When I started at Google in 2004 this model was already outdated. I built the news bit of Google trends in 20% and this was subsequently launched on labs, but only in the context of a larger project; as a rule things launched on labs were built by fully staffed teams. More and more labs became a marketing thing; something to point at to prove continuous innovation at Google. Also since Google products typically remain in beta for a long time, the company needed a different label for products that might not be quite ready. Labs more and more became that label.
Not everybody agreed and there was a lot of grumbling about how hard it was to launch new stuff. As a reaction, individual products like GMail, Youtube, Maps and Search launched their own labs that allowed for the launching of experimental features . Soon another effort came under way to make it a lot easier to launch experiments on labs by using standard components like AppEngine. After Wave got canceled, me, pamela and tirsen got together and built Shared Spaces that way in our 20% time and launched in on Labs.
And now labs is going to die. No doubt this has to do with the founders desire to focus on Big Problems. Like I wrote elsewhere, I am skeptical you can get a grip on the Big Problems without starting small; Labs seemed like a good place to start small. Bits of Shared Spaces reappeared in the Hangouts of Google+. Last week Google announced also the closing of Slide, another bit of the empire meant to experiment. All signs point to that in Larrys new Google innovation is top-down rather than bottom-up.
Enough history. Google Labs might be gone, but I am happy to announce Triposo Labs. We collect data from all over the web and use clever algorithms to produce travel guides. Obviously before we get something working we play around a lot with that data. Quite often we hit upon something that is interesting, but maybe not immediately applicable, but if they make us go, ooh, that's cool, why not share it with the world?
The first experiment we're publishing, tracks the development of the wikipedia on a world map. Each geocoded article is plotted in order of appearance showing how the Wikipedia initially focussed on the US and partly on Europe and later spread to cover the world.
So if you're bored, go checkout Wikigrowth