The discovery to delivery hook-line has been used for a while to describe a goal of those information services which support the academic researcher. The challenge to academic libraries, national information services etc. has been to support the researcher from the moment they begin the process of searching to the delivery of the digital or physical artefact which satisfies their enquiry.

Lately, I've been thinking about discovery to delivery, wondering why it just doesn't quite work for me. I've been preoccupied with this mainly because I was invited to devise a diagram to express discovery to delivery - an architecture if you will - and found myself either focussing on discovery or delivery, but not really both together.

I think it is the way in which it implies a 'round-trip' which bothers me. It sounds synchronous - almost 'client-server'. What it is missing particularly I think, is any notion of how the researcher has registered their interest. There is, perhaps, an implication that the researcher has just initiated a search operation.

At a meeting last week involving, among others, JISC and some of the services they fund, as well as the British Library and CURL, I gave a short presentation (working to a 'maximum 5 slides' rule) outlining the idea that it might be interesting to consider the proposition that more and more of our information is being delivered to us without it having been explicitly asked for, and that there might be an interesting model in this for the next generation of services supporting scholarly research.

I considered the fact that, like many, I'm engaged in a sort of continuous, low-level, background research activity. Firstly, I have registered my interests explicitly with a number of online services, and receive regular deliveries of content which is often useful. Secondly, I have registered interests in a less explicit way by choosing to subscribe to the output of a number of academic and non-academic bloggers. Thirdly, some of the systems with which I am registered are starting to make recommendations to me about stuff I might want to look at. This is the technique used most prominently by Amazon, where the system offers suggestions of other items I might be interested in ('recommendations') using its database which relates me to other users and to our respective activities in the system.

Note that I'm not explicitly seeking particular content here - I'm establishing finely-tuned-antennae to catch useful intelligence. The fine-tuning is a continuous ongoing activity but, importantly, not all of it is conscious, and not all of it is initiated by me. Currently this sort of thing is still done in a fairly passive way - I go to Amazon for example with the intention of making a purchase and Amazon tries to tempt me with what is, essentially, targeted advertising. We might not want Amazon to actively 'send' us suggestions when it's algorithms detect a possible sale to be made. But imagine this model applied to a repository, or a library system system. At the meeting last week we considered this scenario - and what might be possible if various services were 'joined up' and able to share networks of users and preferences. It seems to me that the ultimate utility of this kind of system is when it feeds useful stuff to me that I didn't previously know I was interested in. I sometimes discover the things that it turns out I should be interested in this way.

I like the notion of 'gestures', recently popularised by Steve Gillmor to describe these ways in which our interests are communicated to others, registered by systems, or mined from transaction logs. As I go about my professional life, I make these gestures or indications of interest, and I ensure that my personal information system is tuned to catch the responses from these gestures. My current toolset for this is based primarily around RSS-based harvesting and subscription, but it is not limited to this.

Of course, if this type of activity continues to grow apace, then the problem of managing information discovery remains, it is just transferred closer to the researcher/user. In fact, the activity of discovery follows the (semi)automatic delivery.

Perhaps there is a new model, complementary to the first:

gesture -> delivery -> discovery

where the different elements happen asynchronously.

Is 'discovery to delivery' sufficient any more?