Author Topic: Earth as a sim  (Read 3819 times)

Offline shvarz

  • Moderator
  • Bot God
  • *****
  • Posts: 1341
    • View Profile
Earth as a sim
« on: December 11, 2006, 02:14:42 PM »
I've seen on this board some people complaining that bots don't evolve complex genes or genes that would use conditions and then argue that this indicates that the DB program is somehow failing to simulate the reality.

While it is certainly true that DBs is not perfect, the argument above is totally out of context.  There is almost no way to compare DB to real biological evolution.  The numbers just don't match.  The time, the number of organisms, the complexity levels...

One thing that people may be missing is that the simplest organism that we can find on Earth today is still the product of evolution that's been going on for the last three billion years.  Convert that to "DB time" and you'll get ~4x10^12 cycles.  And then try to estimate the number of living organisms on Earth at any given time, considering that you may have 200 million bacteria in a single ml of liquid under good growth conditions.  And each of these bacteria has genome of 100 million bases!  The numbers are totally mind-blowing!  

Of course, the DB language is designed to have much higher-order commands than regular DNA.  A single command may do something that cells would use millions of base-pairs of DNA to code.  And that evens out the numbers a bit.  But real life is still going to be tens of orders of magnitude more complex than anything we may have in DBs.  

So, if you don't see genes with conditions in your sims, it does not mean that DB language is not working the way it is supposed to work.  It's just that you need to run your sims with million-times more bots for million-times more cycles
"Never underestimate the power of stupid things in big numbers" - Serious Sam

Offline EricL

  • Administrator
  • Bot God
  • *****
  • Posts: 2266
    • View Profile
Earth as a sim
« Reply #1 on: December 11, 2006, 03:44:34 PM »
I whole heartily agree with everything you say here.  That said, it is difficult to judge at this stage whether DB actually has all the mechanisms necessary for eventually evolving something interesting/complex and we just need to run larger and longer sims (perhaps larger and longer than is possible with today's hardware and human lifetimes) or whether we are yet missing some critical mechanism without which nothing of merit will ever emerge no matter how long we wait.

As you say, it's not just the organisms themselves which are the product of billions of years of evolution but the genetic mechanisms they and evolution use and rely upon.  Everything from DNA itself to chromosones, to gene structure to the copying machinery and the mutation patterns, types and probabilities that go with it evolved to and were selected for evolvability.  The thing about DB is that a lot of this evolution machinery is hard coded in the simulator and thus not subject to selection.  We can't evolve the right mutation/copying/gene structure machinery as nature did (or perhaps we can, but have yet to try).  We have to hand code it and it has to be right or nothing of merit will ever emerge.

While it may be beyond our current patience and capabilites to evolve something truly meritious, I do think it within our capabilites to recognize when we have hit upon the necessary mechanisms and to know that something interesting can and would eventually emerge.  It's just my opinion and intuition at this stage, but I think we are still missing some of those critical mechanisms (cheif amoung them, BP sequence relative mutation probabilites exposed to selection) and I hope that once we have them, we will notice the difference in the organisms we can evolve in reasonable time frames.
Many beers....

Offline Jez

  • Bot Overlord
  • ****
  • Posts: 788
    • View Profile
Earth as a sim
« Reply #2 on: December 11, 2006, 05:29:33 PM »
I'm gonna skip your biological and mathematical statements here on the premise that they are true and I'm not worthy to judge any different.

I do like the idea that my constant whine's to people complaining about improper evolution can be put down to sim size.

I think a lot of the evo problems we run into today could be put down to, as discussed with Num's earlier, the advantages we give the bots being balanced with the correctly weighted disadvantages.

It (may be/is) worth remembering Moores Law, it may not be possible today but we are working toward a better tomorrow.

EDIT

Thanks for the info on numbers btw Shvarz
« Last Edit: December 11, 2006, 05:49:09 PM by Jez »
If you try and take a cat apart to see how it works, the first thing you have in your hands is a non-working cat.
Douglas Adams

Offline Endy

  • Bot Overlord
  • ****
  • Posts: 852
    • View Profile
Earth as a sim
« Reply #3 on: December 13, 2006, 07:56:51 PM »
Had an idea about possibly allowing the bots to have some control the types of dna that is inserted into their existing dna. I haven't quite figured out how it'd all look/work, but if the dna could select for what is added it should enable the dna language evolve itself over time.