Corewar idea regarding a self-resetting core

[fizmo]

Hi all,

I would like to present you an idea regarding a self-resetting core, that might have the potential to add new behaviors to corewar without changing the instruction set and the game-play.

Let's assume that each cell in the core possess a counter. Whenever a cell is executed, modified or used as a pointer the counter is reset. After each process execution all counters in the core are decremented by 1. If a counter reaches zero, then the cell is converted back into its initial state "dat.f $0, $0" (which don't need a counter then).

The implication would be that the core automatically "clean" itself from unused "core coloring".

How distinctive this effect is would depend on the value for the counter (could be a pmars option -t). As smaller the value for the counter as more pronounced would be the self-cleaning effect.

It could affect end-game strategies, e.g. code lines that are normally executed only at a later stage in the battle might already resetted by the core before used.

Alternatively one could think of activating the counters of the warrior code just when it is the first time executed, modified or used as a pointer (could be a pmars option -T). This would continue to enable end-game strategies.

Paper warriors might not reach anymore the upper process limit, because the self-resetting of the core would interfere at some point the replication.

Well, that are just some first theoretical considerations.

Any opinions and impressions on that?

How difficult would it be to implement that into a pmars version? If possible it would be a nice topic for a new Courewar tournament ;-)

Fizmo

[P. Kline]

On May 13, 6:45 am, fizmo <fizmo_mas...@yahoo.com> wrote:
> Hi all,
>
> I would like to present you an idea regarding a self-resetting core, that might have the potential to add new behaviors to corewar without changing the instruction set and the game-play.

Just random thoughts...

Code evaporates if not used :-)

To make a difference in the play, the limit would be set fairly low.

You would need include reading in the list of actions that reset the
counter - basically anything that references a byte. Otherwise things
like dat-bombs and increments would disappear.

For replicators, it would be a bit similar to setting the number of
processes low, with the addition of un-coloring the core.

The Five Musketeers might make a comeback, it is designed to monitor
and rebuild itself indefinitely.

Scanners would seem to gain an advantage and decoys would be less
effective.

If reading a byte does not reset the counter you could use flags as
absolute timers, something we've never had before.

P. Kline

[fizmo]

> Just random thoughts...
>
> Code evaporates if not used :-)
>

Yes :-)

> To make a difference in the play, the limit would be set fairly low.
>
> You would need include reading in the list of actions that reset the
> counter - basically anything that references a byte. Otherwise things
> like dat-bombs and increments would disappear.
>

You're right. Reading should be included as well.

>
> Scanners would seem to gain an advantage and decoys would be less
> effective.
>

Yes, might be true.

I guess one can find a magic value range for the counter where the normal corewar behaviour is tilting into something unknown.

Fizmo

[Skybuck Flying]

Hi Fizmo,

1. If this is a request for a different visualization then it can be
implemented differently, to code in the cells does not need modification.

If such a visualization is actually handy remains in doubt, I think I tried
it in my own corewar simulator... however I might try it again some time
with the counter idea... the cells would slowly fade away for example... it
sounds more usefull then it really is... at a certain point nothing would be
visible anymore.

So for visualization purposes the idea sounds nice but in reality really
isn't it, sorry to burst your bubble a bit ;)

2. If this is a request to change gameplay then I think this is kinda lame,
it would not be fair to random injectors/spawners, trying to run some
random/left over code somewhere in the core ;)

Bye,
Skybuck.

[fizmo]

> 2. If this is a request to change gameplay then I think this is kinda lame,
> it would not be fair to random injectors/spawners, trying to run some
> random/left over code somewhere in the core ;)
>

I think it would be kinda cool that random/left over code disappears after certain times. That would reduce the entropy (disorder) in the course of the battle ;-)