After writing the other day about service discovery and the big library [1], a colleague followed up over email about whether it really might be that straightforward. I responded that yes, I think it might be, but the main issues would likely be (a) convincing people to futz with DNS records and (b) the aggregation bits.

By "the aggregation bits", I meant "once we can painlessly and automatically connect to our institutions' and friends' public and personal libraries and all the services each provides, what will we do with all the data?"

To which I think the answer is: we all become greedy librarians.

What does a greedy librarian look like?

My greedy librarian will be a web application with a friendly little desktop companion that speaks to the webapp's API. When new students show up on campus every fall we will give them greedy librarian accounts that can easily and automatically and instantaneously search all of the library's holdings -- and I'm not just talking metadata here -- as well as some useful net engines and all of their soon-to-be-peers' personal holdings.

The handy desktop companion will essentially sit on their personal machines and will, after a quick authentication step, basically suck down all the data that's available from all those sources onto the desktop so that whenever the student might be disconnected, they have everything with them wherever they go.

The handy desktop companion will also check for updates whenever a connection is available.

With the handy desktop companion, the student will *always* have the ability to search, use, pore over, and naively misinterpret nearly everything at once. And even in disconnected mode, the companion will present a web frontend just like the online version so the student only has to get used to one consistent interface to the greedy librarian at all times.

Hmm, I'm getting ahead of myself. Let's spell it out a little more.

Student Foo arrives at Womb University. Foo gets a wombcard and wombid and password. Foo installs Womb Library desktop companion. Foo connects to wombnet. Desktop companion connects to the Womb Greedy Librarian and downloads the catalog.

The whole catalog.

All of it. And full-text resources (enough of this surrogation nonsense). And a high-quality web archive snapshot. And their course reserve resources. All of 'em.

I said "greedy", and I meant "greedy".

The greedy librarian webapp knows what Womb Library subscribes to and greedily sucks it all up. And thirstily checks for updates and greedily sucks all those up too. All the time. Greedily. And indexes it all all the time and provides a lovely and usable interface to it all, through which Foo tags, shares, queries, combines, mixes, mines, semantic webifies, remixes, and generally makes magic with it all at all times.

The desktop companion greedily checks the Womb U. Greedy Librarian all the time and greedily sucks up all of its recently updated materials. All the time. And indexes it all all the time and provides a lovely and usable interface to it all.

Handy, right?

Over the coming years, as Student Foo surfs the net, stumbles through courses, writes term papers, finds and loses love, tries political affiliations on for size, dips toes into untoward behavior that Mom and Pop Foo wouldn't approve of, and otherwise bombards his Foo mind with endless data, constant information, occasional knowledge, and rare-but-undeniably-time-hewn wisdom (which typically arrives sometime during Junior year, from what I can tell), the greedy librarian and desktop companion grow, shift, segment, concatenate, and resegment right along with Foo's own physiological neural network. Foo certainly is one greedy librarian.

As are all of Foo's friends. He's a social bugger, Foo is, so he has many friends, and all his friends are greedy librarians too. And when greedy librarians get together, they get greedy together, if you know what I mean. His desktop companion knows about all his friends and can talk through their desktop companions via local networks and through the central greedy librarian over the big network. It all just discovers each other like in some late 1990s Jini marketing copy from Sun, and just works, like the O'Reilly zeroconf book cover page promises.

Just like Foo and his friends, we are all already becoming greedy librarians.

It's just that there's a little issue of licensing to work out. Oh, and, data packaging and harvesting and indexing and synchronization and usability.

So we're all becoming greedy librarians, as is our software, but there are some kinks to work out along the way. Which is why the aggregation bit will be the main issue, and if we don't get on it soon, the greedy librarian in all of us is going to get very angry.

You won't like the greedy librarian when he gets angry.