10 July 2006
Usability testing has shown that people prefer indexes over search when it comes to accuracy and comprehensiveness, and yet the same tests show that people prefer search as a technology. In other words, people prefer search engines even though they unanimously agree that indexes are more accurate. This is not that different from the person who insists on lifting the heavy box himself, even though someone stronger and better able has volunteered. "No, no, I'll do it myself," says the searcher.
And then he injures himself. Silly, silly person.
What is at stake, apparently, is something more psychological or emotional. Search engines may offer users a sense of power and control, or a sense of speed, that indexes don't. Further, indexes seem so much more complicated when you glance at them -- words, words everywhere -- and in comparison search is so simple: an empty box. Just type a word and bingo! If you were to stop and look at this behavior you'd realize that there's something subsconscious going on; rationality is losing to some deeper sense of emotionality and self. Search simply feels right in a way that using an index does not, at least not instinctively.
Some indexers take this news with a strong sense of pessimism, seeing this "shift toward the emotional" as paralleling our current lifestyle of sensationalist news and entertainment. They believe that indexes will become extinct in most practical circumstances, because search engines are psychologically preferred -- not to mention faster, cheaper, online, and scalable.
These pessimists aren't wrong.
However, I contend that the pessimists are also looking at the situation completely upside-down. Ask yourself what makes a search engine effective or likeable at all -- that is, what does Google have that seems to draw a majority of Web users not only to the Google.com website but also to license Google technology at their own sites -- and you'll realize that there's indexing on the back end. People don't call it "indexing," necessarily, but the intellectual, rational processes that comprise indexing are still taking place.
The difference, however, is that a search company like Google doesn't really look at the individual words and their instances. Instead, the designers of Google search (and other tools) are looking at how people respond to these words. They are looking at behavioral patterns, and using those patterns to do the indexing for them.
My brother and I used to play a game at ballparks. One of us, when it was his turn, would attempt to turn as many heads as possible without speaking. My brother would turn his head and look over his shoulder casually, then allow his eyes to lock on something imaginary but far behind all the people sitting behind us. He'd tap my on the shoulder and get me to look; I'd play along. He'd point. I'd point, and he'd correct me. Then he'd stand up. And so on. After a while, some of those people who can see us directly in front of them would be curious to know what we're looking at, and they'd turn their heads to see. This would inspire other people to turn their heads, and so on. If we'd done our job well -- it was a game of timing as well as body language -- we could get hundreds of people to look behind them, at nothing.
This kind of behavior explains the popularity of some really stupid websites. Get enough people to visit your website, and Google will acknowledge that there's something about this website worth looking at. Then more people will look at it. Internet-based fads occur weekly, from paparazzi photos to cool advertisements.
If a human were indexing this, the indexer might think, "This isn't so important that it needs to be found a million times." That human is right. But the meta-human looks at what all the humans are already doing and thinks, "There is a cultural need for this content."
For a back-of-the-book indexer to break into the world of mass search, he'll have to give up the words and instead figure out the rules -- linguistic as well as social -- behind how these words are being used. Those rules, which govern how we find things (and not what we find), don't describe the indexing we know at all.
If indexing as we indexers know it is going to survive, we'll have to find that nifty middle ground between the words and the people. It should be easy, given that we already do this, but so far we haven't managed to break into this field at all. Hopefully we'll evolve.
In the next generation, we'll index the indexes.
(Brian Pinkerton developed the first full-text retrieval search engine back in 1994. "Picture this," he explained. "A customer walks into a huge travel outfitters store, with every type of item, for vacations anywhere in the world, looks at the guy who works there, and blurts out, 'Travel.' Now where's that sales clerk supposed to begin?")