Death of the Desktop: The Movie

Commentary

Aza recently gave a talk first at CHI 2007 conference as well as a TechTalk at Google’s main campus. Google kindly videotaped this session so now everyone can see Aza talk about Enso, zooming user interfaces, the death of the desktop metaphor, and how he was carried around in the Mac classic bag instead of a stroller.

Here’s the abstract for his talk:The computer desktop metaphor is ubiquitous, but how much work do we get done there? None! Our time is entirely wasted navigating or shuffling content to the application in which we can finally work. What lessons can we learn from designing interfaces without the desktop and without applications? Is it even possible? And how does this apply to the Web? Currently, Web applications are often more usable than their desktop-based counterparts because each one does one thing and does it well. Desktop applications used to be the same way, but over time — as applications grew to support the the users in the long tail — each became a complex portmanteau of all possible features. If we are not careful, our Web apps will suffer the same conglomerated fate. Mashups and services help to solve the problem on the development end by freeing functionality from any particular application. But, there is currently no way to offer that wealth of possible functionality to users in a scalable way. Would it be nice to embed a dynamic map into your Gmail message? Sure. A Flickr slideshow? Sure. But for Google to offer those in addition to the hundreds of other possible options, would clutter the interface beyond usability. What’s needed is a universal method of accessing functionality: a way of harnessing the power of services without the need for application developers to explicitly support them. I’ll be demonstrating such a method.

The talk demonstrates that a ZUI plus a universal method of accessing functionality spells the death of the application-centric computing model and the desktop-design paradigms.

I can’t believe I watched the whole thing. Very intriguing. Actually, quite inspiring. I have been thinking about a lot of the ideas you discussed and have been coming to similar conclusions but you made it less abstract to me.

I hate the fact that there is so much waste (in code and data) due to the application-centric model. Even looking at it in terms of data. Because my data is tied to the application I’m using, it’s a hassle to move from one app to another. In some cases it is pretty easy to export and then import but it certainly isn’t seamless. The same is true for web apps.

I am also very interested in the ZUI. I am experimenting with using different ZUI implentations as the user interface of a web app I’m working on.

Keep up the good work. I’ll definitely be watching.

For anyone interested in this stuff, I recommend you check out “Beyond the Desktop Metaphor: Designing Integrated Digital Work Environments“. Lots of interesting ideas in there along the same lines.

My personal interest is specifically in the area of personal information management. The book describes several interesting projects in the PIM space, such as Lifestreams and Haystack.

The final chapter is available online, and is a pretty good summary of the general flavour of the book.

Thanks Patrick, I’ll look into that.

In the part of the video where Aza shows his javascript ZUI, I’m guessing he is using IE’s style.zoom function. Is that right? Because I’m experimenting with javascript ZUI’s and I’m trying to make it compatible with most browsers. It is a little difficult without the zoom function. Does anybody know of a solid cross-browser zoom technique?

Hello, very nice presentation. It’s mentioned that a Linux version of Enso is coming… is there progress on that or is it just a very long term goal?
I’m a Linux user and I can tell you that there is nothing like Enso there, I also believe that Linux user would agree to pay for good software like Enso.

Hi Carl, There’s Katapult on Linux with similar motives, but not quite there yet. Enso Launcher is actually a Windows version of the excellent Mac utility, Quicksilver, which came first. ZUIs are also nothing new, here’s a survey paper on zoomable user interfaces from 2002. I feel Aza fails to give credit to the giants whose shoulders he’s standing on, including but not limited to his father, Jef.

Oh and BTW, I attended his talk at CHI and am at Google.

Not wanting to start a row on which came first, Enso or Quicksilver, I’d just like to point out that to my understanding Enso developed out of the Archy project, which itself was Jef Raskins continuation of the abandoned Canon Cat interface (1987).

That there are two great pieces of software doing similar things doesn’t necessarily mean one must have copied from the other.

Manas: Quicksilver may have been around for a while, but it is little more then an application launcher that happens to do more, same goes with katapult. Enso is everything else that happens to also to launch applications. I have used Quicksilver for quite a while and do not think that they are the same.

Aza’s talk inspired me to do a couple of interface redesigns:

Inline Spellcheck presents a more streamlined version of Enso’s spellcheck mode.

Gestural Zoom and Pan addresses some shortcomings in typical ZUI navigation.

(Incidentally, I’m the author of the train trip planner that Aza mentions in the talk, as well as the paper that he tells you to read, Magic Ink.)

Hi Manas,

Be fare, Katapult is not Enso, it is mostly just a launcher and an amarok song switcher. I tried Quicksilver on a firends Mac and it is amazing, although Enso feels way more deeply integrated to me. I really seem like you are “controlling” your computer from pretty command line when using Enso. I also prefer how Enso puts the verb before the name, more natural to me than quicksilver.

The Linux app Deskbar has similar aims. It is a panel applet, and after a query is typed, a menu pops up, for treating the query as a command, a desktop search, web search, dictionary look-up, window switching, etc. The menu shows the command, the desktop search results, web search results, etc.

I think the newest version allows the user to specify which plug-in is to be invoked, without selecting from the menu.

I would love to see some integration with selected objects in applications, like Aza inserted that map in an email. But I am guessing that object insertion requires more integration than most apps currently provide. Is it implemented yet in Enso?

Also, don’t forget the efforts to improve the traditional Unix command shells. The Friendly Interactive Shell and Zsh have very good suggestion mechanisms. All that is needed now are command names that are easier to remember.

The ZUI Aza demoed is not actually implemented — it’s a mock-up available at http://rchi.raskincenter.org/demos/zoomdemo.swf

Some how randomly I found the Death of the Desktop movie on google video and I’m so grateful I did. It took me a few searches to find Enso but now that I got the trial i’m 100% certain I’m going to buy BOTH of these product after ONE night of use.

I can’t wait for the ZUI to become a real implemented project!

Near the beginning of the talk Aza asks the audience how many are familiar with log base 2, and apparently many people in the audience raised their hands.

What is the connection between log base 2 and good user interfaces? I tried googling this but didn’t get a good answer.

Personally, I don’t believe that the web offers much of a opportunity to leave the desktop interface.

A more realistic inflection point where the UI could be changed would be with a new generation of screens/input devices. For example, suppose Apple starts putting the multi-touch screens used in the iPhone into the laptop line. It seems like a multi-touch environment would be a natural for a Zooming UI.

I think your right Charlie. I don’t think the mouse/keyboard combination is completely out of the picture for ZUI’s but a multi touch interface ZUI would be so much more natural. At this point I can’t see a mouse/keyboard ZUI replacing the typical desktop experience but I can see a multi touch ZUI replacing it.

I just copied it from gmail to spellcheck :)

I’m watching this video and I think, again, about giving enso a try. Than I see the price. 20$, for a subset of the cool things I’m seeing on the video. Than another 20$ for a spellchecker (You want to sell me another spell checker, after whining how You have 8 on You computer)
Yeah I’m cheap, but the desktop fantasy, I’d buy that for way more :)

Where I’d personally could like to see enso go is being an interface for a bunch of web based services. Even for things like cropping and resizing photos. That maybe a bit on the edge, sending a photo there and back…

(is there any reason for Your site not being encoded as utf8?)

Someone needs to fill this hole: http://en.wikipedia.org/wiki/Cognetics

@Charlie: I /think/ that when he’s talking about log base 2 he is referring to information theory’s applications to interface design. If you search this blog for the post `Know when to stop designing, quantitatively’ I think you’ll see what he’s talking about.

Very nice talk, and I do think Aza fully acknowledged that he was merely putting forth his own version of existing ideas. It wasn’t a history lesson, but that’s ok.

Also, if you aren’t sure you want to drop the $20 on Enso, try Launchy it is certainly a bit more simple, but open source and free (win only, but being ported to mac and linux as we speak), and has the option of being customized with your own scripts.

Aza, I think your right that to really be a powerful tool these things need to be integrated into an OS, so why not design your own OS with sort of a Linspire model?

Along the same sort of lines as Kevin, have you considered the application of natural-language CLIs and zooming to software development itself? That would “close the loop”, in a sense, and further blur the artificial line between developers and users. That’s what seems to happen to command languages as they get extended, anyway: they become scripting languages (… and eventually used to develop new applications. Aagh!) Your Developer Prototype is very interesting, but I’m thinking of something more tightly coupled. Perhaps a new sort of Python IDE?

Smalltalk, and especially Squeak seems like an ideal complement to these interface technologies. Squeak basically is most of an OS, among other things: yet it is wonderfully low on code duplication, and fully “live”. Of course it has its own problems, and not a very big user base compared to Python, which is also more often used as a “glue language”. What do you think of Slate? Just curious.

Excellent work. I can’t wait until Enso works with linux.