yubnub.org: an Internet OS?

Screenshot of Aquino’s yubnub

Originally uploaded by sogrady.

A while back Kottke picked up on a post of Rich Skrenta’s that thoughtfully pontificated on the possibility that Google has something grander in mind than scalable web point applications. Kottke (calling it GooOS) and Skrenta cited the growing accumulation of both PhD’s and operating system experts at the Big G as evidence that huge things were in the works.

Google, somewhat predictably, denies such assertions. Just last week, Eric Schmidt said the following in an interview just last week:

So the evolution of Gmail, the evolution of search…I think it’s up to you and others to speculate about what’s going on. That’s not how we think. We don’t sit there and say, ‘Wait! We want to build an Internet operating system! Would you like to join me?’ That conversation does not occur…We delight in the lack of such strategy. We’re very careful to say we’re not trying to build one thing. We’re trying to innovate in all these interesting spaces.

You can take that at face value, if you wish, but I personally don’t buy it. But whether or not Google is consciously trying to build an internet OS, we already have one. Want proof?

Look no further than the pictured yubnub.org. Before you get carried away with that absurd introduction, let me be clear that this is not some massive project designed to overlay the web with some sort of OS like layer. It is, rather, a small Ruby project built by Jonathan Aquino specifically for the 2005 RoR programming competition. All it is, essentially, is a means of shortcutting to various web resources (Amazon, Google, etc) via a CLI-like mechanism. Its like having specialized, CLI bookmarks. Its very simplicity and lack of scope is (along with its extensibility), however, what makes it interesting.

For example, if one begins at yubnub.org and wants a map of, say, Denver, you merely enter a very Unixlike “gm Denver, CO” and yubnub takes that entry and feeds it into Google Maps. Want to search for the new White Stripes album on Amazon? “am White Stripes” feeds you into an Amazon search. How about MSDN? “msdn Mono.” And on and on. What’s the list of available commands? “ls.” Now this would be interesting in and of itself, but what makes yubnub.org even more compelling from where I sit is the ability to create new commands (command is “create”).

Now frankly I have no idea if there’s prior art here; it may well be that somebody’s done something similar before (though I don’t consider Google’s site: search functionality comparable). John Battelle would be the one to ask on that score. But I do think that in a world where search has become the starting point on the web for a great many people, the ability to search one level up from even a universal search engine like Google is an interesting notion. Yubnub’s key insight is the idea of transforming individual web assets into targetable resources via a keyboard shortcut, just as an operating system marshalls services like filesystems, runtimes, etc. Sort of like the semantic web, as well. In another operating system parallel, yubnub can be extended to perform new and interesting tasks (as an aside to Jonathan, this is not likely to scale indefinitely, because as the volume of available commands grows, their usefulness will decline).

At this point yubnub is little more than a thought experiment, but it’s definitely an interesting one from where I sit, one that gives us hints at what an internet OS might look like. If I was working for Google, MSN, Yahoo or whomever I’d certainly give it a look.

Thanks, Jonathan for taking the time to send over the link.