There is a definitely an air of optimism in the conference - we know the taks are hard and very very diverse but it's clear that many of them are understood. The morning plenary was Carole Goble, Manchester - who has been a central pillar of the eScience community and one of the relatively few computer scientists who have really got immersed in the needs and the culture of the scientists, rather than the abstractions of the CS. (Of course she does the cutting edge stuff there as well, but hides it from us). She'd just come from Amsterdam ("where people smoke in bars") and gave her natural bouncy presentation - today on workflows.
The great thing about Carole is she's honest. Workflows are HARD. They are expensive. There are lots of them. Not of them does exactly what you want. And so on. [We did a lot of work - by our standards - on Taverna but found it wasn't cost-effective at that stage. Currently we script things and use Java. Someday we shall return.]
So random jottings - mainly about stuff which was new to me:
myExperiment.org. A collaborative site for workflows. You can go there and find what you want (maybe) and find people to talk to. "- bazaar for workflows, encapsulated objects (EMO) single WFs or collections, chemistry data with blogged log book, encapsulatd experimental objects Open Linked Data linked initiative...
facebook-like but based round the object, not person
from me-science to we-science. Crossing tribal boundaries
new project WS4LS (web services for life sciences - complete catalogue.)
Scientists do not collaborate - scientists would rather share a toothbrush sharing rather than gene names (Mike Ashburner)
who gets the credit? - who is allowed to update?. Changing metadata rather than data. Versioning. Have to get credit and reputation managed. Scientitsts are driven by money, fame, reputation, fear of being left behind
Web 2.0, perpetual beta, users add value. blogs, viral marketing
Needs to ba an Ontology dictatorship.
RDF, OWL, SKOS, ORE, Open Linked data
Annotations are first class citizens
Nice Interfaces with simple functionality matter more than sophisticated reasoning engines
2 Design patterns for repositories in response to LIz Lyon's request:
Long Tail, Data is next Intel Inside, Users add value, network effects, Some rights reserved, Perpetual beta, cooperate, don't control
Mass community annotation,
React to changes
Timely (just enough)
Review (rather than control)