Breathing Life into Scriptbots

[Julian Hershey]

Hi all!

It has been quite a while since any posts were made, and most of them have been questions from non-coders. I miss the exchange of ideas between those of us who know what is possible, and the git seems to have largely quieted down. I feel that ScriptBots had a lot of potential, as I was intent on unlocking. I altered color schemes to make it more interface friendly, I allowed users to control bots, I added ears and allowed eyes to see cells, and I tried improving the learning code of the brains. I want to soon implement sticks (once I decide how bots will be able to see them...) and I'm working on allowing various food types easily (later I hope to allow inputs, brain sizes, and ear and eye counts to be solely bot-based, allowing much greater variation and complexity, but let's take baby steps, folks). I even added the ability of bots to jump, but I've been considering removing that feature.

So I guess my point is, has anyone else been working on ScriptBots, and if so, what have you tried?

I know that I've not been very active either, but I recently became interested in this project again (as I periodically do), and was wondering where we all are. Hope to hear from some of you soon!

[Andrej]

Hey Julian,

I haven't looked into this for a while because I've become swamped with other projects that have actual deadlines :(

It's great to see various new features, but for me the more interesting part would be instead going in opposite direction, in finding the simplest sets of rules/senses/brains that still yield interesting behavior, dynamics, etc. I also think there is room for better logging, quantifying populations, and measuring the "health" of an ecosystem. For example, there should be some predator prey oscillations in populations, there shouldn't be extinction events, there should be progressive change in genomes, etc.

I'd be interested in seeing at these global ecosystem health patterns and changing between sexual/asexual reproduction, etc. And of other physical features. For example, I suspect that physical barriers are probably very important for speciation. Perhaps some parts of the map should be inaccessible, and the entire simulation should be in a bit of a maze, or something like that. I just wonder if the simulator can be less of a game and more of a tool for science, for studying evolution, and for seeing what is important for a healthy ecosystem. In a semi-scientific way.

Finally, in order to get there, there is a problem that populations can't be too large in current implementation. I would definitely want to look into SIGNIFICANTLY speeding up and expanding the world. For example, kd-trees should be used for speeding up distance calculations. Everything should be threaded and profiled carefully. It would be great to see thousands of bots among different evolved species, and having interface for visualizing all of these graphs and metrics, etc.

Anyway, just some thoughts. There is a lot of work I guess, and I'm sorry that I can't spend too much time going forward, in short term future :(
Good luck!
Andrej

--
You received this message because you are subscribed to the Google Groups "Scriptbots" group.
To unsubscribe from this group and stop receiving emails from it, send an email to scriptbots+...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.