Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I have always been fascinated with alife, the first one I created was something like 20+ years ago after I read an article about Tierra in a magazine.

For anyone not familiar with tierra, peek into the rabbit hole :-)

https://web.stanford.edu/class/sts129/Alife/html/Tierra.htm

http://life.ou.edu/pubs/doc/index.html

https://ai2020.org/tom-rays-tierra

So... Genesys I linked is more than just basic CA, it consists of independent agents with each "creature" using a lisp-like instruction set.



Oh man, I spent a lot of time in high school playing around with Tierra. It is a lot of fun. A while back I started to reimplement it in rust, just to learn it, but never got very far. About a year ago, give or take, I did do a reimplementation of another alife/GA style problem, https://gitlab.com/eythian/tracker - this evolves DFSAs to follow lines. I have some more ideas that I want to test out with it actually, it's time I got back into it.


My main idea behind creating the Genesys project was to create a network of worlds - I have full intentions to make the worlds connected in the future via internet and have functionalities to share worlds, creatures etc..

I just have too much more work to do and other more pleasant blockers like children :-)


In my case I'm doing it because I really enjoy thinking about representations of solutions that make evolutionary operators more effective - it was the topic of my masters thesis. This is more "pure" GA than alife though, mostly because alife was a bit too open-ended to structure a hypothesis around for a thesis, at least as far as I could figure out, so I had to compromise.

Multi-world with migration between them was my PgDipSci thesis :)

But I also really like the results that come out of alife in the broader sense, synthetic evolution, world/environment-building, and so on. It's quite a rich interesting field that's pretty easy to dabble in with some good ideas and some programming ability.

As an aside, be aware that "genesys" is already taken in the field: http://web.cs.ucla.edu/~dyer/Papers/AlifeTracker/Alife91Jeff... (this is the HTML of the paper the thing I linked above is based on.) This may cause google-collisions.


> As an aside, be aware that "genesys" is already taken in the field

Thanks for notification, I was already aware that it was used as a brand for anoher kind of service and was pondering if I should change it but this resolved that question :-) So many thanks for that.

I would be interested in reading your thesis, is it available somewhere?


http://www.kallisti.net.nz/wikifiles/RobinsStuff/ga_thesis.p... (multipopulation) and http://www.kallisti.net.nz/~robin/thesis.pdf (representation)

Warning, I wouldn't say they're great, but if you're having trouble sleeping... :)


> So... Genesys I linked is more than just basic CA, it consists of independent agents with each "creature" using a lisp-like instruction set.

Sure, I do understand that Genesys is an agent-based simulation rather than being ‘just’ a CA. Still, at heart it seems to be a CA with adjustable rules, and as such there is a similarly to my own program.

(I’ve never heard of Tierra before, but it looks interesting; I shall have to look into it!)




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: