ConwayLife.com

On psoup sink listed an A-life collection called "Zooland" at Caltech. The link is not working. Anyone any idea where is it now?

This now appears to live here. archive.org also still has a copy fo the old site; the last capture's from mid-2001, though.

Apple Bottom wrote:This now appears to live here. archive.org also still has a copy fo the old site; the last capture's from mid-2001, though.

This helps a lot! Thank you.

shouldsee wrote:This helps a lot! Thank you.

*tips hat* You're welcome, sugarcube!

Hi mates,

1.Is there any publication that attempted to systematically classify rulespace other than Andy Wuensche?

2.Is there a summary of current knowledge of B/S family and its extension in terms of the complexity of the behavior exhibited,(also including starwars/generation rules, life-like rules, daynnight,)?

shouldsee wrote:1.Is there any publication that attempted to systematically classify rulespace other than Andy Wuensche?

2.Is there a summary of current knowledge of B/S family and its extension in terms of the complexity of the behavior exhibited,(also including starwars/generation rules, life-like rules, daynnight,)?

Nothing very thorough, I don't think. If you haven't run into them, you might find it interesting to investigate some of David Eppstein's articles. There's a chapter by Eppstein in Andy Adamatzky's book from a few years ago, also available from arxiv.org -- Growth and Decay in Life-Like Cellular Automata. The introduction talks about the difficulty of doing any reasonable kind of systematic classification, given that pockets of complexity can turn out to be hiding in very unlikely places.

Short and incomplete summary: you can't really tell what a rule can do just by seeing what happens in random starting conditions. Very often you can build structures that make the rule behave like a completely different rule, and may support computation after all, even if the original rule looks hopelessly explosive or hopelessly quick to fade. Ideas along these lines have recently come up again on the space fabrication thread.

An extreme and probably overly simplistic example might be AntiLife, B0123478/S01234678. The rule looks like a really bad candidate for stability, because almost every cell in the universe turns on at the first time step. But beyond T=1, it turns out that any remaining OFF cells are just doing regular old Conway's Life.

dvgrn wrote:

shouldsee wrote:1.Is there any publication that attempted to systematically classify rulespace other than Andy Wuensche?

2.Is there a summary of current knowledge of B/S family and its extension in terms of the complexity of the behavior exhibited,(also including starwars/generation rules, life-like rules, daynnight,)?

Nothing very thorough, I don't think. If you haven't run into them, you might find it interesting to investigate some of David Eppstein's articles. There's a chapter by Eppstein in Andy Adamatzky's book from a few years ago, also available from arxiv.org -- Growth and Decay in Life-Like Cellular Automata. The introduction talks about the difficulty of doing any reasonable kind of systematic classification, given that pockets of complexity can turn out to be hiding in very unlikely places.

Short and incomplete summary: you can't really tell what a rule can do just by seeing what happens in random starting conditions. Very often you can build structures that make the rule behave like a completely different rule, and may support computation after all, even if the original rule looks hopelessly explosive or hopelessly quick to fade. Ideas along these lines have recently come up again on the space fabrication thread.

An extreme and probably overly simplistic example might be AntiLife, B0123478/S01234678. The rule looks like a really bad candidate for stability, because almost every cell in the universe turns on at the first time step. But beyond T=1, it turns out that any remaining OFF cells are just doing regular old Conway's Life.

Hi dvgrn,

Got it. Thanks for your reply.

I completely agree that the rule itself does not specify the complexity of the dynamics. The dynamics is also affected by the wiring, the mode of updating (probabilistic,sequential,synchronous). But I want to make the point that the complexity itself is visualisable, regardless of how and when it happens,(for example, we shall agree systems with gliders are more complex than white noise). The problem lies in that we can not use a single parameters to distinguish whether a dynamics is complex or not. And often different people use different criteria to describe complexity. Since cellular automata exhibit a range of complexity, it should be a fertile ground for some new measurements.

In short, the problem I want to solve essentially is why emergent structures emerge, and how we can detect these structures automatically. ATM emergent structures is largely a human assertion, and this limit the area we can explore (since the rulespace is way too big)

BTW, I don't think it is suitable to tell "antilife" is explosive from the ruletable. I find it hardly possible to predict the effect of a single mutation to any ruletable and essentially we can only try. Unless we run some random soups and grab some statistics, it would be way too hard for human to tell whether a rule guide complex dynamics or not.

B/S notation is useful to align the picture of CA with real cell/life phenomena. However it sometimes causes people to overlook the symmetry between state0 and state1. They are essentially just two possible states that cell can take, and sometimes one of them become dominative and establish order. The B/S notation has its limit, and the "growth" idea is largely limited to the engineered pattern and fragile when subject to randomness.

On the other hand, I guess this topic is pretty neglected because game-of-life enthusiasts have established a strong enough sub-topic of B3/S23 and its relatives. So it's not really essential to classify rule space once we've fixed the rule to explore. But still, people are looking at variant spaces other than B3/S23 these days, so I guess it would probably a good time to revive this topic.

Kind regards
Feng Geng