Hacker News new | comments | show | ask | jobs | submit login
Enhancing Multitasking to Enhance our Minds (einfall.wordpress.com)
22 points by DaniFong on Aug 25, 2008 | hide | past | web | favorite | 9 comments



Right now, I use Tree Styles Tabs as a sort of "temporary history", since I got tired of thinking about which tabs I should keep open. I'm also using the Ctrl-Tab extension just like Alt-Tab on windows (can't live without it). With a vertical tab list, I open almost everything in a new tab. My depth first search habits are limited by seeing the tab indentations and knowing where the root is.

My ideal tab behavior would involve merging browser history and tabs into an infinite vertical scroll. Pages with unsaved state (form data, important flash objects, AJAX "state") would stay in Firefox's memory until explicitly closed. All "boring" pages would stay in memory for a few hours, then convert to a URL+scroll-position+selected-text object. Combine this with single-keystroke tab tagging and I think that tabs will finally feel like a reliable part of the browser, not just a taskbar-inspired hack.

edit: Ideally, desktop applications would work in a similar way. Right now I have to keep ~30 windows open because I can't remember which serialize state perfectly, and which don't.


I like Tree Styles Tabs, but it seemed to take up too much space for me to consider using. And it wasn't nearly so good at showing the temporal dimension of browsing.

Do you think the proposed system would satisfy your desires? It includes all the data shown by Tree Styles Tab, and merges history and open tabs quite seamlessly, with the 'history gutter' system.


I like some parts of it (like how it detects typing activity), but I think it adds more visual complexity than needed. The temporal information might cause confusion: what if almost all the pages in tab were accessed 14 hours ago, but then you clicked a link in it and went back? what time would you show? There's just so many time variables that I don't know which would be useful. In either case, I would prefer a more subtle approach to showing age, like fading tab color.

I think tabs have been broken from the beginning, though. My goal is to traverse information as rapidly as possible, tagging things I like (today I do this by manually with Save Page As). I simply visit too many pages to know if a tab/page is a true "task" or not; this is why I don't see the purpose of distinguishing between currently open tabs and history. And this is why the gutter bothers me: why distinguish between what's open and what's closed (but is still useful)?

I realize many people use just four or five browser tabs, but that kind of workflow is more suitable to the OS taskbar itself. Often times, half these tabs are a stateful application like gmail and the purpose of the tab is to separate workflows. Automatically identifying tabs with stateful activity is the key to this. Right now, users are doing it manually and keeping those tabs open all the time. It should be the browser's job to know what is "important" and what is "just a page". This is one of the true advantages of the iPhone (I haven't used other smartphones) - all of the built-in applications preserve state perfectly and I never have to fiddle with a separate "app launcher" and "window switcher". In Firefox, the tabs are the unnecessary "window switcher". Of course I'm ignoring the behavior of MobileSafari because it's a memory/space-limited device.

This is somewhat separate from your design, I'm just finally writing my complete rant on tabs. I realize I might be browsing the web very strangely. Sorry to thread-jack :-)


No, don't apologize, this viewpoint is absolutely intriguing, and I greatly appreciate the discussion. The goal is to make the best tab navigation system, not to trumpet any particular design :-)

I especially like your point about identifying tabs with state, and identifying tabs that are basically applications. Perhaps we can actually accomplish much of this automatically? We can at least detect the presence of things entered into text-fields. We can at least save a state of the DOM and Javascript namespace instead of wiping everything else out.

If a tab is an application, it perhaps doesn't make sense to bind it to the same horizontal time axis; perhaps instead to do as OS X does, by loading applications into a doc. On the other hand, though, on OS X I just navigate around with Quicksilver, which has a quasimodal 'task-leaping' behavior that's totally supported by this version of FoxGlide.

It's certainly true that it's unclear which time variables are the most important to display. At the moment, I suggest a hybrid approach -- organize the tabs horizontally by order of opening, as is the default in Firefox. At the same time, indicate on the activity ticker, for each time, which tab you have opened, and what you were doing. This keep the horizontal ordering of tabs static, aiding spatial memory, while allowing you to recall what you were doing at any given moment, using the activity ticker.

Regarding the distinction between open and closed tabs, I think that most of the population likes the feeling of 'finality' achieved once they close a tab. They like knowing that they could recover it, if need be, but mostly, they appreciate tabs being "out of sight, out of mind". It's always going to be important to be able to close tabs (for stopping persistent behavior: eg. Meebo, Pandora, Youtube).

I guess you would advocate the graying out of closed tabs, while perhaps lowering their priority more rapidly. Is this correct?

Is it possible that the gutter can be adapted for your needs as well? Say, it only shows up if you actually close something. We need to strike some compromise between people who want closed tabs out of their face and people who want tabs around, like you.


I somehow forgot about the true need to close a tab sometimes. This problem is much harder than I thought. But, one thing that I don't think will ever work is a large intermediate screen (as the mockup) as the primary method of switching tabs/pages. Most users (especially those on Windows) don't expect to see large intermediate screens (things like alt-tab are tiny). And I'm wary of two or more methods of switching, such as regular tabs plus the full-screen approach. It would kind of bring the full-screen method into an "analytics" category that most users would never use to their advantage. Is a full-screen tab switcher intended to be the primary method of switching?

The problem is especially tricky because users have already adapted their mental models of browsing to non-ideal tabs. This might require a very slow transition into an ideal navigation method. The "find a tab using location bar" approach is heading in the right direction.

I'm much more interested in improving the usability of existing horizontal (and vertical) tabs, since that's what users are used to. I don't have a compromise yet (if ever) but I would base it on these assumptions, which may not apply to everyone:

1) Just because a page has a tab, doesn't mean it needs to stay in memory (for most pages)

2) Keeping tabs, in-tab page history, and global history fragmented is awful

3) Firefox's history pane is unusable, which partly accounts for tab proliferation (for many users - how many?)

As a side note, one of the reasons I like Tree Style Tabs is that it brings the webpage text closer to the center of the screen. As long as the screen is at least 1280 pixels wide, it's usually better for me. edit: but, it is kind of useless if the visual noise bothers you. It doesn't bug me; I like to see a lot of information on multiple monitors, as long as nothing changes in the background.


One thing that I don't think will ever work is a large intermediate screen as the primary method of switching tabs/pages.

It's interesting that you'd say that, because it's not at all what I'd expect. There are two factors you might not be accounting for here: that the transparency of the tab display would make clear that the screen is a temporary one, intended to aid navigation, and that the quasimodal interface, where you have to hold some hotkeys down, keeps transparent navigation overlays from distracting from the browsing. Check out Enso for a live example: www.humanized.com

Additionally, the tab display can be scaled down to whatever level the users end up desiring: recall that navigation is done by dragging and zooming.

I will admit, I'm coming at this from the direction of 'what is the ideal navigation system', not 'what will, for the immediate term, make the most people comfortable'. I suspect that the system is fairly easy to learn, unobtrusive, and that in time greater efficiency will win out.

I agree with your assumptions. In particular, keeping the tabs system on different screens from history is terrible. But I must not fully appreciate your sense that the history gutter will fragment history from tabs. It seems, instead, that since it's part of the same screen, and the same unified system, that it's a step in the right direction. And I think there simply have to be some compromises for people who want their finished tabs done away with. Visual clutter is a serious impediment for many people, me included.


Interesting ideas. I've been thinking along really similar lines.

Good point about how a large number of tabs on the screen can be distracting. One of my friends actually moved his tab bar to the bottom of the screen, and he said that he finds it much less distracting. It's there if you need it, but stays out of your way when you don't.


I like this quite a lot. I find I can not work if I have too many tabs open, I wonder if this will help. But even if not, it's still a great idea.


Consider trying the 'full screen browsing' approach. It's not perfect, but it's helped me a lot, and best of all, it's implemented.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: