This is the subhead for the blog post
I got a terse email yesterday that said this: “As of September 17, 2007 we will no longer be supporting the Watson contextual search software. All subscription programs will be canceled as of this date.”
My bet is that 99% of you have no idea what the Watson software is, er, was. Basically, it was a desktop app that showed you alternatively search results for any page that you visited online. For example, if I went to ESPN.com, the Watson software would pop up and show a combination of sports news, other sports Web sites, and sports-related videos.
It was a cool idea, but plagued by two primary problems. First, the UI was annoying – it slowed down my computer and took up too much space on my desktop. Second, the company wanted to charge for it. I don’t remember the exact cost, but I think they wanted something like $20/month.
Hearing that second problem you might instantly conclude that the failure of this software was inevitable. After all, who pays for search results these days? But I actually think that Watson could have survived and people might have paid for it.
As I have mentioned numerous times before, I still believe that people will pay for a superior search service. And by superior, I don’t mean a slightly better algorithm, less ads, or a better UI. I mean a service that gets to know everything about you – through monitoring your search behavior, through user-inputed information, and through collaborative filtering.
A search engine, in short, that isn’t so much a search engine but more of a personal assistant (and no, I don’t mean “Jeeves”) is valuable enough that people would be willing to pay $20/month for the incredible time and frustration savings.
What Watson learned the hard way is that creating such a search engine is really hard, and if you fail to really create something revolutionary, no one will want to pay for your software. So though we all must say goodbye to Watson, I don’t think the concept of paying for search software is also riding off into the sunset.