Yes, Yes and Yes

1. Google is important, dominant, and good at what it does – we need to have an approach to Resource Discovery which accepts this, rather than tries to build something ‘better’. I’d argue that if we build our approach into the warp and weft of the web then the ability to exploit Google will start to emerge naturally, but if it doesn’t we still have to ensure Google can surface results from our resources.

2. I agree the reasons RSS works don’t matter particularly – in a sense this was partly my point. We need to exploit RSS/Atom because it is there, works and is popular. What we should avoid is looking at RSS as a model of what works, or as good use of web architecture, and then inventing something similar and thinking that will work.

3. OK, OK, I’ve already accepted this (in my blog post)! The reason I made the distinction was that I was trying to emphasise the importance of the hypertext approach to information, but I am persuaded I was making a too subtle, and not very useful, distinction between the the theoretical and the real world. As I said in my blog post any linked environment ‘would absolutely have to be the web’.

1/. If we take ‘Google’ to mean global-class full text indexer and search engine, then I agree with Andy. At the moment, when I say Google, I mean Google!

2/. Doesn’t matter much to me how RSS got popular – it’s ubiquitous & worth exploiting. By RSS I pretty much mean all versions and I include Atom.

3/. Owen – I maintain my opposition to your abstracting the ‘linked’ information from the particular implementation which is the Web. That way lies Xanadu….

The Web works. RSS works. We could do so much more with these than we are in this sector.

First of all differentiating between ’service’ and ‘infrastructure’ – not always clear

Secondly, even if we agree on ‘web’ infrastructure then this still doesn’t get down to brass tacks – we could all agree on this but go away with entirely different ideas of what it would look like.

The people in the room were from a variety of backgrounds, and could engage with different levels of detail. I don’t think it would have been useful to start talking about RESTful approaches, because it would not have been meaningful to many in the room. Once this group has had some of the basic discussions some of the detail will have to be approached by smaller/other groups I think.

I think there is a “Google in the (virtual) room”. I’m torn on this one – on the one hand I absolutely agree we need to find ways of ensuring the content we care about is surfaced via Google. On the other, I’m wary of building explicitly with this in mind. This is partly why I stressed the idea of a ‘linked environment’ in my talk, without being very explicit about it – because I believe that if we can embrace the idea of links as an integral part of how we present data, then some of that Google goodness will inevitably flow in our direction – but I think the strengths of using the linked environment go beyond Google – Google just exploits this.

Finally a sidebar on RSS. I’m not convinced that RSS actually did anything particularly right, except become popular. Well, this clearly isn’t true – RSS became popular for specific reasons – but there was a lot of stuff wrong as well – it is a miracle it ever took off! Clearly being transported over http was key, but apart from that? I suspect that it was as much a question of being the right thing at the right time as much as the detail of how it worked. In essence it succeeded because enough people saw it as solving a problem they had without getting in the way too much. While these are key lessons to learn, I’m not sure this moves us forward – I think we know these things, we just often fail to achieve them for one reason or another.

Re: your suggestion that “OAI-PMH builds on web architecture”… I basically disagree. I mean, it’s debatable, clearly – because we could probably have an argument about what “builds on the web architecture” means in practice – but 1) OAI-PMH isn’t RESTful in its approach and 2) the way it models resources and assigns URIs to those resources is, shall we say, confusing and unhelpful.

As I’ve argued before, approaching harvesting via sitemaps and content negotiation (to serve appropriate representations of the resources listed in the sitemaps) and making significantly more use of RSS/Atom would have been significantly more in line with the Web architecture and would have played to (rather than against) Google’s strengths.

If we agree that ‘the web/web architecture’ is the infrastructure, I don’t think that says enough about how we integrate library resources into it to be useful. When I think about the need for ‘infrastructure’ as regards discovery of ‘library related’ resources, the challenge for me is to work out the most successful way to exploit the web in this context.

To take an example I would say that OAI-PMH builds on web architecture – but it hasn’t (yet?) been successful in making it easy to discover resources which are exposed via OAI-PMH compliant interfaces.

I don’t agree (and again, this comes back to ‘what do you mean by infrastructure’) that Google (at least the search engine bit) is infrastructure. Google is a service that sits on top of an infrastructure – which is ‘the web’ (clearly this is shorthand, but I think understandable)

I argued in my presentation that ‘if we were starting from scratch’ we would build a ‘linked environment’ for our resources – but I also said that this wasn’t necessarily the web – which Paul challenged me on. I clarified in my blog post, that in terms of where we are today that means the web.

It’s not clear from your description what the scope of ‘resource discovery’ is here? I’m guessing it’s in some way library-related (books and journals) rather than general resource discovery – not least because I can’t imagine we would presume to attempt to build an infrastructure more ‘world class’ than Google?

Whatever… doesn’t matter really… the infrastructure is the Web and the Web Architecture – full stop. End of story.

(PS. Not quite ‘end of story’ I’ll grant you, since there are minor issues to resolve like what the hell that means in practice – but it’s a better starting point than we’ve had for some time).

Yesterday I attended the first meeting of the new JISC Resource Discovery Infrastructure Taskforce. I enjoyed the day, and the Task Group is bringing together a great set of people who all had an incredible amount to contribute to the……