multicellularity and findbest controls

[Botsareus]

there should be a way to measure multi-celled robots accurately with the find best control and select them appropriately and not select only one cell; The idea is to limit this by first calculating if all the parts of the multi-bot have a direct family link.

Example:

A reproduced to B
B reproduced to C
B dies
A attaches to C 'no direct family link

Example:

A reproduced to B
B reproduced to C
A attaches to C , B attaches C 'direct family link because B is still alive...

NEVER MIND:

I just realised that a robot is inserted into a new simulation as a single celled robot , not as a multibot, so the above will be pointless... but if someone can find a way arround that... like calculating the original robot before it became a multibot and selecting it... good luck!

[Numsgil]

Bots that are inserted in the species tab begin life as a single bot, but if you have a multibot and you save it as a .dbo file (Darwinbots organism), it saves the entire multibot.

[Botsareus]

The problem here is: There is no way to start a new simulation with a multi-bot... even if the system can be configured to select multi-bots. The solution I was thinking about is tracing the source robot and selecting it... but what if the source robot is dead? Then, we must keep it in memory, thats the best thing I can think of, any further suggestions?

Thank You for your time. (btw I am working of my uncles computer in Florida, I will upgrade my pc and download the latest db version and source code soon, the problem is my motherboard is 3 and 1/2 years old...

[Numsgil]

You can't start a sim with a multibot, but you can add one in using the "insert organism" control.

[EricL]

Just to nit pick, you can't actually start a new sim with a single bot either.  What you do start a new sim with is un-executed DNA encased in a default cell with a few default properties (like body and nrg).  It's a genetype, not a phenotype.  It does not constitue a single bot at cycle 0, since amoung other things, the memory array is uninitialized.

[Botsareus]

Thats nice. But the more you guys "nit pick" the more I am convinced that the find best control does not work properly.
I will try to stick to a modified virsion of my first idea (no need to memorize dead bots)

Here it is:
A reproduced to B
B reproduced to C
A attaches to C , B attaches C 'direct family link because B is still alive...

***A is selected***

Any reason why the above will not work? (Thats all I am looking for)

[Numsgil]

Well, as with all pick best schemes, you're selecting based on a bots ability to do something very specific.  So you'll end up with bots that do that very specific thing to the detriment of all other things.

[Botsareus]

Does sea-snake follow the scenario above? (I can check myself, but that will take me some time , since my computer is still in upgrading)

[Numsgil]

Find best works on single bots, IIRC, but once you find a best bot, you can save the entire sea snake with the save organism.  However, there's no guarantee that the best bot belongs to the best sea snake organism.

[Botsareus]

sorry , I was not specific enough:

lets say the following is a seasnake:

can this happen:


A reproduced to B
A reproduced to C

A is dead
B attaches to C

sea snake lives on without A

[Numsgil]

Yes, sea snake builds itself as a string of individual bots.  If a specific bot dies, the snake will either reconnect itself or just form two daughter snakes.  I think.

[Botsareus]

This will take some time... I will figure it out; Thank anyway...