Comments on: Queues: A Browsing Paradigm for Power Surfers

By: Adam Richard

Adam Richard — Thu, 19 Jul 2007 00:18:14 +0000

I really like this idea. I'm wondering whether tabs are really the best interface for a tree traversal of your web-pages, though. That is, if my browsing history is a tree, why not just have the little down arrow on the back and forward buttons bring up a graph of the web pages connected to the current one? It's very often that I go to a web page (call it A), then go "back" a web page or two, then click another link, and now neither the back or forward buttons can get me back to A. With a tree, I would be able to. Since I'd be able to find all the webpages I've just been to, I would be able to stop worrying about starting new tabs for new threads of browsing. Tabs might still be useful for working with multiple web pages at once, but since that can already be done with separate windows, I'd almost wonder if they're worthwhile at all in the long run.

Further, it may happen that I'm on a webpage and want to remember to visit a bundle of links in the future. So there should be a feature to Ctrl-click or Alt-click or something on a link to add it further along in the graph of the forward/backward buttons. (I think you kind of discussed this in relation to tabs though.)

By: Mike Purvis

Mike Purvis — Wed, 11 Jul 2007 22:13:51 +0000

Terrill: I’ve wondering about similar things as well, but always with more of an emphasis on the present, rather than on historical data… it’s the present that concerns me in the everyday. But as you say, the real problem is finding a way of elegantly collapsing the stretches of linear browsing so as to emphasize the interesting bits. Perhaps you could simply throw the whole thing on a logarithmic time scale, so that the farther in the past you go, the less long the node connection lines become.

Anyhow, the other issue is that SVG and Canvas are both ridiculously slow for doing serious graph stuff. I used Canvas for the clock and SVG for the six-stroke wankel, and in both cases I found the performance pretty disappointing. So you could always experiment, but if you mean to package it as a Firefox plugin, expect to waste time getting it up to speed.

Cam: Really? I think the tabs as implemented are the least surprising approach. Most of the features I’m describing here would catch a user off guard if they weren’t expecting them. Also, this is heavily optimized for a very specific kind of browsing strategy. I’d still like to try implementing these ideas as a plugin, but I don’t they’d ever make sense in a browser core.

By: Cam Turner

Cam Turner — Wed, 11 Jul 2007 17:30:05 +0000

This is an awesome idea. I know many people (who are not power users) who simply assume that this is the way tabs should work and I agree. have you submitted this to the mozilla team? They should at least read this article.

As for Terrill's idea, personal browsing statistics would be really cool, but I'd expect that this kind of feature would be released by some sort of spyware company because that data would be really useful to them too... or a search engine company even. I'd be weary of anything that already does what you suggest as there is a good chance there are alternative motives.

By: Terrill

Terrill — Wed, 11 Jul 2007 14:43:24 +0000

This is slightly off topic, but anyway…

Seeing your hint at trees above I have always wanted a visual tree view of my browsing. Even though it would be more like a forest with lots of root nodes, and very tall trees, it would still be very interesting.

Also, I would really like to see a different view of my browsing over the day/week/month. Something more web like, showing only unique nodes (pages) and the connections I’ve made between them. The line weight of the connections would denote the number of times I had made that connection. The size of the node would denote the number of times I viewed the page (that would be a good indicator of importance).

Since the web would be a real mess when viewed (lines crossing each other, going under nodes, etc) when you click on a node it would ‘organize’ the web around it, to provide at least a bit of clarity to that specific spot. I’m kind of envisioning Digg Swarm, but personalized, and persistant over sessions. http://labs.digg.com/swarm/

Do you know if anything like this exists? I am very tempted to try making something. Use some of these data structures and algorithms I’ve been learning this term.