's not rocket science; it's just simple graphs. Give people the ability to arrange things as they see them, whether it's tags or folders, and then for any particular item, just sort of click on the one, so it's basically you're using the graphical arrangement of the tags as a way of -- as a menu. Yeah.
>>: I mean, it seems like tasks really limit themselves to this kind of thing because -- at least the ones I remember seeing where we want to do this and there's like an end goal associated with it --
>> Williams Jones: Yeah.
>>: -- all the things you do leading up to this goal. And so --
>> Williams Jones: So it's sort of workflow in some cases.
>>: Yeah, yeah. So, I mean, it seems like your tasks tend to maybe push toward that end, where imagine other kinds of information tasks --
>> Williams Jones: Well, I can imagine for -- well, what I would say is I can imagine for other kinds of topics you see a different representation. But I'll wager for any of them, you give people an ability to simply draw something and they're going to come up with something a lot richer than anything they could realize, even with the most creative use of tag names or folder names. Yeah.
>>: I guess I was wondering that, too, but you could also imagine that a person would draw you a list, and that's clearly not what they did. So it wasn't sort of a step-by-step, here's what I need to do to write a book.
>> Williams Jones: No. It's not -- yeah. It's kind of --
>>: It's two-dimensional rather than just one.
>> Williams Jones: But the other interesting thing about that is they draw that and that's theirs. I mean, they have some understanding of that. It made me start thinking about, well, suppose you had -- you presented them with a "here's your list of folders, here's your list of tags, now just do a simple draw, click to arrange them a little bit like, you know" -- well, there's some commercial applications that do something like that with your folders. But the point would be not necessarily to get too carried away about it, but then present that instead of a simple drop-down. Why not? I mean, how much more expensive could it be? And then they can click on that for any particular item as a way to quickly -- to get -- anyway, that's something that -- and we sort of carry that through in a slightly different way in our thinking about the Personal Project Planner. Anyway that's one study. Any other questions, comments?
So obviously there's a lot more to do there, but we thought -- we thought we got a lot of mileage out of ten participants through this within-subjects manipulation. And none of the things that we list there qualitatively -- with all of them it was sort of, like, yeah, that makes sense. I mean, they're all sort of had -- they seem clear after the fact, but we would not have gotten half of them if we just sort of sat down ahead of time. Maybe if we called all of you into a room we could have gotten some -- a larger percentage. But the fact is some of those things are only clear over time. Like, oh, it's really cool to be able to do apples and oranges. Well, not after the third or fourth or the fifth time. Okay.
Some things are only evident after a period of time.
>>: (Inaudible) question, if you could do it again kind of thing, is these were sort of special accounts that they had set up (inaudible) getting other kinds of stuff --
>> Williams Jones: Yeah.
>>: -- in the flow.
>> Williams Jones: That's right. These were special accounts.
>>: And so were you feeding them noise e-mail as well?
>> Williams Jones: No, we thought about doing that. We thought about doing that.
But --
>>: Do you think that would have changed how you -- I mean, I guess if it's pure noise, then --
>> Williams Jones: Yeah.
>>: -- they just pitch it. (Inaudible) because it seems like that's one of the interesting things you hear people talk about when they make these things, has to do with scales --
>> Williams Jones: We thought about doing that. I mean, we had some real problems just getting past the junk filters to begin with. And so -- and it's sort of indeterminate what's going to happen. We don't know. So we just get -- but we did think about that. It was just too problematic. Okay.
So conclusions, don't leave the good stuff behind. Filter -- for example, something simple like in Gmail, give them the option of showing only untagged items, okay, if that's the way people like to manage their workflow, then support it. Support hierarchy was really -- really became clear both with tags and with folders. People need hierarchy.
>>: Did they ask for it?
>> Williams Jones: Well, we did have some quotes that they asked for it, and we also see it in the choice of words, that they're really trying to jury-rig that. And also if you go back, I think we were talking about -- we were talking -- back to your workflow thing, we didn't always get a workflow; we got here was someone else doing publishing a book and they sort of had this Venn diagram, okay, instead. So that worked for them. That's what they did. And over here they had a hierarchy. I mean, I would call this a hierarchy.
Marathon, and then under that diet and events. And you can see that up here. It's basically that's what they're trying to do and can't with -- given the constraints of the UI.
Yeah.
>>: (Inaudible) lot of literature on how people organize e-mail I think that's qualitative
(inaudible) how you organize other kinds of information (inaudible) e-mail, so not quite sure that you should generalize tagging versus folders to other -- I know you said this in the beginning, but to other tasks, like labeling images on Flickr or --
>> Williams Jones: No.
>>: -- because it's not usually a task management application (inaudible).
>> Williams Jones: Right. We did our best to sort of minimize the e-mail aspect of things, so we sort of -- it was just a -- but -- but --
>>: (Inaudible) definitely an e-mail --
>> Williams Jones: Which one was that?
>>: The one you had it done and processed.
>> Williams Jones: Oh, no, definitely that. Yeah, that is, although -- yeah, that's true.
That's a good point.
>>: (Inaudible.)
>> Williams Jones: That one is, yeah.
>>: The other thing I wanted to say, is when you get to these richer representations, then you're getting into that whole sense-making realm and the -- a lot of the evidence on sense-making is that as people continue the sense-making process, the representation has to change. So if you start giving people really rich representations like that, then they're going to want to probably undo them, so --
>> Williams Jones: That's a very good point.
>>: -- you really have to differentiate which one's from -- is very simple (inaudible) --
>> Williams Jones: No, that's a very good point, and that's -- I mean, I think that's an unsolved problem. You draw something and then obviously over time it's out of date, so how do you -- I mean, it's sort of this raised to a higher degree, really. You have to create folders or tags as you're going along, and at some point you're going to look back and say, wait a minute, that's not really the way I see things anymore. So then you're in a sort of agonizing choice between doing everything over again or sort of living with something that no longer is really reflective of how you're doing things.
>>: (Inaudible) rich, that next slide you have where there's the complex process diagram, which is much, much harder to change (inaudible) --
>> Williams Jones: I don't know if it's much, much harder to change. I don't know about that. Yeah. I mean, I think that's -- in some ways, you know, what you have in something like this if you -- but that's something that maybe we should take offline, but I think -- you know, suppose you can just sort of see this and say, well, that's actually over here now. I mean, it depends on the nature of the changes, obviously. But you might say, well, I don't really call it agent anymore, I call it something else. I mean, some changes are easy; some changes are not so easy.
>>: (Inaudible) they probably are really clueless and they start out completely wrong --
>> Williams Jones: Right.
>>: -- where someone who has written lots of books has (inaudible).
>> Williams Jones: No, no, but what I'm saying is it's not clear to me that a representation like this would be harder to change than a representation based on simple tags or folders. That's not clear to me.
Okay. Anyway, support hierarchy, support collections of tags or me see. That was another thought. If you're going to go with tags, maybe you should have -- I mean, essentially you're also saying there are relationships between tags, one of them is maybe a hierarchical relationship, another is a relationship that says these tags are -- they establish a partition, okay. And so, you know, male or female, animal, vegetable, mineral. Those are sort of partitions.
And then active engagement of people and direct comparison of conditions can be very useful, so that we're thinking about using that in other places, too, where we would like to -- I mean, obviously you're never going to find existing systems out there that -- they're probably not going to come as neatly close to direct comparison as Hotmail and Gmail.
But the question's can you sort of -- can you instruct the jury to disregard these irrelevant things and focus on these key differences and then try it over time and get their insights.
Okay. Now we're on to the mapping, and we're already into discussion period, so I don't know what to do about this. So mapping, this is a -- this is also from the book, but it -- the point of this is to say a lot of our lives are -- a lot of our daily lives which are at the level of keeping and finding activities. I give you something. I give you a business card or I say, here, you know, check out this article, that's a keeping -- you have to make a keeping decision; do I pay attention to this or not. You get new e-mail; do you answer it or not or do you -- you know, that's a keeping decision. It's sort of -- I broadly define as do you attend to it and, if so, what do you do and how do you handle it. On the other hand, you have a need, you have a need to call someone up, so you're doing a finding activity; you're going from need to information.
And a lot of our daily lives are just at that level. One of my colleagues at the university said -- called it -- Karen Fisher called it hook-ups and look-ups. I thought that was nice.
Looking up information and you're trying to hook up the information you do get so that you'll get back to it later. And then there are these mental-level activities that sort of are more focused on the mapping. So aspects of that are like schemes of organization. And we all have them, we all have schemes we're organizing, whether they're with tags or
folders. And we change them over time. The structure is the result from that as well.
There's some strategies for managing time and energy and information flow.
I had a colleague at the Royal Institute of Technology in Stockholm who said, I don't keep any e-mail messages even though that's his area, Ulla Galta [phonetic], that's his area of expertise. But he says, I don't keep any e-mail messages. If I need an e-mail message, I'll ask one of my colleagues for it later. Well, that might work if you're an engaging guy like Ulla and you don't ask too often, people are sort of going to say, sure, I can forward that to you.
>>: We worked with somebody else like that.
>> Williams Jones: Pardon?
>>: We worked with somebody else like that.
>>: (Inaudible.)
(Laughter.)
>> Williams Jones: Yeah. It works if you're in a position of power, too. Not that Tom ever -- so, at any rate, there's strategies. You know, strategies like I close my door.
Somebody at the university says I never deal with e-mail on Tuesday and Thursdays.
That's a strategy for managing flow.
Supporting tools and a system that brings them all together, so we've done this -- this is another -- a short paper being presented at ASIST by Elisabeth Jones, no relation to me.
And so she's a Ph.D. student at the university. And we're really -- it's a part of a larger study with NSF looking at projects as a unit of analysis for PIM research. And there -- obviously it's too big to just sort of say, well, how do you do PIM? So that doesn't work.
But, on the other hand, if you partition along the lines of applications, that's not a good idea either. E-mail, well, why -- that's sort of buying into current partitions that -- when that's not necessarily -- that sort of perpetuates the problem. So we looked at projects as a way to bring together lots of different kinds of information. And, good grief, we've got this long procedure over many, many days, weeks, where we pick a project in somebody's life and we have several different ways of asking them about the project and what's going on.
And I'm just going to gloss over this study, because it's such a provocative thing. So we found different ways -- basically why do systems fail, why does a system fail in your life: the New Year's Eve resolution system that you had such great hope for in January and now it's July and you're not using it anymore. And if it's a physical system, for filing papers, that's painfully obvious. You see a growing stack of things waiting to be filed, for example. It's not so obvious in the digital case, but there are signs that it's breaking down. And so, anyway, gosh -- so we've got various factors. One is visibility, for the failure on the short run to get -- to do something. I need -- so basically if it's not there, they won't use it. So I need something constantly visual to me, so I have that with a paper pad but not with the task pad in Outlook. That's what one person -- a comment.
Integration is another thing. If it sort of integrates into something else you're already doing, you're more likely to use it. Over the long run, there are things like scalability.
And we can all relate to that. This person talked about in the context of Microsoft
Project. She said it's good for working on one particular project -- and that's a very generous comment; I haven't heard many people say that even -- but when you get four or six, then it's really hard to balance between them.
And then return on investment. People sort of start off with something and then after a while they gradually start to say, you know, I'm just not getting anything out of this; I'm putting a lot into it. We thought -- we call it co-adoption, but it's really a social factor that really goes on in a lot of systems, and this is probably something near and dear to your heart, Jonathan. This is a great quote: It's one of those deals where at some point you realize you're throwing effort after nothing and you stop doing it. It means nothing to other people, so if I'm not doing it for myself, why bother. You know, sort of people are -- there's a little bit of edge in this quote.
And we've also interviewed other people who, for example, do an elaborate tagging of things on Flickr, on photographs, that will ultimately be beneficial to them, but it's not motivated by any thoughts that they want to get back to the information later; it's motivated by a social aspect. They want to bring people to their photographs to get comments. So we're very intrigued by the social aspect of PIM and how to motivate people to do things.
So everyday living, what do we say there. Well, there are various ways you could do that. You can sort of record what people are doing and make use of it. Or sort of do things case by case, so that smaller pieces -- if you have an onerous task you can break it into smaller pieces. But what we're really interested in is -- in the Planner is people don't necessarily like to -- most people don't think it's a lot of fun to organize their information, but some people like to plan -- all of us have to plan whether we like it or not. Planning is a part of life. So the thought is let's support people in planning and have information organized as an incidental by-product. And so this is something we're going to be submitting to WIST. There's also a paper that we've presented at CHI, a short paper, a
CHI influence.
So new ways of working with the same old file system, I'll just -- so I'll go to the demo now. Let's see. The question we ask is can an effective organization just happen as a part of project planning. So what we have, I'll just show it to you. Actually, guess I need to -- so I'm -- I'm actually using this now, the Planner, even though it's still very much a prototype. Let me just find -- so this is a view of all of my projects. And if I -- I can look at one of them, submit to WIST, for example, and -- okay. I think I just opened two different views of that, but this will probably get the point across.
So what is this? This is basically -- looks like a mess to you, probably, but it's basically a view of my files and folders, but it's a document-like view. It sort of looks like an outline view in Word. And as I create new headings in here, so there are headings and subheadings, what I basically see are underneath are new folders and subfolders being created. So the idea is that I'm kind of planning the video demo, for example, that we're going to do for WIST, and I do things like I've got this hints for making a good video. I can do things like drag and link to bring things in here. I can drag information in and it's linked back to the source and I can get back to the source easily enough.
The same thing with e-mail messages. So here I have an e-mail message to somebody about -- we're talking about using Adobe Captivate, so I just drag that in there. I can either drag the whole e-mail message in or I can sort of select the whole item and drag it in and have a hyperlink back to it, or I can just drag in a few key sentences. So one idea here is that I can take an e-mail thread that maybe has gone on and on and on and just pick the resolution. For example, Jonathan, you and I were involved in a threaded conversation like that where we finally came to a resolution, and I can drag that in and say, okay, now we're going to meet at 5:30 at such and such a place and have that resolution, but also have the link back to the source so that I can go back and say, well, why didn't we consider some other -- why are we doing that again. So that's sort of the idea is to -- so that's drag and link. That's one way to sort of -- the overall overriding idea here is to have all of our information, regardless of its form, whether it's e-mail or Word documents or Excel documents or Web pages, all of it together. Everything that you need to have in order to do a project is right there in front of you.
>>: (Inaudible.)
>> Williams Jones: Pardon me?
>>: (Inaudible) this language (inaudible).
>> Williams Jones: Oh, that is the most recent document. So we sort of have in this view one problem we have is we can move things around in this view, because it's a document view. So we have this -- for example, we have this sort of home remodeling file. And we can move that up and down as we want to. But one problem with that is that we can't easily flip a switch and say show us the most recent, so this is our way of doing that.
>>: There are lots of views you might want on it, and so one thing you might want is pending tasks, another thing you might want is the order in which things should have happened.
>> Williams Jones: Right.
>>: So if you use it for other things.
>> Williams Jones: The thing -- what this sort of got us to thinking was -- I mean, this sort of says now once you sort of open the door a little bit on this idea that you can have this overlay to the file system, breathe new life into a tired file system, and that through that you can do a lot of things. You can still manage your information, but you can also do other things, like we also support calendar reminder, for example, so video, three minutes, maybe I'll say it's due by -- it is due by tomorrow, so I'll put that in there. I think
I already had done that in one other place, but I'll sort of say, yeah, due, three minute video. So now that's in here.
And if I go -- so one thing you can also start to do is you can start to log and sort of -- basically is you can sort of log what you're doing. So if I go back there on Friday, yeah, I see I've got that, the video demo. And if I open that up, what we have now is it's -- task
reminders are great, but they don't give you much by themselves; they're sort of inert.
What we have is a link back to the information you need in order to complete that.
Now, we don't have a way to go back to the plan now, so then we go back to the underlying folder. But you can also -- this is a good place also to sort of show you that everything you saw on that plan is reflected here as a folder. And if I create a new folder, like maybe I'll say -- you know, if it's -- if it -- yeah, it's going to complain about me -- okay, so if I -- there's a document open, so -- but if it's accepted, we're going to have to think about travel, for example, and hotels, for example.
So if I -- and then I might make some notes to myself, so I might say, oh, I don't know, send an e-mail for recommendations on hotels to so-and-so. And what you see here is -- sure enough, there's travel, okay, and there's a folder -- and hotels is under that. And we also have this thing called and XML file. How do we realize this reach -- this sort of document-like view, we basically distribute XML files, one per folder, and we dynamically reassemble them to create the view. You don't see the notes here.
One of the frustrations I often experience is if I put everything project related into a folder, I want to make notes about that too. The only way I can do that is to create a file in the folder. But that's awkward for me sometimes, but here I can type notes in freehand, as many as I want. And I can also -- I could also send an e-mail to someone in here. Let's see if this works. All right. I'll e-mail myself asking for recommendations.
Well, I'm not going to bother spelling it, blah, blah, blah. So I do that and now later I can go back and say, well, did I ever get back to myself on recommendations? That's a common thing we have in e-mail. And I sent an e-mail message, did they ever get back to me?
Well, I can troll through my sent mail folder or maybe I blind-copied myself, but neither of those is a very effective way to do that over time. It's just tedious. But in the context of a plan, that's when I could say, oh, remember sending it as I was thinking about hotels, okay. It's a natural thing here to go back and I can leverage a feature of Outlook to say, well, find related messages. Okay. And so you get the idea.
That's sort of -- that alone is an extremely powerful feature, in my opinion, just to sort of be able to get back to those responses. Now, over time you can obviously think of other things. Well, if you do that, why not bring that information in. So wouldn't it be cool if you saw this view and there was a little -- a little icon saying -- indicating that there has been a response and you could click on that to get the responses. That would be pretty cool. So another thing that we'll do, I'll sort of show you the sequence. Another thing we're doing is a lot of times we exchange documents back and forth. We do that all the time in academia, people do it in business too sometimes, even if they have a central
SharePoint system. They'll sort of say, okay, here, I marked it up and now it's your turn, back and forth, back and forth.
What do we do, then, with -- we got an e-mail message, and this is something -- it's not a new problem. Victoria Bellotti tried to address it in -- what's the name of the system? I can't think of it now. Anyway, she tried to address it in that bringing e-mail to task and --
Taskmaster, that was it. Great name. This idea that you want the attachment to be there as sort of as its own file. You want to have a location for that, but you don't want it to be completely separate from the e-mail message, so you sort of want both. So what we do
is -- this is sort of illustrated here, you can just -- first of all, I sent and e-mail message to somebody and I got back responses, so here's the response. And when I drag that response in, it preserved a shortcut back to Outlook and to the message in Outlook, but it also created a separate note for the attachment itself, and that actually now resides in the file system. Okay.
And then what I did here is I created a new version of that, because I'm very -- maybe a little bit compulsive about this, but I created a new version with sort of -- indicating my changes, so I had WJ1 on the end of it, just so that it -- and then I make my modifications and then I send it back in another e-mail message, okay, back to the person. So I can sort of show this if -- that's sort of the way it should work, and I can show you -- well, I'll say yes to that. Let's see. I'll just show you how it actually works so you can -- if it works, it works; if it doesn't, it doesn't. I know that it worked just a few minutes ago or an hour or so ago. Let's find a good attachment here. So you sort of have to -- yeah, here's one.
So this is from one of my developers. So if I drag this in here, what you see there is that it has both the message and the attachment above it. We decided to go up instead of down. We had a big discussion about this and we decided you have to go up. You want everything to -- you know, these are meant to be sort of sometimes roughly corresponding to tasks, subtasks, or tasks of a project, and we wanted to be able to go up to a task and then immediately know what's the current status of things, where's the action right now.
So this is sort of the idea. We have drag and drop, we have in-context create. I showed you in-context create of an e-mail message. You just click here. One basic advantage of that is that you -- if you want to send an e-mail message, you don't have to go to the inbox to do that; you can just create it here. And if you go to the inbox, you're inevitably going to get distracted by other e-mail messages. Oh, well, so-and-so, what do they want.
It will only take a minute; I'll just respond. And then you're taken out of the planning that you're doing for this project. And then I also told you about the other side of that, that once you send it you also have a link to it in the plan so if you want to you can go back to it later without having to troll your sent mail folder. In-context create also works for documents as well, so you can do an in-context create of a Word document or an Excel document. And then drag and link works for a variety of different kinds of information, to bring things in.
And I think that's it. I'm just going to wrap up now. Sorry for going over. But let's see if
I can resume the -- is there any option to -- from the current slide. Let's hope that works.
No, that doesn't work. All right. Okay. Let's see. It's in this -- I just wanted to go to the concluding slides.
So we've talked a bit about drag and link is sort of outside in and in-context create is inside out. Two design principles, no new organization. Well, in a sense, it is a new organization, but it's just -- it's direct mapping out of your file system. Ultimately we would imagine this is just being a part of the file system. You'd have a classic views, details, list, and you'd have the project view. And obviously you can imagine something beyond just an outline view, you can imagine workflow view, you can imagine and/or decision trees. All sorts of things are possible. The point is they can just be an overlay onto the file system. And the other thing is organize incidentally. Project-related information is organized as a by-product of a plan's elaboration.
Got some nice comments from participants in some evaluations we've done. It is out there for a download. We're doing a major new release in the next few days, so I would invite you to give it a try. It's out on the Keeping Found Things Found Web site.
To conclude, we sort of said "P" is a destination, better personal mapping, and the starting point is information readily available, which is good, but very fragmented, which is bad. And what we're saying is why not provide tools that enable us to sort of make sense of our information as a integral part of our daily lives. And so I guess that's the conclusion.
(Applause.)
>>: You're around for another 15 minutes.
>> Williams Jones: Yeah, and if you want to read on, there's the book Jaime and I did,
Keeping Found Things Found , and there's also Tales of PIM is a Web site we set up for discussion, and then there's also the Keeping Found Things Found Web site at the university where you can download the planner if you want to give it a try.
>>: You said the green line was the most recent thing.
>> Williams Jones: Um-hmm.
>>: But it wasn't the most recent; the one above it.
>>: Right, so as you were adding stuff in there, those things --
>> Williams Jones: Yes. So basically if we -- I can show you. It should work. You've got good eyes there. And I think what happens -- I mean, obviously we have a problem updating it as you're doing things, but I think what would happen is -- so we can do this -- basically this refresh, and let's see what happens now. That's still not the most recent. So that's a bug.
>>: Okay.
>> Williams Jones: That's a bug. We're working on it. We have a fairly crude way of determining most recent, because there's some obvious questions about that most recent, with respect to what, you know? And we do it just very crudely with just respect to file type now. Any other questions?
>>: I'm just curious, are you familiar with a product called PersonalBrain?
>> Williams Jones: Yeah.
>>: Because over the years I've used it, abandoned it, used it, abandoned it, used it, abandoned it --
>> Williams Jones: Well, you know, that's the funny thing about it; there's something there that brings you back, but it isn't -- doesn't quite work. I know when I try it it's like I end up making a big mess of things.
>>: What is the (inaudible)?
>> Williams Jones: Well, it's -- you can -- why don't you describe it.
>>: Well, it actually -- the latest versions of it actually meet a lot of the criteria you're talking about. It's like a mind map, except instead of a central bug that's static, anytime you click on something it goes to the center. So instead of being hierarchical, you can link anything to anything else. and I found that it's great, on the one hand, if you approach it in that way. But naturally I want to put everything into a structure that I can remember. And in your book -- I never heard the term before -- I think you say "personal taxonomy" or something like that.
>> Williams Jones: Well, I like -- yeah. Personal unifying taxonomy.
>>: And when I read that, I thought I have that in my head and then in a tool that allows me to do anything, I find myself not doing everything because I like to have it in my structure. So that's why I keep abandoning it and going back to more (inaudible) tools.
>> Williams Jones: Yeah. Personally, I find it disorienting in a way that it's shifting the view for you. So that's part of a basic problem that I have is I want to have a sense of going to it rather than having it come into view always in front and center. And then I think it's just too easy to get yourself into knots with it because it's too easy to make links that don't make any sense after a while.
>>: Have you seen (inaudible)?
>> Williams Jones: I did. I saw it. I saw a demo. I think the developer showed me a -- the guy that was working on it.
>>: How long ago was that?
>> Williams Jones: When was that? That was when I came to talk (inaudible) -- yeah.
Um-hmm. Robin. That's right. Yeah. Yeah. He was at -- I remember he was also especially interested in the Outlook stuff because he'd done some work on Outlook, so he was curious on how we'd -- how -- I mean, people that know about Outlook say, well, how did you do that linking into Outlook? Because it's not all that easy and you have to be sort of sneaky about it.
But, yeah. So we can talk about that afterwards. I saw that. And that was certainly an interesting demo.
>>: It's similar to this in the sense that you can create -- it's similar to OneNote in some ways as well where you create links in context and you can also grab information elsewhere and put it --
>> Williams Jones: Right. Yeah.
>>: It doesn't have a tight integration with the file system.
>> Williams Jones: Yeah. Yeah.
>>: It has a different integration with search.
>> Williams Jones: Yeah. We debated that over and over and over again, and we -- because it's not easy to keep integration with a file system. But what we found in our own surveys of people is that when they're required to put all their trust in a separate database, they ultimately abandon it. And the file system is sort of there to stay for better or worse.
>>: How well does this kind of thing work when you've got a -- more like a distributed set of documents, like you've got sort of file servers over here and other kinds of things where maybe you don't have everything stored. Like if my files are -- the structure is located in multiple servers --
>> Williams Jones: Right. Well, you know, I don't know. I mean, a quick answer is that it works pretty well because of the way we've architected it. We had an early -- a debate early on and one of the previous developers said, well, I think we're going to make this database and it will be central. And I actually said, no, that's going to work because you need to have this -- so the way it works now, you can go out to -- we have this thing we call the N drive at the University of Washington that we have stuff on and you can build a plan out there as easily as anywhere else. Now --
>>: Just have to be able to write to it.
>> Williams Jones: You can write to it, yeah. But the other thing -- and, I mean, so where are we going with this? There's some interesting things that once you open the door, there's all sorts of things you can start to do. One thing that people ask for is, well, why not have a little -- if that represents a task, why not have a little flag like you have in
Outlook that sort of represents it's a flag for completion and then a check -- that nice, satisfying check mark when you're done with it. And so that's an easy thing to do. That's like a no-brainer thing to do. And there's just basically cycling through three states.
And then there's also make this go away. Archive it. I'm done with this. But I don't want to move it, because if I move it, I lose a lot of the representational richness that comes with it being where it is. So make it go away means that we just simply add a don't show flag to our XML representation. So it's still there on the file system, but we don't have to continually see it in the plan.
So these are things that in the context of bringing information together and sort of make it active. It's not just a static repository for your information; it's an active area that you can start to use for planning and thinking through things. Very, very exciting to us. And obviously this is only one of many possible overlays. There are many others that we could do: workflow, mind mapping. Those are all possible overlays.
>>: How does it handle versioning, like if you see a file (inaudible) and then later it gets changed, are you going to lose that information?
>> Williams Jones: Well, so --
>>: This is like a bookmark into the file system, so --
>> Williams Jones: Yeah.
>>: -- if the information you want to remember was in a particular version of the file and the file changes to lose that information, are you going to be losing the information?
>> Williams Jones: Yeah, I mean, these are actual pointers to -- that's an actual file, and if you go underneath "video three minute," you'll see there's an actual file there. I mean, we can go out there to the -- so I guess the short answer is it's the same -- it's no different from what -- the problem you already have with -- it's way over there. I have too many things open. But so if you -- so let's see. Where was that -- what was that file called? I don't remember which one it was. But basically it's there. So it's in the file system. And so you have all the problems of versioning. But if you click on that, you'll get back to whatever that file currently is.
But we also do make explicit provision for -- we all know that most of us still do file-level versioning. There have been -- the best attempts to sort of build versioning into, say, Word, have failed. People don't use that; they use file-level versioning and they have sort of ad hoc ways of representing that this is a version of that. Like I put WJ at the end of a file and maybe the date or something like that or version 1 or 2.
So we have a way of doing that here. We sort of say -- I didn't show you that, but as I'm doing -- the scenario I tried to illustrate, and I kind of ran out of time for it, is I give a document to you, you give it back to me, I create a version of it, so I can create a version of this, for example, new version, it prompts me and so I can say, well, this is actually
WJ2, and then it creates that and that's the new version. Okay. And it has a nice little note here. We're still working with the notes. Maybe they're a little bit wordy, but it says version demo script WJ2 created from demo script WJ1 on such and such a date.
But the thing is, if I think that's pretty wordy, and I think I do, I can just change it to new version. Because it's just a document. Okay. So that's the spirit of things. None of us are going to buy into a formal system like Microsoft Project for most of the projects that we actually do, so make it much more like a document. We're comfortable working with documents, and make it like that.
>>: Can I cut and paste this and have it reflected in the file system? Can I open up a new task and cut and paste parts of this (inaudible)?
>> Williams Jones: Well, we've talked about that. That's not a feature that we're supporting yet. We only do text-level cut and paste right now, because you sort get into sort of you have to support the clipboard with a more complicated object. And so we don't do that yet. Yeah.
Okay. Well, it sounds like we're...
(Applause.)
>> Williams Jones: Thank you. Good comments.