Today I have been at an “e-Science Think Tank”, run by our very own e-Science Institute here in Edinburgh. A theme of the day has been debating whether the whole e-Science thing has been a success or not, and where to go next. Two particularly interesting points. Point-1. Nearly everybody agreed that pull is better than push … that is, all the successful examples have been driven by specific disciplines (biology, climate science, astronomy, etc). The alternative – that computer scientists can design generic solutions for grateful application scientists to implement – doesn’t seem to work. Point-2. “The Grid” in the pure sense of Globus-driven CPU-cycle pooling, has not really taken off.
These two points are really the same. Going back to 2001, the Globus-SRB “Grid” thing was a computer-science agenda driven by two specific US labs, and given impetus by a very successful particle physics community campaign. Gradually everybody else realised it wasn’t quite what they wanted, discovered industry standard technology like XML and web services, and worked their own standards. There two really good things about “The Grid”. One was that it rescued LHC computing. The second was that it was a banner under which we marched to the Government Feeding Troughs. The bad thing was the rigidity produced, which thankfully gradually subsided as the more general idea of e-Science emerged.
Before Ian Foster discovers this blog and gets cross with me, I should stress that his vision of “The Grid”, as opposed to the Globus software, was much broader and more general, and quite like what we now mean by e-Science and Web 2.0 – it was about transparency, ubiquitous computing, collaboration, participation, and democracy. Spot on.
Transparency is the key word. HTML/HTTP is all about transparency of documents; XML and SOAP is all about transparency of data; in principle Globus is all about transparency of processing. But possibly transparency of processing is a chimera, for technical reasons ..
I would explain but I am off now to get my free hot dinner, just to annoy Rick Nowell.