Evolution is the process of learning what works and what doesn’t based on feedback of changes in the genome. Describing that process in an algorithm or ecorithm as the author describes it will allow us to create programs that can learn.

I must say I love this book. So many great ideas.

Humans and other creatures react to the world without having to understand how everything works. You don’t have to understand the underlying theories of electro magnetism to use a computer, or the standard model in physics to get through your life successfully.

Worms can burrow through the ground without apparently any understanding for the physical laws they are subject to.

Learning is done by concrete mechanisms that can be understood by the methods of computer science.

Ecorithims are algorithms that derive their power from learning from whatever environment they inhabit.

Ecorithims can be designed using conjunction and de-junctions (simple identifying statements that are stackable) you can identify and classify things as being positive or negative for the simulated organism. Or you can identify and classify things as plants are animals depending on a rule set. You have to have some examples to start with. Hopefully these examples are randomly chosen then you need to establish your rules and then you need to be able to correct your rules when you find them being incorrect.

Humans have the code to generate 20,000 different proteins, then we have regulatory genes that control when those proteins are encoded and how often. Then there are environmental factors and epigenetics. Lots of variables.

Then there is the need for reasoning on top of that.

We don’t have the perfect algorithm yet, but we are getting closer.

Come back for part 2 when I take a closer look at some of the algorithms.