Darwinbots Forum
Code center => Suggestions => Topic started by: Greven on August 21, 2005, 10:44:02 AM
-
Do you have any suggestion for the change of "the ref value thing"?
-
you made it possible to choose more than 1 option
-
it is completly artificial, and should be changed in the future.
but at the moment I think there are just too many other things that are more important than this one.
-
zelos what do you mean?
-
I could have voted for all 4 things at the same time
-
The way they are now doesn't bother me but I do recognize that they are in some way artificial if you think of bots as living organisms.
However if you think of them as nanites or some form of artificial, mechanical life (as I do sometimes) then it is perfectly acceptable that they each broadcast their memory for others to look at. It's just the way they comunicate.
-
Zelos: Oh I forgot that...
PY:
You have a good point it that (AGAIN), but still it seems odd, even if they are "mechanical"...;)
-
I dont think he ahev a point. this is a alife program with the purpose to mimic life as good as possible isnt it?
-
Who says life is organic, and not mechanical? We should keep all options open in DB, and not define life ourselves.
-
life cant be mechanical, cause life need to be adepteble to its surronding, and that need somekinda information that can be slighhtly changed that is bieng passed on from generation to generation. And mechanical creatures like robots are not able to do that. Robots dont fill all needs to be life.
-
yes DB is a program to mimic life, or so I think. But because we also have a "game" within DB it will be hard to code other ways for the ref value etc., without letting it be to difficult for the Game-programmers to find they optimal "killing"-bots strategy.
-
In analogy with reality, the memory locations are somekinda internal parameters - like how much enzym there is in a particular cell for a particular enzym. At least to my knowlegde there are no bacteria that can detect the components of another bacteria, and I dont think we should stick with the definition that bots in DB is like bacteria in reality. Because of the strictness of the DNA structure and all trillion commands ther exists, it is very unlikely, within reasonable time, to get any good evolution going, (I know it is a little out of topic here) but I think we need to reduce the number of commands (like a million different ref commands), and do something about the numbers, so we dont get 5000 .up store... They are no point that a simple and UN-complicated bot, get a reftieblablablabla in it genome (to strange for my taste)
-
persinaly I think we should remove as many artificial commands and replace it with biological replacements. Its more alife. But some is not possible to remove.
-
I think alot of tref variables especially can be done better. Problem is backwards compatibility.
In the end it doesn't matter too much. The point is that the bots adapt to a ruleset, not wether that ruleset has a conterpart in real life.
-
It really doesn't matter at all whether DBs are mechanical life or organic life. The point of the program is that we want to make them able to adapt to their surroundings by whatever means necessary.
What we have are bots that don't really reflect anything. They are uniquely suited for their own environment (a computer) and can be thought of as a completely novel life form. We have no real need to make them conform to anything.
If we want to make robots that totally mimic real life then we will need to start over completely and write the DNA exactly like real DNA. Not likely to happen in DB as that would be a major shift in the direction of the program.
-
if we shall begin using ATCG it will be to hard. But to mimic life as good as possible would require to program all over again. and since thats out of question we cant. So instant we go bit by bit closer to mimic life.
-
Right, and I have no problem with that as long as we don't throw away the functionality that makes DB such a great program already.
Add, modify but don't take away.
-
dont you think that making a few commands and then replace the old ones with it is modefieng? isnt it then waste to have those if they are replaced with better ones?
-
To be quite honest, I would prefer to keep the old ones too, unless we determine them to be completely useless or too unrealistic to leave in the program.
"sun" would be an example of this. Nobody ever uses it. We have better options available. But then what happens if a robot mutates a really neat trick using this supposedly obsolete function?
IMO the more different pathways to any given end that are available, the more potential diversity there is. Just makes it more fun and takes nothing away from the program.
-
are you a conservative person? a out with the new in with the old?. im a out with the old in with the new.
-
I think PY and I probably are both "keep all the old, just add a heck of new crap" philosophy. I don't know that you can descriube something like updates to a computer program in terms of conservative or liberal.
-
I have no problem with them at all as they stand. I also think there are other aspects which need work more than updating variables to more accuratlely emulate biology.
-
I think that the ref's are kind of artificial, it'd be like recognizing someone based on their individual dna. While it's possible for us, using tons of technology, I'm not sure it should be quite so easy for the bots.
I'd be more in favor of expanding on memloc/memval (multiple?) and using them to sense what another bot's value's are. This would be more akin to bacteria sensing each other's chemistry, rather than dna.
The only problem is backward's compatability, I honestly can't think of any way to pull it off :unsure:
There is a bit of a problem starting to arise with using ref's for recognition, I thought I should mention. Using delgene/mkvirus it is possible to remove/add false genetic info(eyes) changing it to whatever numbers are required. It'd be time consuming and ineffective against all bots, but it could be done.
Another possible future problem is being able to define a readback value using a * and some value. This could allow anyone to change to amount of (* eyeX) simply by changing the value of eyeX. (Although a bennefit would be better recognition for new bots with use of a more variable recognition gene)
Alright enough for now, I'd like to hear what everyone else has to say :)
-
I'd be willing to rewrite all the interbot sysvars if people are willing to accept that all the old bots will need to be rewritten.
For instance, instead of refbody we could do a refsize, since eventually body will be split into muscle and fat, and other things may change the size of a bot in the future. Refnrg would disappear. On and on.
Trefs likewise could be rewritten to use the idea of bandwidth. Depending on how large the tie you jsut shot is, you can transfer X information through the tie per cycle.