Path: utzoo!utgpu!news-server.csri.toronto.edu!clyde.concordia.ca!uunet!decwrl!limbo!taylor
From: eugene@wilbur.nas.nasa.gov (Eugene N. Miya)
Newsgroups: comp.society
Subject: Re: The next 15 years
Message-ID: <1031@limbo.Intuitive.Com>
Date: 23 Jul 90 18:34:04 GMT
Sender: taylor@limbo.Intuitive.Com
Organization: NASA Ames Research Center, Moffett Field, CA
Lines: 66
Approved: taylor@Limbo.Intuitive.Com

> They predicted a huge increase of terminals located in homes and connected
> to mainframes, via inexpensive dial-up lines.  The idea that an individual
> might have a computer in his home - a computer as powerful as the 
> mainframes of 1970 - simply did not occur to them. But what would the
> "dedicated amateur" with his "home terminal" then do with it?

Perhaps not in the US, but consider France's Minitel system.  This can
be explained by James Martin's background in strictly IBM mainframe data
processing.  He lacked some foresight in where LSI was going and had a
data processing shop outlook on software.  There are many other
predictions, such as Alan Kay's, much more useful.

> What about our predictions today ?  What will the next 15 years be 
> like - does anyone want to make a similar set of prediction for 
> the year 2005?

The speed of machines will level off.  This will cause more problems
than people would believe.  The average man will still be happy editing
text, but we won't have "AI."  Physics will catch up as we reach certain
fundamental limitations of electronics.  Unfortunately, software
won't improve immensively in this time for several reasons.

"We will still call it Fortran|Unix|C|Pascal, etc."

Communications will be as important as the ability to store or crunch
numbers or symbols.  There will be problems with bandwidth.

What is needed in a next stage of integration.  A level of packaging
which must combine software and hardware into a higher-level module.
Modules will be important because we won't be able to get the speed or
memory we need from serial CPUs.  The modules must hide the
communications details.  It won't be perfect, but it will be
"standard."

These modules must be useable by nearly everyone, not just EEs or S/WE.
The point will be the assembly of these modules (Lego? Erector sets?)
into larger useage compositions: smart machines, smart buldings, etc.
Clearly commodity.  But again, how?  The module must be rugged, but it
must also be inexpensive.  What must the software do?  Limited
programming (but big in contrast to what's done on micros today).

So long as electronics stays "standard" only at the chip level, it won't
progress.

There will be some social problems in the physical sciences.  Too much
computing and not enough contact or understanding of the physical
world.  These problems will require new solutions in the areas of optics
and "superconductivity."

We might be building computers whose cost will be just an order
magnitude less than the most expensive scientific instruments: particle
accelerators, satellites, telescopes, etc.  Just to solve one or two
single purpose problems.  They will be important.

For every day life: there will be "micros" every where in the civilized
world, but we will laugh and cringe occasionally to find the odd 1982ish
PC or VT100 in some critical application which has not been upgraded for fear
changing will cause the system to fall apart.  More people will be dying
of "computer related problems."  But not eneough to cause alarm, just a
part of "modern life."  We will split our society even more into
information haves and have-nots.  Education and unemployment will
continue to plague us.

All of this can be elaborated in more detail.

e. nobuo miya