a very human assumption

As I explored in in which the OPAC was wrongly provided, there are some fundamental library assumptions that simply don't hold in the digital world.

Librarians think mainly in terms of human-intermediated interactions.  Computer Science people think a lot about computer-to-computer (or machine-to-machine) direct communications using APIs.  Basically, as a CS person, I think computers should do what they're good at, and humans should be freed up to do what they are best at.

Free The Humans!

That's why it's very important not to assume things about what computers can and can't do.  CS people learn that computers are dumb.  Very dumb.  Staggeringly dim.  Anything involving any kind of natural language understanding, any kind of linguistic analysis, many types of visual recognition - computers suck at these.  Like in Danny Dunn and the Homework Machine, we learn that in order to get answers OUT of a computer, you first have to put all the answers IN, in incredible detail.  Want a computer to recognize a natural language query?  No problem.  I'll just imagine every possible query variant, and program a decision tree for every single possibility (yes, I did this for an Artificial Intelligence course, in Lisp, for a program that could answer questions about a television schedule).

So the great nemesis: GOOOOOOOOGLE.

Google does relevance-ranked keyword matches against billions of web pages in a fraction of a second.

But that's ALL it does.  Google can only answer questions if you, the intelligent human, have already parsed them into an appropriate set of keywords.  (Other than some very very simple questions that Google can parse.)  If you think choosing keywords is easy, ask an inexperienced Internet search user what keywords they might use.  I know myself that many people without such experience construct queries that I know won't give them good results.

But anyway, think about what this tells us about what Google CAN'T do.  Google can't answer many types of complex questions, or questions that use even basic natural language expressions.  That's why this profusion of AskVille type solutions.  Humans can answer questions.  So let Google do what it does well (and remember "does well" is as defined by the satisfaction of the user, not what you THINK the user should be satisfied with).

But here's the thing: if you let the computers do what they're good at, there are still tons of things for humans to do.  I would argue, more interesting and more creative things, since it's those areas where computers are weakest.  Don't make assumptions about what computers can and can't do, about what processes should be transferred unchanged into the digital world, and about what should and shouldn't be intermediated.  Instead:

1. Redesign online processes to recognize the strengths of computers and the opportunities for non-intermediated machine-to-machine interaction
2. Recognize what computers suck at, and find opportunities for human intermediation there