Author Topic: DB Biochemistry  (Read 3131 times)

Offline Elite

  • Bot Overlord
  • ****
  • Posts: 532
    • View Profile
DB Biochemistry
« on: April 28, 2006, 03:15:27 PM »
For compilation and discussion of our thoughts on DB biochemistry. This includes the chemicals involved in metabolism, the make-up of the bots themselves and other substances (including the chemicals that the envirogrid will have to deal with)

We currently have the metabolism system, that models chemicals as binary strings
eg. 01010001
With 1 signifying the presence of a chemical group and 0 signifying the lack of a chemical group

Can we do better?

How are we going to model the properties of these chemicals?

Maybe we should move towards having everything in DB dependant on a universal biochemistry ie. bot bodies, shots etc.

I also think we should give DNA a physical representation in the DBverse and allow it's movement around the sim, this could allow HGT, viruses etc.

If we have everything in DB coming from a universal biochemistry we may also solve the virus problem - there may be some chemicals that viruses can use to facilitate access to the cell.

Maybe we should have seperate 'behavioral', 'structural' (ie. for cell specialisation) and 'metabolic' DNA?

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
DB Biochemistry
« Reply #1 on: April 28, 2006, 03:24:42 PM »
The physical properties of chemicals are assigned quite randomly.  This random distrivution need not be uniform.  For instance, free nrg will probably be rather uniform except for a small handful that are "super" compact, holding maybe 10 times the energy of other chemicals.

"Emergence" is a popular buzzword, but it doesn't really offer anything to the bots that a well thought out and quite artificial system doesn't.  People just like it because things they don't expect to happen occurr.

What I'm trying to say is that emergence isn't a magic bullet guarenteed to solve all the ills in a simulation.  It just moves the design decisions away from the developers.
« Last Edit: April 28, 2006, 03:25:07 PM by Numsgil »

Offline Welwordion

  • Bot Destroyer
  • ***
  • Posts: 325
    • View Profile
DB Biochemistry
« Reply #2 on: April 28, 2006, 03:54:58 PM »
I disagree a little emergence can be useful when you want to create a system where the number of aspects a chemical etc has is not limited previously.
However this design decision as to be made early in creation, you can not just change a game like darwinbot to such asystem without inventing a complete new game.

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
DB Biochemistry
« Reply #3 on: April 28, 2006, 04:03:31 PM »
I ranted a little about emergence in another thread, I'd cross link to it but I'm sure you'll find it.

The Darwinbots Mebo system needs to be emergent because otherwise it involves us manually creating possibly hundreds of metabolic pathways, and that's not something that's worth doing by hand.

Things in DB which are more high level, shots and ties for instance, are just easier to program as high level concepts instead of trying to break them down into basic rules and having what we're looking for by emergent because these systems need to be balanced.  If -1 shots are always the best strategy, that's not a very interesting simulation.
« Last Edit: April 28, 2006, 04:04:54 PM by Numsgil »

Offline Elite

  • Bot Overlord
  • ****
  • Posts: 532
    • View Profile
DB Biochemistry
« Reply #4 on: April 28, 2006, 04:18:42 PM »
Yeah, you're right Nums

DB is a behavioral evolution sim