Does Google define accuracy for search?

Approaching launch time for a product you want people to listen to you. To encourage this, you promise all sorts of crazy shit; you'll say whatever it takes to get your name out there. Hype is disproportionate publicity combined with ensuing commotion. All your dashboard metrics go through the roof, page hits, visitors, sign ups, external links, twitter followers. Unfortunately none of these metrics really matter and Google Analytics doesn’t have a “quality of product” metric on the dashboard just yet.

Encouraging hype before launch is like borrowing money from the mafia. If you don’t pay up you’re fucked. If you pay on time, in full, in non-marked bills, well, you’re fucked anyway; the mob is never happy. Cuil.com borrowed shitloads from the mob, and as a quick search on Twitter will show you, it’s now paying the price.

But that’s not what I wanted to talk about here.

Cuil.com is an alternative search engine to Google. The majority of people arriving there will do one of three searches. Either their name/company name, a holiday related query or a celebrity name. They do this, because they know what results they want to see. They’re going to compare the output of Cuil.com with that of Google. Now, the problem is that if Cuil.com provides identical results to Google then everyone shrugs their shoulders and adds Cuil to the list of sites they’ll use whenever Google isn’t available i.e. Cuil is then fighting with a9.com and live.com for those 60 seconds a year that Google isn’t immediately available.

The alternative is, well, alternative. If Cuil.com provides different results to Google, people assume it’s wrong. They don’t sit back and think “Hmmm , this is interesting“ They think “WTF? Where is my homepage? Why is that really old page coming up?”

This is familiar to me, as in a previous life I worked on assessment of programming skill. I was the alternative. I developed a suite of software that generated MCQs based around programming concepts. I put a lot of work in to the concept. It worked, and it was in my opinion superior to the traditional assessment in place. However, the one question people always asked me was “How do your scores correlate with the actual class tests?”. The answer was always “Not so well”. I had all sorts of evidence and arguments as to why this was so, but no one cared. If it didn’t correlate with the model in place, it wasn’t considered good enough.

For the record I think Cuil.com is a very questionable offering at the moment. The accuracy is atrocious beyond belief. The layout of results is mental. (Usability tip: Don’t ask users to search multiple columns at once.) Overall I can’t see any competitive advantage it has over any of the search engines. I’ll give it a chance though. It deserves more than 20 hours to define itself.

Having seen the reactions to Cuil.com, I wonder will the next great search engine be ignored simply because it doesn’t tally with Googles results? Is that all we can compare it to? Is that the best we can do? We need to be open to new types of search results.

I believe Google won’t lose it’s throne over one single new search engine, but it could be because of many new search engines, each one specialised for a particular field. A search engine for music, art, guitar tablature, medicines, car parts, you name it. Googles only weakness is that it displays all results in the same manner. Cooking recipes, PDFs, books, films, mobile phones, they all get the same results page. Yet, if you were to design a search results page only for cooking recipes, would you design it exactly like Google? Or would you try to be a bit better?