Building Web applications that are robust and perform well is hard, but trying to promote them, build a market for them, and then make money from them is far harder. And harder still is when you're launching a Web service that competes with an established brand in the same market.

Could anything be harder than that? Well, yes, that's what launching a Web service that competes with Google is and exactly what Cuil attempted when they went live on July 28.

Cuil (pronounced "cool") was created by ex-Google and ex-IBM employees with a US$33 million venture capital investment. It's goal was to out-Google Google with an index that the company implied made it the biggest available (they claimed 120 billion pages, but as Google no longer publishes the size of its index it is hard to verify the claim).

As many commentators have pointed out this is a curious thing for Cuil to focus on as what matters, as we will revisit in a moment, is not the number of results but their relevancy - we want to spend less time looking for stuff and returning more results doesn't help.

Another key marketing differentiator was that Cuil stated it does not retain any personally identifiable information at all, an issue that Google has been criticized for. If they loose a point on the size issue they certainly gain one on privacy. We'll score them a total of zero so far.

Cuil's user interface is on the minimal side, offering only a simple and elegant search form on a black background. Entering a search term such as "web applications restful api" and clicking on submit, renders (at the time of writing) 1,196 results on a page that is good looking and polished. Compare this to Google's claim of "about 317,000" for the same search.

Now except for the results of very narrow search terms the total number of results is meaningless as a measure of a search engine's ability. Even so, Cuil hasn't, as far as I can determine, addressed the issue of the number of results in any written form. This would appear to be a minor but important oversight by Cuil's marketing - when you are claiming that size matters you'd better defend anything that appears to be smaller than expected. I'll give them a minus one for that.

On the plus side the Cuil user interface is nice looking and has a polished feel, but however good it looks that has no bearing on whether people find the service valuable, but I'll give Cuil one point for trying. Their total score is still zero.

So what of the quality of Cuil's results? I'd say they are dismal. The search I used above returns Wikipedia's entry on REST followed by three results for the Google Search API, the NING developer network, and so on ... essentially all technical resources. Compare that to Cuil's results: An XML article that mention REST and a Java Developer article on RESTful Web services from Sun followed six book-related sites discussing RESTful Web services. The relevancy of Cuil's results have been, in my experience, generally disappointing. Score -1 million because that's what search is all about - get it wrong and you aren't going to be a player.

Maybe Cuil can improve the relevancy of its results, but given that the time for it to show just how good it is was at their much ballyhooed launch its have lost the opportunity and probably for good.