United States. Congress. Senate. Committee on the.

The Industrial reorganization act. Hearings, Ninety-third Congress, first session [-Ninety-fourth Congress, first session], on S. 1167 (Volume pt. 7) online

. (page 66 of 140)
Online LibraryUnited States. Congress. Senate. Committee on theThe Industrial reorganization act. Hearings, Ninety-third Congress, first session [-Ninety-fourth Congress, first session], on S. 1167 (Volume pt. 7) → online text (page 66 of 140)
Font size
QR-code for this ebook


assets to portions of World Trade assets, if you contemplate some
horizontal divestiture ?

Mr. DoxAGHUE. We examined that problem and gave a lot of
thought to how it could be accomplished, and came to the conclusion
we just didn't know how to do it.



5481

There are varying reasons for tliis. In tlie first plan that we de-
veloped on the horizontal divestiture Ave knew a fair amount about
their plant stnicture and what was being built in the various plants
here in the United States.

"We didn't have that kind of information readily at hand for Western
Europe and the Far East, so it wasn't as readily apparent to us just
how you could go about this.

I'm not saying it is impossible. I'm just saying that when we exam-
ined it for the IDTO plan we just didn't know liow to do it.

Mr. O'Leary. We have received testimony with respect to the dif-
ferent functions which may or may not be included under one roof.

We earlier received testimony that the CPU and systems software
should be left under one roof, with respect to any reorganization which
is contemplated. What is your reaction to that ?

INIr. DoNAGHUE. There are certain kinds of softAvare that are so
closely tied to the computer itself — that is the basic operating soft-
ware — that the hardware and the software are usually developed along
in parallel. As a matter of fact, in many iristances of computer design
that I can recall, one has influenced the other; the design of software
has influenced the design of hardware and vice versa, and, therefore, I
think that there are certain elements of software that must stav with
the CPU.

INIr. O'Leary. With respect to your 1970 plan, did that contemplate
that each of the computer companies would be capable of providing
both systems and applications software ?

Mr. DoNAGnuE. Yes ; it did.

^Ir. O'Leary. The second relief proposal contains a mixture of di-
A'estiture and divorcement, with respect to the di\'orcement measures
suggested, how long was that divorcement intended to last ?

Mr. DoNAGHUE. Well, we didn't set an actual time limit because
it would vary according to the type of business. Now, the divorce-
ment wasn't forever and so it's not eternal divorcement from the
marketplace.

First, we mentioned that if it was done it had to be policed, and that
part of the policing action would determine the particular form of
divorcement and the timing of its termination.

Mr. O'Leary. One of the measures listed involves mandatory use
by IBM of high order languages and mandatory disclosure of prod-
ucts specifications and design technology leading to hardware and
software compatibility.

Would you exj^lain the significance of that provision ?

Mr. Donaghue. Yes. I think some of the other peo])le that have
testified might have mentioned something about this. Let's take the
software area first.

One of the most difficult problems for a customer who does have
current IBM equipment and contemplates making a change to an-
other vendor is that he is faced with the problem of his current soft-
ware and his capability of being able to take that software, which
is usually a huge investment, and transfer to some other manufac-
turer's.

The higher the level of product language that is used the easier
it is to accomplish this. Therefore, if we all happen to be using a
higher level language — Algo, Cobal, Fortran — the user has the oppor-



5482

tunity to actually compare the new hardware system that he's buying
and its performance and not have to worry so much about the fact
that the software might not run in that machine, because indeed it
should.

Mr. Gkaxfield. Mr. O'Leary, would you yield for a moment ?

I just didn't understand that. Could you please go over that again?

Mr. DoxAGiiuE. Yes. Certainly.

I'm trying to rephrase it so it will come out l)ettor.

I was speaking, primarily, first of higher level languages and stand-
ards involved witli higher level languages.

The concept of developing higher level languages in the first place,
and of evolving standards in the industry, was to enable read}- trans-
ferability of programs written in those liigher level languages from
one machine to another.

The higher the level language, following the set of standards de-
veloped by the industry, the easier it is to write a program in Fortran
and have it run on a Control Data piece of equipment today and a
Univac piece of equipment tomorrow.

The closer your language is to the assembly language of the machine,
or the code of the machine, the more difficult this transferability is.
So, what we are suggesting here is mandatory use of higher level lan-
guages will allow future users an opportunity to consider the hard-
ware he wishes to purchase rather than consider the bulk of software
written in loAver level langTiages which dictate a certain progression
in computer development, going from one model of an IBM machine
to another model, for example.

The other one was in regard to product specifications and design
technology. This is an area that I'm sure you've heard about from the
Computer Industry Association. This involves specifications in input-
output, interface specifications of all kinds; at the device level and at
the channel level.

The problem that lias been pointed out — I think Mr. Sanders also
mentioned this yesterday — is that when you design a device for the
marketplace and you see the biggest part of the marketplace belonging
to IBM, you will design a device that's compatible with IBM's. Unless
you know their design specifications, or unless they are designed to a
standard — one of the two — then you're always behind and have to
wait until their model is in the marketplace before you know what you
should design. Product specifications, if made available to competition
once they're amiounced, or once, as I've heard said, introduced to the
Government, and then released to the public, tend to help a number of
the small companies.

Mr. O'Leary. Does the second relief proposal represent a backing
away from the first, or a backing away fi-om horizontal divestiture?

Mr. DoNAGHUE. It does not preclude horizontal divestiture as sug-
gested in the first one ; but what we're concerned about, and have been
in our company, is to prevent IB]M from continuing its monopolistic
position in the segment of the business that was dying out anyway and
allowing them to move into new market areas and fully dominate them.

We're in a very volatile industry, as you well know. You've heard
presentations on how lapidly the technology has changed over the last
20 years, but segments of the business are also changing.

The marketplace seems to be constantly changing. One part is grow-
ing faster than the other and what our concern is that maybe when and



5483

if we ever achieve some type of relief for the industry, it might just be
the type of relief that I*B]M would welcome, because there are faster
o-rowinff segments of the indust ry that they'd prefer.

One that was mentioned yesterday was the telecommunications in-
dustry. Back in our 197:2 relief plan we recognized that, too. So, we may
not be backing away from the total — the horizontal — divestiture. We
know that it presents a number of problems but we also have said that,
in itself, will not suffice.

Mr. O'Leary. My last question is what do you think is necessary as a
minimum for etfective relief?

Mr. DoNAGHUE. I would have to say onr No. 2 plan, or second plan,
as a minimum.

Senator Hart. Mr. Chumbris ?

]Mr. Chumbris. AVe have no questions, Mr. Chairman.

]Mr. Gr<\xfield. I have just one.

Who is the economist that was your consultant on divestiture ?

INIr. DoNAGiiUE. Joseph Peck.

]Mr. Granfield. Thank you.

Senator Hart. Thank you ^■ery much, Mr. Donaghue.

[The following was received for the record :]

MATERIAL RELATING TO THE TESTIMONY OF HUGH P. DONAGHUE

Exhibit 1. — Prepared Statement of Mr. Donar/hue

I'KKPARED Statement of Hxtgh P. Donaghue, Assistant to the Chairman of
THE Board. Control Data Corp.. Washington, D.C.

Mr. Chairman, I am Hugh P. Donaghue, Assistant to the Chairman of the
Board of Control Data Corporation. I am pleased at the opportunity to appear be-
fore this Ctimmittee to review for you the historical development of Control
Data's anti-trust lawsuit against the International Business Machines Corpora-
tion.

Control Data Corporation is a major manufacturer of computers and related
computer peripheral equipment not for our own systems needs, but we also manu-
facture peripheral equipment for other computer manufacturers in the United
States and abroad. In 1973, our computer operations produced revenues of $948
million. Of this, $330 million was overseas business. We operate in 31 coun-
tries worldwide and employ nearly 35,000 people in our computer business.

The corporation was founded in 1957 by William C. Norris, our Chairman of
the Board. The growth of Control Data since it's formation in 1957 was the result
of careful concentration in selected areas, particularly in scientific computing,
and the avoidance of general competition with the very much larger companies in
the industry. The technological excellence of Control Data's equipment, coupled
with a careful marketing effort, enabled the company to gain an initial foothold
in the industry.

Control Data's first product, the model 1G04 computer, was a great success and
IBM was startled to see us getting orders from the prestigious organizations that
were its first users. The 1604 was announced in 1957 and it gained virtually im-
mediate acceptance. It offered more price /performance to the important scien-
tific segment of the market than any other large computer. Because that type of
customer relied mostly on his own software, it was necessary to offer only a
minimum amount of software and professional service support with the com-
puter. IBM's bundled pricing policy where hardware and software were provided
as a package, and which could be applied to all types of users, made it possible for
(~'ontrnl Data to set a lower price and yet maintain a reasonable profit on the
total 1604 system. This was accomplished by including only a minimum of soft-
ware and support since the scientific customer didn't really need or want the
extra software and support.

Tlie wide acceptance of the 1004 started Control Data on its way to becoming
a major computer company. It also started a chain of events within IBM that



5484

took quite a diffex-ent direction. IBM's reaction at first appeared confused be-
cause there wasn't any legitimate way to compete with the 1(504 without a major
change in IBM pricing policy. Ultimately they reacted violently, principally by the
announcement of new computer models that were only slightly modified versions
of their existing model 7090 computer and with prices well below the 1604. For
almost a year after the announcement of the first modified version of the 7090
in 1961, Control Data was unable to get an order. Finally we succeeded in get-
ting some more orders by lowering our price and increasing performance.

Also, we hastened to bring out a new computer, the model o600. Each time that
Control Data offered improved price/perforaiance. IBM countered by announce-
ment of further reductions in prices of slightly modified existing products or new
"paper" machines. In addition, buy-back arrangements, large educational dis-
counts, and/or excessive free analyst support was provided for those users with
less experienced staffs. Our 3600 was followed by the 6600 and again IBM
countered in the same manner and the struggle was prolonged for more than
ten years.

A more detailed picture r»f various product announcements is interesting :

1957-58. — Philco had announced the model 2000 (first installation 1957), and
Control Data the CDC 1604 (first installation January 1960). both of which
were solid state computers (the first of such to be offered in the commercial
marketplace). IBM then started to talk about a "solid state 709," which they
said would be bigger and faster than the 709 "when delivered."

i.9.5S.— CDC obtained its fourth order for a CDC 1604 from Couvair to be used
in an Air Force installation. Convair's purchasing manager telephoned an IB^I
officer in Xew York just prior to awarding that order to CDC, and a Vice
President of CDC heard him ask if IBM was going to bid a "solid state IBM
709," because the Air Force recently had decided it would only order solid state
computers for the future. IBM declared it would not bid such a machine at that
time. Couvair awarded the order to us for a CDC 1604. About one month lafer
IBM announced its IBM 7090, the long-awaited "solid state IBM 709." with per-
formance that IBM said was equal to or greater than the CDC 1604. which was
then the largest solid state computer being offered for commercial sale in tlip
world.

1!)59.— IBM installed its first model 7090 in November 1959.

J960.— CDC delivered its first model 1604 in January 1960.

1961. — IBM undercut the price of its model 7090 by announcing two modified
versions of it. called the 7040 (first installation 1962) and the 7044 (first in-
stallation 1962) to provide a wider range of performance. By the announce-
ment of what we termed "fighting machines," IB:M killed the CDC 1604 and
almost knocked us out of the big computer market and out of business (we had
a dry spell of many months where we received no new orders). CDC's only
retaliatory capability came from announcing the larger, more eflRr-ient CDP
1604A (first installation 1963), which again was hopefully to be the largest and
most powerful computer in the world.

1962. — IBM retaliated with the 7094, which was a field upgraded version of
their 7090. CDC was forced to respond with the model 3600. which was an up-
graded version of our I604A. The CDC 3600 (first installation 1963) was more
powerful than the IBM 7094.

1963. — IBM responded by announcing and starting deliveries of their model
7094-11, which was an upgraded version of tlie IBM 709J. IB:m was bound and
determined that CDC should not get ahead of IBM in producing tlie largest
computer in the world.

Control Data announced its CDC 6600 early in 1963. and the first CDC 6600
was delivered in September 1964. Meanwhile, in 1963 IBINI talked with a number
of specific large scale computer customers, such as the Atomic Energy Commis-
sion's Livermore Laboratories concerning a whole new line of machines known
by various model numbers in pre-announcement versions, said line of machines
to include very large computers at the level of CDC's 66(K) or higher.

196ff. — April 7. 1964. IBM announced its new 360 series of computers (so-called
third generation) and then announced that the new line lacked an extra large
computer. HoweA'er. by May 19(>4. IBM started making a series of proposals for
giant computers to Livermore Laboratories and others, including the following:

Model nnmher and date inirodured
IBM 360/90. May 1964 (this machine was withdrawn in August 1965—15 months

later).
IB:M 360/92, August 1964.



5485

IBM 360/94, Xovemher 1004.
IBM 360/91, January 1966.
IBM 360/91K, January 1966.
IBM 360/91L, January 1966.
IBM 360/95, August 196.".

The model IBM 360/92 was described by IB:M as providing siieed.s of 3 niilliim
executions per second, making it comparable to the CDC 6600.

The model IBM 360/91 was formally announced on January IS. 1966. and tlie
January 21, 1966, issue of the Wall Street Joiirnuh in announcing that the model
!tl series were the upper end of the System 360 family of computers, said that
these announcements effectively "scorched" rumors that IBM would drop out of
the super computer market. Certainly there could be no doubt that this was
IBM's intent.

Late in 1964, because of the IBM actions, Contr(jl Data found itself in ex-
tremely serious circumstances. It had not fully recovered from the devastating
blow to the 1604. Also the 1604A and the 3600 never gained very profitable posi-
tions because of IBM's C(jntinuing announcements. Control Data had been mak-
ing a number of proposals to the Bettis and KAPL Laboratories of the AEC! for
(>600 computers. The date of the first proi)osal was Ai»ril 1964. Shortly after the
submission of the first proposal, we were ntjtified that contracts were going to
be negotiated. Just prior to the beginning of negotiations, the rumor had it that
IBM had re-submitted their proposal and was offering a computer four times
the capability of the 6600 at a substantially lower price. Soon officials notifica-
tion was made to CDC that both laboratories had selected another manufacturer.
Six months later we learned that negotiations with IBM had not been con-
chuled, and we were invited back for further negotiations. Ultimately, Control
Data received the contracts; however, only after price concessions of over $6
million and costly concessions on performance improvement and guarantees. The
temporary loss of the Bettis and KAPL contracts and the aggressive selling of
large "paper" machines by IBM, again brought Control L^ata orders to a virtual
standstill for many months.

We don't have full information on the number of 360/90 series computers
that IB:\I produced. HoweA'er. we do know that this number was very limited
and. in fact, we believe certain models were never produced at all. Moreover,
we also know that they suffered huge losses on the development and manufacture
of these models — losses far greater than any corporation would normally l)e
willing to accept unless the purpose was to submerge competition.

By 1964 it was apparent that unless the U.S. Government would again bring
anti-trust action against IBM we would have to resort to legal action ourselves
and we began to Ituild up a file of evidence of unfair practices. Early in 1966 we
requested the Oppenheimer law firm in St. Paul to analyze the situation in the
light of the anti-trust laws and to prepare a presentation for Control Data's Board
of Directors.

On July 15, 1965, a presentation was made to the Control Data Board and
the Oppenheimer firm was authorized to begin to collect evidence from Control
Data's marketing stait on IBM's marketing practices, which, when taken to-
gether with its market share, should spell out a violation of the anti-trust laws.
The results of this investigation were to be sulunitted to the Anti-trust Division
of the U.S. Justice Department as grounds for a government investigation of
IBM which might lead to a suit by the government against IBINI. It was fnrtlier
planned that such legal efforts could serve the additional purpose of providing
the foundation for a suit by Control Data in the event the U.S. Government failed
to act.

Starting in January 1966. Control Data submitted a numlier of written memo-
randa to the Anti-trust Division and had numerous conferences in Washington
in an effort to ]»oint out the need for immediate governmental action against
IBM. I have here, and will include for the record, a copy of a memorandum
submitted to the Justice Department in January of 1966. This document de-
scribes in detail the abuses that had occurred up until that time. Tlie Justice
Department displayed considerable interest and even exhibited some enthusiasm
for government action. The Chief of the Anti-trust Division, however, told tis
in October of 1966 that the chances of the government taking action liefore
April 1, 196.'^. were very slight. This date was important to Control Data, since
it was possible that statutes of limitations would expire on certain allegations
concerning IBM's practices vis-a-vis CDC's 1604 computer.

40-927 — 75 42



5486

Oin- c'tillection of data on unfair marketing practices continued tliru 1967 and
into 19(>8. In the fall of 19G8 it was the considered judsnient of Control Data
management that the Department of Justice was not going to take any action
against IBM. Indeed, in early December 1968. our legal counsel visited the De-
partment of Justice and was informed that no legal action was contemplated.
Therefore, on December 11. 1968, Control Data tiled suit in the United States
District Court in Minnesota. I have a copy of tliat complaint for inclusion in
the record if you so desire, Mr. Chairman.

Tlie Cori>oration undertook this action with full realization of the magnitude
of the task that lay ahead. Preparation and aggressive pursuit of the lawsuit
against IBM would obviously require an enormous effort extending over many
years. This wouldn't be an ordinary lawsuit. It would be technically complex
and there would be an enormous number of documents involved. IBM's defense
would be massive and shrewd. Many of our employees, particularly those in the
upper echelon of management would be required to assist the lawyers as well as
participate as witnesses at the trial. Nevertheless, the management of Control
Data felt that the lawsuit was necessary for our survival.

The task was indeed enormous. A para-legal staff of approximately 120 peo-
ple were engaged in the discovery process, in screening l)etween 2.5-40 million
documents in various IBM files throughout the country. Of those, more than one
million documents were copied on microfilm as being relevant to our allegations.
An automated data base was established and software developed for an infor-
mation retrieval system to provide access to relevant documents. Of the one
million documents that were copied into microfilm, between 80-100,000 of
these were put into the automated data base. This also required extensive coding
key punching, verifying, etc. We employed over 10 full time lawyers on the case
and had 20 additional lawyers available on a part time basis. IBM employed
about 5 times as many.

For their part, w^e estimate that IBM reviewed over 120 million documents
of Control Data's and they copied over 6 million of these as being relevant to
either their defense or to a counterclaim that they had filed against Control
Data. Up until the time of settlement in January 1973, our lawyers had taken
over 100 depositions from individuals throughout the United States and at
the time of settlement, the bulk of this effort had not really begun.

As I stated we did settle our case with IBM in January 1973. The settlement
called for the transfer to Control Data of their Service Bureau Corporation
subsidiary, and the payment to Control Data of 101 million dollars. Out of the
101 million dollars, 15 million was for settlement of Control Data's costs in-
curred in the preparation of its lawsuit. It is impossible to estimate what the
total costs would have been if we had proceed to trial. But. when you consider
the additional effort involved in the construction of the case and the fact that
we intended to use our computer data base with direct access by terminal dur-
ing the actual trial, the costs involved would have been very large.

I would now like to address the issue of structural relief for the computer
indiTstry : given, the Government's successful conclusion of its anti-trust suit
against IBM. As the Control Data lawsuit proceeded the Corporation retained
an economist to look into the question of structural relief for the industry with
a view towards increasing competition but without substantial co.st to the econo-
my as a whole. In 1970 a document was produced by this consultant entitled.
"Achieving Effective Competition in the Computer Industry," and subtitled. "A
Plan for Divestitute." I have a copy of this document and would be pleased to
submit it for the record if you so desire. On September 1, 1970, a copy of this
document was submitted to the Department of Justice.

The divestitiire measures in this plan are quite extensive and involve the
formation of five computer systems companies, a Federal Systems Company, a
Terminal Company, a Components Company, and a Service Company engaged
in leasing, maintenance and support of out-of-production .systems. In addition,
four "vertical"' divestitures breaking off the Service Bureau Corporation, which
Control Data has since acquii-ed as part of our settlement with IBM, the Office
I'roducts Division. Science Research Associates, and World Trade Corporation.

With respect to the five compiiter companies, it was realized that this would
require some plant splitting. Specifically, the en.2rineering function and some
manufacturing lines woiild have to be transferred. In addition, the softwai'e
functions would have to be distributed. The plan does not envision giving each
company a full line of peripherals.

Forming the entities contemplated in the 1970 plan in such a fashion as to



Online LibraryUnited States. Congress. Senate. Committee on theThe Industrial reorganization act. Hearings, Ninety-third Congress, first session [-Ninety-fourth Congress, first session], on S. 1167 (Volume pt. 7) → online text (page 66 of 140)