Autonomy CEO Mike Lynch has been quoted by Tom Foremski as saying that the widespread social search phenomenon is misguided, because tags don’t work. I guess that I am getting old, because I have less and less patience for such black-and-white approaches. Of course tags are not the answer. Nothing is the answer.
Search relevance is not about having one thing that works or doesn’t work. If it was, we wouldn’t call it “ranking”—we’d call it “sorting.” The secret sauce of relevance ranking algorithms rely on dozens or even hundreds of factors. Doing content analysis alone doesn’t cut it. Nor links. Nor tags. Nothing works, if you just look at that one factor.
Because I am older than dirt, I remember the debates back in the 1980s about what was the right way to find information: search or hypertext. What do you think that answer was?
Now Mike Lynch is a very smart guy and he knows this very well. He is trying to make the point that tags don’t replace all the rest of this stuff, I hope. And that if we did a better job of semantic analysis, that maybe we wouldn’t need tagging.
But I think that is too black-and-white. It’s clear that tags are helpful for non-text content, such as videos and images. Just as some tags are inaccurate by mistake and some are inaccurate intentionally (spam), the same problems occur with content and with links today. Just as some good content does not get tagged, so it is true that good content doesn’t get linked today. In other words, whatever people complain about with tagging applies to everything else in the secret sauce, too.
No, tags are not the answer. They are not magic. They are part of the secret sauce just like everything else. Perhaps someday we won’t need to do any work to get the right content to appear, but I don’t think that day is all that close. Until then, we should keep an open mind about using as many possible ingredients in the secret sauce as possible, rather than marking things as “working” or “not working.”