Yesterday pipes arrived (1, 2, 3, 4).
I will not describe the service, the picture speaks by itself; one can visually compose procedural modules to compute data from Internet sources. Maybe it is a “a milestone in the history of the internet” as Tim O’Reilly puts it but yesterday it was extremely difficult to make oneself an opinion since they were clogged most of the time:
Today we can, and it is a (small) disappointment. The service only accepts as input Atom, RDF, or RSS, and not arbitrary html (although one could use dapper to expose data). Atom and RSS feeds are only useful for news based services…, and RDF does not work yet! Tried with several foaf files as well as with the planetrdf blogroll . Hopefully this will be fixed soon.
The main problem is still:
- Scarcity: not much data is available on the web, in RSS rdf or other formats.
- Smartness: much of this data is pretty useless without reasoning, beyond very basic mashups.
However some aspects of the tool are astounding:
- Usability: the tool is a pleasure to use, and effort has been made to keep things simple, I reckon even non programmers will be able to give it a go after a screencast.
- Publicity: once a service is published it is automatically public, nothing to do against that.
- Collaboration: once a service has been publish anyone can view its internals, but also copy it (this operation is called “cloning”) to start editing and modifying it.
- Social Orientation: while browsing pipes one can see the number of time it has been run and cloned. I expect comments and ratings to be added in the future.
Composition of pipes is not offered, however each pipe exports an rss feed, therefore one can use it as a source.
I like the name ‘pipes’, even ‘tubes’, because it gives a 3rd dimension to the internet, beyond the deep web made of dynamic content. Indeed, rather than a surface of pages interconnected by hyperlinks, more people will start to consider the web as the result of transformations occurring under its visible surface.
Ok, so most of the blogs I read have a demo of some pipe, why not here? Well, unfortunately, mashups are boring; in a GIS context they are points on a map lacking context and semantics. I tried to play with geonames services given that some return results as RSS, but what do do with them, why would I like all Wikipedia’s entries near my place in Switzerland? Not really sure.
So in my opinion pipes are the end of mashups as we know them; the ease of combining feeds from everywhere will soon reveal the ultimate emptiness of the exercise. And from there on, we’ll step further.
Comments(2)
