My goal with this blog

I write about relevant changes in the way that people use the web and how startups are built to provide services and products for this ever changing wonderful thing we still know as "the web." As a former entrepreneur turned early-stage investor, my greatest hope is for this to be useful to other folks that are like me in the hopes that they can avoid some of the mistakes I've made.

Web browser as Platform

As the first high-profile rich browser app, Google Maps unleashed an explosion of creativity in pushing the web-browser-as-application-runtime. It gave birth to the term AJAX, put drag-and-drop on the list of approved cross-browser features, and seemed to lift the ban on imaginative ways to use Javascript and DOM manipulation to give users a much richer experience.

Proponents of the browser-as-platform have been around since Andresen's famous snub at Microsoft ten years ago ("Windows is gonna be reduced to a buggy bag of device drivers"), but in recently years people have noticed that some apps (i.e., webmail) are making a real run at replacing their desktop cousins.

But what if instead of working on a task-by-task basis, someone came out with a full-blown application runtime for the browser? I'm not sure I buy it, but I was reminded of it yesterday when I saw this Mac OS 7 emulation in the browser. Now it turns out that this is actually very painstakingly constructed Flash movie with a bunch of interactive dead-ends, but it nonetheless got my wheels spinning on what would be required to go further and provide a full OS environment in the frame of the web browser. Are there ways in which this would be more useful than the current stand-alone "rich" browser-based apps, i.e. Gmail, or Writely?

From an implementation point-of-view, I think that the pseudo OS in browser is finally really possible now. In the office, everyone is really tired of my constant touting of the recently open sourced Yahoo User Interface Library which I find to be as well thought-out and documented as any rich GUI library that I've seen. It would not take too much work to wrap something like Amazon's Simple Storage Service, and a few web service APIs for richer interop with "application islands" out there on the Internet (EBay, Google Maps, etc.), and you've got a really rich runtime at your hands.

This morning I saw a glimmer of this in the Y-Combinator backed YouOS. They've got an application installer, cross-application interop, an API for people looking to write new applications for YouOS, and even a set of primitive data structures that can support concurrent network access and be persisted on the server. I think these guys deserve a lot of credit for trying this, and the remarkable thing is that it works almost as well as an early Linux X-Windows window manager.

However, it's just a glimmer because no matter how far the richness of the browser has come, my rich client has gone further. Right now OS X 10.4 on Intel is finally snappy enough to be a first class GUI with a rich UNIX core and there ain't no web-based OS that can come close to that level of productivity.

But there may be a compromise. Apple (to continue the example) provides a hosted-data service called Dot-Mac which supports webmail, calendar, and basic web publishing as well as a very limited backup service for data and metadata. Over the years the web side of Dot-Mac has gotten richer and richer in terms of the functionality exposed. Why couldn't Apple go the whole way and create an AJAX version of their Display PDF engine and Cocoa framework, sync the list of your installed applications, and allow you to run an entire Desktop in the web browser? They've already got key data and metadata stored on the server, and they own the real OS APis that everyone is already programming to. For sure the web experience would be slower and crummier, but for occasional use it would be just fine.

To repeat it for the third time, I am not sure that this whole OS-in-the-web-browser is a better model than finely tuned webapps that use their own native medium better- in fact, if I had to bet on it coming from a startup like YouOS, I'd bet against it. But I do think that if someone could pull off a valuable use-case, it would have to be one of the rich client platform vendors (Microsoft or Apple). Dashboard Widgets was an interesting step in that direction from the rich client out (as was Konfabulator), but the big guys could do much more.

[ As an aside, it is too bad that the guy with enough skill, experience, and muscle to pull it off seems now to be distracted by the kind of cute web hack that should be left to the rest of us not in control of a major platform. ]

As I often think when I finish one of these posts, it's nice to see that we live in interesting times.