Threads : Musings on Crowd-Sourced Evolution

One of the things about blogs, is that as time goes by, the navigation needs change… that massive long list at the side is starting to get a bit unweildy so I need to categorise or tag etc.

Which has meant I’ve had to perform the exceedingly tedious task of tagging about 200 posts.

Out of this patterns have emerged… one of which is that I tend to get hung up on what look like “threads of innovation”. Bristlebots for example. Lego-rubik’s-cube solvers… Rubik’s Cubes. Hexapods. Guerilla-Art Knitting. I’m not sure why I’ve fixated on these, but I have.

And the temptation of course is to try to make genealogies… family-trees of innovation, which is a kindof gene-centric way of interpreting what’s going on. It’s not what happens with Machine Generations. A new generation of a machine doesn’t have two genetic parents… it (potentially) has genotypes of the entire previous generations, and quite a few more besides. A major influence on my cardboard solar collector was (for example) Vinay’s Hexayurt things – rather than the thousands of “how to make a solar collector” instructables and videos currently available on the web.

A while back there was a load of fuss (which I thought was a bit childish TBH) because some book publisher put out a book about bristlebots without citing the original “inventors”. Hell, I’ve written a couple of things about bristlebots – and I didn’t know there WERE any original inventors. I’m guessing the publishers had the same inputs as I did – a huge mass of youtube videos, with no clear paths as to who inspired what… because really, each new generation may or may not have been inspired by any number of disparate sources.

Crowd-sourced Evolution is quite a bit more powerful than biological evolution. Machine Generations are not made up of randomly combining traits from a successful previous generation – the number of inputs can be much greater, and they’re specifically selected for rather than being randomly combined. A bit like an advanced sorting-algorithm… it homes in on the optimal solution far more quicky than the more basica algorthims.