Jeopardy's newest champion didn't have very good stories to tell after the break, but when your entire life is spent inside a metal box there's not much to tell.
I don't know if you managed to catch Watson on Jeopardy last week, but it was of kind amazing. Kind of scary, too, if you believe our destiny involves creating artificial intelligence that eventually overthrows humanity and takes over the world, but amazing nonetheless. IBM created something incredible in Watson, and the ramifications for humanity are huge - well beyond the $1 million that Watson won by defeating Ken Jennings and Brad Rutter, Jeopardy's best ever contestants, in three straight nights of quiz competition.
Watson's ability to hear speech and intuitively understand, most of the time, what a question is asking, then come up with an answer (in the form of a question) - thereby showing a unique ability to extract and interpret the right answer from a pile of data - was nothing short of incredible. This was not intuitive programming, like you'd find built into a search engine that uses algorithms to predict your search terms; this was a context-based, interactive demonstration that approaches true artificial intelligence.
Here's a rundown of how Watson works:
Watson is comprised of about 90 IBM server cores with mulitcore processors, capable of running at combined 80 teraflops. The bandwith on each chip is 500 GB, which means Watson had the ability to move a full terabyte of data in about two seconds. It was connected to the studio terminal by a 10 GB dedicated Ethernet network, and had 15 terabytes of memory and 20 terabytes of disk storage.
To answer quizmaster Alex Trebek, Watson sifted through roughly 200 million pages of content to find the correct answers.
Watson wasn't right all the time, and the computer sometimes lacked the ability to pick up on nuanced questions that employed figures of speech, or were two part questions, or were short questions without a lot of data to go by. It also couldn't hear the wrong answers given by other competitors.
Still, as a starting point, Watson shows a lot of promise for its future potential use in a variety of industries, from the mundane handling of call centre requests to interviewing patients at hospitals, to conducting business queries, to testing legal cases.
There's too much hardware to create a home version, but that will change in time and IBM's technology will be streamlined in future editions. In 20 years every home could have a Watson controller. In which case I give humanity about 21 years.
Redesign nightmares
I've never given up on Facebook, sticking with it despite the occasional redesign I don't like. But as much as I hate adjusting to the new redesign, within two weeks I usually can't even remember what the old layout used to look like. Facebook is forgiven and life goes on.
Not all redesigns fare as well. Take the example of Digg.com, the popular news and web aggregator site. The new version of the site was released in August and it damn near killed the entire website. Traffic, estimated at around eight million users a day, dropped by more than third overnight, boosting the fortunes of other aggregator sites like Reddit.com in the process. The number of users has recovered slightly since then, but overall it's still down around 25 per cent.
In recent weeks we've seen redesigns to other popular sites as well. The Gawker Network, a collection of blog sites, did a massive redesign and traffic dropped significantly as a result at Gawker.com, Gizmodo.com, Deadspint.com, Kotaku.com, Jalopnik.com, IO9.com and Jezebel.com. Visitor numbers were up slightly at Lifehacker.com - one of my favourite sites - but that was the exception.
Also new is the layout of DailyKos.com, which I'd also tend to rate as poor. A redesign may also have killed the online budgeting tool Wesabe.com, driving users over to Mint.com.
What's important to note is that these redesigns are not just an update of look and feel, but complete revisions to the way the sites actually work. People were familiar and now they're confused.
These sites were probably counting on loyalty and content to overcome any misgivings about their new layouts, but the web is vast and there is no shortage of other options; if people don't like one particular site then they can just go to another that does the same thing.
So why update? There are a lot of reasons, such as the rise of smart phones and the need to tweak content so that sites are web-friendly. Web technology is also changing with the shift to HTML 5 and CSS 3, streamlining the amount of coding required for these sites. As well, it makes practical sense for networks like Gawker to adopt a common look and feel for all their sites so they are easier to maintain and manage.
Redesigns can also be good. The Globe and Mail , both print and web, went through a great new redesign. CBC is planning something big as well in the next few months, and I think they'll do a good job. The Province and the Vancouver Sun also made changes a few years ago that improved the layout. CNN's new layout puts more information front-and-centre by not bothering as much with news categories as with the timing and interest of a story.
I guess the lesson is that a well thought out redesign can work, but change for the sake of it is not always good thing. Smaller changes over time are preferable to big changes, and less likely to scare of readers. If it ain't broken then don't fix it.
