: Unless some of the resources he is appropriating are extra coprocessors. I've
: always imagined Marathon's computer network as a massively parallel system
: in every way; banks of processors, memory and storage available for the
: use of any program in the network that needs them (perhaps with certain
: programs limiting other programs' access to these resources and so on). As
: Durandal grows, he would require more processing power and memory to run
: his more complex program at any decent speed, and possibly more storage to
: store all the new data he is acquiring.
: Imagine a program on our normal computers: it's a certain size on disk and
: needs such-and-such much RAM and processing power to run well. Newer
: versions usually require more of all of the above. Now imagine it can
: rewrite and update itself, getting fancier and fancier, requiring more and
: more resources with every new version of itself. Just as running a very
: resource-hungry program on your computer will slow down or otherwise
: interfere with the operation of all other programs running, so too would a
: self-growing program start to cause those same problems. And as space got
: more and more tight, the growing program would also start to suffer
: performance problems. If it could (and was intelligent self-aware of it's
: nature as a computer program), it would probably try to find access to
: better hardware so it could keep on improving itself without suffering
: from all these overhead limitations.
: That is rampancy.
: As this relates to exponential growth: I don't think anybody was talking
: about "space" as in physical size. Rather, it was meant that the
: program grows in it's system resource usage (memory, processing, storage)
: in a fashion best approximated by an exponential curve. Roughly every
: X-period-of-time it requires Y-percent more resources. What X and Y is are
: a mystery; the point is, rampant growth accelerates over time, somewhat
: slowly at first and then much faster later, "exploding" instead
: of just gradually creeping at a linear rate.
: That current programs seem to run at about the same rate on modern hardware
: as older programs run on their contemporary hardware is a pattern of
: exponential growth in our software, as hardware resources (as approximated
: by Moore's Law) follow an exponential growth pattern (roughly 100% growth
: every 18 months).
Memory and disk space are precisely what "space" refers to when you talk about rates of growth in computer science. And I did imagine that as a rampant AI consumed networks, it would also harness the processing power. I don't think that this entirely mitigates that limitation (though maybe they have the year 2400 equivalent of 1 GHz computers with 16 MB of RAM and disk space combined; can't rule that out).
And over the past two or three posts I've given a number of reasons why exponential growth probably is not what is actually going on. And ukimalefu already explained exponential growth to the board; his post was actually the reason that I jumped onto this thread.
Though if you were going for a strange sort of sarcasm in explaining it again, then I /did/ smile at the joke. :)
