Aaron Stanton is poised to announce some sort of major deal for his Book Lamp system. It's a bit like Pandora for books. Apparently, he has an algorithm that can scan a book and then crunch it down into the stylistic and thematic essentials so that it can find other books of a similar kind that you are likely to enjoy.

I hate this for two reasons. Firstly, it's pretty much redundant. Bookshops already categorise books by genre and Amazon has had a 'customers who bought this also bought this' recommendation system since forever. I'm willing to bet that any differences between Stanton's algorithm and Amazon's wisdom of the crowds will be essentially irrelevant.

Familiar stodge

Much worse though is the fact that every recommendation system based on similarity just narrows your horizons. It's the reason that Terry Pratchett sells so well and it's the reason that people visit foreign countries and then eat at McDonalds. When presented with a wide choice, most of you drongos just opt for something familiar. But you only have time to read, maybe 7,000 books in your life and there are 130,000 new books published every year in the UK alone. You can't afford to creep slowly from related author to related author like this – it will take forever to discover anything new.

Naturally, I have a solution. I'm going to launch a rival system that purports to extensively analyse your reading habits, literacy level, body mass index and sexual orientation but in fact is just a java applet to generate a random ISBN number and look it up for you. The placebo effect of thinking that the book was chosen for you will probably get you through the first couple of chapters. After that, you're on your own.