Just ignore me, I'm undergoing a brainstorm of sorts; I had a flash at
work about writing a game based around players writing the sensor and
subsumptive/spreading activation network code to control a robot tank
on realistic terrain, today. It'd make use of pre-coded sensory
modules in a simulated environment and, basically, be the ultimate
geek toy.
Yes, I'm weird. It /does/ have an ObMemetics angle, though. In a
spreading activation network like I'm using, one can see the
activation of competency modules (behaviours) in terms of the conflict
of memes which can be supportive (add activation) or disruptive
(reduce/inhibit activation). Not all of these memes are built into
the network. For instance, the meme `target is within close range of
weapon one' is an environmental state, put into the network's
`perceptual field' by the code of the robotank, but `the achievement
of Goal A conflicts with Goal B' does /not/ exist within the system as
such, /even though the modules that support Goal A inhibit the modules
that support Goal B and vice versa/ (because the state-attributes that
Goal A requires are mutually exclusive with the ones that Goal B
requires their respective supporting modules inhibit their opposite
members at the module level, not the Goal level).
Interestingly, this means that spreading activation architectures can
work toward /conflicting Goals in the same network/, sometimes with
the same competency modules. All this without explicitly defining
confliction but simply defining required states and changed states of
the organism/entity. Could this be a situation that helps us
understand the idea of `meme' in a better or more complete light? It
does hint that memes needn't have other memes that refer to their
interaction in order to have it rectified; that rectification can
occur as a direct result of simpler rules.