Search Isn’t Broken
Posted by Bob Warfield on May 18, 2009
Another next-generation search engine launches. It looks more differentiated than the much-ado-about-nothing known as Cuil, but that’s not saying much.
Our prediction: Wolfram Alpha (terrible name) will see a nice spike in traffic for a few days, then it will disappear unnoticed along with all the other “next-generation” search engines.
Because search isn’t broken. It can be improved, yes, and companies like Wolfram Alpha will show Google how to improve it. But no search engine we’ve seen, including this one, comes close to making the quantum leap in performance required to get real volumes of Internet users to switch.
Others says WA isn’t Google and shouldn’t be compared to Google, but this misses the important point Blodget is making, so we should all stop right here, quit worrying about WA versus Google, and take a deeper look at Blodget’s key proposition, that search isn’t “broken.”
We can put it into more classical business strategy and product marketing lines by restating his proposition more along the lines that there isn’t enough pain associated with Google that any of the alternatives can save to be worth switching.
True or false in your experience? I think I’m with Henry on this one. By and large, Google works pretty darned well. Do I wish it worked better? Yes, but it would have to work a lot better, and with all due respect, I think it’ll take some real hard core AI before that happens.
We’ve been nattering around this problem for ages, but let’s face it. No matter how much semantic slight of hand we throw at the problem, computers don’t understand human languages. They guess with some statistical precision at what we might want. Wolfram Alpha does this, Google does this, and so does every other engine out there.
It’s a messy problem to extract interesting signals from the noise. Google’s essential breakthrough was Page Rank, which was a very clever take on Mark Twain’s approach of getting someone else to paint the fence. It was a breakaway heuristic, because it was so broadly applicable. It’s also very vulnerable. Google is involved in a continuous arms race to stop the SEO’s from totally gaming the value of the algorithm away.
We could use some more breakaway heuristics while we wait for the Singularity (I’m not holding my breath!) to deliver true AI.
Meanwhile, I read Don Dodge’s post about Microsoft Live Search. Don says Live Search will surprise you in a good way, so you should try it out. He writes about how he searched for a local restaurant and was surprised at how easy it was to get great directions. I typed “Gayle’s bakery Santa Cruz” into Google and proceeded to get directions, a link to Gayle’s site, and reviews just like Don got from Live Search.
Don, it’s like Henry was saying: search isn’t broke, and even if it is, LiveSearch doesn’t fix what’s broken.