Amazon’s web services are an intriguing hotbed of innovation. As an example take Mechanical Turk a “crowdsourcing marketplace that enables computer programs to co-ordinate the use of human intelligence to perform tasks” [Source: Wikipedia].
In essence small repetitive tasks that computers either do badly, slowly or not at all, are available for people to complete in exchange for a small financial reward. The service is setup so that programmers can incorporate people-power seamlessly into existing code. Some examples include:
- · Analysing human perceptions of colour
- · Determining the relevancy between web pages and keywords
- · Paraphrasing and rewriting product descriptions
- · Collecting 12,000 drawings of sheep facing left for your thesis
Don’t give up the day job just yet though, rewards for completing a Human Intelligence Task (HIT) range from $0.02 for 5 minute tasks to $75.00 for an 8 hour brain-marathon.
Why do people complete the tasks if the rewards are minimal? You’re familiar perhaps with Dr Kawashima’s recent Brain Training success – it’s made Mr Nintendo a lot of money. Where there once were crosswords and then Sudoku now there are HITs. Be it for momentary diversion, light entertainment or simple self-satisfaction, the fact is that people are actively contributing to Amazon’s hive-mind starter-kit.
It interests me because Efficient Frontier have completed tens of thousands of repetitive SEM-related tasks (eg assessing keyword relevancy) using the service, and SkyPromote apparently have a staff of 120 ‘virtual workers’. I welcome such intelligent competition and all-out innovation in our industry, but above all I’m happy to see humanity realise that for some tasks silicon still just can’t compete with a good old human brain.
Related Links:
Katharine Mieszkowski’s article posted on Life (not) as We Know It
Dolores blog – a blog all about MTurk experiments
[Photo via bistrosavage on Flickr]
