Interesting Innovation - The Anti-Search.

I just came across an interesting site today, a kind of an anti-search engine, the LibraryThing Unsuggester.

If you have been to (and who hasn't?) you will see that they have a suggestion system that basically tells you that other people who bought this book, also bought these other ones - helping you find other books that you might be interested in.

Well, the Unsuggestor reverses this and allows you to type in a book you like, then tells you books you probably *won't* like. It's kind of fun to play with.

It analyzes the twelve million books LibraryThing members have recorded as owned or read, and comes back with books least likely to share a library with the book you suggest.

For example, if I type in one of my all time favorite books, "Friday" By Roberta A. Heinlein, the Unsuggestor comes back with a list of books I probably won't like, in this case, mostly Christian-related ones. Although I don't consider Friday to be directly anti-Christian, I can see how Heinlein's portrayal of future society would be at odds with traditional Christian views. Interesting.

The problem, of course, is that I have a degree in Religious Studies (combined with Anthropology) and although I'm not a Christian, I certainly am not anti-Christian. As a matter of fact, I have a large number of Christian history books in my own library (along with Buddhist, Islamic and even Wiccan books). I admit that I'm unusual in this regard, but still it shows that it's not nearly a flawless system (to be fair, it doesn't claim to be).

This highlights both a flaw and an interesting innovation, IMO. The flaw, of course, is that just because other people like (or don't like) a book, doesn't mean that their recommendations (or anti-recommendations) are going to be accurate.

But it's good information to have, since if you are starting from nowhere, and really have no clue, then you are better of taking advice from others than by choosing something blindly. I could see the possibility that a more complex system would actually combine the suggestions and un-suggestions together and apply them to searches for books for consumers.

From a research standpoint, it's even more interesting. There is a difference between learning about a subject and real research. If you are learning about a subject, then you want to have the most agreed-upon information, because at that stage you don't know enough to decide what's good information and what's bullshit.

But once you know a subject well, reading more and more of the same stuff won't help you at all. You need to move outside of what's recommended or agreed upon. You can't take your knowledge to the next level by reading what got you to your current level, almost by definition.

I'm finding this in the SEO field. If I read one more article about how title tags or links are important, I'm gonna throw up. Where is the NEW information and insights?

I think a research tool that not only shows you the recommended results, but ALSO high quality, related, but non-recommended results would be a huge benefit to research and innovation.

Let's say you did a search for a material to be used to make a widget. Let's say you have a generation 3 search engine, that uses personal and aggregate information as well as generation 1 (content analysis) and generation 2 (link/reference analysis) to provide results.

The generation 1 system chooses sites that are on-topic, the generation 2 system chooses ones that others feel are important, and then you add the generation 3 on top - which uses social information like what sites people who like this topic are likely to visit, bookmark or recommend to others.

For most searches, that would be pretty good. You'd probably find a list of the most recommended materials for that widget pretty quickly. Most searches are about learning and buying.

But for innovation and research, you want to leave the beaten path. What if you are looking for new ideas for materials? That's not going to come from the old "tried and true" collection of resources.

What I'd like to see is a research tool that showed me sites that were high quality but NOT well referenced or even known to "other people". It seems to me that the Unsuggestor tool may be a step in that direction.

Who knows? Maybe the cure to cancer or the Next Big Thing is buried in the bowels of the Googleplex somewhere in the Supplemental Results database, languishing due to the fact that it's too innovative or off the beaten path to get enough links to be considered "important".

Perhaps it's only linked to by people I have irreversible and fundamental differences of opinion with. Just because you don't agree with someone doesn't mean they don't have knowledge of information that you don't. As a matter of fact, it's much more likely they do.

Something to think about. For now, I think I'm going to read "Designing with Web Standards" by Jeffrey Zeldman, which apparently I'm going to hate because I like Robert Jordans "Wheel of Time" series.

We'll see...


No comments: