Jonathan Diehl

Three years ago, Noriyasu Vontin from our research lab developed the workstation-independent desktop environment (WIDE). He started with a simple observation: desktop application store your documents in your file system, while web applications (like Google Docs) store your documents in their (file) system. This is really unfair treatment, because after all these documents are yours and you should be in control of them. Admittedly, you can download the web documents and store them on your computer, but this is cumbersome and eliminates the benefit of having documents (and applications) in the cloud to begin with.

Nori’s solution was to create another web service that acts as a file system to store documents from web applications. This service is separate from the applications such that, in theory, all web application share a single file system, just like desktop applications.

He spent quite a bit of effort on getting this system secure, such that malicious applications cannot break anything. By doing so, he actually invented what is now the sandbox of Mac applications: no application can directly access the file system; instead, the applications ask the file system for storage, which is granted on a single file basis only after confirming this directly with the user.

When Nori finished his thesis, we severely struggled with placing his work anywhere. There was no obvious benefit for the research community, because everything that came out of it in terms of technology was already known, and the system was not suitable for running user studies to extend our knowledge of  human behavior. At the same time, it did not really fit an industrial context either, because for them having the data of their users is very beneficial.

Recently, almost three years later, the unhosted project has re-developed this idea and started an open source initiative to push this concept into the world wide web. In hindsight, I believe this is really the right way to go, and I am excited about what will come out of it. However, I doubt that the big players (Google) will go along with it, so it’s success mostly depends on whether they can build up enough momentum (apps) to actually become a useful service for us.

Currently, the unhosted project does not appear, yet, to be ready for wide-spread adoption. However, when the time comes, I urge you to give it a try and prefer those web apps that are ‘unhosted’ over those that are not.

By the way, can you spot the similarity between web applications today and iPhone applications? Hint: it is related to storing documents…

If you are interested in the details of Nori’s work, please read his Diploma Thesis or contact me.

Siri is Apple’s latest marketing gig, along the lines of the buttonless iPod shuffle and the original iPhone: a thing that will solve all of your problems while you show it off to your overly-impressed friends. But what can it really do?

In reality, Siri is a new entry point to your mobile device. Instead of picking an App from the overloaded home screen, you tell the device (in voice) what you want to do. On the desktop, we already have this in the form of Quicksilver and it’s copy cats, where you can access an application by simply starting to type its name. This actually has been around on the iPhone as well in the form of the search screen that always shows up when you just wanted to get back to the first screen. However, without voice input, this feature is really useless, as typing on mobile devices is and always will be painful.

But Siri is more. It not only enables you to access Apps, it integrates with these Apps and lets you perform the most common tasks without actually starting the App. This reminds me a lot of AppleScript, which existed back in the days to allow geeky users to automate tasks on the Mac. While this does not sound like much, it actually is the first step to a whole new world of interconnected applications that goes way beyond how we connect Apps today, like Android Intents. Imagine writing a blog article with Pages including a hand-drawn sketch and a link to your favorite YouTube video without ever leaving the publishing app or having to deal with copy&paste…

What is missing in Siri today, is the opportunity to integrate and control custom Apps, just like the built-in Apps. I personally believe this will come very soon and unleash the true power of Siri. Right now, Siri is like the iPhone 1: Apple dictates what can be done and everything else is unimportant. Soon the Siri team will learn what the core team has learned back than: true power comes from custom Apps.

Technically, integrating custom Apps into Siri is actually quite easy. Each app must define a set of terms and actions, which Siri can respond to. A simple note-taking App, for instance, would define the term ‘note’ with ‘list’, ‘take’, and ‘edit’ as possible actions. These would trigger the appropriate functionality to list, take, or edit notes in the App without actually launching it (to be precise, the app is launched but without its UI so it goes unnoticed by the user). The App can either pass back a list of things to chose from (in Siri), perform the action, or launch itself at the appropriate screen.

But where to go from there… As I mentioned earlier, if App developers can access each others’ task interfaces, they can interconnect their Apps and allow users to use all Apps at the same time. While this is not exactly document-centered computing, it is a major step towards it that is still compatible with our notion of applications. I sincerely hope that Apple understands this and will support innovation in this area.

By the way, do you know what Siri looks like? Google knows…