Wolfram Alpha: I can’t believe it’s not Cuil!

The problem with an intense hype campaign is that high expectations are often hard to meet in reality. When the search engine Cuil launch in 2008, it claimed among other things it was going to be a Google killer, and instead became a noun used to describe a hyped service that fails on execution.

This years Cuil will be decided at the end of the year, but the leading candidate by a long way in undoubtedly Wolfram Alpha.

The “computational knowledge engine” that promised to revolutionize the world launch FridaySaturday…actually, I’m not sure. It didn’t launch Friday night despite a live web cast that dragged on for hours that was suppose to detail the launch. It possibly launched Saturday, least it’s working now….well, if you call it that.

The idea behind Wolfram Alpha is a sound one: an engine that can deliver usable data, as opposed to search results like Google. However, the execution lacks on a couple of fronts. The main one is data, or specifically lack of data. Simply data queries come up blank, for example asking for the population of Victoria, Australia delivers the population of Hong Kong and Australia…but not Victoria. The capital of Victoria, Australia…nothing (Google: Melbourne). Victoria Australia as a query…nothing (Google: Wikipedia entry, official tourist site, pictures). Where you do get data, it’s old. For example Population of Canberra offers stats from 2004 (Google 2008). It’s not an Australian stats thing either: population of Palo Alto came back with 2004 figures as well (Google no direct answer, did link though to sites.)

Then there’s the depth problem. If statistics and numbers are where Wolfram is suppose to shine, you’d expect there to be decent time depth to numbers. There isn’t. GDP for the United States stops dead at 1970. Stock prices don’t seem to go back far at all (a search for IBM at a set date in 2006 came up blank.) Where I’d hoped this would work was in statistical analysis, but you can’t compare stats that aren’t in the system.

There are some things it does reasonably well. Straight mathematical calculations are easy, and obtaining basic information to basic questions, things like a weather forecast, star charts etc. The thing though is that although Wolfram presents some of this data in a more interesting way, many of these things can be obtained from Google. In fact, Google gives nicer graphics on a weather forecast.

Then there’s the uptime/ speed issue. As I type this at 1:30am Pacific time, I’m constantly hitting error messages on Wolfram telling me they’re overloaded. When I don’t get the message, it’s slow. Maybe not tediously slow, but a lot slower than Google. When playing with Wolfram for fun, not a big issue, but if you tried to use it regularly, I’m betting it would become painful.

It is early days yet with Wolfram Alpha, but it’s important to remember that 12 months later Cuil is only remembered for the complete let down their hype was. Wolfram so far is heading in the same direction.