Earlier this month while working on a research paper I became frustrated with the tools available to me. I was conducting research in on-line, web-based databases on campus, knowing full well I would need the information I found later that day at home. After collecting my inch thick stack of output from the printer, I wrote “two”:http://www.yergler.net/averages/archives/2004/04/12/blogging_is_hard “different”:http://www.yergler.net/averages/archives/2004/04/12/better_blogging_or_its_more_than_just_a_blog_virginia posts in quick succession describing my frustration and expressing my desire for a better way to manage the information I found. This post is an attempt to clarify that vision and promote the idea of a better way to manage information found on the web. More simply put, a vision for better bookmarks.
The Internet has fundamentally changed the way people communicate. You can debate whether we’ve seen the full extent of that change yet, but the fact is our lives have changed. And mostly for the better. Students today are more likely to use Internet based sources for research, as well as use Internet tools such as instant messaging and e-mail for communication. However, this change is not without it’s own problems. People are increasingly relying upon multiple sources of information to form aggregate opinions and stay informed. Additionally, people are faced with mentally organizing information from a vast array of sources. It seems that ten, twenty years ago, people obtained much of their information from books and periodicals. If they needed to recall a fact or story they had previously read, there was a limited number of sources they needed to search. Today, information workers may read dozens of news sites or blogs daily, in addition to any mailing lists they may participate in. This rise in volume means it’s much more difficult to recall and locate individual pieces of information on demand.
The simple fact is this: information workers are faced with a growing volume of information, and a corresponding need to categorize, quantify and assimilate this information for future use. The focus of this discussion is web-based information, which is probably second only to e-mail in quantity. The traditional method for recording relevant web pages, bookmarks, is woefully inadequate for the task today.
Bookmarks, as we have traditionally known them, are inadequate for several reasons. First, there is the manual nature. A method for marking pages as relevant, important, or note-worthy is definitely important, but often users would like to recall information from pages that were on the cusp on the importance threshold: not quite important enough to warrant their own bookmark, but relevant to some task none the less.
Second, and most importantly, is the task of organization. Major web browsers today allow for the creation of folders and sub-folders to organize bookmarks, but the problem remains that this, too, is a manual time-consuming process. It’s important to note two additional failings of this model. First, a bookmark exists in a single place. A political scientist may wish to track information on both national legislatures and copyright law; a web page describing EUCD implementation falls under both categories. True, this user could create two bookmarks, one in each folder, but the semantic difference is important. Second, if a user commits to organizing their bookmarks, the process of maintaining that organization is onerous and time consuming. A tenant of Extreme Programming is that you probably won’t be right the first time. Jim’s Second Law of Engineering states that you can’t solve a problem unless you know the answer. These two principles imply that it is unlikely a user will know the “correct” or optimal filing hierarchy at the start of a project. Only as users work with information and evolve their ideas and “mental map” do they develop this optimal hierarchy. However, given the manual nature of bookmark management, it is not unlikely that sub-optimal organizations persist simply out of inertia.
So what features would improve the usefulness of bookmarks? I can think of three. First, I want them to store more information. More than a URL, I need information (annotations or metadata) that describes what a page means to me. Is it vacation planning, business relocation, or both? This purpose is currently served in a limited fashion by folders. However, I envision a system where not only is the URL and “topic” (folder name) available, but also arbitrary annotations (Wiki-style, anyone?), and possibly even a cached copy of the content.
Second, I need a system that’s portable. It is telling to me that a “feature” of Apple’s dotMac service is the ability to syncronize bookmarks across multiple machines. There’s absolutely no reason I shouldn’t be able to maintain a single set of “bookmarks” at school, at home and at work. Whether I store them on a web server or USB memory is irrelevant; the information needs to be portable. Note that this also ties in with my first point: if I’m working on a paper at school and want to use information from a database that’s only available to computers at school, the ability to transport my “bookmarks”, annotations and a cached copy of the page home in one package is very valuable to me. It opens up the list of places where I can work with that information, and frees me to work where I want to.
Finally, the system needs to be integrated. A “comment”:http://mt.yergler.net/mt-comments.cgi?entry_id=1181 to one of my “previous posts”:http://www.yergler.net/averages/archives/2004/04/12/better_blogging_or_its_more_than_just_a_blog_virginia suggested that what I really want is XiTouch, which the commenter bills as a “web based PIM/blogger.” No, that’s not what I want. I want my browser to be smarter, not some web service where I have to type in notes. I’m not saying XiTouch isn’t useful, but looking at the web page, I know it’s not for me. To me, the advantage to searching for “better bookmarks” instead of an electronic Notepad is that bookmarks are integrated, first-class (well, that may be arguable) citizens of the browser world. And by making them better or smarter, we can enable better use of information.
I want my “bookmark” information to be smarter, portable and completely integrated with my browser. Shouldn’t it be possible to pull a USB memory device from my machine at work, drive home, plug it in and start where I was? Or hop on a plane, plug it into my laptop (at the appropriate altitude, of course), and work with the cached content? These two use cases really highlight what I consider to be “optimal” uses.
I don’t think I’m alone in identifying this as a problem. I haven’t done an exhaustive search, by any means, but there are some projects underway which I believe share at least the principles I’ve outlined here. In the category of information management, both “Chandler”:http://osafoundation.org and “Haystack”:http://haystack.lcs.mit.edu/ come to mind. Chandler, while written in Python and cross-platform, is very e-mail and PIM centric. I like what they’re doing, but it fails the integrated test. Haystack, well, it’s a research project, and as cool as it may be, waiting 15 minutes for my app to start really doesn’t fly for me.
I also like “Dashboard”:http://www.nat.org/dashboard/, but I think it’s trying to solve slightly different problem. Instead of helping organize incoming information, Dashboard tries to show you information relevant to your current task, which is definitly cool. While Dashboard is Gnome only, it is written in C# (Mono), so I hold out hope that it could run on other platforms at some point.
Finally, “TrailBlazer”:http://www.acm.uiuc.edu/macwarriors/projects/trailblazer/ is a Mac OS X browser put together by the UIUC chapter of ACM. TrailBlazer builds on Apple’s WebCore framework, so it inherits the excellent rendering engine from Safari. Of course, that also means it’s Mac OS X specific, a killer for me since I use Linux at work, Mac OS X at home, and Win32 at school. Even with that “problem” it’s worth examining because it introduces a new paradigm for browsing history in the context of paths instead of just a list of pages.
So now that I’ve recorded my opinion for posterity, what’s next? I’d like to know if I’m way off base or right on or somewhere in between. Is this really a problem, or just the perceptions of a psuedo-crack-pot? And if it is a problem, does anyone have any ideas that I’ve overlooked? “Must-have” features that will make life better? As far as implementation goes, I lean towards implementing this as a Mozilla extension. I have some experience with them, and it seems like the only way I know of to satisfy the “cross-platform” and “integrated” litmus tests. Even with the Gecko platform to build on, this may prove to be a larger task than I expect, but I’m willing to take that chance. Anyone else?
It seems like there’s two separate problems here – creating/storing/organizing bookmarks and then later accessing them. This kind of stuff fascinates me.
Do you think some kind of algorithm, maybe to analyze the links you click and the contents of the page, maybe combined with keyword-matching (so that, maybe it wouldn’t bother doing much with news items, but would if it saw, say, that I was looking at biology-related stuff, for my major?). And then, somehow weighting importance depending on how long I linger on a page? For me, it’d have to include PDF’s (viewed in-browser, via that Safari plugin whose name I can’t remember), too. And organizing by date / link-thread / subject?
If it’s got to be manually-made bookmarks, I’d like a project-style way of managing them. I’m picturing an Exposé-style array of thumbnailed webpages, possibly with the ability to divide off subsets, and maybe star-rate for importance. But then you’re getting into far too much work (but I think it’d get natural very quickly). I know that I think very visually, and seeing a thumbnail like that would appeal to the way I’d like to organize data in my head… have it all out in front of me. Maybe this whole business could store scraps, relevant quotes, too? You could even extend it to a collaborative tool, like SubEthaEdit, in a way, with people all contributing to the well of references.
Maybe combining the two ideas wouldn’t be terrible: an automatic breadcrumb trail of where you’ve been (with each page automatically, intelligently rated for importance), which a little quick glance over would let you refine for your own use? I don’t know.
As for retrieving traditional bookmarks, this reminds me of what I consider an odd design decision of Safari – the whole “bookmarks as a webpage” idea – I feel like I’m encouraged to use that full-screen bookmark view in Safari, but that completely takes me off the task at hand, makes my lose my train of thought.
On the other hand, the History menu wins back some of my favor. It’s completely simple: recent web pages are all right there, none of this browsing-by-date bullshit. All you care about is that they’re the ones you were just looking at. I’ve often accidentally closed a window (and all eight of its tabs), and just had to go back to that menu to find what I’d been looking at.
It’s crazy that every browser doesn’t just have a “locate bookmarks” option. Jeez – they’re a standard enough format, usually, right? .url files? Your bookmarks really are your bookshelf in physical terms. Well, except that they only refer to the info. Umm. You know what I mean.
So, my UI development experience is limited non-existant. But these are ideas. And it’s 2AM. First final tomorrow!
Oh. Umm. So, Trailblazer is pretty amazing. Damn. Wish I’d looked at that first.
So you raise some very interesting points that I’ve been thinking about, too. It may be two different problems, but I think the solutions need to be intertwined. I don’t doubt that different UIs will be necessary for marking relevant information (ie, creating “bookmarks”) and displaying them (ie, Safari’s bookmarks-as-webpage), but the two need to be aware of each other. I think they need to be aware because I envision the use of bookmarks to be a much more iterative process than it is now. Currently I create a bookmark and occassionally use it to access the target page. But more often I end up bookmarking a page, looking around for a while and find a piece of information that’s much more valuable to me than the initially bookmarked page. However, given my “lazy” nature, I don’t update my bookmark; instead I continue to use the intial bookmark, knowing that the page I want is “nearby”. If my browser stored my breadcrumbs trail (ala TrailBlazer) with the initial target, it’d be that much more useful to me. Or if I was able to quickly update the bookmark (without going Tools –> Manage Bookmarks, etc), maybe it’d be more useful.
I think the idea of tracking links, etc, is worth exploring although I have doubts about how effectively it could be applied. For your example, is it possible to accurately discriminate between news stories about biological weapons and journal articles related to bacteria? It may be possible, however, to apply some sort of Bayesian analysis to the links, similar to what Mozilla Mail/Thunderbird uses for classifying Spam. That is, after some training on what’s “important” (to you, the user) and what’s not, it’s able to make mostly accurate guesses.
Finally, in regards to the idea of a “locate bookmarks” feature: the real problem is that browsers *don’t* store bookmarks in the same format. Internet Explorer uses some proprietary (shocking!) binary file, if I recall correctly, while Mozilla and it’s derivatives use an RDF file. I don’t know what Safari uses.
At this point I plan to work on the Mozilla platform, mostly because it’s cross platform and explicitly supports extensibility. I’d love to extend something like Safari, but it fails on both counts: it’s not cross-platform, and I’ve found no documentation on extending it. I feel pretty confident in saying this, since I searched several times when I was considering porting mozCC (http://yergler.net/projects/mozcc) to Safari.
Anyway, thanks for the feedback, Andy; I hope you’ll give some more as this progresses.
Great article, I’m searching now for something like too. Just now I found a sort of software called mind-mapping, and an example of it at http://freemind.sourceforge.net/ For the first look it looks great. And it manages bookmarks. Maybe it can help you
have you tried furl.net? it goes some way in solving some of the problems described and it’s a great way to track your history of interesting pages. there’s a fast furl browsers button that only takes only click to store interesting pages.