[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Machine storytellers, was: process vs. data
> Do other people on the list disagree with me about the deficiencies in existing computer power in relation to what is required to create meaningful stories? To interact with the audience so as to create new plot lines?
Well, yes and no. As for processing power, we had plenty five years ago
to work such miracles. The problem of natural language is one of
development time; although there may be thousands of people working on
natural language parsers (I find this doubtful but suppose it's true),
surely it's not a concerted effort for more than a handful. But
cataloguing and defining every quirk of language is a monumental task.
So, step one will likely tae decades -- just creating the database and
defining the rules. After that the code-jockeys can start working at
optimization. In the end, I have no doubt that a viable, multilingual
natural language parser could run flawlessly, in real-time, with less
than a 200MHz processor. The data shuffling would benefit from a good
amount of RAM and a fast system of hard storage, but that's another
Where we are more likely to run into problems with lack of processing
power is the expectations of the user. A natural language text parser
lone, astounding as that would be, would likely elicit yawns from the
average consumer. I can hear it now: "Yeah, but if you can do all this
with text why doesn't it read it for me with voice synth?" or "Okay, so
it can understand and construct language, but that isn't very
interesting unless it' coming from an avatar in a highly detailed 3d
world that competes with the brainless shooting game I have here."
In that case, the computer can do with some evolving. The trend for
memory and processing hardware to become departmentalized for different
multimedia (as with video and sound cards) will probably continue until
hardware adjuncts handle all those ancillary tasks and the central
processor can concentrate on other things, like natural language parsing
and story-relative data processing. Not that this will make the
development tasks any less monumental.