3 ms·

> I find it fascinating that we have not witnessed self reproduction and evolution in a simulated world as of now "The evolution of self-replicating computer o

> I find it fascinating that we have not witnessed self reproduction and evolution in a simulated world as of now "The evolution of self-replicating computer organisms" (1996) http://diyhpl.us/~bryan/papers2/ai/The%20evolution%20of%20self-replicating%20computer%20organisms%20-%20A.%20N.%20Pargellis.pdf http://diyhpl.us/~bryan/papers2/ai/The%20evolution%20of%20se... 4096x4096 grid. 16 opcodes. Randomly seed the field. Observe string replicators at surprisingly high frequency. Much simpler than y'all think.

Florin_Andrei 8y ago
Right, but then you need a universe where string replicators then proceed to evolve to the point where they paint the Mona Lisa and come up with general relativity. That might be the much harder problem.
TekMol 8y ago
Hmm... They created random computer programs and say that one in 10000 turned out to be self replicating. But that is just because they gave them all the necessaray tools to do so. So the program only has to 1) Set pointer A to its beginning 2) Set pointer B to somewhere 3) Copy from pointer A to pointer B 4) Increase pointer A 5) Increase pointer B. 6) GOTO 1 If you look at the actual instruction set they use, it's even worse. They served the opcodes needed for replication on a silver plate. That does not count. You could just as well give the programs an opcode that says "copyMyself" and say that creating life is easy because many programs used that opcode.
- rwallace 8y ago
  This is true, but not unusual. Any time you ask for the simplest/earliest/minimal X, you quickly get bogged down in the gray area of what exactly counts as an X. So the above doesn't count, okay, make it a bit stricter and it takes a few orders of magnitude more computation to generate self replication. A bit stricter again, another few orders of magnitude. Then we run out of computing power, but can conjecture that by the time you get to the physics of our universe, self replication will presumably still appear at least once in, say, every 1e1000 Hubble volumes. Which is still silver plate territory compared to e.g. Conway's game of life, which has no uphill paths. So you get to pick what level you're satisfied with.
derefr 8y ago
You forgot the "evolution" part. Also, I think most people would expect that "organisms" (as opposed to the more usual A-Life term "agents") are capable of homeostasis—that random bit-flips of their internal structure, or even another replicator attempting to "eat" into them, won't break their replicating ability. That's one of the (more implicit, oft-left-unstated) bars by which we usually measure whether something is alive. (E.g. it's the one that allows us to separate viruses from dormant bacteria.)

Florin_Andrei 8y ago

Right, but then you need a universe where string replicators then proceed to evolve to the point where they paint the Mona Lisa and come up with general relativity. That might be the much harder problem.

TekMol 8y ago

Hmm... They created random computer programs and say that one in 10000 turned out to be self replicating. But that is just because they gave them all the necessaray tools to do so. So the program only has to 1) Set pointer A to its beginning 2) Set pointer B to somewhere 3) Copy from pointer A to pointer B 4) Increase pointer A 5) Increase pointer B. 6) GOTO 1 If you look at the actual instruction set they use, it's even worse. They served the opcodes needed for replication on a silver plate. That does not count. You could just as well give the programs an opcode that says "copyMyself" and say that creating life is easy because many programs used that opcode.

rwallace 8y ago
This is true, but not unusual. Any time you ask for the simplest/earliest/minimal X, you quickly get bogged down in the gray area of what exactly counts as an X. So the above doesn't count, okay, make it a bit stricter and it takes a few orders of magnitude more computation to generate self replication. A bit stricter again, another few orders of magnitude. Then we run out of computing power, but can conjecture that by the time you get to the physics of our universe, self replication will presumably still appear at least once in, say, every 1e1000 Hubble volumes. Which is still silver plate territory compared to e.g. Conway's game of life, which has no uphill paths. So you get to pick what level you're satisfied with.

derefr 8y ago

You forgot the "evolution" part. Also, I think most people would expect that "organisms" (as opposed to the more usual A-Life term "agents") are capable of homeostasis—that random bit-flips of their internal structure, or even another replicator attempting to "eat" into them, won't break their replicating ability. That's one of the (more implicit, oft-left-unstated) bars by which we usually measure whether something is alive. (E.g. it's the one that allows us to separate viruses from dormant bacteria.)