REMINISCENCES ON THE HISTORYOF

TIME SHARING

John McCarthy Stanford University

Winter or Spring

I rememb er thinking ab out timesharing ab out the time

of my rst contact with computers and b eing surprised

that this wasnt the goal of IBM and all the other man

ufacturers and users of computers This mighthave b een

around

By timesharing I meant an op erating system that p er

mits each user of a computer to b ehave as though he were

in sole control of a computer not necessarily identical with

the machine on which the op erating system is running

Christopher Strachey maywell have b een correct in saying

in his letter to Donald Knuth that the term was already

in use for timesharing among programs written to run

together This idea had already b een used in the SAGE

system I dont knowhow this kind of timesharing was

implemented in SAGE Did each program havetobesure

to return to an input p olling program or were there in

terrupts Who invented interrupts anyway I thoughtof

them but I dont b elieve I mentioned the idea to anyone

b efore I heard of them from other sources

My rst attempts to do something ab out timesharing

was in the Fall of when I came to the MIT Com

putation Center on a Sloan Foundation fellowship from

Dartmouth College It was immediately clear to me that

the timesharing the IBM would require some kind of

interrupt system Iwas very shy of prop osing hardware

mo dications esp ecially as I didnt understand electron

ics well enough to read the logic diagrams Therefore I

prop osed the minimal hardware mo dication I could think

of This involved installing a relay so that the could

b e put into trapping mo de by an external signal It was

also prop osed to connect the sense switches on the ccnsole

in parallel with relays that could b e op erated by a Flex

owriter a kind of teletyp e based on an IBM typ ewriter

When the machine wentinto trapping mo de an inter

rupt to a xed lo cation would o ccur the next time the

machine attempted to execute a jump instruction then

called a transfer The interrupt would o ccur when the

Flexowriter had set up a character in a relay buer The

interrupt program would then read the character from the

sense switches into a buer test whether the buer was

full and if not return to the interrupted program If the

buer was full the program would store the current pro

gram on the drum and read in a program to deal with the

buer

It was agreed I think I talked to Dean Arden only

to install the equipment and I b elievethat p ermission

was obtained from IBM to mo dify the computer The

connector to b e installed in the computer was obtained

However at this time we heard ab out the real time

package for the IBM This RPQ request for price

quotation was IBM jargon for a mo dication to the com

puter whose price wasnt guaranteed which rented for

p er month had b een develop ed at the request of

Bo eing for the purp ose of allowing the to accept in

formation from a wind tunnel Some element of ordinary

timesharing would havebeeninvolved but we did not

seek contact with Bo eing Anywayitwas agreed that the

real time package whichinvolved the p ossibilityofinter

rupting after any instruction would b e much b etter than

merely putting the machine in trapping mo de Therefore

we underto ok to b eg IBM for the real time package IBMs

initial reaction was favorable but nevertheless it to ok a

long time to get the real time package p erhaps a year

p erhaps two

It was then agreed that someone p erhaps Arnold Siegel

would design the hardware to connect one Flexowriter to

the computer and later an installation with three would

b e designed Siegel designed and build the equipment the

op erating system was suitably mo died I dont remem

ber by whom and demonstration of online LISP was

held for a meeting of the MIT Industrial Aliates This

demonstration which I planned and carried out had the

audience in a fourth o or lecture ro om and me in the com

puter ro om and a rented closed circuit TV system Steve

Russell who worked for me organized the practical details

including a rehearsal This demonstration was called time

stealing and was regarded as a mere prelude to prop er

timesharing It involved a xed program in the b ottom

of memory that collected characters from the Flexowriter

in a buer while an ordinary batchjobwas running It

was only after each job was run that a job that would deal

with the characters typ ed in would b e read in from the

drum This job would do what it could until more input

was wanted and would then let the op erating system go

back to the batch stream This worked for the demonstra

tion b ecause at certain hours the MIT Computation

Center op erated at certain hours a batch stream with a

time limit of one minute on anyjob

Around the time of this demonstration Herb ert Teager

came to MIT as an assistant professor of Electrical Engi

neering and expressed interest in the timesharing pro ject

Some of the ideas of timesharing overlapp ed some ideas he

had had while on his previous job but I dont remember

what they were Philip Morse the Director of the Compu

tation Center asked me if I was agreeable to turning over

the timesharing pro ject to Teager since articial intelli

gence was my main interest I agreed to this and Teager

underto ok to design the three Flexowriter system Im not

sure it was ever completed There was a prop osal for sup

p ort for timesharing submitted to NSF and money was

obtained I dont rememb er whether this preceded Tea

ger and I dont rememb er what part I had in preparing

it or whether he did it after he came This should b e an

imp ortant do cument b ecause it will contain that years

conception of and rationale for timesharing

Besides that IBM was p ersuaded to make substantial

mo dications to the IBM to b e installed at the MIT

Computation Center These included memory protection

and relo cation and an additional words of memory

for the timesharing system Teager was the main sp ecier

of these mo dications I remember my surprise when IBM

agreed to his prop osals I had supp osed that relo cation

and memory protection would greatly slow the addressing

of the computer but this turned out not to b e the case

Teagers plans for timesharing were ambitious and it

seemed to manyofusvague Therefore Corbato under

to ok an interim solution using some of the supp ort that

had b een obtained from NSF for timesharing work This

system was demonstrated some time in but it wasnt

put into regular op eration That wasnt really p ossible

until ARPA supp ort for Pro ject MAC p ermitted buying a

separate IBM

Around I b egan to consult at BBN on articial

intelligence and explained my ideas ab out timesharing

to Ed Fredkin and J C R Licklider Fredkin to my

surprise prop osed that timesharing was feasible on the

PDP computer This was DECs rst computer and

BBN had the prototyp e Fredkin designed the architecture

of an interrupt system and designed a control system for

the drum to p ermit it to b e used in a very ecientswap

ping mo de He convinced Ben Gurleythechief engineer

for DEC to build this equipment It was planned to ask

NIH for supp ort b ecause of p otential medical applications

of timesharing computers but b efore the prop osal could

even b e written Fredkin left BBN I to ok technical charge

of the pro ject as a onedayaweek consultant and Shel

don Boilen was hired to do the programming I redesigned

the memory extension system prop osed by DEC and

p ersuaded them to build the mo died system instead of

the two systems they were oering but fortunately hadnt

built I also sup ervised Boilen

Shortly after this pro ject was undertaken DEC de

cided to give a PDP to the MIT Electrical Engineer

ing Department Under the leadership of this

computer was installed in the same ro om as the TX ex

p erimental transistorized computer that had b een retired

from Lincoln Lab oratory when TX was built Dennis

and his students underto ok to make a timesharing system

for it The equipmentwas similar but they were given less

memory than the BBN pro ject had There wasnt much

collab oration

My recollection is that the BBN pro ject was nished

rst in the summer of but p erhaps Corbato remem

b ers earlier demonstrations of CTSS I left for Stanford in

the Fall of and I hadnt seen CTSS and I b elieveI

hadnt seen Denniss system op erate either BBN didnt

op erate the rst system and didnt even x the bugs They

had few computer users and were contenttocontinue the

system whereby users signed up for the whole computer

They did undertakeamuch larger followon pro ject in

volving a timeshared PDP that was installed in Mas

sachusetts General Hospital where it was not a success

The computer was inadequate there were hardware and

bugs and there was a lack of application pro

grams but mainly the pro ject was premature

At the same time that CTSS the BBN system and the

EE Department systems were b eing develop ed MIT had

started to plan for a next generation computer system

The management of MIT evidently started this as an

ordinary university planning exercise and app ointed a high

level committee consisting of Philip Morse Alb ert Hill and

Rob ert Fano to sup ervise the eort However the actual

computer scientists were p ersuaded that a revolution in

the way computers were used to timesharing was called

for The lower level committee was chaired byTeager but

after his ideas clashed with those of everyone else the

committee was reconstituted with me as chairman The

disagreementcentered around howambitious to b e and

whether to go for an interim solution Teager wanted to

be very ambitious but the rest of us thought his ideas

were vague and he wanted MIT to acquire an IBM

Stretch computer as an interim solution As it turned

out acquiring a Stretchwould have b een a go o d idea

Our second rep ort to MIT prop osed that MIT send

out a request for prop osals to computer manufacturers

On the basis of the resp onses wewould then ask the Gov

ernment for the money The RFP was written but MIT

stalled p erhaps for two reasons The rst reason was that

our initial cost estimates were very large for reasons of

conservatism Secondly IBM asked MIT to wait saying

that they would make a prop osal to meet MITs needs

at little or no cost Unfortunately the design to ok

longer than IBM management exp ected and along ab out

that time relations b etween MIT and IBM b ecame very

strained b ecause of the patentlawsuit ab out the invention

of magnetic core memory

As part of the stall President Stratton prop osed a new

study with a more thorough market survey to establish the

demand for timesharing among MIT computer users I

regarded this as analogous to trying to establish the need

for steam shovels bymarket surveys among ditch diggers

and didnt want to do it Ab out this time George Forsythe

invited me to come back to Stanford with the intention

of building a Department and I was

happy to return to California

In all this there wasnt much publication I wrote a

memo to Morse dated January prop osing that we

timeshare our exp ected transistorized IBM It has

b een suggested that the date was in error and should have

b een I dont remember now but I b elievethatif

the memo had b een written at the end of it would

have referred to the b ecause that name was bythen

current In that memo I said the idea of timesharing

wasnt esp ecially new I dont knowwhy I said that except

that I didnt want to b other to distinguish it from what

was done in the SAGE system with whichI wasnt very

familiar

Most of my argumentation for timesharing was oral

and when I complained ab out Fano and Corbato cred

iting Strachey with timesharing in their Scientic

American article Corbato was surprised to nd my

memo in the les Their correction in Scientic American

was incorrect b ecause they supp osed that Strachey and

I had develop ed the idea indep endently whereas giving

each user continuous access to the machine wasnt Stra

cheys idea at all In fact he didnt even like the idea when

he heard ab out it

Teager and I prepared a joint abstract for an ACM

meeting shortly after he arrived and I gave a lecture in

an MIT series called Management and the Computer of

the Future In this lecture I referred to Stracheys pap er

Timesharing of large fast computers given at the

IFIP Congress in Paris I had read the pap er carelessly

and supp osed he meant the same thing as I did As he sub

sequently p ointed out he meant something quite dierent

that did not involvealargenumb er of users each b ehav

ing as though he had a machine to himself As I recall

he mainly referred to xed programs some of whichwere

compute b ound and some inputoutput b ound He did

mention debugging as one of the timeshared activities

but I b elieve his concept involved one p erson debugging

while the other jobs were of the conventional sort

My memo advertised that users generally would

get the advantage of online debugging However it said

nothing ab out howmany terminals would b e required and

where they would b e lo cated I b elieve I imagined them to

be numerous and in the users oces but I cannot b e sure

Referring to an exchange suggests that I had in mind

many terminals I cannot now imagine what the eect

was on the reader of my failure to b e explicit ab out this

p oint Im afraid I was trying to minimize the dicultyof

the pro ject

The ma jor technical error of myideaswas an un

derestimation of the computer capacity required for time

sharing I still dont understand where all the computer

time go es in timesharing installations and neither do es

anyone else

Besides MITs NSF prop osal there oughttobesome

letters to IBM and p erhaps some IBM internal do cuments

ab out the prop osal since they put more than a million

dollars worth of equipmentinto it discusses

DECs taking up timesharing in Bell and Newell b o ok

but I dont recall that they discuss Ben Gurleys role Fred

kin and p erhaps Alan Kotok would know ab out that

After I came to Stanford in I organized another

PDP timesharing pro ject This was the rst timesharing

system based on display terminals It was used until

or for Supp ess work on computer aided instruction

note Then it was donated to the Indian Institute

of Technology at Kanpur where it was used for ab out

years

Appp endix

Don Knuth who was curious ab out who had done what

wrote to Christopher Strachey and got the following reply

OXFORD UNIVERSITY COMPUTING LABORATORY Banbury Road

PROGRAMMING RESEARCH GROUP Oxford OX PE

st May

Professor D E Knuth

Stanford University

Computer Science Department

Stanford California

USA

Dear Don

The paper I wrote called Time Sharing in Large Fast

Computers was read at the first pre IFIP conference at

Paris in l It was mainly about multiprogramming to

avoid waiting for peripherals although it did envisage this

going on at the same time as a programmer was debugging his program

at a console I did not envisage the sort of console system which

is now so confusingly called time sharing I still think my use

of the term is the more natural

I am afraid I am so rushed at the moment being virtually

alone in the PRG and having just moved house that I have no

time to look up any old notes I may have Ihopetobeableto

do so while settling in and if I find anything of interest I

will let you know

Dont place too much reliance on Halsburys accuracy He

tends to rely on memory and get the details wrong But he was

certainly right to say that in l time sharing as a phrase

was much in the air It was however generally used in my sense

rather than in John McCarthys sense of a CTSSlike object

Best wishes

Yours sincerely

C Strachey

Professor of Computation

University of Oxford

steamstanfordeduuftptimesharingtex b egun Dec latexed

Sep at pm