WEBVTT Kind: captions; Language: en

00:00:16.001 --> 00:00:22.001
This film is about an idea.

00:00:23.001 --> 00:00:30.001
An idea we call the polymorphic concept
of data processing system design,

00:00:31.000 --> 00:00:35.000
which is a complicated name for an idea,
we admit.

00:00:36.000 --> 00:00:40.000
But the idea itself isn't so complicated
when you think of it in terms of how

00:00:40.000 --> 00:00:46.000
people get machines to help solve their
problems. Now if you have a problem to be

00:00:46.000 --> 00:00:53.000
computed, you take it to a computer,
and as your problems get bigger,

00:00:53.001 --> 00:01:00.000
you need a bigger computer.
As more and more guys like you have problems to

00:01:00.000 --> 00:01:05.001
solve, they take turns on the computer,
which isn't so bad if the computer is

00:01:05.001 --> 00:01:11.001
fast enough, which it can be if it is
big enough and expensive enough. But

00:01:11.001 --> 00:01:18.000
as the number of guys and the number of
problems increase, the machine gets huge.

00:01:20.000 --> 00:01:24.001
Isn't that a waste? Because while one guy
is using the computer to solve an easy

00:01:24.001 --> 00:01:30.000
problem, which may only be taking one
thousandth of the computer, everyone else

00:01:30.000 --> 00:01:35.001
has to wait. Unless you spend still
more money and buy a still bigger

00:01:37.000 --> 00:01:44.000
computer. But worse, if one piece fails,
the whole computer will probably

00:01:44.000 --> 00:01:49.000
quit, even though most of it is perfectly good,
ready and willing to work.

00:01:50.001 --> 00:01:57.000
What's wrong here? To start with,
if you look inside, you see that it

00:01:57.000 --> 00:02:02.001
isn't just one solid hunk of computer,
but a lot of individual parts and circuits

00:02:02.001 --> 00:02:09.000
which each have specialized functions.
Now it would be nice if someone would

00:02:09.000 --> 00:02:16.000
divide the huge computer into little parts,
and solder the connections, so that

00:02:16.000 --> 00:02:20.001
everybody could get a chance to use
the parts he needs for his particular

00:02:20.001 --> 00:02:27.000
problem. Next, you will want to
organize things a little. Not in the

00:02:27.000 --> 00:02:33.000
usual way, where the brainy computer
part not only supervises everything, but

00:02:33.000 --> 00:02:39.000
also insists on doing a large share of
the detailed work. What you want is more

00:02:39.000 --> 00:02:45.001
delegation, a more easygoing cooperative
relationship between all the parts, so

00:02:45.001 --> 00:02:50.000
that they can associate freely and directly
with one another and in various ways.

00:02:52.000 --> 00:02:57.000
Isn't this a lot of fun? Actually,
of course, the electronic boxes don't move.

00:02:58.000 --> 00:03:02.001
You just make a nice, flexible automatic
switchboard connection, like the

00:03:02.001 --> 00:03:07.001
telephone company does. That way,
the computer boxes, the real brains of the

00:03:07.001 --> 00:03:12.001
system, don't have to waste their time
and talents acting like telephone

00:03:12.001 --> 00:03:17.000
operators trying to handle every single
message personally. They can stick to

00:03:17.000 --> 00:03:22.000
computing and let the other boxes do the
detailed work, such as remembering large

00:03:22.000 --> 00:03:26.001
number of facts or sending messages
from one box to another or carrying out

00:03:26.001 --> 00:03:33.000
simple logical routines. And so,
each computer box doesn't have to be so big

00:03:33.000 --> 00:03:39.000
now. This is distributed control,
distributed memory,

00:03:40.000 --> 00:03:45.001
distributed arithmetics, distributed
everything. And it makes sense, because now

00:03:45.001 --> 00:03:50.000
more than one guy can be using the system
at the same time to solve a number of

00:03:50.000 --> 00:03:56.000
easy problems. And when those occasional
crash problems come along, your system

00:03:56.000 --> 00:04:01.000
can reconnect itself in a matter of
microseconds to get the big urgent jobs done,

00:04:01.001 --> 00:04:08.001
as well as an ordinary giant
computer could do them. And

00:04:08.001 --> 00:04:15.000
if one box quits, it's no great catastrophe.
The system just automatically

00:04:15.000 --> 00:04:21.001
switches the job to another box. If you
find yourself getting the busy signal too

00:04:21.001 --> 00:04:28.000
often, you buy more of exactly the kind
of boxes you need. That's like adding on

00:04:28.000 --> 00:04:33.001
new subscribers as they move in. Or you
can add more displays so that more people

00:04:33.001 --> 00:04:39.001
can see what goes on. Suppose you have
problem calls for a lot of detailed work

00:04:39.001 --> 00:04:44.000
to do most of the time. Then you buy
boxes to give you all the memory or

00:04:44.000 --> 00:04:49.001
messenger service or whatever else you need.
On the other hand, if your problems

00:04:49.001 --> 00:04:54.001
are mostly heavy duty brain work,
you can load your system up with computer

00:04:54.001 --> 00:05:01.001
boxes. When you really need it and can
prove you need it, you can put together

00:05:01.001 --> 00:05:08.000
a big system or even a really,
really big one without over-investing early or

00:05:08.000 --> 00:05:11.001
throwing anything away or
starting from scratch again.

00:05:12.001 --> 00:05:18.001
So you save money. And all the while you
have better reliability and multiple use

00:05:18.001 --> 00:05:25.001
and better access. You can even switch
together two complexes or three

00:05:25.001 --> 00:05:32.000
or four, like a nationwide telephone system.
Or you can connect in one of those

00:05:32.000 --> 00:05:35.001
ordinary giant computers as a subscriber if

00:05:35.001 --> 00:05:41.001
you end up with one big party line.

00:05:42.001 --> 00:05:47.001
You can keep some of the lines private
by opening a few switches permanently so

00:05:47.001 --> 00:05:54.001
that certain information won't be available
to nosy subscribers. And so now you

00:05:54.001 --> 00:06:00.001
see that it wasn't just exactly a computer
you wanted, but a group of guys with

00:06:00.001 --> 00:06:06.001
brains and a group of boxes with electronic
brains and consoles and displays all

00:06:06.001 --> 00:06:11.000
connected together in varying ways and
growing together so that the partnership

00:06:11.000 --> 00:06:16.000
can solve more problems per dollar
than each group could solve by itself.

00:06:19.000 --> 00:06:24.000
And this is all there is to the data
processing concept known as polymorphic,

00:06:24.001 --> 00:06:30.001
many shaped, or multiple mode,
or the matrix concept,

00:06:32.000 --> 00:06:38.001
or maybe just the common sense
switchboard systems. Now it just so

00:06:38.001 --> 00:06:44.000
happens that Thomson-Reymoor-Woolbridge
has been working on the polymorphic idea

00:06:44.000 --> 00:06:51.000
for several years.
And in case you hadn't already guessed, we

00:06:51.000 --> 00:06:56.000
have built the first actual polymorphic system.
If you are

00:06:56.000 --> 00:06:58.001
interested, let us know.

00:07:06.000 --> 00:07:21.000
[...]