Conceptual clutter of the Bad Kind.

If there’s anything like Good Clutter.
Have you ever used Ubiquity for Firefox? It’s quite nice, yet somewhat inaccessible for most people. The idea is that with an easy key combination on your keyboard you can bring up a text field that can understand a large number of commands and bring you information related to them. For example you can calculate stuff, show a google map, translate text etc. It’s arguable how useful such an interface is for the majority of people, in the context of a browser add-on, but Ubiquity won’t be the focus of this article. While it certainly has its merits, it’s cousin, Taskfox, a project aiming to bring Ubiquity to Firefox [as in: the core of the browser] is so bad — even at a conceptual level — that’s raising questions as to whether its authors have a solid grasp of usability and the role of an address bar in browser design.
So what’s wrong with Taskfox [as shown in the prototype]? Well, for starters: it — in many cases — embeds web content in the address awesome bar drop-down window at the same time when a fully functional, familiar interface for viewing web pages exists just pixels below. Take for example, the Wikipedia search, shown on the demonstration video. It strikes me as particularly problematic that the awesome bar drop-down menu is inherently abused, turned into a movable widget that persists even after the text-entry operation is complete. Besides distasteful, it’s also problematic; what is the purpose of the Firefox window underneath the awesome bar if we’re only going to be using its crammed popup windows for an increasing number of tasks?

Yet, the problems behind Taskfox are more fundamental than that. To some extent I can understand the confusion of the Taskfox authors. There seems to be a underlying problem with the way data is provided and presented online; a problem that makes the task of writing systems like Taskfox for the internet difficult, and the lives of those that try to design them harder.

Mass Confusion

One major problem with the way data and services is the need for a presentation-semantic dichotomy; so much information nowadays is delivered in human-readable form online in various web sites using HTML. Sometimes this information can be easily parsed by computers using either standard scrapers or custom software; others, in most cases that is, it’s very hard. This lack of the separation of form and content in HTML — or, in other words, the combination of information with its presentation that’s inherent in the modern web and that the elusive Semantic web aims to solve — is, in the case of Taskfox, complemented by an unjustifiable eagerness to do everything and anything through a text interface and a text-field, even if this results in the ridiculous concept of showing web content in a small drop-down window that appears under the address bar of a browser, while the browser content pane remains unused.
Interfaces like Ubiquity, the LaunchBar, Quicksilver, Gnome DO and to some extent even Apple’s Spotlight or the Google search engine do promise a radical change (for the better) in human computer interaction and certainly any work on them is very welcome. When they originally appeared launchers such as LaunchBar or the more sophisticated Quicksilver really provided a fantastic new way to access information; but those applications differed from both Ubiquity and Taskfox. They tried hard to guess what you wanted without expecting a ‘command’. Then it was a matter of how information was handled: both applications were launchers which means that once the required information was identified they offloaded its actual presentation and processing to other applications that could present it best; for example, once you found that word file you wanted to edit, Microsoft Word would be launched with it. LaunchBar never tried to process text, Quicksilver never tried to send email. Taskfox seems to try so hard to replace everything and this — eventually — is what hurts the project both conceptually and practically from a usability point of view. Further to this, LaunchBar, QuickSilver, GnomeDO etc. all share a unique place in one’s computer. They represent a focal point through which interaction should take place. This is not immediately shared by other applications of the sort: Google search for example, or even Apple’s Spotlight, the former in lieu to its web nature the latter due to its limited functionality. If anything, the ‘shell’ (and I don’t mean this in a unixy sense) is — still — the place to innovate in this respect, not the browser (or the web).
While someone might expect the appearance of the interface elements to improve (after all this is only a prototype) I’m afraid that the conceptual problems behind Taskfox and the underlying issues with information online make it somewhat pointless as part of — what effectively is — a web browser.