I’ve started an interesting book called Kluge- The Haphazard Construction of the Human Mind by Gary Marcus. In the opening chapter, Marcus talks about how biological systems evolve with a “good enough” threshold, not always with the most elegant design imaginable, for a myriad of reasons we can talk about at a later date. What struck me was his discussion of memory. Computers, he points out, use something of a zip code type of order- everything is assigned a number or address, and can be found accordingly. The human mind, however, makes all sorts of memory errors all the time- we can’t find our keys, or our glasses resting on our heads, and according to one survey Marcus quotes, the average person spends close to 55 minutes a day looking for stuff they own but can’t locate. Part of it is because our memories work in a largely contextual fashion, and we depend on other clues to help us remember where we put all our stuff. (I don’t doubt any of this, having recently purchased a second audio recorder, having lost the first, thinking I must have lost it on a trip, only to have the thing surface in a weird spot about a week later.)
So, how does this apply to search online? I love Google, and it does a good enough job 99% of the time. If I am not finding what I want, I assume it’s a human problem, meaning that my search query was bad, and all I need to do is reformulate a better boolean search to get what I want. Having used boolean search in one form or another for almost 2 decades, first with Lexis-Nexis in law school and now daily with Google, I understand how the information I get is limited by the way I ask the question.
But in all this time, I miss what I call the “serendipity of the library” method of search. I often have found that the books in the same neighborhood as the one I thought I wanted contain better information and are more to the point than what I was initially searching for. The things further along on the shelf, or in the neighborhood (based on the good old Dewey Decimal System) sometimes are much better than what I thought I needed to know, and help me get the background knowledge I need to really solve my problem.
This is essentially because the mind works on contextual search. I am crappy at giving directions, because I get places based more on landmarks and making turns in context (making it more difficult for me to find someplace new in the dark, of course) than by “Go south for three miles and then make a left.” But I can give you all sorts of landmarks along the way to keep you from getting nervous that you missed a turn- driving and directions, for me, is totally contextual. If they start building new stuff or tear things down, I will probably get lost, so this is not particularly adaptive, but it works good enough for me, most of the time.
Google and most online search engines are like the zip code directory- they are great at taking you from Point A to Point B, and suggest similar destinations, but they don’t have the fuzzy logic of humans built in. This is why Mahalo, a human powered search engine, is such an intriguing idea- it fits humans and fuzzy logic back into the equation. It’s also why Bing may not be the answer to search, even if it gives different results than Google- it’s still boolean based, and it’s results are only going to be as good as your queries.
Google serves up ads I rarely pay attention to while typing emails, based on keyword searches around what I am writing. In some ways, these could be really intriguing ads- they are probably offering suggestions to solve problems I’m only just contemplating, without relying on the context of an exact query, but by trying to guess the gist of what I might be most interested in. I can say that after thinking about the difference between logic based and contextual based search, I’m certainly going to pay them more attention than before as a result.
I don’t know how to solve the search problem, other than to teach kids (and adults) how to better construct searches in the first place, and that over time, maybe human based search, like Mahalo, will fill this gap by introducing important “neighborhood” fuzzy logic recommendations into the mix.
I know this is incredibly geeky, but as long as humans use contextual based memory to keep things organized; as long as we look for things based on context more than specific discreet pieces of information- only search involving more of that fuzzy, quirky logic will fill the bill, and look for what we want, decode what we really meant by the request, and simply begin to understand us when what we really want to ask is “I want to find that christmas song about the Abuella and the icicles, but I don’t remember the name of it or the artist.” Not enough information for a good search string, but wow, I’d love the answer to that one.