Compare these two songs, with otherwise identical lyrics:
Do you hear a difference? One was written and performed by musicians. The other by an algorithm (Microsoft’s Songsmith). Which would you prefer to listen to?
Okay, this is an extreme example. But I think it’s a great comparison. Art, music, human expression, these things are so complex that the current state of the art hasn’t a chance in hell of even barely being able to recreate them.
There are a lot of economic benefits to automating everything via algorithms. Primarily, it’s a question of scale. With automation, you can get orders of magnitude improvements in productivity. With a tool like Songsmith, perhaps with some autotune, you could produce reams of music, an album a week? A day maybe? Charge a nice fee per album and you make a mint, with minimal labor involved. Unfortunately for this hypothetical music business, humans are a little more discerning about music then other things.
Rise of the algorithms
Automation and algorithmic models have been successfully deployed in many places. Particularly for information organisation. Possibly the most well known is search. You know, Google Search. Well, if you’re old enough, you may remember a time when links were curated. Yahoo had a directory service you could advertise your website on, organised by category. I think this approach was really nice – but obviously didn’t scale as the internet exploded. But you could rely on it to find decent websites and use the categories to quickly get an idea of the places to go for content on just about anything.
Even better than this was del.icio.us – and by far the best search experience I’ve ever had. Curated bookmarks with tagging that you could search! The search results I got on del.icio.us in it’s heydey were far better than Google’s at the time. The results got ranked by the number of times a site were tagged. I was able to find esoteric stuff really easily, stuff that would never even surface on Google.
Google Search isn’t bad. Just that I think, like Songsmith trying to create music, it is a poorer form of web categorization than what is possible. But you only really notice if you’ve used something better. News aggregation is another example of this.
The future?
It’s a bit of a downer to think that algorithmic models may be framing all information consumption. If people are used to it, they won’t realize how bad things are. Perhaps it’s up to us to notice.
James Ravn 