It’s not the widgets that suck, it’s the widget delivery

I agree with Fred Wilson’s points in his fantastic post Widgets Is The Wrong Word For What We’re Doing. But perhaps it’s not really the widget that’s the problem. Perhaps it’s the widget delivery architecture.

Early last year I said widgets would be big in terms of usage and adoption. Brad Feld agreed that they’d be big as an “application container for publishers.”

And widgets are clearly an important part of the web today. They help us distribute content where we want it and where it’s of the most use to us. They’re so rich, so personal. Yet still, now in the second half of 2008, the experience of widgets still undoubtedly sucks on so many levels.

This seems fixable to me. The beauty of it is that I’m not so sure we need to throw the baby out with the bathwater. Some of these widgets are pretty amazing. They’re skinnable, flexible, and intelligent. Best of all, they give developers a simple model for distribution. It’s just that the delivery model is fundamentally flawed.

So I think what we really need are improvements in the widget delivery model. There’s simply no quality of service, not enough caching going on, poor design in some cases, and no graceful failure modes. It seems like some kind of “widget-CDN” could be developed.

Why is it that a single injection of Javascript couldn’t be the last one I’d ever need?

If there was a widget distribution intermediary, installing a widget could be a simple dashboard-based drag and drop operation for the publisher. There would be almost no pain to remove/add/replace a widget. If a widget wasn’t available (e.g. the widget publisher was down/overloaded – think Twitter), then that page view would simply not contain that widget. Or perhaps it would contain a cached (and working) version – wouldn’t that be nice? The distribution point would understand availability. It would be efficiently caching widget views with publisher control. It would be ensuring a reasonable experience.

I think this is fairly close to what Fred is actually calling for: a “more integrated model of mashing up web services.“. When I first heard about WidgetBox, I figured this was essentially what they were going to go and do. But alas, services like this simply exacerbate the central problem.

So, how is a widget like a toilet?

Maybe it’s not all that bad if you have the right plumbing.

file under: Blog, Startups