9 months and almost 21 days ago a new search engine was born. The hopes for this search engine were as Earth-shattering as the hype surrounding it - it was a Google Killer. That website was Cuil; according to the press release cuil is Gaelic for "knowledge", in actual fact cuil means "Insect" - "Eòlas" would have been the apt Gaelic word. I'm sure any French speaking Gaels would have cracked a smile at hearing where this search engine claimed it's knowledge came from. At launch its servers went down, and then it was terribly slow with results that often weren't as accurate as they could have been and of course there was the incident with an inappropriate image appearing when searching for a Grenoble based quantum computer researcher. It was more akin to a mid 90's search engine than cutting edge and it was due to mid 90's search engines that Google found it's place as king of search engines - with proper search algorithms that ranked on relevance through metadata, content, hyperlinks and popularity - it didn't get cheated by 'invisible text' either. Due to the family friendly nature of this blog I've not provided links to substantiate these claims but they are freely available through basic searches. Whilst this may sound like an attack on Cuil it is not. The people behind Cuil are some of the major developers that helped Google through it's infancy. These are knowledgeable people who perhaps could have planned ahead better and were hit with some very bad luck on their day of launch. Cuil is actually quite a respected, polished and effective search engine now - with better privacy policies over your searches, but I'm getting ahead of myself. I just thought it important to enlighten you of what happened last Summer. This brings me nicely to the title of this piece - “The annual killing of Google”. On Friday evening this year's Google Killer was launched to the public 3 days early and if I thought last year's hype on Cuil was overbearing, this year for Wolfram|Alpha it's claustrophobic. What I've written of Cuil's shortcomings come nowhere near the downfalls I've witnessed with the new 'Answer-Engine' that's blown the doors from the safe of the bank that holds the Emperor’s clothes. In short, I'm almost shocked into eternal silence. Almost. On May 1st I watched this 1 hour, 45 mintue long video which details an elaborate infomercial lecture on how useful Wolfram|Alpha is going to be to everyone from pioneering scientific researcher to an everyday Joe that wants good, clean facts. I'd heard about this project back at the start of March, and before long it had snowballed into every tech blog, feed and IT news piece going as the future of the web. I was quite excited to say the least. A compilation of curated peer-reviewed datasets that allowed for computational analysis at an academic level can only be a worthwhile, and potentially amusing, pursuit. Imagine having easy access to the monthly rainfall from a particular location since records began there and comparing it with natural death of livestock in that region. Think of the fun of correlating the demographics of accountants born in the sign of Pisces against those of manual labourers born in Aquarius. Rejoice in the ability to delve into the spending habits of people of differing religions during certain months, in certain years and then projecting what they could be in the future. Sadly the best Wolfram|Alpha was able to provide me with was the GPD of certain countries or the past few months share index of a few Blue-chip companies, granted they were with very nice graphs - but I can find these results elsewhere and with much richer related content.
Don't get me wrong it's a whiz at solving equations, just like my spreadsheet program already is.
It's got some nice information on countries, just like the CIA World Factbook does (it cites it too).
There is plenty of scientific data available, just nowhere near the wealth that is on Wikipedia.
It does nothing that sets it apart from existing avenues of knowledge gathering - other than guarantee its results are accurate.
If this was an app for the iPhone or a quirky program for my PDA it'd get the thumbs up. For being cute and trying to be important.
It's not though, it's being contented by it's founder and the mainstream media as being something above it's station.
I'm sure if those datasets which could have answered my questions had been in place for Wolfram|Alpha when I was using it I would have tried harder to outfox it - knowledge is eternal and our satisfaction at being content is not.
I'm not content with what is on offer at a fundamental level. I've played with Wolfram|Alpha a lot over the past 40 hours or so and I think the following benchmark might help illustrate my frustrations with it.
I am able to get some basic info on China, or a place in the USA called Tea from Wolfram|Alpha but that's it
Google doesn't give me an answer, but puts me on to some places that might.
Answerbag got me a nice answer, and a link.
Wired sorted me out too.
Sadly, the one time I could actually use a citation of Cecil Adams of The Straight Dope's knowledge, it's not there (although the usual wry wit and nuggets of wisdom and joy are).
Now I understand that Wired and Answerbag are fallible and their answers aren't necessarily correct. I do however have a ballpark figure and sources to base my investigation upon now. I have enough to get going. Add to this the newest features Google has to cut down results via dates and the subsets of forums, reviews, video, images, news, books, scholar, maps etc. and I'm pretty sure I can get it all worked out for myself - which is how all good research should be done.
If Wolfram|Alpha had either done it's homework or been honest with what it could be used for out-the-box I'd be a lot happier with it. But it did neither, and surrounding it are the "Yes!" men and women of today's tech media ( and IT Guru wannabes) regaling it with undue credit as is the moments fashion.
I am certain that given the right datasets this could be a very powerful tool, the one I dreamed of when I first heard about it. However if these datasets were to exist I doubt their only room to manoeuvre would be within the confines of Wolfram|Alpha's servers. I'd imagine the eternal battle for them would take place between those who live in the harrowing depths of social control and those who believe in sharing all knowledge. Just like it's always been.
What of next years Google Killer? Well no news as yet. Whether or not Google should be killed is another matter and one which will come up no doubt year after year. Cuil does have better practises with regards to users privacy. Something Google will probably need to look into as laws evolve. Wolfram|Alpha will probably get more datasets and will work out the kinks of it's launch which will make it better.
Is it this years Google Killer though?
I'm off now to have a nice cup of West Lake Dragon Well green tea. I can't have all the tea, so some of the best will make do.