Full text of "Interactive Computing newsletters"

See other formats

M/A '77

The following articles are reprinted solely as items of interest for the independent evaluation by members of
A TSU. The opinions, statements of fact, or conclusions expressed herein are not those of the Association.

Mini Revolution in the Computer World

Time-Sharing Loses

By STANLEY KLEIN

Among tl^e nation’s technology lead¬
ers/Bell Laboratories is highly regarded
for its expert management, and mis¬
takes in decision-making at this major
subsidiary of the American Telephone
and Telegraph Company are considered
rare.

But missteps occasionally do occur,
and William O. Baker, the Bell Labora¬
tories president, admits that one of
them involved minicomputers—-the
compact, task-oriented machines that
are suddenly taking over many of the
functbns drice performed only by big
data processing systems.

■\ # We‘€fcigi»afly. thought -tfeat' big cen¬
tral computers could handle all of our
management information work,” said
Mr. Baker. That was back in the mid-
1960’s, when most experts believed
that the big computers n^de by the
International Business Machines Cor-

Ground As Small

poratidn, the Control Data Corporation
and the Univac division of the Sperry
Rand Corporation, among others, would
provide all the data, processing power
needed by any organization.

In those years, before the potential
of the minicomputer was recognized,
the concept of big, central computers
providing service akin to an electric
utility was in vogue. Business offices,
laboratories, factories, even private
houses would have access to computers
through terminals connected to tele¬
phone lines.

Such computer time-sharing services
do exist, but now the minicomputer has
relegated that grandoise concept to a
niche in the data processing industry.
The Bet System, having caught up on
its miscalculation, now deploys thou¬
sands of minicomputers to keep track
of equipment orders, telephone net¬
work usage and many other opera¬
tions. "This is what we though we
could do using just large, central com¬

its Boom in Sales

puting systems/' Mr. Baker said.

“The future belongs to the minis and
the micros” said Eric D. Wolfe, a
senior computer scientist who heads
vhe Washington office of Bolt Beranek
and Newman, a consulting firm. In
scarcely more than a decade, mini¬
computers have turned into a $3 billion
industry, and no end to the growth is
in sight

At a cost of a few thousand dollars,
a minicomputer can be dedicated to
carrying out a single job—doing a
payroll, running a laboratory experi¬
ment, controlling a machine and so on
—in lieu of using a central computer
to handle all of the same tasks or any
combination of them.

The Digital Equipment Corporation
in Maynard, Mass., pioneered the con¬
cept of small computers in the early
1960’s. After the company introduced
the first mass-produced commercial
minicomputer in 1965, the company’s
annual report boasted: “In the eight

months between our announcement and
the end of June, more than 200 of the
computers were sold.”

Installations of that basic Digital
Equipment model, the PDP-8, have
soared to 37,000 worldwide. Moreover,
the company, which accounts for about
35 percent of the minicomputer mar¬
ket, has sold some 72,000 minis of all
types, including newer systems that
would have been considered super¬
computers if they had been introduced
10 years ago. Digital Equipment’s an¬
nual sales volume has climbed from
$15 million in 1965 to the $1 billion
that the company is expected to report
for the fiscal year ending next June 30.

Such growth has become almost
typical throughout the field of mini¬
computers and associated products—
printers, terminals, communications
gear and the like. Data General, Micro¬
data Computer Automation, Applied
Data Systems, Sycor, Datapoint, Codex,
Basic Four and scores of other rela¬
tively new companies have joined in
the boom.

Such well known concerns as Texas
Instruments, Burroughs, Honeywell,
the NCR Corporation, I.B.M., Control
Data, Sperry Rand, Hewlett-Packard
and Perkin-Eimer • are also involved.
Moreover, the consensus is that the
industry is still in its infancy. '‘For
this industry to stop growing, a dis¬
aster would have to strike the coun¬
try,” said Roland Thomas, a Data
General vice president.'

By any of a variety of measures,
Mr. Thomas’s perception seems justi¬
fied, Price-earnings ratios that measure
investor expectations are typically
higher for the shares of small com¬
puter companies than they are for
any other stock group, despite the re¬
cent big tumble in the group due to
investor concern over the fierce com¬
petition in the field. The I.B.M. Series/1
minicomputer introduction last fall is
an even more telling commentary on
ithe outlook for the small computer.
“I.B.M.’s timing could not have been

better,” exclaimed one industry source,
citing the stretched-out deliveries
from Digital Equipment, running six
months and longer on certain product
lines.

Such strong demand , results from
a continuing decline in the cost of the
technology. The PDP-8 that cost
$18,000 at the time of its introduction
12 years ago now costs $2,000. Be¬
cause of such favorable economics,
some big companies have begun to
break out part of the data processing
load that traditionally was concen¬
trated in one central computer and,
instead, are turning jobs over to mini¬
computers.

The lustiest of all the minicomputer
growth markets, however, is now the
small business application that makes
it economic for companies in the $1
million to $10 million sales range to
afford a computer on their premises to
perform payroll, billing, accounting and
other functions. A study by market
researchers Frost & Sullivan in New
York forecast that such systems, now
totaling about 100,000 installations na¬
tionwide. will soar to 460,000 installa¬
tions by 1984. Total value of the new
equipment and software to be in¬
stalled; $17 billion.

The reason for the minicomputers’
popularity is that they can do almost
anything that the programming in¬
structs them to do. Nevertheless, such
small computers do have limitations.
They cannot store as much data as the
big machines, nor can they process their
jobs as fast, and they are limited also
to the number of jobs that they can
handle at one time. According to Mr.
Wolfe of Bolt, Beranek and Newman,
“there will always be a place for the
big central processors.”

But it’s from the opposite end of the
size scale, the microcomputer, that
trouble signals can be seen. The same
semiconductor technology that makes

up much of the minicomputer’s innards
and that rendered the small computer
an effective rival to the big main-frame
computer continues to work its magic.

Engineers can now cram all of
the computer circuitry onto a silicon
chip that measures about the size of a
pinky nail. This is the so-called micro¬
computer and it could some day im¬
pinge on the growth of its bigger mini¬
computer brethren. Indeed, an annual
survey conducted by Modern Data
Sendees Inc., a research and publish¬
ing concern based in Hudson, Mass.,
already shows that about 16 percent of
potential minicomputer buyers opted in
1975 for the computer on a chip
instead.

For the moment, however, the per¬
formance of the microcomputer is too
limited to handle most of the jobs the
minicomputer is called upon to do, so
the micro devices are finding a home
in video games, electronic timepieces,
appliance controls and other applica¬
tions where, in essence, they supplant
traditional integrated-circuit electronic
technology.

Nevertheless, all of the technology
is changing so swiftly that anything
can happen. As a hedge, Digital Equip¬
ment, Data General and Texas Instru¬
ments have all devised microcomputers
to complement their minicomputer
product lines. And at this level the
mini companies will some day buck up
against such powerful semiconductor
companies as Intel, Fairchild Camera
and Instrument, National Semiconduc¬
tor. Motorola and others.

The market growth of the minicom¬
puter will expand some 12-fold over
the next 10 years, according to another
Frost & Sullivan study, but it will be
exceeded by that of the microcom¬
puter: its market growth over the same
time frame will be an astounding 150-
fold.

news in perspective

Education _

Time-Sharing in Education
Going, Going, But Not Gone

Time-Sharing Confronted with Standalone Computers As
Educational Institutions Examine One-on-One Approach

When time-sharing was aborning, it was
explained that the technique would
allow many users to share the processor
simultaneously as though each had ex¬
clusive use of the large machine. Termi¬
nals subsequently became almost as
commonplace as the telephone. But in
the 10 or so ensuing years, dramatic re¬
ductions in the cost of electronics have
brought to the marketplace computers
that are cheaper than some terminals.

What follows, then, is a natural con¬
sequence. Do away with the terminal
and modem, avoid the phone line
charges and, instead, get your own ma¬
chine. Some people call this one-on-
one.

And it’s exactly what the University
of California currently is looking into
for instructional computing. Not a com¬
puter for each student, alas, but a suf¬
ficient number of standalone machines
to accommodate students who must
write, debug, and test programs to satis¬
fy classroom assignments, as well as
those who just want to learn what a
computer can do.

Terminals out at Pasadena

At the Pasadena Polytechnic school,
a high school in southern California,
they’ve just replaced their three termi¬
nals, which were time-sharing the Cal
Tech computer across the street, with
three small systems. It’s not such a new
or startling idea, says supplier Gene
Murrow. “It predates time-sharing. Be¬
fore they had time-sharing, they had
one-on-one—one user and a $450,000
computer!” says Murrow, who is sup¬
plying three $1800 computers.

Murrow is president of fledgling
Computer Power & Light Inc., Studio
City, Calif. His microcomputer-based
Compal-80, priced at a mere $1,863,
comes equipped at that price with 12K
bytes of read-write storage, IK of
prom, a nine-inch crt and keyboard.
Extended Basic, and the facility to at¬
tach an ordinary audio cassette recorder
for auxiliary storage. The system has

been purchased by a number of schools
in the Los Angeles area.

The concept of one-on-one is as natu¬
ral to youngsters as the personal auto¬
mobile is to their parents. It’s the age
of transistorized radios for bicycles and
of supermarkets selling pocket calcula¬
tors for less than $10. The home elec-

THE $1800 Compal-80 system being used
by a number of schools for instructional
purposes is shown by Gene Murrow of
Computer Power & Light.

tronic video game, which attaches to the
tv set, found its place under many a
Christmas tree last year. Indeed, Crea¬
tive Strategies Inc. has projected sales
of such games at 17 million by 1980,
saying that prices for such games “will
fall more rapidly than most industry
members recognize.” The San Jose,
Calif., research firm foresees prices for
the low-end, limited-feature games
dropping from $35 last year to $20 this
year.

Reeducate users

Murrow, a former math teacher who
has been writing instructional programs
for years, is convinced there’s a place
for small computers, too. But he finds
he must reeducate users to the idea that
three machines, or 32 of them, are better
than an equal number of terminals shar¬
ing a larger machine. For one thing, a

32-machine facility can have one system
crash and still have 31 running. “Also,
it’s much less likely to crash because it’s
such a simple operating system,” he
says. “The one-on-one is so much
simpler than these complex time-shar¬
ing systems.”

He adds that this is especially appeal¬
ing to school systems, which are deli¬
cious targets of bright students whose
sole aim is to crash the system. Students
love getting into the system exec and
into restricted memory. They can do the
same with a Compal-80, but then
they’ve crashed only one system, which
can be reloaded in 30 seconds.

Educational consultant LeRoy Finkel

STANDALONE computer systems for less
than $5,000 each are considered ideal by
the Univ. of California’s Charles W. Ste¬
venson. He prefers them over time-sharing
terminals.

of Menlo Park, Calif., recalls the big
debate in schools not so long ago. It was
over whether students should be al¬
lowed to use handheld calculators dur¬
ing exams. One of the strong arguments
against it was that many students
couldn’t afford those $200 calculators.
“With the prices down to $25,” he says,
“that argument gets destroyed. Maybe
the concept of the future will be that
everyone will have to have a home com¬
puter.”

Continued . . .

Stanford’s approach

The problem of delivering computing
services to students is being addressed
anew this year by Stanford University,
which has installed a Digital Equipment
Corp. 2040 with 48 terminals. This is
being called lots, for low overhead
time-sharing. The system can be staffed
by only four persons, costs $440,000 to
install, and perhaps another $200,000 a
year to operate. It will be for students
and faculty only, not for those perform¬
ing sponsored research work, and it will
take an estimated 85% of such workload
off the large campus computing facility.

Stanford’s progression, from a large
mainframe supporting both sponsored
research work and students and faculty
to the installation of a smaller machine
to serve only the latter category, stands
in interesting contrast to the program
being anticipated by the University of
California.

UC’s problem with a 6400

This institution has 120,000 students
and nine campuses, all with their own
computing facilities for administrative
dp, for research, and for instructional
purposes. On the Berkeley campus, for
example, students formerly submitted
jobs in a batch mode to be run on the
campus ( DC 6400. But the first run. after
an anxious wait, would only serve to in¬
form the student that he was a terrible
keypuncher—which he already knew.
And subsequent submissions reinforced
what he also suspected, that there were
errors in his program. More recently the
campus added two Digital Equipment
pdp-1 1 /70s.

“On either system, a student can get
computing for a dollar an hour; and
those are pretty capable systems,” says
Charles W. Stevenson, manager of com¬
puter planning for the university’s Sys¬
temwide Administration. The two
11 /70s offer two different operating sys¬
tems and several languages. When the
first machine was installed, it was pre¬
dicted that terminal usage would range
between eight and ten hours a day.
“Their experience in the first six months
was an average of 15 hours of use a day,
seven days a week, on each of those
terminals,” Stevenson says. Based on
the use of 25 terminals on each system,
he says they can amortize the equip¬
ment over a two- to three-year period
at a dollar an hour. “They’re getting a
lot of good computing at a buck an
hour.”

At that price, he adds, departments
prefer the small systems. One professor
predicts that within the next six months
the campus will have three 11/70s and
one 11/34. “And he considered that to

be roughly equal to the capabilities of
the 6400” at approximately the same
price. “And yet one is far more ap¬
proachable and has a lot more interest
for instructional purposes.”

Take that a step further and you have
an even more approachable situation,
the fabled one-on-one—a standalone
machine for instructional computing.
It’s like a terminal, except that while the
student sits at the keyboard he has the
entire machine to himself. It would be
portable—cumbersome, perhaps, but
portable—and could be taken into a
classroom for use there. “All you’d need
is a three-wire outlet in the wall.”

Why a terminal?

“Why buy just a terminal,” Stevenson
asks rhetorically, “and then add insult
to injury by paying phone charges and
the cost of modems and that sort of
thing, when you can get what you really
need for instructional computing in a
standalone version for under five
grand?”

He expects soon to be issuing an rfp
for this machine. It would have a floppy
disc or cassette, graphics capability,
maybe in color, an interactive program¬
ming capability in some language such
as Basic or apl, at least as much user
space as in a multi-user time-sharing
system, full typewriter keyboard, and
some form of coursewriting software or
firmware, such as Pilot or Dialog. And
it should be able to communicate with
a host computer, should that be desired,
and to other devices, such as in labora¬
tory experiments.

“My target price is five grand for
that,” Stevenson says. “1 think it’ll come
in for less than that from what I’ve seen
so far.” In a few' years, he adds, such
a computer should be available for a
third to a half of that price. At a unit
price of $5,000. it’s easy to figure how'
many can be purchased for the cost of
installing and operating the larger
time-shared systems over, say, a tw-o- or
three-year period.

The Compal-80, for example

To show that this is not an idle dream,
he produces a brochure on the Com¬
pal-80 system, saying, “It doesn’t take
very many hours of use at a dollar an
hour to pay for the use of that.” He
figures it would be less than 18 months.
“Probably less than that if you have
them in public areas w'here they’re used
to the extent that those terminals on the
dec machines are used on the Berkeley
campus. And no phone charges.”

He has also visited with other pro¬
spective vendors, some that regretfully
can talk only in terms of a hierarchy of

machines. “If you’re talking about 200
of these standalone machines, they w ant
you to get one of their big machines.
If you w-ant 50 of them, then they want
to talk about their medium-sized ma¬
chines, and if you need one to four, then
they w ant you to get this other (smaller)
one. But I think they’re missing the
point. Because in no case are those
systems completely symmetric, and I
think it’s a mistake to go into this kind
of a thing and assume that all systems
will be in one particular kind of envi¬
ronment.”

Stevenson wants to be able to allow
professors to write their courseware,
whether it’s called computer-assisted in¬
struction or computer-managed instruc¬
tion, on the same type of machine. And
allow them to do this in the privacy of
their offices, at home, or in the same
setting where students congregate. Ven¬
dors, unfortunately, want to provide
profs with an expensive one-on-one ma¬
chine to develop their courseware, over¬
looking the fact that the programs they
develop w'on’t necessarily run on the
students’ machines, and vice versa.
There’s no symmetry there. What’s re¬
quired is complete program transferra-
bility without change.

Books vs. courses

He regrets the fact that few academic
institutions give a prof equal credit for
writing good courseware as for writing
a book or technical paper, though hope¬
ful that this will begin to happen. “Well
done courseware may involve far more
thought and careful preparation and
money than writing a book,” he says.
Whbn this is recognized, it may be pos¬
sible to store the courseware on a floppy
or cassette and sell it in campus book¬
stores with its companion workbook,
thus providing a royalty to its creator.
“People sell audio courses now; why not
also sell digital courses?” Stevenson
asks.

Nor w'ould this be restricted to
campus bookstores. What with the cur¬
rent boom in the hobbyist market and
the anticipated proliferation of home
computers, one could also sell, say, in¬
come-tax preparation programs, games,
checkbook-balancing programs, and
personal records inventory programs.

“I’m not saying these things will re¬
place computing everywhere,” Steven¬
son adds. “Not by a long shot. But it’s
for a class of instructional use that cur¬
rently is not being handled very well by
anybody—because it’s either too expen¬
sive or it’s unapproachable or you can’t
get money to pay for it.”

—Edward K. Yasakl