comments about Brians definition of efficiency (Re: holistic code-generation model)

RE01 Rice Brian T. EM2
Fri, 4 Dec 1998 13:13:57 +0800

> since Brian talked about efficiency and there is no definition of it in
> the glossary (although it is used at several other points) i would like to
> comment on this.
> Defining efficiency by minimising overall time seems plausible but is not
> always the right choice. It is true if you have a symmetrical system of
> equivalent "costs" for human- and computer-time. I think this is not the
> case and you clould argue that at a much less price for computer-time
> efficiency would be defined as the minimum of time needed for the user to
> communicate his needs to the computer.
Thanks.  I thinks that the user-to-computer communication time is important,
but there is a real problem concerning _what_ the user spends time
communicating, how much information is redundant, the kind of impact that
the mode of communication has on the user's life, as well as the amount of
interactivity (i.e. continued discussion) needed between the user and the
computer for a certain process (say, incremental development).

The question, of course, is that with computer resources up for grabs these
days for simple reasoning activities, how do we pick a scheme that makes the
computer fill just the right role in a user's life, by removing redundant
reasoning activities (habits)?