Thats also assuming Von Neumann probes don't replicate wrong like DNA as it clones, etc. Too many variables for it too be feasible
back
2 comments
If I could attach a software program to every replicating chunk of DNA, I'm pretty sure I could write a checksum and diagnostic and instructions to shut down if they failed. That's not a variable that does anything to feasibility.
What would this "software program" be made of?
Millions of years of evolution have already granted DNA the capacity for diagnostics, error correction and self-repair. Despite that, genetic replication is still error prone, and everything still fails, sometimes catastrophically. DNA is made of molecular bonds, and that "software" is necessarily also made of molecular bonds, there is no distinction between "software" and "hardware" at that level of physical granularity.
You're not going to get perfect replication of anything over millions of generations. That violates the laws of thermodynamics. The system has to change, randomly and unexpectedly, and it has to fail over time. Entropy must accumulate. And that's not even taking into account the radioactive hellscape of interstellar space.
This is the difference between mathematics and physics. Mathematicians can handwave away or ignore inconvenient or uninteresting complexities, and thus something like the infinite exponential progress of self-replicating probes across the galaxy seems obvious because the math is obvious, but reality doesn't allow that.
> The system has to change, randomly and unexpectedly. Entropy must accumulate.
Yes to the first two, but "has to fail over time" is something you are making up--not a law of physics.
Local decreases of entropy happen continually. Resilient, error-checking, self-healing systems are possible.
Self-replicating probes aren't closed systems. They exist within the universe, consume matter (which is how they replicate) and emit waste heat (as all physical systems must,) and are bound to the laws of physics which, yes, include the second law of thermodynamics. There may be short term local reductions in entropy, but eventually, inevitably, entropy must increase.
>Resilient, error-checking, self-healing systems are possible.
You cannot have such systems be perfectly efficient. That isn't physics, it's magic.
No one suggested perfect efficiency. I'm debating with you, a product of a similar reproductive process (evolution) where a system of throwing away the bad offspring has worked just fine. Magical indeed!
In fairness, that's a bad argument because this same process also gives us cancer.
I'd instead point out that while "perfect" isn't possible, we can relatively simply design the system to have an error rate such that there's less than a 1e-12 chance of an error occurring anywhere in the universe even if you did turn the entire mass of the universe into probes.
I'd counter that with the point that people are very bad at accounting for all the possible ways that systems can fail, and that while it's easy to create an error correction code that good, the actual failure rate of the system as a whole is likely to be much, much worse.
But the reason DNA is an apt analogy is because as we see in reality cancer does exist. So a self replicating probe would in theory be following same principles of DNA, it would develop a cancerous probe, that in theory could defeat the original design plan and maybe kill all previous probes. Or say by a bacteria on a planet laying latent that causes unexpected issues, etc.
Why do you assume DNA replication is "wrong"? Maybe it's just part of the exploring. It needs to adapt to new environments to explore them after all.