1
00:00:00,000 --> 00:00:10,490


2
00:00:10,490 --> 00:00:13,820
The history of time-sharing
is the subject of the lecture.

3
00:00:13,820 --> 00:00:17,780
And I'm not going to
try to address that.

4
00:00:17,780 --> 00:00:22,250
Instead, I want to address the
context of our speaker, who's

5
00:00:22,250 --> 00:00:23,960
played a key role.

6
00:00:23,960 --> 00:00:28,550
And I want to go back
to his roots a bit,

7
00:00:28,550 --> 00:00:33,230
because I want to explain why
he's in a position to give

8
00:00:33,230 --> 00:00:36,170
this distinguished lecture.

9
00:00:36,170 --> 00:00:38,780
I won't discuss his books.

10
00:00:38,780 --> 00:00:39,710
He has several.

11
00:00:39,710 --> 00:00:41,570
I won't discuss his
long list of papers.

12
00:00:41,570 --> 00:00:44,150
I won't discuss his
patents, the awards,

13
00:00:44,150 --> 00:00:50,200
or the many academies he
belongs to, has been elected to.

14
00:00:50,200 --> 00:00:53,600
He has a truly
distinguished record.

15
00:00:53,600 --> 00:00:55,790
As the program notes
briefly outline,

16
00:00:55,790 --> 00:00:59,480
Bob Fano was born
in Torino, Italy.

17
00:00:59,480 --> 00:01:02,180
He came to this country
when Europe came

18
00:01:02,180 --> 00:01:04,760
under the totalitarian grip.

19
00:01:04,760 --> 00:01:08,390
He came to the US,
and MIT in particular.

20
00:01:08,390 --> 00:01:12,180
He got his bachelor's in 1941.

21
00:01:12,180 --> 00:01:15,740
And he went on from there
as a student engineer

22
00:01:15,740 --> 00:01:18,410
at General Motors
whose specialty

23
00:01:18,410 --> 00:01:19,850
was power engineering.

24
00:01:19,850 --> 00:01:22,610
And he tells me that he was
in charge of maintaining

25
00:01:22,610 --> 00:01:24,080
the welding machines.

26
00:01:24,080 --> 00:01:27,830
And he lasted all of six months.

27
00:01:27,830 --> 00:01:30,860
And he decided that
was not for him.

28
00:01:30,860 --> 00:01:34,460
And came back to MIT to
enter graduate school--

29
00:01:34,460 --> 00:01:38,360
clearly a great decision.

30
00:01:38,360 --> 00:01:43,560
As a graduate student,
he went through a--

31
00:01:43,560 --> 00:01:46,340
had a nice career of being
a TA and then an instructor

32
00:01:46,340 --> 00:01:48,530
when that was a meaningful
rank in those days.

33
00:01:48,530 --> 00:01:49,460
It still is.

34
00:01:49,460 --> 00:01:52,250


35
00:01:52,250 --> 00:01:54,950
He became a staff member
of the radiation laboratory

36
00:01:54,950 --> 00:02:01,340
during wartime, moved into
working with microwaves,

37
00:02:01,340 --> 00:02:07,010
antennas, basically
traditional electromagnetic

38
00:02:07,010 --> 00:02:10,880
or modern electromagnetic
applications.

39
00:02:10,880 --> 00:02:16,640
He got his doctorate in
1947, just after the war.

40
00:02:16,640 --> 00:02:21,380
And he joined the faculty
as an assistant professor.

41
00:02:21,380 --> 00:02:24,920
At that point, he moves
into his second career,

42
00:02:24,920 --> 00:02:27,290
because one of the amazing
things about Bob Fano

43
00:02:27,290 --> 00:02:29,210
is he's had about--
at least four careers,

44
00:02:29,210 --> 00:02:31,370
depending on how you count.

45
00:02:31,370 --> 00:02:33,740
The first was this
power machinery.

46
00:02:33,740 --> 00:02:39,350
The second was communication
theory and information theory.

47
00:02:39,350 --> 00:02:42,610
And that was a very
fruitful period.

48
00:02:42,610 --> 00:02:45,630
It lasted for
approximately a decade.

49
00:02:45,630 --> 00:02:48,170
And he also, at that
time, became very active

50
00:02:48,170 --> 00:02:49,430
as an educator.

51
00:02:49,430 --> 00:02:51,890
And he was part of the
curriculum revolution

52
00:02:51,890 --> 00:02:58,760
that MIT led and which led to
our current notions of a core,

53
00:02:58,760 --> 00:03:01,850
and in particular, led to
a couple of books, which he

54
00:03:01,850 --> 00:03:05,360
and Dick Adler and
Lan Chu authored.

55
00:03:05,360 --> 00:03:07,940


56
00:03:07,940 --> 00:03:12,410
By the early 1960s, the time
was ripe for time-sharing.

57
00:03:12,410 --> 00:03:15,050
But it hadn't yet
happened I'm not

58
00:03:15,050 --> 00:03:18,350
going to discuss that,
because I know our speaker is

59
00:03:18,350 --> 00:03:20,090
going to talk about that.

60
00:03:20,090 --> 00:03:21,650
Suffice it to say,
it was a very--

61
00:03:21,650 --> 00:03:25,760
it was a watershed period,
a very important period.

62
00:03:25,760 --> 00:03:30,620
And Bob was right in
the thick of that.

63
00:03:30,620 --> 00:03:33,230
But he also played an
even more important role.

64
00:03:33,230 --> 00:03:36,320
And that was, he played
the role of establishing

65
00:03:36,320 --> 00:03:39,330
computer science at MIT.

66
00:03:39,330 --> 00:03:40,790
And this is not as well known.

67
00:03:40,790 --> 00:03:45,290
So let me spend a
moment to outline it.

68
00:03:45,290 --> 00:03:48,190
The situation in
the early 1960s was

69
00:03:48,190 --> 00:03:51,290
that we had all the ingredients
of a computer science

70
00:03:51,290 --> 00:03:51,920
department.

71
00:03:51,920 --> 00:03:53,270
But we didn't have the focus.

72
00:03:53,270 --> 00:03:56,060
And we certainly didn't
have the leadership.

73
00:03:56,060 --> 00:03:59,540
Most of the participants
were relatively young.

74
00:03:59,540 --> 00:04:03,500
And we were missing
the leadership.

75
00:04:03,500 --> 00:04:05,997
Bob Fano was the person that
recognized that, and stepped

76
00:04:05,997 --> 00:04:06,830
in, and supplied it.

77
00:04:06,830 --> 00:04:10,760
And the major outgrowth
of that was Project MAC,

78
00:04:10,760 --> 00:04:13,400
which has in turn evolved.

79
00:04:13,400 --> 00:04:19,790
And it now comes in two flavors,
the Laboratory for Computer

80
00:04:19,790 --> 00:04:21,778
Science and the AI Laboratory.

81
00:04:21,778 --> 00:04:24,530


82
00:04:24,530 --> 00:04:28,160
But out of that same framework,
which the time-sharing served

83
00:04:28,160 --> 00:04:34,580
as kind of a nucleating base,
came the first beginnings

84
00:04:34,580 --> 00:04:39,350
of a computer science
major and what

85
00:04:39,350 --> 00:04:41,960
has eventually
evolved to be a de

86
00:04:41,960 --> 00:04:45,870
facto department within EECS.

87
00:04:45,870 --> 00:04:48,350
The computer science
undergraduate curriculum

88
00:04:48,350 --> 00:04:50,630
was basically started
and established

89
00:04:50,630 --> 00:04:53,060
under Bob Fano's urging.

90
00:04:53,060 --> 00:04:55,850
And it now accounts
for approximately 10%

91
00:04:55,850 --> 00:04:58,250
of our majors--

92
00:04:58,250 --> 00:05:01,160
undergraduate majors, of course.

93
00:05:01,160 --> 00:05:05,000
A few years later, Bob Fano
became the first associate head

94
00:05:05,000 --> 00:05:06,800
for computer science.

95
00:05:06,800 --> 00:05:08,870
And that in turn
was a recognition

96
00:05:08,870 --> 00:05:12,710
that in fact, the subject had
become sufficiently important

97
00:05:12,710 --> 00:05:15,380
and required enough
attention, that in fact, it

98
00:05:15,380 --> 00:05:19,490
needed further management.

99
00:05:19,490 --> 00:05:23,360
And finally, let me--

100
00:05:23,360 --> 00:05:25,910
before introducing
Bob, let me just

101
00:05:25,910 --> 00:05:29,900
leave you with one brief
anecdote about names.

102
00:05:29,900 --> 00:05:35,330
We all know names are
terribly important as symbols.

103
00:05:35,330 --> 00:05:39,200
And I hope Bob will go into
the name of how Project MAC got

104
00:05:39,200 --> 00:05:40,790
to be what it is.

105
00:05:40,790 --> 00:05:42,990
But I won't mention--

106
00:05:42,990 --> 00:05:47,120
but he probably won't
mention that when Bob first

107
00:05:47,120 --> 00:05:51,170
came to this country, he
was named Roberto Fano.

108
00:05:51,170 --> 00:05:55,130
And when he finally
came to be naturalized,

109
00:05:55,130 --> 00:05:57,500
he was debating
whether to change it

110
00:05:57,500 --> 00:06:02,300
or not to Robert Fano, said
to be a little more American.

111
00:06:02,300 --> 00:06:06,140
And he got to talking with
his friend, Professor Al Hill.

112
00:06:06,140 --> 00:06:08,900
And apparently, Al Hill came
up with a brilliant thought,

113
00:06:08,900 --> 00:06:11,660
that since he was
in the Boston area,

114
00:06:11,660 --> 00:06:15,380
that the way he could really
get ahead was to make a change,

115
00:06:15,380 --> 00:06:18,973
but not to change the letters,
but to change the punctuation,

116
00:06:18,973 --> 00:06:20,390
and that all he
really ought to do

117
00:06:20,390 --> 00:06:22,370
is change his name
to Robert O'Fano.

118
00:06:22,370 --> 00:06:26,744
[LAUGHTER]

119
00:06:26,744 --> 00:06:29,180


120
00:06:29,180 --> 00:06:33,800
Well, the loss to Boston
politics has been our gain.

121
00:06:33,800 --> 00:06:37,289
Let me introduce Robert Fano.

122
00:06:37,289 --> 00:06:41,726
[APPLAUSE]

123
00:06:41,726 --> 00:07:06,420


124
00:07:06,420 --> 00:07:09,180
Thank you very much, Corby.

125
00:07:09,180 --> 00:07:14,730
Al Hill still calls me O'Fano.

126
00:07:14,730 --> 00:07:15,585
That's a true story.

127
00:07:15,585 --> 00:07:19,000


128
00:07:19,000 --> 00:07:21,780
I will be talking
tonight about events

129
00:07:21,780 --> 00:07:28,200
that took place in the
decade of the '60s, 20 years

130
00:07:28,200 --> 00:07:30,990
plus or minus 5.

131
00:07:30,990 --> 00:07:34,890
I will say that was a
golden year for computers,

132
00:07:34,890 --> 00:07:36,450
for computation, at MIT.

133
00:07:36,450 --> 00:07:39,420


134
00:07:39,420 --> 00:07:42,540
A lot of things happened
beside time-sharing.

135
00:07:42,540 --> 00:07:49,470
And more importantly,
as Corby mentioned,

136
00:07:49,470 --> 00:07:53,640
the foundation of the
educational and research

137
00:07:53,640 --> 00:07:56,520
program in the
department were indeed

138
00:07:56,520 --> 00:07:59,070
established at that time.

139
00:07:59,070 --> 00:08:00,960
There is no question
that time-sharing

140
00:08:00,960 --> 00:08:05,970
was at the center of the
scene, both as a research topic

141
00:08:05,970 --> 00:08:09,060
and also as a tool for
a lot of other work.

142
00:08:09,060 --> 00:08:12,885
But it was by no means
alone on the scene.

143
00:08:12,885 --> 00:08:15,840
A lot of other computer
science research

144
00:08:15,840 --> 00:08:20,850
was going on in that
period, particularly

145
00:08:20,850 --> 00:08:24,480
the artificial intelligence
group, which was originally

146
00:08:24,480 --> 00:08:26,230
part of Project MAC.

147
00:08:26,230 --> 00:08:28,770
I was very active
in that period.

148
00:08:28,770 --> 00:08:32,909
And I hope that some of the
other topics in computer

149
00:08:32,909 --> 00:08:35,200
science will be discussed
at further meeting.

150
00:08:35,200 --> 00:08:37,409
I have barely,
barely enough time

151
00:08:37,409 --> 00:08:40,610
to talk about time-sharing.

152
00:08:40,610 --> 00:08:45,540
Now, I'm not going to make
this an historical talk.

153
00:08:45,540 --> 00:08:48,570
I won't talk about
facts and dates.

154
00:08:48,570 --> 00:08:51,450
It's not my style.

155
00:08:51,450 --> 00:08:54,150
And I would be very boring.

156
00:08:54,150 --> 00:08:59,220
What I'm going to focus on
is the ideas, the interaction

157
00:08:59,220 --> 00:09:01,950
of ideas that took
place at that time,

158
00:09:01,950 --> 00:09:05,700
and the people in whose
mind those ideas were

159
00:09:05,700 --> 00:09:09,480
born and evolved.

160
00:09:09,480 --> 00:09:11,880
So that's going
to be the subject.

161
00:09:11,880 --> 00:09:14,290
I wish I could do
something more.

162
00:09:14,290 --> 00:09:18,360
I wish I could bring to you
the excitement of the period.

163
00:09:18,360 --> 00:09:19,785
But that's very
hard to recreate.

164
00:09:19,785 --> 00:09:23,220


165
00:09:23,220 --> 00:09:28,560
Now, to start with, I would
like to describe to you what

166
00:09:28,560 --> 00:09:34,620
computation was around
1960 so that you can see

167
00:09:34,620 --> 00:09:39,060
how time-sharing came about.

168
00:09:39,060 --> 00:09:42,450
While there was a
computation center at MIT,

169
00:09:42,450 --> 00:09:48,990
it had been established
in 1956 by Philip Morse.

170
00:09:48,990 --> 00:09:53,670
Morse was, I should say, yes,
a very distinguished physicist.

171
00:09:53,670 --> 00:09:56,610


172
00:09:56,610 --> 00:10:02,460
He's also well known
because his work

173
00:10:02,460 --> 00:10:05,970
during World War II
laid the foundation

174
00:10:05,970 --> 00:10:10,050
for a new discipline that got to
be called operations research.

175
00:10:10,050 --> 00:10:12,930
And as a matter of fact, he
established the Operations

176
00:10:12,930 --> 00:10:15,570
Research Center at MIT.

177
00:10:15,570 --> 00:10:25,860
Now, in 1956, he persuaded IBM
to establish at MIT a IBM 704

178
00:10:25,860 --> 00:10:34,890
installation to serve one shift
MIT research and one shift

179
00:10:34,890 --> 00:10:37,980
research in other
New England colleges.

180
00:10:37,980 --> 00:10:41,040
This was established,
of course, charge free.

181
00:10:41,040 --> 00:10:45,540
But MIT promised in return
to put up a building.

182
00:10:45,540 --> 00:10:49,260
And building 26 was
then being built.

183
00:10:49,260 --> 00:10:53,580
So they decided in a
hurry to add a wing to it.

184
00:10:53,580 --> 00:10:58,140
And that wing is where the
physics library is now.

185
00:10:58,140 --> 00:11:02,500
That was the computation
center, established in 1956.

186
00:11:02,500 --> 00:11:08,400
Now, originally, the
machine was a 704.

187
00:11:08,400 --> 00:11:14,610
It became a 709 in 1960.

188
00:11:14,610 --> 00:11:19,620
7090, which was the first IBM
transistorized machine, that

189
00:11:19,620 --> 00:11:25,470
is the 709 transistorized,
came in, I believe, in 1962.

190
00:11:25,470 --> 00:11:29,220
And the 794, which
was just the addition

191
00:11:29,220 --> 00:11:32,670
of a floating point,
arithmetic unit,

192
00:11:32,670 --> 00:11:35,110
came in shortly thereafter.

193
00:11:35,110 --> 00:11:38,520
Now, I want to show you what--

194
00:11:38,520 --> 00:11:41,280
well, first of all, let me
say, a computation center

195
00:11:41,280 --> 00:11:43,020
was something special
in those days.

196
00:11:43,020 --> 00:11:46,920
It was both a
showpiece in the sense

197
00:11:46,920 --> 00:11:50,610
that every group touring
MIT will stop in front

198
00:11:50,610 --> 00:11:51,960
of the computation center.

199
00:11:51,960 --> 00:11:53,100
There were big windows.

200
00:11:53,100 --> 00:11:56,520
And you still see them
over in building 26.

201
00:11:56,520 --> 00:11:59,700
And you saw all the
time crowds gaping

202
00:11:59,700 --> 00:12:03,750
at this marvel of modern
engineering, the computer.

203
00:12:03,750 --> 00:12:08,820
But they couldn't get in,
just look through the window.

204
00:12:08,820 --> 00:12:14,400
It was also a temple
with his own high priest.

205
00:12:14,400 --> 00:12:18,210
Now, I want to show you
what it looked like then.

206
00:12:18,210 --> 00:12:22,290


207
00:12:22,290 --> 00:12:29,470
Well, let me see if it works.

208
00:12:29,470 --> 00:12:32,345


209
00:12:32,345 --> 00:12:37,265
I'd like to get a
picture if I could.

210
00:12:37,265 --> 00:12:39,174
We're being observed.

211
00:12:39,174 --> 00:12:39,970
Huh?

212
00:12:39,970 --> 00:12:41,140
We are being watched.

213
00:12:41,140 --> 00:12:42,670
OK.

214
00:12:42,670 --> 00:12:43,540
It will appear.

215
00:12:43,540 --> 00:12:54,710


216
00:12:54,710 --> 00:12:57,350
Well, that is the
computation center.

217
00:12:57,350 --> 00:12:59,750
And right in the
middle of it, there

218
00:12:59,750 --> 00:13:06,620
is the associate director,
Dr. Fernando Jose Corbato.

219
00:13:06,620 --> 00:13:10,040
Of course, he didn't have
a beard in those days.

220
00:13:10,040 --> 00:13:17,030
And you look a little
bit younger, I must say.

221
00:13:17,030 --> 00:13:19,430
Now, he was associate director.

222
00:13:19,430 --> 00:13:22,910
And I must mention that one
of the major contributions

223
00:13:22,910 --> 00:13:26,630
that Philip Morris
made to sharing

224
00:13:26,630 --> 00:13:31,190
was that somewhere
in the early '50s,

225
00:13:31,190 --> 00:13:38,690
he gave an assistantship in
computation to Corby, Fernando

226
00:13:38,690 --> 00:13:39,530
Jose Corbato.

227
00:13:39,530 --> 00:13:41,045
He's generally known as Corby.

228
00:13:41,045 --> 00:13:44,420


229
00:13:44,420 --> 00:13:47,810
And that's how he cut
his teeth on software,

230
00:13:47,810 --> 00:13:50,180
by working on Whirlwind.

231
00:13:50,180 --> 00:13:56,180
Now, when the computation
center was established in 1956,

232
00:13:56,180 --> 00:13:59,090
Morse had the
further good judgment

233
00:13:59,090 --> 00:14:03,290
of offering a
research associateship

234
00:14:03,290 --> 00:14:06,260
to Corby, who had just
completed his doctorate

235
00:14:06,260 --> 00:14:09,320
in theoretical physics.

236
00:14:09,320 --> 00:14:11,510
So without knowing
at the time, he

237
00:14:11,510 --> 00:14:15,290
made sure that the
future Mr. Time-Sharing

238
00:14:15,290 --> 00:14:16,610
would stay in computation.

239
00:14:16,610 --> 00:14:20,180
So that's a significant
contribution to time-sharing,

240
00:14:20,180 --> 00:14:22,760
because of course, you
all know, there really

241
00:14:22,760 --> 00:14:24,030
is Mr. Time-Sharing.

242
00:14:24,030 --> 00:14:26,510
I'm just talking
about what he did.

243
00:14:26,510 --> 00:14:29,840
It's a long-established
tradition, as you will hear,

244
00:14:29,840 --> 00:14:32,360
that he does a thing,
and I talk about it.

245
00:14:32,360 --> 00:14:36,430
It's kind of an act we go about.

246
00:14:36,430 --> 00:14:40,070
I'm the front man.

247
00:14:40,070 --> 00:14:45,020
Well, now, the
computation center

248
00:14:45,020 --> 00:14:47,600
was operating in
what was then called

249
00:14:47,600 --> 00:14:54,140
a closed shop, which meant
that people who had a program--

250
00:14:54,140 --> 00:14:56,990
and incidentally,
the program were all

251
00:14:56,990 --> 00:15:01,280
punched on so-called IBM cards.

252
00:15:01,280 --> 00:15:05,090
Each of these--
well, let me say,

253
00:15:05,090 --> 00:15:12,350
this box here contains 1,000 of
these devices, the IBM cards.

254
00:15:12,350 --> 00:15:15,380
And there are four
more down there.

255
00:15:15,380 --> 00:15:18,730
Now, altogether, they
have a storage capacity

256
00:15:18,730 --> 00:15:22,700
essentially equal
to a garden-variety

257
00:15:22,700 --> 00:15:26,186
double-sided diskette, roughly.

258
00:15:26,186 --> 00:15:29,180
Now, that gives you some
proportion, you see.

259
00:15:29,180 --> 00:15:32,820
Now, all program had to
be punched on these cards.

260
00:15:32,820 --> 00:15:35,070
So people arrived at
the computation center.

261
00:15:35,070 --> 00:15:36,080
There was a counter.

262
00:15:36,080 --> 00:15:39,290
And you gave them
a batch of cards.

263
00:15:39,290 --> 00:15:40,670
They will take them.

264
00:15:40,670 --> 00:15:43,670
They will go-- very
efficiently operated--

265
00:15:43,670 --> 00:15:47,210
will go to a machine that will
translate, read the cards,

266
00:15:47,210 --> 00:15:48,950
and put them on tape.

267
00:15:48,950 --> 00:15:52,070
Then the tape was
run on the computer.

268
00:15:52,070 --> 00:15:56,120
Then the output, again, went
into another similar machine

269
00:15:56,120 --> 00:15:58,940
of the same, which will
generate some printout of some

270
00:15:58,940 --> 00:16:00,590
of the cards.

271
00:16:00,590 --> 00:16:05,220
All right, now this whole-- the
whole process turnaround time

272
00:16:05,220 --> 00:16:08,660
was pretty long, something
like one or two days, often.

273
00:16:08,660 --> 00:16:10,400
Am I correct?

274
00:16:10,400 --> 00:16:11,960
It was about that.

275
00:16:11,960 --> 00:16:20,160
Now, and people printed each
printout for two reasons.

276
00:16:20,160 --> 00:16:24,980
One was that to
debug your program,

277
00:16:24,980 --> 00:16:30,340
you had to get what was called
a memory dump, all right?

278
00:16:30,340 --> 00:16:31,650
That's one reason.

279
00:16:31,650 --> 00:16:35,330
The second reason, that
you could get at a computer

280
00:16:35,330 --> 00:16:39,140
so seldom that
when you got to it,

281
00:16:39,140 --> 00:16:41,780
you'd better compute anything
that you could possibly

282
00:16:41,780 --> 00:16:46,880
think and print out anything
that you could possibly think,

283
00:16:46,880 --> 00:16:47,870
you see?

284
00:16:47,870 --> 00:16:49,900
To make sure--
because otherwise,

285
00:16:49,900 --> 00:16:52,760
it would take a while
before you got another shot.

286
00:16:52,760 --> 00:16:56,030
So that was the
type of operation.

287
00:16:56,030 --> 00:16:58,670
Now, to give you another
order of magnitude,

288
00:16:58,670 --> 00:17:03,200
the 7094 that existed in that
period-- that's the 7094.

289
00:17:03,200 --> 00:17:04,619
I read it carefully.

290
00:17:04,619 --> 00:17:09,380
You will see 7094 written
on the console over there.

291
00:17:09,380 --> 00:17:10,940
But you are too far away.

292
00:17:10,940 --> 00:17:15,109
7090, I'm sorry.

293
00:17:15,109 --> 00:17:26,720
And a core memory of
32,000 36-bit words, right?

294
00:17:26,720 --> 00:17:35,855
Now, that memory rented, list
rent, for $17,000 a month,

295
00:17:35,855 --> 00:17:38,210
all right?

296
00:17:38,210 --> 00:17:44,510
The arithmetic-- no, I'm sorry,
the instruction processing unit

297
00:17:44,510 --> 00:17:48,380
rented for about that
much too, $17,000.

298
00:17:48,380 --> 00:17:51,620
The arithmetic unit,
which was a separate unit,

299
00:17:51,620 --> 00:17:52,820
was half that much.

300
00:17:52,820 --> 00:17:56,990
That gives you some idea
of the cost of those days.

301
00:17:56,990 --> 00:18:00,080
And it's important
to keep in mind that.

302
00:18:00,080 --> 00:18:03,000
Now, the speed, I believe
that the memory cycle

303
00:18:03,000 --> 00:18:05,630
was something like 2
microseconds, something

304
00:18:05,630 --> 00:18:06,710
like that.

305
00:18:06,710 --> 00:18:14,600
So you realize that it wasn't
quite the horsepower of a PC,

306
00:18:14,600 --> 00:18:15,530
see?

307
00:18:15,530 --> 00:18:20,310
Now, keep it keep that fact in
mind, because it's important.

308
00:18:20,310 --> 00:18:23,510
Now, there was a
lot of frustration,

309
00:18:23,510 --> 00:18:26,840
as you can well expect,
because if you misplaced

310
00:18:26,840 --> 00:18:32,990
a comma in any program, well,
it was two days that have gone.

311
00:18:32,990 --> 00:18:37,250
You know, you corrected
the comma quickly.

312
00:18:37,250 --> 00:18:41,000
But then you had to wait two
days to have another shot.

313
00:18:41,000 --> 00:18:43,670
So that was very frustrating.

314
00:18:43,670 --> 00:18:47,157
Now, this is the typical mode
of operation of university

315
00:18:47,157 --> 00:18:47,990
computation centers.

316
00:18:47,990 --> 00:18:51,410
And in industry, it was the same
thing, was typical operation.

317
00:18:51,410 --> 00:18:54,920
But at MIT, however,
there was another style

318
00:18:54,920 --> 00:18:56,180
of computing as well.

319
00:18:56,180 --> 00:18:58,610
And that is very important.

320
00:18:58,610 --> 00:19:00,980
It has a lot to do
with time-sharing.

321
00:19:00,980 --> 00:19:03,200
This alternate
style of computing

322
00:19:03,200 --> 00:19:06,530
that I will call
it online computing

323
00:19:06,530 --> 00:19:10,550
really dated back to
the Whirlwind computer.

324
00:19:10,550 --> 00:19:12,770
Since it was a
home-grown computer,

325
00:19:12,770 --> 00:19:15,230
people were not so
worried about how much

326
00:19:15,230 --> 00:19:19,190
it cost if you sat there and
fiddled around for a while.

327
00:19:19,190 --> 00:19:21,930
So people started
using it online.

328
00:19:21,930 --> 00:19:24,650
Now, the Whirlwind
computer was ready to be

329
00:19:24,650 --> 00:19:30,740
demolished around 1951.

330
00:19:30,740 --> 00:19:33,980
And suddenly, a
competitor discovered

331
00:19:33,980 --> 00:19:35,720
that you could
build an air defense

332
00:19:35,720 --> 00:19:37,940
system around a computer.

333
00:19:37,940 --> 00:19:40,580
That's how Lincoln
Laboratory was born.

334
00:19:40,580 --> 00:19:44,270
So the Whirlwind computer
became the primary tool

335
00:19:44,270 --> 00:19:51,090
in the original experimentation
on the air defense system.

336
00:19:51,090 --> 00:19:52,500
But other things happened.

337
00:19:52,500 --> 00:19:56,570
You see, the whole
group under Forrester--

338
00:19:56,570 --> 00:19:58,640
I'm sure many of
you heard this talk.

339
00:19:58,640 --> 00:20:00,890
It was the first
talk in the series--

340
00:20:00,890 --> 00:20:03,440
moved to Lincoln
Laboratory with him.

341
00:20:03,440 --> 00:20:08,240
So a tradition of online
computation and development,

342
00:20:08,240 --> 00:20:10,670
Lincoln Laboratory
got established.

343
00:20:10,670 --> 00:20:15,200
Now, there was a person that was
very critical in that respect.

344
00:20:15,200 --> 00:20:18,170
His name was Wes Clark.

345
00:20:18,170 --> 00:20:22,370
He was a top computer designer.

346
00:20:22,370 --> 00:20:25,610
And he designed
the TX-0 computer,

347
00:20:25,610 --> 00:20:29,810
which was shortly thereafter
loaned, permanent loan,

348
00:20:29,810 --> 00:20:33,050
to electrical engineering
department and started students

349
00:20:33,050 --> 00:20:34,550
here working online--

350
00:20:34,550 --> 00:20:39,120
very important-- and then
developed the TX-2 computer,

351
00:20:39,120 --> 00:20:43,940
a much more powerful
computer, which,

352
00:20:43,940 --> 00:20:51,080
because Clark believed
in that, was used online.

353
00:20:51,080 --> 00:20:53,000
That is, you signed
up on the board.

354
00:20:53,000 --> 00:20:55,010
You wander for half
an hour, for an hour.

355
00:20:55,010 --> 00:20:57,290
And as long as there
were slots available,

356
00:20:57,290 --> 00:21:00,500
you would have had the computer
totally at your disposal.

357
00:21:00,500 --> 00:21:03,620
That was a computer with
64,000 words of memory.

358
00:21:03,620 --> 00:21:08,870
It was the first big core
memory ever built, you see?

359
00:21:08,870 --> 00:21:13,580
He also built another little
computer this time called LINC.

360
00:21:13,580 --> 00:21:16,160
And that was a computer
that was specifically

361
00:21:16,160 --> 00:21:23,330
designed for real-time
laboratory experiments.

362
00:21:23,330 --> 00:21:25,490
And that had tremendous
influence too.

363
00:21:25,490 --> 00:21:29,750
So there was this whole
tradition of online computation

364
00:21:29,750 --> 00:21:31,820
and computers in
Lincoln Laboratory.

365
00:21:31,820 --> 00:21:38,450
Now, let me-- now,
I may mention also

366
00:21:38,450 --> 00:21:43,700
that one of the people that were
in the team that built the TX-2

367
00:21:43,700 --> 00:21:49,220
computer happened
to be Ken Olson, who

368
00:21:49,220 --> 00:21:54,840
left Lincoln Laboratory shortly
thereafter and started DEC,

369
00:21:54,840 --> 00:21:55,340
you see.

370
00:21:55,340 --> 00:21:59,510
And he's still chief
executive officer of DEC.

371
00:21:59,510 --> 00:22:03,050
Now, that tradition
went into DEC.

372
00:22:03,050 --> 00:22:06,395
That's why they came out with
the PDP-1 computer around 1960.

373
00:22:06,395 --> 00:22:08,000
Was it '60?

374
00:22:08,000 --> 00:22:09,930
Something like that, yeah.

375
00:22:09,930 --> 00:22:12,560
All right?

376
00:22:12,560 --> 00:22:14,060
For online use.

377
00:22:14,060 --> 00:22:17,150
So a whole tradition
of online use

378
00:22:17,150 --> 00:22:22,580
small computer, minicomputer,
grew out, really, of Whirlwind,

379
00:22:22,580 --> 00:22:23,270
right?

380
00:22:23,270 --> 00:22:25,630
Now, there was
another path, however.

381
00:22:25,630 --> 00:22:26,600
Well, let me show.

382
00:22:26,600 --> 00:22:32,990
One of the very important pieces
of work done on TX-2 was what

383
00:22:32,990 --> 00:22:37,190
was called a program
called Sketchpad by--

384
00:22:37,190 --> 00:22:40,880
that was a doctorate
thesis by Ivan Sutherland.

385
00:22:40,880 --> 00:22:44,240
Now, Ivan Sutherland
did this work--

386
00:22:44,240 --> 00:22:49,040
his thesis was
completed in 1962.

387
00:22:49,040 --> 00:22:52,310
And you see him there at
a console of the TX-2.

388
00:22:52,310 --> 00:22:56,845
Now, a scope was the routine
thing for those computer--

389
00:22:56,845 --> 00:22:58,345
I wouldn't call
that a minicomputer.

390
00:22:58,345 --> 00:23:00,620
It was a big one in those days.

391
00:23:00,620 --> 00:23:02,480
You see, and he's there.

392
00:23:02,480 --> 00:23:07,460
He developed the
first that I know

393
00:23:07,460 --> 00:23:11,030
honest-to-goodness
computer-aided design graphical

394
00:23:11,030 --> 00:23:12,260
system.

395
00:23:12,260 --> 00:23:15,470
And that's it, you see?

396
00:23:15,470 --> 00:23:20,030
Now, but there was
another ramification.

397
00:23:20,030 --> 00:23:24,110
You see, Whirlwind was--

398
00:23:24,110 --> 00:23:28,700
grew up in one of the
department laboratories, which,

399
00:23:28,700 --> 00:23:31,235
at that time, was called the
Servomechanisms Laboratory.

400
00:23:31,235 --> 00:23:35,210
It changed name when
Frank Reintjes over there

401
00:23:35,210 --> 00:23:39,810
became director and became the
Electronic System Laboratory.

402
00:23:39,810 --> 00:23:41,270
And now it's called LEDS.

403
00:23:41,270 --> 00:23:46,070
So it's changed name, but
still the same laboratory.

404
00:23:46,070 --> 00:23:51,860
You see, since the Whirlwind was
born there, several people who

405
00:23:51,860 --> 00:23:54,530
stayed, who didn't
go to Lincoln,

406
00:23:54,530 --> 00:23:56,560
were permeated with
the same ideas.

407
00:23:56,560 --> 00:23:59,270


408
00:23:59,270 --> 00:24:03,500
And there were some very
significant computer projects

409
00:24:03,500 --> 00:24:04,670
that took place there.

410
00:24:04,670 --> 00:24:10,010


411
00:24:10,010 --> 00:24:13,760
First of all, there was a
numerically controlled machine

412
00:24:13,760 --> 00:24:20,930
tool very early, you see,
and then the APT language

413
00:24:20,930 --> 00:24:24,890
for programming that machine--

414
00:24:24,890 --> 00:24:28,370
for generating the tapes
that control the machine.

415
00:24:28,370 --> 00:24:31,130
And then, finally,
the laboratory

416
00:24:31,130 --> 00:24:33,860
got together with the mechanical
engineering department.

417
00:24:33,860 --> 00:24:37,430
And by 1959 or '60,
they had a project

418
00:24:37,430 --> 00:24:39,410
on computer-aided design.

419
00:24:39,410 --> 00:24:47,070
Now, one of the people who cut
his teeth in APT is over there.

420
00:24:47,070 --> 00:24:48,410
His name is Doug Ross.

421
00:24:48,410 --> 00:24:50,330
And I think I've
seen him over there.

422
00:24:50,330 --> 00:24:52,725
Stand up, Doug, so that
people can compare.

423
00:24:52,725 --> 00:24:56,030
[LAUGHTER]

424
00:24:56,030 --> 00:25:04,220
Now, Doug, continuing his
work on the language that

425
00:25:04,220 --> 00:25:11,060
was called automated engineering
design, and then, not long

426
00:25:11,060 --> 00:25:15,260
thereafter, left MIT, was
the founder of SoftTech,

427
00:25:15,260 --> 00:25:17,210
and is still chairman
of the board,

428
00:25:17,210 --> 00:25:23,600
but now is part-time lecturer
in the department again.

429
00:25:23,600 --> 00:25:29,690
Now, at about that time--

430
00:25:29,690 --> 00:25:33,440
I'm talking now '62--

431
00:25:33,440 --> 00:25:37,760
no, 1960, there was a
very important paper

432
00:25:37,760 --> 00:25:40,820
that was written and
that created a great deal

433
00:25:40,820 --> 00:25:43,430
of interest in the community.

434
00:25:43,430 --> 00:25:49,430
The author was JCR Licklider,
who wrote a paper called

435
00:25:49,430 --> 00:25:52,190
"Man-Computer Symbiosis."

436
00:25:52,190 --> 00:25:57,800
Now, you see, from
Lincoln Laboratory,

437
00:25:57,800 --> 00:26:02,330
these ideas permeated the firm
of Bolt, Beranek, and Newman,

438
00:26:02,330 --> 00:26:04,610
where Licklider
was vice president.

439
00:26:04,610 --> 00:26:08,840
Now, I must also
say that Licklider

440
00:26:08,840 --> 00:26:13,940
was a distinguished
experimental psychologist

441
00:26:13,940 --> 00:26:19,100
and was on our faculty just
around 1950 for a short period

442
00:26:19,100 --> 00:26:23,120
before going to BBN, so
was a member of the family.

443
00:26:23,120 --> 00:26:28,430
And he got enamored of computers
and wrote this very influential

444
00:26:28,430 --> 00:26:31,070
paper called
"Man-Computer Symbiosis,"

445
00:26:31,070 --> 00:26:36,110
where it was emphasized the
advantage that one could draw

446
00:26:36,110 --> 00:26:38,660
psychologically from
man-machine interaction.

447
00:26:38,660 --> 00:26:40,500
That was a very
influential paper.

448
00:26:40,500 --> 00:26:43,700


449
00:26:43,700 --> 00:26:47,600
All right, now this
is the environment

450
00:26:47,600 --> 00:26:51,890
that existed in 1960.

451
00:26:51,890 --> 00:26:55,640
All right, who
invented time-share?

452
00:26:55,640 --> 00:27:01,370
Well, obviously, it was a
very favorable environment.

453
00:27:01,370 --> 00:27:07,340
Far as I know, the first written
mention of something that I

454
00:27:07,340 --> 00:27:12,350
would call a time-sharing system
was a memorandum written--

455
00:27:12,350 --> 00:27:17,450
dated January 1,
1959, and written

456
00:27:17,450 --> 00:27:21,770
by John McCarthy, who was
a member of the faculty

457
00:27:21,770 --> 00:27:23,850
of the department at that time.

458
00:27:23,850 --> 00:27:28,430
A couple years, three years
later, he left for Stanford,

459
00:27:28,430 --> 00:27:31,100
where he's been ever since.

460
00:27:31,100 --> 00:27:33,605
Now, let me say a few
words about John McCarthy.

461
00:27:33,605 --> 00:27:36,330


462
00:27:36,330 --> 00:27:39,630
John McCarthy is well known
for a number of things

463
00:27:39,630 --> 00:27:41,440
at about the same time.

464
00:27:41,440 --> 00:27:43,800
As a matter of
fact, a year later,

465
00:27:43,800 --> 00:27:48,912
he invented LISP,
just plain that,

466
00:27:48,912 --> 00:27:54,550
and wrote a classic paper
with a theoretical foundation

467
00:27:54,550 --> 00:27:55,090
for LISP.

468
00:27:55,090 --> 00:27:57,940


469
00:27:57,940 --> 00:28:00,940
At the same time,
he was a member

470
00:28:00,940 --> 00:28:05,830
of the international committee
that defined ALGOL 60.

471
00:28:05,830 --> 00:28:11,350
So he was a major figure
in the computer field.

472
00:28:11,350 --> 00:28:12,700
So he wrote this memorandum.

473
00:28:12,700 --> 00:28:14,720
This was a memorandum
to the director

474
00:28:14,720 --> 00:28:16,990
of the computations
center, which

475
00:28:16,990 --> 00:28:20,050
was Professor Philip Morse.

476
00:28:20,050 --> 00:28:23,140
Shortly thereafter, in
June of the same year,

477
00:28:23,140 --> 00:28:29,290
1959, a Britisher by the name of
Christopher Strachey presented

478
00:28:29,290 --> 00:28:32,440
a paper at an
international meeting,

479
00:28:32,440 --> 00:28:35,680
essentially made the
similar, same suggestion.

480
00:28:35,680 --> 00:28:39,580
Now, a couple of words
about Christopher Strachey.

481
00:28:39,580 --> 00:28:48,040
Christopher Strachey, I think,
was an EE by trade, I believe,

482
00:28:48,040 --> 00:28:48,930
or a physicist.

483
00:28:48,930 --> 00:28:52,060
I'm not sure-- one or the other.

484
00:28:52,060 --> 00:28:55,620
But he became an exponent,
a very important figure

485
00:28:55,620 --> 00:28:59,460
in computer science in England.

486
00:28:59,460 --> 00:29:04,170
He worked on
theoretical linguistics.

487
00:29:04,170 --> 00:29:07,950
And as a matter of fact,
he was, in the '60s,

488
00:29:07,950 --> 00:29:11,100
a visiting professor
here for a year.

489
00:29:11,100 --> 00:29:17,070
And we gained a great deal in
the development of a curriculum

490
00:29:17,070 --> 00:29:18,690
from his teaching.

491
00:29:18,690 --> 00:29:22,150
Unfortunately, Christopher
died relatively young.

492
00:29:22,150 --> 00:29:26,610
I think he was still
in the 50s in 1975.

493
00:29:26,610 --> 00:29:29,700
I had the privilege of
spending an evening with him,

494
00:29:29,700 --> 00:29:31,560
I believe, something
like a couple of months

495
00:29:31,560 --> 00:29:33,120
before he died in Oxford.

496
00:29:33,120 --> 00:29:36,889
He was at a chair in Oxford.

497
00:29:36,889 --> 00:29:38,220
All right.

498
00:29:38,220 --> 00:29:42,690
Now, what were the
goals of time-sharing?

499
00:29:42,690 --> 00:29:45,810
Obviously, one
strong motivation,

500
00:29:45,810 --> 00:29:47,850
certainly on the
part of McCarthy,

501
00:29:47,850 --> 00:29:50,250
is to eliminate the
frustration that

502
00:29:50,250 --> 00:29:53,670
resulted from missing
or misplaced commas

503
00:29:53,670 --> 00:29:56,070
that he was after.

504
00:29:56,070 --> 00:30:00,110
Then he shortened the turnaround
time to practically nothing.

505
00:30:00,110 --> 00:30:02,820
So that was his goal.

506
00:30:02,820 --> 00:30:08,490
He felt also that big
computers were not efficiently,

507
00:30:08,490 --> 00:30:10,830
in a real sense, utilized.

508
00:30:10,830 --> 00:30:13,500
He wanted to have an
efficient utilization.

509
00:30:13,500 --> 00:30:16,860
You see, you got to be very
careful about efficiency.

510
00:30:16,860 --> 00:30:21,870
When the community of users
puts out large print-out

511
00:30:21,870 --> 00:30:24,390
and computer everything
that you can possibly

512
00:30:24,390 --> 00:30:29,790
compute to save themselves time
because of the long turnaround,

513
00:30:29,790 --> 00:30:33,240
it is not an
efficient operation.

514
00:30:33,240 --> 00:30:35,375
So if you look at the
overall efficiency,

515
00:30:35,375 --> 00:30:36,750
there was a great
deal that could

516
00:30:36,750 --> 00:30:43,410
be gained by what got to be
known as time-sharing system.

517
00:30:43,410 --> 00:30:48,780
Now, he also had another
agenda that had to do with AI.

518
00:30:48,780 --> 00:30:51,420
John McCarthy was an AI guy.

519
00:30:51,420 --> 00:30:53,940
He had invented LISP.

520
00:30:53,940 --> 00:30:59,220
Of course, he used a lot
of memory, all right?

521
00:30:59,220 --> 00:31:02,070
So he wanted a big memory.

522
00:31:02,070 --> 00:31:04,560
But you know, big memory,
particularly then,

523
00:31:04,560 --> 00:31:07,060
very, very, very expensive.

524
00:31:07,060 --> 00:31:12,210
So he was looking for a way of
building economic justification

525
00:31:12,210 --> 00:31:14,490
for a large memory.

526
00:31:14,490 --> 00:31:18,340
When he didn't use it,
somebody else could use it.

527
00:31:18,340 --> 00:31:19,460
So that was clear.

528
00:31:19,460 --> 00:31:23,280
I remember very clearly
that was part of his agenda.

529
00:31:23,280 --> 00:31:24,240
And he was right.

530
00:31:24,240 --> 00:31:27,990


531
00:31:27,990 --> 00:31:32,100
Well, generally speaking,
what he had in mind

532
00:31:32,100 --> 00:31:34,730
was to combine the power--

533
00:31:34,730 --> 00:31:39,070
in those days, it was
great-- of the 7090.

534
00:31:39,070 --> 00:31:42,480
Even if I denigrate it
now by comparing to a PC,

535
00:31:42,480 --> 00:31:45,990
it was a very powerful
machine for those days.

536
00:31:45,990 --> 00:31:52,500
And combine that with the
easy access that had become

537
00:31:52,500 --> 00:31:56,710
characteristic of the TX-0
computer that was more or less

538
00:31:56,710 --> 00:31:58,770
next door-- it
was one floor up--

539
00:31:58,770 --> 00:32:00,480
which was an 18-bit machine.

540
00:32:00,480 --> 00:32:03,615


541
00:32:03,615 --> 00:32:07,370
It had the power that an 18-bit
machine and a small core memory

542
00:32:07,370 --> 00:32:10,890
had, although it was
far from trivial.

543
00:32:10,890 --> 00:32:15,080
So he wanted to combine
the two, all right?

544
00:32:15,080 --> 00:32:21,480
Now, John McCarthy
also gave a lecture.

545
00:32:21,480 --> 00:32:24,900
As part of the centennial
celebration of MIT,

546
00:32:24,900 --> 00:32:28,930
there was a series
of lectures in 1961.

547
00:32:28,930 --> 00:32:31,470
And he came out in that
lecture with the idea

548
00:32:31,470 --> 00:32:33,070
of a computer utility.

549
00:32:33,070 --> 00:32:36,410
Now, this is a very
important notion.

550
00:32:36,410 --> 00:32:42,800
You see, it's one thing to think
of time-sharing a computer,

551
00:32:42,800 --> 00:32:47,150
another to think
about a computer being

552
00:32:47,150 --> 00:32:50,540
a sort of a public
utility, to supply

553
00:32:50,540 --> 00:32:54,110
computer power to all people.

554
00:32:54,110 --> 00:32:56,450
And as a matter of
fact, it's this notion

555
00:32:56,450 --> 00:32:59,480
that really got me interested.

556
00:32:59,480 --> 00:33:02,780
And in those days,
I used an awful lot

557
00:33:02,780 --> 00:33:06,620
the analogy in
talking and writing

558
00:33:06,620 --> 00:33:10,200
or computer power
and electric power.

559
00:33:10,200 --> 00:33:14,900
The real big difference,
that a distribution system,

560
00:33:14,900 --> 00:33:17,810
as you could do with
electric power made,

561
00:33:17,810 --> 00:33:22,700
was that power tools were
available to individual

562
00:33:22,700 --> 00:33:28,820
as individual, amplifying
the capability of the hands.

563
00:33:28,820 --> 00:33:31,400
Without an electric
distribution system,

564
00:33:31,400 --> 00:33:33,740
you could have power
tools only in the faculty,

565
00:33:33,740 --> 00:33:36,530
with a steam engine
running the whole thing.

566
00:33:36,530 --> 00:33:42,050
And my vision of what a
computer utility meant,

567
00:33:42,050 --> 00:33:46,700
and John McCarthy's vision,
was that you could distribute

568
00:33:46,700 --> 00:33:51,620
computer power to amplify
the intellectual capability

569
00:33:51,620 --> 00:33:57,470
of people, just like power tools
amplify the physical and manual

570
00:33:57,470 --> 00:33:59,930
capability of individual.

571
00:33:59,930 --> 00:34:04,190
That was kind of the image
that arose from his talk.

572
00:34:04,190 --> 00:34:07,940
Furthermore, MIT
was very worried

573
00:34:07,940 --> 00:34:11,120
about its
computational facility,

574
00:34:11,120 --> 00:34:17,090
because computation was just
booming around 1960, '61.

575
00:34:17,090 --> 00:34:20,060
So a committee was
appointed to study

576
00:34:20,060 --> 00:34:21,710
the computer needs of MIT.

577
00:34:21,710 --> 00:34:23,570
And that committee,
which, in the end,

578
00:34:23,570 --> 00:34:27,920
was chaired by John
McCarthy, wrote a report

579
00:34:27,920 --> 00:34:30,530
which strongly recommended
the establishment

580
00:34:30,530 --> 00:34:33,860
of a time-sharing
computer system.

581
00:34:33,860 --> 00:34:36,500
By that time, work
in the development

582
00:34:36,500 --> 00:34:39,080
of a computation-- a
time-sharing system

583
00:34:39,080 --> 00:34:41,210
was already going on, of course.

584
00:34:41,210 --> 00:34:44,780
But that was a very
strong recommendation.

585
00:34:44,780 --> 00:34:49,650
Now, before I get to the
time-sharing system itself,

586
00:34:49,650 --> 00:34:56,000
let me point out that while
the notion of time-sharing

587
00:34:56,000 --> 00:35:02,900
gained quickly a lot of support,
that was by no means unanimous,

588
00:35:02,900 --> 00:35:03,920
by no means.

589
00:35:03,920 --> 00:35:07,520
And some very outstanding
people in the computer field

590
00:35:07,520 --> 00:35:09,440
were dead set against it.

591
00:35:09,440 --> 00:35:16,490
I'm mentioning this because it
clarifies some of the problems.

592
00:35:16,490 --> 00:35:21,680
Well, your position largely,
but not exclusively,

593
00:35:21,680 --> 00:35:25,700
was based on
efficient utilization

594
00:35:25,700 --> 00:35:28,805
of this very precious
resource called computer time.

595
00:35:28,805 --> 00:35:32,210
It was a very precious resource.

596
00:35:32,210 --> 00:35:35,450
And many people were
so worried about it

597
00:35:35,450 --> 00:35:37,700
that it didn't occur
to them there was also

598
00:35:37,700 --> 00:35:42,590
something else called
precious human time,

599
00:35:42,590 --> 00:35:45,800
and that one had to make
a reasonable compromise

600
00:35:45,800 --> 00:35:48,950
between using the
two efficiently.

601
00:35:48,950 --> 00:35:50,810
It was not all one way.

602
00:35:50,810 --> 00:35:53,870
Well, let me mention some
examples to be very clear.

603
00:35:53,870 --> 00:35:57,710
I remember visiting Dick
Hamming of Hamming Code

604
00:35:57,710 --> 00:36:01,130
fame and the Bell
Telephone Laboratory.

605
00:36:01,130 --> 00:36:04,880
And he took the attitude that
it was essentially morally

606
00:36:04,880 --> 00:36:10,010
wrong to allow somebody to type
in instructions to a computer.

607
00:36:10,010 --> 00:36:12,260
He was serious.

608
00:36:12,260 --> 00:36:15,680
He was serious, all right?

609
00:36:15,680 --> 00:36:19,370
Well, a little bit
at a lesser scale,

610
00:36:19,370 --> 00:36:23,840
there was Gene Amdahl,
who gained a lot of fame

611
00:36:23,840 --> 00:36:28,340
as the designer of the
360 system for IBM,

612
00:36:28,340 --> 00:36:33,410
who just felt it was a bad
idea to interrupt a computer

613
00:36:33,410 --> 00:36:34,340
while it was running--

614
00:36:34,340 --> 00:36:36,440
I mean a program
while it was running.

615
00:36:36,440 --> 00:36:38,570
He just thought it was bad--

616
00:36:38,570 --> 00:36:41,720
again, an efficiency
reason, and for reasons

617
00:36:41,720 --> 00:36:45,200
the designer of the
360 showed very clear

618
00:36:45,200 --> 00:36:48,600
this very strong feeling
that he had about that issue,

619
00:36:48,600 --> 00:36:49,100
all right?

620
00:36:49,100 --> 00:36:53,240


621
00:36:53,240 --> 00:36:59,540
I remember sitting
next to Clark--

622
00:36:59,540 --> 00:37:02,930
Wes Clark, I mean, the
Lincoln Laboratory Clark--

623
00:37:02,930 --> 00:37:05,270
at a committee meeting
in the department.

624
00:37:05,270 --> 00:37:07,430
And somebody mentioned
time-sharing.

625
00:37:07,430 --> 00:37:12,410
And he turned to me and
said, what there is to share?

626
00:37:12,410 --> 00:37:16,010
That is, his view was
that computers were--

627
00:37:16,010 --> 00:37:18,140
the power of
computer was so small

628
00:37:18,140 --> 00:37:22,100
that there was not
enough to share.

629
00:37:22,100 --> 00:37:26,500
It didn't make any sense
to talk about sharing.

630
00:37:26,500 --> 00:37:32,555
Now, I know the view was that it
was somehow intellectually bad.

631
00:37:32,555 --> 00:37:37,180


632
00:37:37,180 --> 00:37:40,570
Jay Forrester, for instance,
who is, of course, an expert,

633
00:37:40,570 --> 00:37:44,620
I remember him saying, well,
when somebody makes a computer

634
00:37:44,620 --> 00:37:48,070
run, he ought to sit down
and think about the result

635
00:37:48,070 --> 00:37:50,800
before he makes another run.

636
00:37:50,800 --> 00:37:53,140
He shouldn't be in a
position-- it's a bad habit

637
00:37:53,140 --> 00:37:54,790
to keep trying things.

638
00:37:54,790 --> 00:37:57,590
He should sit down and think.

639
00:37:57,590 --> 00:38:01,600
Well, he changed his mind
awfully quickly once he tried.

640
00:38:01,600 --> 00:38:03,425
I'll tell you later the story.

641
00:38:03,425 --> 00:38:06,610
[LAUGHTER]

642
00:38:06,610 --> 00:38:10,780
So it was by no means-- these
are not inconsequential people.

643
00:38:10,780 --> 00:38:12,520
I want to mention
their objective,

644
00:38:12,520 --> 00:38:16,180
because it gives you some idea
of the thinking of the time.

645
00:38:16,180 --> 00:38:21,760


646
00:38:21,760 --> 00:38:27,610
Well, time-sharing-- the
first time-sharing system

647
00:38:27,610 --> 00:38:30,610
was developed by
none less than Corby

648
00:38:30,610 --> 00:38:35,500
with his associate in the
MIT Computation Center.

649
00:38:35,500 --> 00:38:39,580
And the first version
was still on the 709

650
00:38:39,580 --> 00:38:46,050
and was demonstrated
in November 1961.

651
00:38:46,050 --> 00:38:48,590
Now, a much improved version--

652
00:38:48,590 --> 00:38:52,190
the improvement continued.

653
00:38:52,190 --> 00:38:54,560
This was for the 7094--

654
00:38:54,560 --> 00:38:58,760
was in regular operation
by the summer '63.

655
00:38:58,760 --> 00:39:00,710
But a lot of problems
had to be solved.

656
00:39:00,710 --> 00:39:04,480
I believe that it could handle
something like 10 or 15 users

657
00:39:04,480 --> 00:39:05,540
at the time.

658
00:39:05,540 --> 00:39:08,360
There were a lot of
problems to be solved.

659
00:39:08,360 --> 00:39:12,770
Now, people have often asked,
what made time-sharing possible

660
00:39:12,770 --> 00:39:13,670
at that time?

661
00:39:13,670 --> 00:39:15,500
And my answer is two things.

662
00:39:15,500 --> 00:39:19,440
First of all, you needed
a transistorized computer.

663
00:39:19,440 --> 00:39:22,940
A vacuum tube computer,
they were just not up long

664
00:39:22,940 --> 00:39:26,840
enough to make time-sharing
worthwhile, just simply that.

665
00:39:26,840 --> 00:39:30,950
You had to have a
transistorized computer.

666
00:39:30,950 --> 00:39:36,470
The other thing was the
fact that IBM came out

667
00:39:36,470 --> 00:39:41,630
with a very nice device
called a disk file in

668
00:39:41,630 --> 00:39:44,900
just about 1962, '63.

669
00:39:44,900 --> 00:39:46,940
And now let me show
you what it was.

670
00:39:46,940 --> 00:39:49,710


671
00:39:49,710 --> 00:39:53,550
That huge box was
the 1301 disk file.

672
00:39:53,550 --> 00:39:55,020
It was a marvelous device.

673
00:39:55,020 --> 00:40:00,360
It had a stack of
disks and arms that

674
00:40:00,360 --> 00:40:05,550
move independently in and out to
get on the tracks on the disk.

675
00:40:05,550 --> 00:40:11,280
I think the 1301 was 9 or
18 million words, something

676
00:40:11,280 --> 00:40:14,370
of that order of magnitude.

677
00:40:14,370 --> 00:40:16,740
But that made
time-sharing worthwhile.

678
00:40:16,740 --> 00:40:19,035
Before that, mass
storage was tapes.

679
00:40:19,035 --> 00:40:21,680


680
00:40:21,680 --> 00:40:25,460
And we had a beautiful
demonstration of the difference

681
00:40:25,460 --> 00:40:27,440
that the disk file meant.

682
00:40:27,440 --> 00:40:31,640
That means relatively
quick access

683
00:40:31,640 --> 00:40:35,210
to programs and data in
a mass storage device.

684
00:40:35,210 --> 00:40:37,580
That made a
tremendous difference.

685
00:40:37,580 --> 00:40:40,370
Now, another thing that
was developed at that time

686
00:40:40,370 --> 00:40:42,440
was a fast ROM.

687
00:40:42,440 --> 00:40:43,910
That was also big box.

688
00:40:43,910 --> 00:40:46,540


689
00:40:46,540 --> 00:40:49,070
But it was essential
for swapping programs

690
00:40:49,070 --> 00:40:50,612
in and out of memory.

691
00:40:50,612 --> 00:40:52,070
Now, you know what
time-sharing is.

692
00:40:52,070 --> 00:40:59,420
So I won't try to describe
the time-sharing technique.

693
00:40:59,420 --> 00:41:08,020
Now, this wasn't--
during that period,

694
00:41:08,020 --> 00:41:12,420
a lot of other time sharing
projects got started.

695
00:41:12,420 --> 00:41:18,450
At MIT, there was a time
sharing system on a PDP-1

696
00:41:18,450 --> 00:41:21,090
that Jack Dennisd where is Jack?

697
00:41:21,090 --> 00:41:22,050
Some place there.

698
00:41:22,050 --> 00:41:24,210
I'll come back to that later--

699
00:41:24,210 --> 00:41:26,296
developed.

700
00:41:26,296 --> 00:41:28,230
At Bolt, Beranek,
and Newman, there

701
00:41:28,230 --> 00:41:32,010
was another time-sharing
system on the PDP-1.

702
00:41:32,010 --> 00:41:36,450
There was a very serious
time-sharing effort

703
00:41:36,450 --> 00:41:39,960
at the System Development
Corporation in Santa Monica,

704
00:41:39,960 --> 00:41:44,700
using a surplus computer
of the state system.

705
00:41:44,700 --> 00:41:46,890
I don't remember
what the number was.

706
00:41:46,890 --> 00:41:47,890
Q32.

707
00:41:47,890 --> 00:41:48,790
Q--

708
00:41:48,790 --> 00:41:49,350
Q32.

709
00:41:49,350 --> 00:41:50,400
Q32.

710
00:41:50,400 --> 00:41:52,080
I thought it was
something like that.

711
00:41:52,080 --> 00:41:54,660
All right.

712
00:41:54,660 --> 00:41:58,600
Then-- very interesting and
had a lot of consequences.

713
00:41:58,600 --> 00:42:02,580
You see, that machine did
not-- that system did not

714
00:42:02,580 --> 00:42:03,720
have a disk file.

715
00:42:03,720 --> 00:42:08,670
And you could tell the
difference between CTSS,

716
00:42:08,670 --> 00:42:12,060
the Compatible Time-Sharing
System that Corby developed

717
00:42:12,060 --> 00:42:15,930
and the system of
SDS because of that.

718
00:42:15,930 --> 00:42:18,690
Everything was stored
on drums or tapes.

719
00:42:18,690 --> 00:42:22,390
And it takes forever
to get to data program.

720
00:42:22,390 --> 00:42:27,600
Now, the other very important
time-sharing actually

721
00:42:27,600 --> 00:42:30,810
was at Dartmouth
College, originally

722
00:42:30,810 --> 00:42:33,540
spurred by Professor
Kemeny, who eventually

723
00:42:33,540 --> 00:42:35,880
became president of
the Dartmouth College

724
00:42:35,880 --> 00:42:39,210
and the creator of BASIC.

725
00:42:39,210 --> 00:42:41,940
That was a little different
time-sharing system

726
00:42:41,940 --> 00:42:44,760
in the sense that it was a
one-language time-sharing

727
00:42:44,760 --> 00:42:45,660
system.

728
00:42:45,660 --> 00:42:48,210
You could only
program and use BASIC.

729
00:42:48,210 --> 00:42:51,390
It was built essentially
around a BASIC interpreter,

730
00:42:51,390 --> 00:42:54,750
while the systems that I talked
about were multi-language,

731
00:42:54,750 --> 00:42:57,150
were really truly
general-purpose time-sharing

732
00:42:57,150 --> 00:42:59,520
system--

733
00:42:59,520 --> 00:43:00,690
and a lot of others.

734
00:43:00,690 --> 00:43:04,020
I won't get into that.

735
00:43:04,020 --> 00:43:07,620
Now, at that time,
that's the time

736
00:43:07,620 --> 00:43:10,920
when Project MAC got started.

737
00:43:10,920 --> 00:43:13,275
And I want to tell you
how it got started.

738
00:43:13,275 --> 00:43:19,170


739
00:43:19,170 --> 00:43:26,880
Licklider, about whom I
spoke before, in 1962,

740
00:43:26,880 --> 00:43:31,200
became director
of the information

741
00:43:31,200 --> 00:43:36,450
processes branch of the
Advanced Research Project Agency

742
00:43:36,450 --> 00:43:39,690
of the Department of Defense.

743
00:43:39,690 --> 00:43:44,850
Now, he decided that he was
going to start a program.

744
00:43:44,850 --> 00:43:47,280
He used the term
centers of excellence

745
00:43:47,280 --> 00:43:54,450
in computing in
several universities

746
00:43:54,450 --> 00:43:58,800
to push man-machine
interactions with time-sharing.

747
00:43:58,800 --> 00:44:01,230
That was the goal.

748
00:44:01,230 --> 00:44:04,470
Well, he started running
around the country,

749
00:44:04,470 --> 00:44:07,140
trying to build up
interest in doing that.

750
00:44:07,140 --> 00:44:08,640
And of course, he came to MIT.

751
00:44:08,640 --> 00:44:16,200
And he wanted very much to
have MIT start the big research

752
00:44:16,200 --> 00:44:17,850
effort on that basis.

753
00:44:17,850 --> 00:44:19,740
The trouble was
that the people that

754
00:44:19,740 --> 00:44:24,220
knew anything about
computers were kind of young.

755
00:44:24,220 --> 00:44:25,630
You saw the picture of Corby.

756
00:44:25,630 --> 00:44:30,090


757
00:44:30,090 --> 00:44:33,900
Professor Morse, who was
director of the computation

758
00:44:33,900 --> 00:44:38,130
center, had already too many
fingers in too many pies.

759
00:44:38,130 --> 00:44:42,420
And he didn't want to stick
his finger in another pie.

760
00:44:42,420 --> 00:44:45,300
So nothing was happening.

761
00:44:45,300 --> 00:44:47,310
That's where I got in.

762
00:44:47,310 --> 00:44:54,180
I felt very strongly that MIT
had to have a big research

763
00:44:54,180 --> 00:44:55,920
effort in the computer field.

764
00:44:55,920 --> 00:45:01,200
And I couldn't see
anybody ready to do it.

765
00:45:01,200 --> 00:45:10,260
And well, I hated the
idea of managing anything

766
00:45:10,260 --> 00:45:15,810
in those days, and
still do, but kind of

767
00:45:15,810 --> 00:45:23,040
got caught by Licklider's what
I call contagious enthusiasm.

768
00:45:23,040 --> 00:45:28,260
So what happened is that there
was a meeting in Virginia

769
00:45:28,260 --> 00:45:31,050
in which I spent quite a bit
of time talking with Licklider.

770
00:45:31,050 --> 00:45:36,060
As a matter of fact, there
was a long, long train ride

771
00:45:36,060 --> 00:45:37,090
back to Washington.

772
00:45:37,090 --> 00:45:38,340
That's when I talked with him.

773
00:45:38,340 --> 00:45:40,233
It was the day
before Thanksgiving.

774
00:45:40,233 --> 00:45:42,150
I want to tell you the
story, because it shows

775
00:45:42,150 --> 00:45:44,340
you something about the MIT.

776
00:45:44,340 --> 00:45:48,270
Things can go very
fast at MIT sometimes.

777
00:45:48,270 --> 00:45:53,220
Well, the next day,
it was Thanksgiving.

778
00:45:53,220 --> 00:45:55,260
And I kept thinking about this.

779
00:45:55,260 --> 00:45:59,940
And I said, oh, hell, I'll
dive in and close my eyes.

780
00:45:59,940 --> 00:46:02,430
So I decided to start Project
MAC, what became Project

781
00:46:02,430 --> 00:46:04,080
MAC, on Thanksgiving day.

782
00:46:04,080 --> 00:46:07,620
The next day, I had a date with
the provost, who at that time

783
00:46:07,620 --> 00:46:12,160
was Charlie Townes, the
Nobel Prize, the laser guy.

784
00:46:12,160 --> 00:46:15,840


785
00:46:15,840 --> 00:46:18,000
So after talking about
the other business,

786
00:46:18,000 --> 00:46:21,540
I said, hey, I got
this crazy idea.

787
00:46:21,540 --> 00:46:24,940
Would you think about it,
see if it makes sense to you?

788
00:46:24,940 --> 00:46:26,280
He said, oh, go right ahead.

789
00:46:26,280 --> 00:46:27,880
It makes sense.

790
00:46:27,880 --> 00:46:34,470
So over the weekend, I
wrote a two-page memorandum.

791
00:46:34,470 --> 00:46:38,010
The next Tuesday, I had a date
with the president, who was,

792
00:46:38,010 --> 00:46:42,570
that time, Julius Stratton.

793
00:46:42,570 --> 00:46:44,490
I told him what I had in mind.

794
00:46:44,490 --> 00:46:48,030
And he said, where are
you going to do it?

795
00:46:48,030 --> 00:46:52,780
Well, as Professor Morse
said at that time, MIT

796
00:46:52,780 --> 00:46:59,700
he had been caught with its
building down, quote unquote.

797
00:46:59,700 --> 00:47:01,470
There was no place.

798
00:47:01,470 --> 00:47:03,840
But by that time,
I heard a story--

799
00:47:03,840 --> 00:47:07,230
you know, Technology Square
was just going up at that time.

800
00:47:07,230 --> 00:47:10,770
And the top two
floor of the building

801
00:47:10,770 --> 00:47:17,640
had been leased by CEIR, which
was a computer service company.

802
00:47:17,640 --> 00:47:20,910
They intended to put a stress
computer on the ninth floor.

803
00:47:20,910 --> 00:47:24,330
That was the dream in those
days-- the most powerful

804
00:47:24,330 --> 00:47:26,220
machine in the world.

805
00:47:26,220 --> 00:47:31,110
But by then, business
had not gone to well.

806
00:47:31,110 --> 00:47:34,980
And they were instantly getting
rid of the lease commitment.

807
00:47:34,980 --> 00:47:36,480
That was the eighth
and ninth floor.

808
00:47:36,480 --> 00:47:37,500
So I jumped in.

809
00:47:37,500 --> 00:47:39,090
I knew already about it.

810
00:47:39,090 --> 00:47:41,700
So I told Stratton, yeah,
the eighth and ninth floor

811
00:47:41,700 --> 00:47:43,650
look like they may be available.

812
00:47:43,650 --> 00:47:45,550
Fine, he said.

813
00:47:45,550 --> 00:47:49,020
Well, on Thursday, Licklider
happened to be at MIT.

814
00:47:49,020 --> 00:47:53,580
So we all gathered in Stratton's
office and shook hands.

815
00:47:53,580 --> 00:47:54,780
And that was it.

816
00:47:54,780 --> 00:47:56,790
Then, of course, you
had to write a proposal.

817
00:47:56,790 --> 00:48:00,240
But everything was decided
within a week, really.

818
00:48:00,240 --> 00:48:04,350
There was a commitment of all
those people within a week.

819
00:48:04,350 --> 00:48:08,199
That's kind of unusual,
but it does happen.

820
00:48:08,199 --> 00:48:11,040
Well, all right.

821
00:48:11,040 --> 00:48:16,550


822
00:48:16,550 --> 00:48:23,300
Well, we wrote a proposal
by January 14, I believe.

823
00:48:23,300 --> 00:48:26,592
If you want to see
it, here it is.

824
00:48:26,592 --> 00:48:29,640
It was presented and
approved in principle.

825
00:48:29,640 --> 00:48:33,110
And we were given
the authorization

826
00:48:33,110 --> 00:48:36,560
to start spending
money by March 1.

827
00:48:36,560 --> 00:48:38,210
Well, I'd been planning.

828
00:48:38,210 --> 00:48:40,760
So I had to hire
somebody who was

829
00:48:40,760 --> 00:48:43,910
going to be assistant
director for administration.

830
00:48:43,910 --> 00:48:47,460
And he showed up on March 1.

831
00:48:47,460 --> 00:48:49,850
So I told him, you know,
you'll go around the institute,

832
00:48:49,850 --> 00:48:52,340
get your keys, get
a parking permit,

833
00:48:52,340 --> 00:48:54,660
get everything you need.

834
00:48:54,660 --> 00:48:57,690
So he showed back in my
office shortly thereafter.

835
00:48:57,690 --> 00:49:00,740
He said, they won't give
me a parking permit.

836
00:49:00,740 --> 00:49:02,960
They asked me what I
was going to work for.

837
00:49:02,960 --> 00:49:04,907
And there is no name.

838
00:49:04,907 --> 00:49:08,723
[LAUGHTER]

839
00:49:08,723 --> 00:49:11,120


840
00:49:11,120 --> 00:49:15,050
So I realized that
either I had to create

841
00:49:15,050 --> 00:49:19,010
a name, or otherwise,
have the project go

842
00:49:19,010 --> 00:49:22,220
by the name that has been used.

843
00:49:22,220 --> 00:49:24,620
You see, when I wrote
that memorandum,

844
00:49:24,620 --> 00:49:26,480
as a matter of
courtesy, of course,

845
00:49:26,480 --> 00:49:30,740
I gave that memorandum to
the dean of engineering, who

846
00:49:30,740 --> 00:49:33,290
was Gordon Brown at that time.

847
00:49:33,290 --> 00:49:38,060
And not long later, I went
to see him about this.

848
00:49:38,060 --> 00:49:42,650
And he told his secretary
to get the memo out.

849
00:49:42,650 --> 00:49:44,270
You know what he told her to do?

850
00:49:44,270 --> 00:49:47,435
To file under FF, Fano's Folly.

851
00:49:47,435 --> 00:49:50,360
[LAUGHTER]

852
00:49:50,360 --> 00:49:52,610
So I either was going to
be called Fano's Folly,

853
00:49:52,610 --> 00:49:54,500
or I had to think
of another name.

854
00:49:54,500 --> 00:49:56,660
That's how Project MAC--

855
00:49:56,660 --> 00:50:02,360
within a day, Project
MAC, the name was coined.

856
00:50:02,360 --> 00:50:05,450
Now, MAC stood
originally for two--

857
00:50:05,450 --> 00:50:06,920
had two meanings.

858
00:50:06,920 --> 00:50:12,080
One, the goal, Machine-Aided
Cognition, and the other,

859
00:50:12,080 --> 00:50:17,690
the tool to achieve the
goal, Multiple-Axis Computer.

860
00:50:17,690 --> 00:50:21,260
Now, of course, a lot
of meanings were coined.

861
00:50:21,260 --> 00:50:24,050
One that was coined
pretty shortly was--

862
00:50:24,050 --> 00:50:27,410
and came from the West
Coast, obviously--

863
00:50:27,410 --> 00:50:29,510
was More Assets to Cambridge.

864
00:50:29,510 --> 00:50:33,430
[LAUGHTER]

865
00:50:33,430 --> 00:50:38,820


866
00:50:38,820 --> 00:50:47,160
Now, by October 1 of
1963, a copy of CTSS,

867
00:50:47,160 --> 00:50:50,400
as it existed in the
computation center,

868
00:50:50,400 --> 00:50:53,700
was installed and in
operation under Project MAC.

869
00:50:53,700 --> 00:50:56,130
And that becomes
the basis of what

870
00:50:56,130 --> 00:50:58,450
was later called a MAC system.

871
00:50:58,450 --> 00:51:04,230
It was originally just
plain a copy of CTSS,

872
00:51:04,230 --> 00:51:09,910
and then evolved with time.

873
00:51:09,910 --> 00:51:14,070
Now, I won't go through
all the evolution.

874
00:51:14,070 --> 00:51:19,440
By the end, it could support
something like 30 users.

875
00:51:19,440 --> 00:51:24,720
It rented for about
$62,000 a month

876
00:51:24,720 --> 00:51:29,250
with a 60% educational
discount, which

877
00:51:29,250 --> 00:51:36,090
meant a rental for something
like $155,000 per month.

878
00:51:36,090 --> 00:51:41,790
Now, let me show you
what it looked like.

879
00:51:41,790 --> 00:51:49,025


880
00:51:49,025 --> 00:51:50,790
Oh, I haven't done things right.

881
00:51:50,790 --> 00:51:53,640
I should have
switched these papers.

882
00:51:53,640 --> 00:51:59,660


883
00:51:59,660 --> 00:52:12,500
Now-- oh, up instead of down.

884
00:52:12,500 --> 00:52:15,200
Now, an interesting thing--
it has two banks of memory.

885
00:52:15,200 --> 00:52:20,750
Each one had 32,000
36-bit words.

886
00:52:20,750 --> 00:52:23,480
One was used for the
system supervisor.

887
00:52:23,480 --> 00:52:26,160
And the other was
available to the user.

888
00:52:26,160 --> 00:52:28,700
Now, you see, there is
a lot of data channel.

889
00:52:28,700 --> 00:52:30,340
There was a disk file there.

890
00:52:30,340 --> 00:52:34,940
There were drums to speed up the
swapping between core memory,

891
00:52:34,940 --> 00:52:37,220
in and out of core memory.

892
00:52:37,220 --> 00:52:41,870
There was also a card
punch, card reader for--

893
00:52:41,870 --> 00:52:46,580
to accommodate the past,
and magnetic tapes.

894
00:52:46,580 --> 00:52:48,322
There were two
graphical displays.

895
00:52:48,322 --> 00:52:50,280
And as a matter of fact,
it was more than that.

896
00:52:50,280 --> 00:52:52,140
And I'll come to
that in a moment.

897
00:52:52,140 --> 00:52:54,800
And then there was a
transmission control unit,

898
00:52:54,800 --> 00:52:58,100
which was called 7750.

899
00:52:58,100 --> 00:53:01,130
It was a huge thing.

900
00:53:01,130 --> 00:53:04,730
There was one person that
knew how to program it.

901
00:53:04,730 --> 00:53:06,755
And I still remember
his bald head.

902
00:53:06,755 --> 00:53:10,190


903
00:53:10,190 --> 00:53:15,140
Terribly difficult to program,
but it did the job beautifully

904
00:53:15,140 --> 00:53:16,550
when it worked.

905
00:53:16,550 --> 00:53:23,930
And that connected to the MIT,
a private branch exchange,

906
00:53:23,930 --> 00:53:27,980
so that we could dial in
from all over the place

907
00:53:27,980 --> 00:53:31,700
into the computer, and
30 lines eventually going

908
00:53:31,700 --> 00:53:33,410
to the private
branches exchange.

909
00:53:33,410 --> 00:53:39,920
We also connected to the Telex
international network operated

910
00:53:39,920 --> 00:53:46,610
by Western Union and TWX
national network operated

911
00:53:46,610 --> 00:53:53,710
by AT&T. Now, I've talked about
the computer, the operating

912
00:53:53,710 --> 00:53:54,210
system.

913
00:53:54,210 --> 00:53:57,223
But remember, to have good
time-share, you need terminals,

914
00:53:57,223 --> 00:53:58,390
and you need communications.

915
00:53:58,390 --> 00:54:00,490
And let me talk a
little bit about it,

916
00:54:00,490 --> 00:54:03,040
because it was
quite interesting.

917
00:54:03,040 --> 00:54:06,220
Well, originally,
we used something

918
00:54:06,220 --> 00:54:09,910
that was called the
Model 35 Teletype, which

919
00:54:09,910 --> 00:54:15,640
was used by the TWX AT&T
system for telegraphy.

920
00:54:15,640 --> 00:54:22,540


921
00:54:22,540 --> 00:54:24,360
Now, what you see
there, sitting--

922
00:54:24,360 --> 00:54:26,980
that's a Model 35 Teletype.

923
00:54:26,980 --> 00:54:30,940
And you see Dick Mills,
who was assistant director.

924
00:54:30,940 --> 00:54:33,340
He was a graduate of
this department that

925
00:54:33,340 --> 00:54:37,140
got a master's in management.

926
00:54:37,140 --> 00:54:41,980
After being in
Project MAC, he became

927
00:54:41,980 --> 00:54:46,030
head of information
processing services at MIT.

928
00:54:46,030 --> 00:54:48,970
And then he left for a
big bank job in New York.

929
00:54:48,970 --> 00:54:51,100
And he's now a
private consultant.

930
00:54:51,100 --> 00:54:56,320
Next to him is Oliver Selfridge,
who was, for a couple of years,

931
00:54:56,320 --> 00:54:58,780
associate director
of Project MAC.

932
00:54:58,780 --> 00:55:02,380
He was interested in
artificial intelligence.

933
00:55:02,380 --> 00:55:03,790
He was at Lincoln Laboratory.

934
00:55:03,790 --> 00:55:08,910
And he was loaned to Project MAC
to serve as associate director.

935
00:55:08,910 --> 00:55:12,750
Now, those are
the Model 35 type.

936
00:55:12,750 --> 00:55:14,620
There were other
people there also.

937
00:55:14,620 --> 00:55:16,930
You see, there is a
strange guy over there

938
00:55:16,930 --> 00:55:19,120
sitting at a Model 35.

939
00:55:19,120 --> 00:55:22,540
And that happened to be
Moses without a beard.

940
00:55:22,540 --> 00:55:25,840


941
00:55:25,840 --> 00:55:31,570
And I don't know
whether you were still

942
00:55:31,570 --> 00:55:34,870
a graduate student or just an
assistant professor-- graduate

943
00:55:34,870 --> 00:55:36,160
student still.

944
00:55:36,160 --> 00:55:38,830
Well, I stole it from your
book, so you ought to know.

945
00:55:38,830 --> 00:55:41,470


946
00:55:41,470 --> 00:55:44,410
So there were all sorts of
people in those terminals.

947
00:55:44,410 --> 00:55:45,820
[LAUGHTER]

948
00:55:45,820 --> 00:55:47,290
Everything.

949
00:55:47,290 --> 00:55:48,730
They went places, all right.

950
00:55:48,730 --> 00:55:53,170


951
00:55:53,170 --> 00:56:00,730
Now, I said that we were
connected to the Telex they

952
00:56:00,730 --> 00:56:01,780
brought it abroad.

953
00:56:01,780 --> 00:56:08,350
And in April '63, really shortly
after Project MAC was started,

954
00:56:08,350 --> 00:56:12,700
we had the first, I
believe, operation

955
00:56:12,700 --> 00:56:16,150
of a time-sharing system
across the Atlantic.

956
00:56:16,150 --> 00:56:18,670
And this is an
historical document.

957
00:56:18,670 --> 00:56:20,110
So let me show it.

958
00:56:20,110 --> 00:56:25,180
Sitting at the other end was
Professor Maurice Wilkes.

959
00:56:25,180 --> 00:56:31,270
Now, Wilkes is the man who built
at Cambridge the first storage

960
00:56:31,270 --> 00:56:33,100
program computer.

961
00:56:33,100 --> 00:56:38,170
He didn't invent it,
but he built it first.

962
00:56:38,170 --> 00:56:42,460
He retired from Cambridge when
he reached the critical age

963
00:56:42,460 --> 00:56:46,210
of 65, which was
a few years ago,

964
00:56:46,210 --> 00:56:51,508
and is now a part-time
professor in the department.

965
00:56:51,508 --> 00:56:53,050
I hope that some of
you have met him.

966
00:56:53,050 --> 00:56:55,110
He is called an
adjunct professor.

967
00:56:55,110 --> 00:56:56,710
The "adjunct" simply
means part time.

968
00:56:56,710 --> 00:56:59,750


969
00:56:59,750 --> 00:57:05,260
Now, this is what
it looked like.

970
00:57:05,260 --> 00:57:09,570
This is actual from the
printout that came--

971
00:57:09,570 --> 00:57:12,000
I don't know whether this
was the printout here

972
00:57:12,000 --> 00:57:13,530
or the printout in England.

973
00:57:13,530 --> 00:57:15,030
I'm not sure.

974
00:57:15,030 --> 00:57:17,370
But that's what it did.

975
00:57:17,370 --> 00:57:18,780
It gave a demonstration.

976
00:57:18,780 --> 00:57:21,150
And then at the end--

977
00:57:21,150 --> 00:57:25,350
this was at the meeting of
the British Computer Society

978
00:57:25,350 --> 00:57:28,260
and was an
enthusiastic reception.

979
00:57:28,260 --> 00:57:32,430
The president ran a program
himself and was all excited,

980
00:57:32,430 --> 00:57:35,040
all psyched up,
particularly because there

981
00:57:35,040 --> 00:57:37,290
was an electrical engineer
in the program who

982
00:57:37,290 --> 00:57:39,180
designed filters.

983
00:57:39,180 --> 00:57:40,640
And this is the end.

984
00:57:40,640 --> 00:57:44,280
And you see the message, all
this absolutely magnificent

985
00:57:44,280 --> 00:57:46,840
Edinburgh Computer Conference,
so forth and so on.

986
00:57:46,840 --> 00:57:50,010
So it was a great show.

987
00:57:50,010 --> 00:57:53,370
Now, you remember
those displays?

988
00:57:53,370 --> 00:57:56,070
Well, there were various
things connected there.

989
00:57:56,070 --> 00:57:59,490
One was a PDP-1.

990
00:57:59,490 --> 00:58:03,360
And look who is
sitting at the PDP-1--

991
00:58:03,360 --> 00:58:06,450
nobody less than Marvin Minsky.

992
00:58:06,450 --> 00:58:12,230
See, that PDP-1 was
connected to the computer.

993
00:58:12,230 --> 00:58:14,550
Now, accidentally,
I'm on the side there.

994
00:58:14,550 --> 00:58:15,600
But that doesn't matter.

995
00:58:15,600 --> 00:58:18,120


996
00:58:18,120 --> 00:58:20,450
Now, there was another terminal.

997
00:58:20,450 --> 00:58:24,270
Now, let me explain
a little bit here.

998
00:58:24,270 --> 00:58:26,310
Well, you see me
there without glasses,

999
00:58:26,310 --> 00:58:29,010
because the TV
didn't like glasses.

1000
00:58:29,010 --> 00:58:36,300
But what you have on the
right is a very, very powerful

1001
00:58:36,300 --> 00:58:38,760
display terminal for
computer-aided design,

1002
00:58:38,760 --> 00:58:42,060
very powerful in those days,
which was affectionately

1003
00:58:42,060 --> 00:58:45,120
known as the kluge.

1004
00:58:45,120 --> 00:58:47,250
John Watt, sitting
right over there,

1005
00:58:47,250 --> 00:58:50,100
was inside of that
whole development.

1006
00:58:50,100 --> 00:58:54,600
Now, the kluge was a
really wonderful machine.

1007
00:58:54,600 --> 00:58:58,590
You see that globe
on the side there?

1008
00:58:58,590 --> 00:59:00,870
Yeah, it's hard to see.

1009
00:59:00,870 --> 00:59:03,450
You could turn it
just like a joystick.

1010
00:59:03,450 --> 00:59:07,650
And what it did, if you
define in the computer

1011
00:59:07,650 --> 00:59:11,590
a three-dimensional object,
would make that object rotate.

1012
00:59:11,590 --> 00:59:14,790
That is, it had hardware
to make the transformation

1013
00:59:14,790 --> 00:59:18,250
of coordinates so that
you saw the object rotate

1014
00:59:18,250 --> 00:59:19,290
in front of you.

1015
00:59:19,290 --> 00:59:23,520
This was 1963.

1016
00:59:23,520 --> 00:59:29,040
Now, one of the people that
was quite effective this device

1017
00:59:29,040 --> 00:59:36,900
was professor Cy Leventhal in
the biology department, who

1018
00:59:36,900 --> 00:59:40,777
just one day came over to see me
and wanted to have some advice.

1019
00:59:40,777 --> 00:59:42,610
And I said, well, we'll
talk about it later.

1020
00:59:42,610 --> 00:59:44,820
I had something in mind.

1021
00:59:44,820 --> 00:59:45,810
Let me show you around.

1022
00:59:45,810 --> 00:59:46,770
So I took him here.

1023
00:59:46,770 --> 00:59:51,240
And as I expected, he
caught fire, real fire.

1024
00:59:51,240 --> 00:59:53,130
I was going skiing the next day.

1025
00:59:53,130 --> 00:59:55,890
And I had to make all the
arrangements so that he

1026
00:59:55,890 --> 00:59:57,870
could start the next day.

1027
00:59:57,870 --> 01:00:00,330
This was just before Christmas.

1028
01:00:00,330 --> 01:00:06,060
What he saw was the opportunity
to study complex molecules

1029
01:00:06,060 --> 01:00:08,080
in three dimensions.

1030
01:00:08,080 --> 01:00:09,320
And he did a lot of work.

1031
01:00:09,320 --> 01:00:11,160
He left for Columbia,
unfortunately,

1032
01:00:11,160 --> 01:00:12,520
because he's a New Yorker.

1033
01:00:12,520 --> 01:00:13,950
And he had an offer
from Columbia.

1034
01:00:13,950 --> 01:00:16,470
He had to go there.

1035
01:00:16,470 --> 01:00:18,180
But he kept doing
this sort of work.

1036
01:00:18,180 --> 01:00:22,160


1037
01:00:22,160 --> 01:00:25,710
Now, let me get back a bit
to the communication problem.

1038
01:00:25,710 --> 01:00:31,770
See, the Model 35 Teletype
had only capitals.

1039
01:00:31,770 --> 01:00:35,280
Everything in the computer
field was only capitals.

1040
01:00:35,280 --> 01:00:38,670
The big printers
were all capitals.

1041
01:00:38,670 --> 01:00:39,870
Well, we wanted lowercase.

1042
01:00:39,870 --> 01:00:42,810
[LAUGHTER]

1043
01:00:42,810 --> 01:00:45,210
Well, fortunately,
there was a little group

1044
01:00:45,210 --> 01:00:48,945
of people in Cambridge
area that had

1045
01:00:48,945 --> 01:00:54,660
formed something that was called
the reactive typewriter circle.

1046
01:00:54,660 --> 01:00:56,190
Now, I never know--

1047
01:00:56,190 --> 01:00:58,260
I don't know why they call it.

1048
01:00:58,260 --> 01:01:03,460
Was a man by the name of Calvin
Moore, who was a computer type.

1049
01:01:03,460 --> 01:01:05,790
And then there was
another man by the name

1050
01:01:05,790 --> 01:01:12,960
of Stewart Ferguson, who was
a a computer type, all right?

1051
01:01:12,960 --> 01:01:15,520
And they were beating
the drum for lowercase

1052
01:01:15,520 --> 01:01:17,470
and for other things.

1053
01:01:17,470 --> 01:01:20,820
So among other things,
they arranged for a visit

1054
01:01:20,820 --> 01:01:24,360
to the Teletype Corporation
in Skokie, Illinois.

1055
01:01:24,360 --> 01:01:29,190
And I asked the chief engineer
about getting lowercase in it.

1056
01:01:29,190 --> 01:01:33,190
He said, you know, we made
recently a survey of people,

1057
01:01:33,190 --> 01:01:36,411
whether they wanted lowercase.

1058
01:01:36,411 --> 01:01:37,680
It was negative.

1059
01:01:37,680 --> 01:01:38,970
They didn't want lowercase.

1060
01:01:38,970 --> 01:01:41,040
Of course, if you
ask anybody what

1061
01:01:41,040 --> 01:01:44,070
he would do with something
that that he never had before,

1062
01:01:44,070 --> 01:01:45,840
he can't think about.

1063
01:01:45,840 --> 01:01:48,210
Very few people
can envision what

1064
01:01:48,210 --> 01:01:50,300
they can do with a new tool.

1065
01:01:50,300 --> 01:01:58,980
So well, eventually, Teletype
produced the Model 37

1066
01:01:58,980 --> 01:02:00,930
that had also lowercase.

1067
01:02:00,930 --> 01:02:05,700
But in meantime, mother
IBM came to help.

1068
01:02:05,700 --> 01:02:10,170
IBM had gotten out, not
long before, the Selectric

1069
01:02:10,170 --> 01:02:13,090
typewriter-- you know,
the golf ball machine.

1070
01:02:13,090 --> 01:02:17,220
So it was not very hard
to put some electronics

1071
01:02:17,220 --> 01:02:19,705
and make it into a
computer terminal.

1072
01:02:19,705 --> 01:02:20,580
That's what they did.

1073
01:02:20,580 --> 01:02:22,830
And it was called
a 250, then 50.

1074
01:02:22,830 --> 01:02:25,470
And that was really
a major tool.

1075
01:02:25,470 --> 01:02:28,920
But let me get back
to communication.

1076
01:02:28,920 --> 01:02:32,490
Yeah, when we first approached
New England Telephone

1077
01:02:32,490 --> 01:02:36,090
about connecting a computer
to the switch network,

1078
01:02:36,090 --> 01:02:37,710
you can imagine what they said.

1079
01:02:37,710 --> 01:02:38,580
It's possible.

1080
01:02:38,580 --> 01:02:40,650
It's never been done.

1081
01:02:40,650 --> 01:02:43,200
In fact, that's what they said.

1082
01:02:43,200 --> 01:02:48,480
Fortunately, we got to some
very helpful people in AT&T.

1083
01:02:48,480 --> 01:02:52,320
And they really were very
helpful in organizing things.

1084
01:02:52,320 --> 01:02:55,600
One of the problems was
that the telephone business,

1085
01:02:55,600 --> 01:02:59,730
particularly in those
days, was highly regulated.

1086
01:02:59,730 --> 01:03:03,630
You couldn't do just
what you wanted.

1087
01:03:03,630 --> 01:03:06,330
So they found a way
of getting around.

1088
01:03:06,330 --> 01:03:08,790
And we started using
the WX network.

1089
01:03:08,790 --> 01:03:11,340


1090
01:03:11,340 --> 01:03:13,980
I gave demonstrations
from all over the place.

1091
01:03:13,980 --> 01:03:16,800
And they always worked,
except at Brandeis,

1092
01:03:16,800 --> 01:03:18,540
where they had the
wrong character set.

1093
01:03:18,540 --> 01:03:19,830
I don't know why.

1094
01:03:19,830 --> 01:03:23,580
But now, I have to
tell you a story.

1095
01:03:23,580 --> 01:03:29,520
And pardon me if it's
just slightly sexy--

1096
01:03:29,520 --> 01:03:31,440
sexist, I should say, not sexy.

1097
01:03:31,440 --> 01:03:33,660
[LAUGHTER]

1098
01:03:33,660 --> 01:03:38,910
Well, a person from AT&T that
was helping us a great deal

1099
01:03:38,910 --> 01:03:45,420
was a fellow by the name
of Earl Vaughn, who was

1100
01:03:45,420 --> 01:03:49,260
pretty high in AT&T, long line.

1101
01:03:49,260 --> 01:03:52,110
And I remember, I spoke
at one of the meetings

1102
01:03:52,110 --> 01:03:54,510
of the reactive
typewriter circle.

1103
01:03:54,510 --> 01:03:57,900
And he and my assistant
director, Dick Mill,

1104
01:03:57,900 --> 01:04:00,210
was sitting in the front row.

1105
01:04:00,210 --> 01:04:03,750
In those days, I made speeches
about the marriage of computers

1106
01:04:03,750 --> 01:04:05,490
and communication,
because that's

1107
01:04:05,490 --> 01:04:08,940
what we were trying to do, marry
computer and communication.

1108
01:04:08,940 --> 01:04:11,410
And Earl Vaughn, I
understand, turned

1109
01:04:11,410 --> 01:04:14,670
to Dick Mill, who was sitting
next to him, and said,

1110
01:04:14,670 --> 01:04:17,460
yeah, sure, a marriage of
computer and communication.

1111
01:04:17,460 --> 01:04:19,530
But who is going to
be the groom, and who

1112
01:04:19,530 --> 01:04:22,350
is going to be the bride?

1113
01:04:22,350 --> 01:04:24,550
Of course, that question
has never been settled

1114
01:04:24,550 --> 01:04:27,180
and will never be settled,
because the rules of the game

1115
01:04:27,180 --> 01:04:28,420
have been changed now.

1116
01:04:28,420 --> 01:04:34,510


1117
01:04:34,510 --> 01:04:38,223
All right, now let me get
to something, in a sense,

1118
01:04:38,223 --> 01:04:38,765
more serious.

1119
01:04:38,765 --> 01:04:42,090


1120
01:04:42,090 --> 01:04:48,060
Now, the goal of Project MAC was
to create an online community

1121
01:04:48,060 --> 01:04:52,170
and somewhat simulate
a computer utility.

1122
01:04:52,170 --> 01:04:54,360
That was really the goal.

1123
01:04:54,360 --> 01:04:58,560
Well, within six months,
we had achieved that goal.

1124
01:04:58,560 --> 01:05:01,350
That was incredible.

1125
01:05:01,350 --> 01:05:05,430
By spring '64,
there were something

1126
01:05:05,430 --> 01:05:12,240
like 100 terminals around
MIT, some 200 users

1127
01:05:12,240 --> 01:05:13,920
from 10 different departments.

1128
01:05:13,920 --> 01:05:15,820
All sorts of things
were happening.

1129
01:05:15,820 --> 01:05:19,770
One thing that I am very
sorry that I never did

1130
01:05:19,770 --> 01:05:23,550
was to hire real
smart psychologists,

1131
01:05:23,550 --> 01:05:27,330
social psychologists,
to really see

1132
01:05:27,330 --> 01:05:30,240
what was happening, because the
phenomena that were happening

1133
01:05:30,240 --> 01:05:33,540
were very, very interesting
from a sociological and

1134
01:05:33,540 --> 01:05:35,040
psychological standpoint.

1135
01:05:35,040 --> 01:05:37,620
Everything happened, from
people making friendship

1136
01:05:37,620 --> 01:05:43,410
to the computer, to
keyboards smashed by fists.

1137
01:05:43,410 --> 01:05:46,880
All sorts of things
happened, all right?

1138
01:05:46,880 --> 01:05:51,565
It was extremely,
extremely interesting.

1139
01:05:51,565 --> 01:05:55,660


1140
01:05:55,660 --> 01:05:59,930
Now, we also had regular users
from other universities coming

1141
01:05:59,930 --> 01:06:02,190
in from the Teletype network.

1142
01:06:02,190 --> 01:06:07,310
Now, let me show
you a paragraph.

1143
01:06:07,310 --> 01:06:13,230


1144
01:06:13,230 --> 01:06:15,035
Oh, there is a picture
that I didn't show.

1145
01:06:15,035 --> 01:06:18,570
And it's going to
mess up the sequence.

1146
01:06:18,570 --> 01:06:21,480
But I'll show it in a moment.

1147
01:06:21,480 --> 01:06:24,780
This is a paragraph from
a report that I wrote,

1148
01:06:24,780 --> 01:06:29,520
a paper that I wrote
in the spring '64,

1149
01:06:29,520 --> 01:06:31,380
just about six months.

1150
01:06:31,380 --> 01:06:33,480
It is becoming
increasingly clear

1151
01:06:33,480 --> 01:06:36,420
that the system's ability
to provide the equivalent

1152
01:06:36,420 --> 01:06:39,030
accessibility of
a private computer

1153
01:06:39,030 --> 01:06:43,770
is a secondary, although
necessary, characteristic.

1154
01:06:43,770 --> 01:06:46,270
What users find most
helpful is the fact

1155
01:06:46,270 --> 01:06:49,380
that the system places
literally at their fingertips

1156
01:06:49,380 --> 01:06:51,480
a great variety of services.

1157
01:06:51,480 --> 01:06:55,980
That was the important thing
that became clear immediately.

1158
01:06:55,980 --> 01:06:58,200
The system users
themselves are beginning

1159
01:06:58,200 --> 01:07:01,860
to contribute to the
system in a substantial way

1160
01:07:01,860 --> 01:07:05,250
by publishing their work in
the form of new commands.

1161
01:07:05,250 --> 01:07:06,870
Now, this was an amazing thing.

1162
01:07:06,870 --> 01:07:09,390
Before that time,
nobody, nobody,

1163
01:07:09,390 --> 01:07:13,600
ever wrote a program expecting
anybody else to use it.

1164
01:07:13,600 --> 01:07:16,140
It was too awkward.

1165
01:07:16,140 --> 01:07:18,210
Suddenly, it became easy.

1166
01:07:18,210 --> 01:07:22,110
And people started looking
at a program as a publication

1167
01:07:22,110 --> 01:07:25,587
and make it-- specifically
try to design it so that they

1168
01:07:25,587 --> 01:07:26,670
could use by other people.

1169
01:07:26,670 --> 01:07:30,510
It was just a phenomenon
that occurred very quickly

1170
01:07:30,510 --> 01:07:33,390
and was very important.

1171
01:07:33,390 --> 01:07:36,330
Now, as a matter of
fact, an editorial board

1172
01:07:36,330 --> 01:07:38,700
is being established
to review such work

1173
01:07:38,700 --> 01:07:41,280
and formally approve its
inclusion in the system.

1174
01:07:41,280 --> 01:07:44,340
We established an
editorial board

1175
01:07:44,340 --> 01:07:45,840
and worked very hard on it.

1176
01:07:45,840 --> 01:07:48,630
Thus, the system is
beginning to become

1177
01:07:48,630 --> 01:07:51,990
the repository of the
procedural and data knowledge

1178
01:07:51,990 --> 01:07:53,790
of the community that it serves.

1179
01:07:53,790 --> 01:07:58,890
This last sentence is indeed
the purpose of time-sharing.

1180
01:07:58,890 --> 01:07:59,890
That is it.

1181
01:07:59,890 --> 01:08:03,310
It's a communication
system between users.

1182
01:08:03,310 --> 01:08:04,890
That's the primary function.

1183
01:08:04,890 --> 01:08:07,560
The process is almost
secondary to it.

1184
01:08:07,560 --> 01:08:10,860
That's a point I want to
make very, very strongly.

1185
01:08:10,860 --> 01:08:15,810
Now, lots of things
were developed.

1186
01:08:15,810 --> 01:08:18,850
I want to mention
a particular one.

1187
01:08:18,850 --> 01:08:21,930
And let me show you a picture.

1188
01:08:21,930 --> 01:08:24,510
Well, I have to skip
a couple of them.

1189
01:08:24,510 --> 01:08:28,250


1190
01:08:28,250 --> 01:08:29,460
What is this?

1191
01:08:29,460 --> 01:08:31,350
Oh yeah.

1192
01:08:31,350 --> 01:08:32,670
All right.

1193
01:08:32,670 --> 01:08:37,800
The fellow on the side is
Professor Jerry Saltzer.

1194
01:08:37,800 --> 01:08:40,350
He was a graduate student there.

1195
01:08:40,350 --> 01:08:42,585
And when he was a
graduate student,

1196
01:08:42,585 --> 01:08:46,140
he wrote two very
important programs

1197
01:08:46,140 --> 01:08:48,660
called Typeset and Runoff.

1198
01:08:48,660 --> 01:08:55,680
Typeset was the first context
editor ever written, right?

1199
01:08:55,680 --> 01:08:56,760
It was fantastic.

1200
01:08:56,760 --> 01:09:07,590
And Runoff was a primordial
scribe, you know?

1201
01:09:07,590 --> 01:09:09,930
But was quite capable already.

1202
01:09:09,930 --> 01:09:14,550
In fact, we used that system
to write all our reports

1203
01:09:14,550 --> 01:09:16,270
and for a lot of other purposes.

1204
01:09:16,270 --> 01:09:20,310
I even used Typeset for
accounting purposes.

1205
01:09:20,310 --> 01:09:22,319
I won't bother to tell you how.

1206
01:09:22,319 --> 01:09:24,149
But it was very
powerful in many ways.

1207
01:09:24,149 --> 01:09:28,560
Now, it's interesting
how he wrote it.

1208
01:09:28,560 --> 01:09:31,512


1209
01:09:31,512 --> 01:09:36,060
All of you who have done any
writing or any work that really

1210
01:09:36,060 --> 01:09:40,529
requires a lot of concentration
know that when, at some point,

1211
01:09:40,529 --> 01:09:44,130
you say, I got to do it, I
got to do it, I got to do it,

1212
01:09:44,130 --> 01:09:46,529
then you pick out
papers, a pencil.

1213
01:09:46,529 --> 01:09:49,260
And you go and
sharpen one by one,

1214
01:09:49,260 --> 01:09:53,490
trying to delay the time when
you really have to think.

1215
01:09:53,490 --> 01:09:55,890
[LAUGHTER]

1216
01:09:55,890 --> 01:09:58,290
I've done that, I know.

1217
01:09:58,290 --> 01:10:02,920
Jerry Saltzer wrote Typeset
and Runoff in that spirit.

1218
01:10:02,920 --> 01:10:05,220
He had to write his
thesis proposal.

1219
01:10:05,220 --> 01:10:09,020
[LAUGHTER]

1220
01:10:09,020 --> 01:10:11,400


1221
01:10:11,400 --> 01:10:13,740
Now, this is real history.

1222
01:10:13,740 --> 01:10:14,980
It's a fact.

1223
01:10:14,980 --> 01:10:23,260
Well, now, let me show
you something else.

1224
01:10:23,260 --> 01:10:25,600
Time is flying, but let me
show you something else.

1225
01:10:25,600 --> 01:10:29,920


1226
01:10:29,920 --> 01:10:30,420
Oh, no.

1227
01:10:30,420 --> 01:10:32,870
I want this.

1228
01:10:32,870 --> 01:10:38,180
Now, there was this user
community around the computer.

1229
01:10:38,180 --> 01:10:42,200
And in a paper that Corby and I
wrote for Scientific American,

1230
01:10:42,200 --> 01:10:45,830
I tried to make a
sketch of the user's

1231
01:10:45,830 --> 01:10:48,770
view of this marvelous
time-sharing system.

1232
01:10:48,770 --> 01:10:51,900


1233
01:10:51,900 --> 01:10:55,290
You see what I see?

1234
01:10:55,290 --> 01:10:57,390
That's a network of
personal computers.

1235
01:10:57,390 --> 01:11:00,550


1236
01:11:00,550 --> 01:11:02,560
Well, they are called
pseudo-processor and

1237
01:11:02,560 --> 01:11:03,820
pseudo-memory.

1238
01:11:03,820 --> 01:11:06,370
Take the "pseudos" out.

1239
01:11:06,370 --> 01:11:08,980
They have a file
server, of course.

1240
01:11:08,980 --> 01:11:11,560
They have a message center.

1241
01:11:11,560 --> 01:11:15,040
They have also some services.

1242
01:11:15,040 --> 01:11:20,620
What DAEMON was was like
a user, automatic user.

1243
01:11:20,620 --> 01:11:23,920
What it did, when anybody
logged out and went

1244
01:11:23,920 --> 01:11:26,310
to look at the file, if
anything has changed,

1245
01:11:26,310 --> 01:11:30,730
it was copying it in background,
was copying it and back up

1246
01:11:30,730 --> 01:11:32,740
automatically.

1247
01:11:32,740 --> 01:11:40,060
FIBMON Was Foreground
Initiated Background.

1248
01:11:40,060 --> 01:11:43,030
What it did, from
a typewriter, you

1249
01:11:43,030 --> 01:11:45,850
could write a program
when the computer had time

1250
01:11:45,850 --> 01:11:47,460
and put it in my file.

1251
01:11:47,460 --> 01:11:49,720
And I'll look at it tomorrow.

1252
01:11:49,720 --> 01:11:55,330
I combined online use
with batch processes.

1253
01:11:55,330 --> 01:11:59,170
This was more traditional for.

1254
01:11:59,170 --> 01:12:00,310
No, DAEMON was that?

1255
01:12:00,310 --> 01:12:02,110
Yeah.

1256
01:12:02,110 --> 01:12:03,983
What did I have
in the BACKGROUND?

1257
01:12:03,983 --> 01:12:05,650
BACKGROUND was a
traditional background.

1258
01:12:05,650 --> 01:12:08,110
Was a traditional
background, yeah,

1259
01:12:08,110 --> 01:12:12,190
where you just ran the program
when the computer had time,

1260
01:12:12,190 --> 01:12:13,690
right?

1261
01:12:13,690 --> 01:12:20,810
Now, I was surprised
when I saw it.

1262
01:12:20,810 --> 01:12:27,560
It is, has the geometry of a
network of personal computers.

1263
01:12:27,560 --> 01:12:32,750
As far as I know, none
with those features

1264
01:12:32,750 --> 01:12:35,120
are in operation today.

1265
01:12:35,120 --> 01:12:38,810
I'll come back to that, although
I understand from Professor

1266
01:12:38,810 --> 01:12:42,980
Saltzer that indeed,
that is sort of a goal

1267
01:12:42,980 --> 01:12:45,980
he has for Project Attila.

1268
01:12:45,980 --> 01:12:50,910
Now, let me move on.

1269
01:12:50,910 --> 01:12:53,950


1270
01:12:53,950 --> 01:12:56,290
Very early in Project
MAC, we started

1271
01:12:56,290 --> 01:12:59,290
planning for the next system.

1272
01:12:59,290 --> 01:13:03,250
It was quite clear that it
was unbelievable what you

1273
01:13:03,250 --> 01:13:06,490
could do with your old 7094.

1274
01:13:06,490 --> 01:13:08,680
But that machine was
certainly never designed

1275
01:13:08,680 --> 01:13:09,970
for time-sharing.

1276
01:13:09,970 --> 01:13:11,860
So we wanted to get a new one.

1277
01:13:11,860 --> 01:13:14,030
So we started talking
with manufacturers

1278
01:13:14,030 --> 01:13:18,510
and so forth to find
the right machine.

1279
01:13:18,510 --> 01:13:22,770
Well, during one of
the visits at UNIVAC,

1280
01:13:22,770 --> 01:13:26,730
you see some strange character
over there, Professor Dennis.

1281
01:13:26,730 --> 01:13:30,720
Stand up so they can
see you in the corner.

1282
01:13:30,720 --> 01:13:34,780
There's Corby, whom you know.

1283
01:13:34,780 --> 01:13:39,870
And that is Arnold
Cohen from UNIVAC.

1284
01:13:39,870 --> 01:13:42,315
We invested quite a bit
of time in looking around.

1285
01:13:42,315 --> 01:13:44,860


1286
01:13:44,860 --> 01:13:48,370
We also did a lot of thinking.

1287
01:13:48,370 --> 01:13:52,470
And from our
experience with CTSS,

1288
01:13:52,470 --> 01:13:56,970
it became quite clear what
the general organization

1289
01:13:56,970 --> 01:13:59,010
of a computer
utility ought to be.

1290
01:13:59,010 --> 01:14:02,520


1291
01:14:02,520 --> 01:14:10,140
First of all, let me
give you a sketch.

1292
01:14:10,140 --> 01:14:13,680
This is what is in the
program that was given to you.

1293
01:14:13,680 --> 01:14:15,040
There were several ideas.

1294
01:14:15,040 --> 01:14:17,650
First of all, it is a
memory-centered system.

1295
01:14:17,650 --> 01:14:23,010
This, incidentally, was also
in the 1965 Scientific American

1296
01:14:23,010 --> 01:14:25,290
article.

1297
01:14:25,290 --> 01:14:28,980
That's where it comes from.

1298
01:14:28,980 --> 01:14:30,810
It's a memory-centered system.

1299
01:14:30,810 --> 01:14:32,870
All the communication
goes through the memory.

1300
01:14:32,870 --> 01:14:35,550
That was a very
important view that

1301
01:14:35,550 --> 01:14:37,110
had come out of experience.

1302
01:14:37,110 --> 01:14:40,890
That's how the user viewed
the system, as a memory,

1303
01:14:40,890 --> 01:14:44,490
with a processor poking into it.

1304
01:14:44,490 --> 01:14:46,440
That was very important.

1305
01:14:46,440 --> 01:14:50,460
Also, as you see, there
are pools, more than one,

1306
01:14:50,460 --> 01:14:51,870
every single box.

1307
01:14:51,870 --> 01:14:54,690
There are two processors,
four banks of memory,

1308
01:14:54,690 --> 01:14:57,090
two drum controllers,
two disk controllers,

1309
01:14:57,090 --> 01:14:58,660
two general controls.

1310
01:14:58,660 --> 01:15:00,250
But it was a pool.

1311
01:15:00,250 --> 01:15:04,410
And the idea was that if
any one went bad, well,

1312
01:15:04,410 --> 01:15:06,030
you could run it.

1313
01:15:06,030 --> 01:15:08,440
This was, again, the
idea of a public utility.

1314
01:15:08,440 --> 01:15:11,250
You don't stop to fix things.

1315
01:15:11,250 --> 01:15:12,960
You keep running.

1316
01:15:12,960 --> 01:15:18,000
Now, as a matter of fact, we
established certain goals.

1317
01:15:18,000 --> 01:15:19,690
And I want to put
them on the slide

1318
01:15:19,690 --> 01:15:22,680
so you can see what they are.

1319
01:15:22,680 --> 01:15:26,020
Continuous operation--
that's public utility.

1320
01:15:26,020 --> 01:15:28,530


1321
01:15:28,530 --> 01:15:31,140
Pools of identical units.

1322
01:15:31,140 --> 01:15:35,610
We wanted to reconfigure the
system online without stopping.

1323
01:15:35,610 --> 01:15:38,910


1324
01:15:38,910 --> 01:15:42,700
Communication to memory--
this was a technical point.

1325
01:15:42,700 --> 01:15:45,720
You want a clean
communication system.

1326
01:15:45,720 --> 01:15:50,145
Now, the hierarchy of memories
with automated migration

1327
01:15:50,145 --> 01:15:51,780
of program-- these
are all things

1328
01:15:51,780 --> 01:15:58,020
that you see now, but
didn't exist in those days.

1329
01:15:58,020 --> 01:16:04,140
Now, to save time, I want to
go to this point, execution

1330
01:16:04,140 --> 01:16:05,400
of unbound modules.

1331
01:16:05,400 --> 01:16:09,000
This is what is often
called segmentation,

1332
01:16:09,000 --> 01:16:21,460
or the two-dimensional
addressing.

1333
01:16:21,460 --> 01:16:27,460
Now, that idea, if
you really look back,

1334
01:16:27,460 --> 01:16:31,870
came up in conjunction with the
design of the B-5000 machine

1335
01:16:31,870 --> 01:16:33,040
way, way back.

1336
01:16:33,040 --> 01:16:36,370
But as far as I know, the
person who really understood

1337
01:16:36,370 --> 01:16:42,160
the implication of this was
Jack Dennis, I mentioned before.

1338
01:16:42,160 --> 01:16:44,290
As a matter of fact,
I remember vividly

1339
01:16:44,290 --> 01:16:50,770
a scene that took place
toward late spring

1340
01:16:50,770 --> 01:16:54,460
or early summer '63.

1341
01:16:54,460 --> 01:16:58,510
Jack was giving a talk
on program segmentation

1342
01:16:58,510 --> 01:17:01,660
for an [INAUDIBLE] program.

1343
01:17:01,660 --> 01:17:05,950
And I was sitting next to
Corby in the front row.

1344
01:17:05,950 --> 01:17:08,350
I was the chairman
of the affair.

1345
01:17:08,350 --> 01:17:10,180
At a certain point,
Corby started

1346
01:17:10,180 --> 01:17:12,070
getting all excited--
yes, yes, that's

1347
01:17:12,070 --> 01:17:13,420
right, what you're doing.

1348
01:17:13,420 --> 01:17:15,780
Yes, yes, yeah--
really got excited.

1349
01:17:15,780 --> 01:17:20,820
That that's how segmentation
got into the Multics system.

1350
01:17:20,820 --> 01:17:25,780
Sometimes communication takes
place in a strange occurrences.

1351
01:17:25,780 --> 01:17:29,740
Well, I think that
some of you may not

1352
01:17:29,740 --> 01:17:33,310
know what a segment
of visual memory is.

1353
01:17:33,310 --> 01:17:36,070
It's a great idea.

1354
01:17:36,070 --> 01:17:38,620
I believe so.

1355
01:17:38,620 --> 01:17:41,650
Segments are distinct
procedural data modules.

1356
01:17:41,650 --> 01:17:46,170
They are the basic
unit of computation.

1357
01:17:46,170 --> 01:17:48,497
Segments are files.

1358
01:17:48,497 --> 01:17:50,580
There's no distinction
between segments and files.

1359
01:17:50,580 --> 01:17:51,570
A file is a segment.

1360
01:17:51,570 --> 01:17:53,310
A segment is a file.

1361
01:17:53,310 --> 01:17:57,630
It's choice, individual.

1362
01:17:57,630 --> 01:18:01,650
You are using, essentially,
two-dimensional addressing.

1363
01:18:01,650 --> 01:18:06,490
That is, one dimension is
the name of the segment.

1364
01:18:06,490 --> 01:18:11,610
The second dimension is the
location, just like a book.

1365
01:18:11,610 --> 01:18:13,680
If you want to make a
reference to a book,

1366
01:18:13,680 --> 01:18:16,170
you give the name of
the book and the page

1367
01:18:16,170 --> 01:18:18,930
in the book-- you know,
location in the book.

1368
01:18:18,930 --> 01:18:21,060
Seems very straightforward.

1369
01:18:21,060 --> 01:18:26,610
Then segments are divided into
pages, just like a book, right?

1370
01:18:26,610 --> 01:18:32,060
That's the paging mechanism
that you are all familiar with.

1371
01:18:32,060 --> 01:18:34,950
Now, paging is not
seen by the user.

1372
01:18:34,950 --> 01:18:38,870
It just goes on
automatically in the machine.

1373
01:18:38,870 --> 01:18:42,680
But the users are very
conscious of segment.

1374
01:18:42,680 --> 01:18:44,735
There are names that
they know and they use.

1375
01:18:44,735 --> 01:18:47,910


1376
01:18:47,910 --> 01:18:50,850
The very important
thing is that segments

1377
01:18:50,850 --> 01:18:55,080
are bound to one another in a
computation and execution time.

1378
01:18:55,080 --> 01:18:58,680
You don't have to
pre-prepare anything.

1379
01:18:58,680 --> 01:19:03,590
You don't have to load
programs together.

1380
01:19:03,590 --> 01:19:05,700
They're separate things.

1381
01:19:05,700 --> 01:19:10,500
Now, the analogies that I often
use is, if you read a book,

1382
01:19:10,500 --> 01:19:14,400
it makes reference to
other books to paper.

1383
01:19:14,400 --> 01:19:17,880
What is the traditional
way of loading program

1384
01:19:17,880 --> 01:19:20,400
will correspond that
you give a book.

1385
01:19:20,400 --> 01:19:21,990
You have a book.

1386
01:19:21,990 --> 01:19:25,530
You give it to somebody to
Xerox all the paper, and all

1387
01:19:25,530 --> 01:19:30,930
the pages, and everything
that the book refers to.

1388
01:19:30,930 --> 01:19:33,630
Then they are all
assembled together.

1389
01:19:33,630 --> 01:19:36,180
And the references are
changed to make reference

1390
01:19:36,180 --> 01:19:37,480
to the various things.

1391
01:19:37,480 --> 01:19:39,720
And then you read it.

1392
01:19:39,720 --> 01:19:42,150
That's the normal loading.

1393
01:19:42,150 --> 01:19:48,660
This is what we actually do
with books and papers, right?

1394
01:19:48,660 --> 01:19:50,340
Now, there are lots
of other things

1395
01:19:50,340 --> 01:19:51,820
that I don't want to get into.

1396
01:19:51,820 --> 01:19:55,290
But that, putting paging
and segmentation together

1397
01:19:55,290 --> 01:19:58,710
was a real difficult goal.

1398
01:19:58,710 --> 01:20:01,440
Well, Multics project.

1399
01:20:01,440 --> 01:20:04,160


1400
01:20:04,160 --> 01:20:08,940
Well, we talked with all
sorts of manufacturers.

1401
01:20:08,940 --> 01:20:13,280
Eventually, three companies,
IBM, DEC, and General Electric,

1402
01:20:13,280 --> 01:20:14,720
made a proposal.

1403
01:20:14,720 --> 01:20:18,770
It turns out that the
architecture of the new 360

1404
01:20:18,770 --> 01:20:23,180
system and the architecture
of the PDP-6 just plain

1405
01:20:23,180 --> 01:20:25,220
were not suitable for our goal.

1406
01:20:25,220 --> 01:20:27,740
We wanted the
memory-centered system.

1407
01:20:27,740 --> 01:20:30,810
And those were
processor-centered system.

1408
01:20:30,810 --> 01:20:32,060
So we went General Electric.

1409
01:20:32,060 --> 01:20:35,510
It was a memory-centered system.

1410
01:20:35,510 --> 01:20:38,300
But a lot of hardware
still had to be

1411
01:20:38,300 --> 01:20:43,140
designed to get what we wanted.

1412
01:20:43,140 --> 01:20:47,900
And I wanted to mention another
name, the name of Ted Glaser,

1413
01:20:47,900 --> 01:20:50,600
who was a member of our
faculty at that time.

1414
01:20:50,600 --> 01:20:54,110
He was a superb designer.

1415
01:20:54,110 --> 01:20:55,830
Incidentally, he was also--

1416
01:20:55,830 --> 01:20:58,790
he is blind.

1417
01:20:58,790 --> 01:21:01,490
But he had overcompensated--

1418
01:21:01,490 --> 01:21:05,960
I tell you, really
overcompensated for being

1419
01:21:05,960 --> 01:21:06,540
blind.

1420
01:21:06,540 --> 01:21:09,120
He had a memory
that was fantastic

1421
01:21:09,120 --> 01:21:12,260
and an ability to find--

1422
01:21:12,260 --> 01:21:15,110
I went with him to
visit the company.

1423
01:21:15,110 --> 01:21:17,750
And he was-- the
way he said it, he

1424
01:21:17,750 --> 01:21:22,190
was looking at a
mechanical device.

1425
01:21:22,190 --> 01:21:24,230
Somebody opened it
and just went out,

1426
01:21:24,230 --> 01:21:25,820
where they said, ah-ha, ah-ha.

1427
01:21:25,820 --> 01:21:28,370
Oh, that's what you do, yeah.

1428
01:21:28,370 --> 01:21:32,750
He had understood the
operation of that machine--

1429
01:21:32,750 --> 01:21:33,770
just incredible.

1430
01:21:33,770 --> 01:21:37,250
Well, Ted Glaser was
a major contributor

1431
01:21:37,250 --> 01:21:40,670
to what turned out to
be the Multics system.

1432
01:21:40,670 --> 01:21:43,230


1433
01:21:43,230 --> 01:21:47,090
Unfortunately, he was
not here to the end.

1434
01:21:47,090 --> 01:21:50,300
He got offered chairmanship
of the computer science

1435
01:21:50,300 --> 01:21:50,910
department.

1436
01:21:50,910 --> 01:21:53,210
So he left.

1437
01:21:53,210 --> 01:21:59,940
Well, the Multics project, which
had those ambitious goals--

1438
01:21:59,940 --> 01:22:04,610


1439
01:22:04,610 --> 01:22:05,960
I don't know when it started.

1440
01:22:05,960 --> 01:22:10,130
I would say '64, we decided
we'd get that equipment.

1441
01:22:10,130 --> 01:22:14,120
In 1965, we had the four
joint computer conferences.

1442
01:22:14,120 --> 01:22:17,030
A series of papers on
the design of the system

1443
01:22:17,030 --> 01:22:19,220
were presented-- the same four.

1444
01:22:19,220 --> 01:22:21,320
The hardware was frozen.

1445
01:22:21,320 --> 01:22:25,430
To make a long story
short, it finally

1446
01:22:25,430 --> 01:22:28,640
got into operation
in November '69,

1447
01:22:28,640 --> 01:22:33,440
two years later, because you
have no idea what a project

1448
01:22:33,440 --> 01:22:34,730
it was.

1449
01:22:34,730 --> 01:22:38,840
Everything had to be
done from scratch.

1450
01:22:38,840 --> 01:22:40,670
There were no software tools.

1451
01:22:40,670 --> 01:22:44,210
When the machine arrived,
it had to be debugged.

1452
01:22:44,210 --> 01:22:47,270
A compiler had to
be written for PL1,

1453
01:22:47,270 --> 01:22:51,150
because it had to be
written all in PL1

1454
01:22:51,150 --> 01:22:53,870
It was a major undertaking.

1455
01:22:53,870 --> 01:23:00,560
It was completed, I say, quite
honestly, because of Corby.

1456
01:23:00,560 --> 01:23:03,470
He is the only leader
in the team involved

1457
01:23:03,470 --> 01:23:06,770
with three organizations,
Project MAC, the Bell Telephone

1458
01:23:06,770 --> 01:23:09,320
Laboratories, and General
Electric, and later,

1459
01:23:09,320 --> 01:23:10,760
Honeywell--

1460
01:23:10,760 --> 01:23:13,160
was the only leader
of that effort

1461
01:23:13,160 --> 01:23:19,190
that was there at the beginning
and was there at the end.

1462
01:23:19,190 --> 01:23:21,720
I want to tell you
something else about Corby,

1463
01:23:21,720 --> 01:23:24,260
because it's important.

1464
01:23:24,260 --> 01:23:28,850
When people design
complicated things,

1465
01:23:28,850 --> 01:23:31,190
they say, now-- they scratch
their head, and they say,

1466
01:23:31,190 --> 01:23:33,470
now, suppose if this happens.

1467
01:23:33,470 --> 01:23:34,805
What should I do?

1468
01:23:34,805 --> 01:23:37,560
Corby never does that.

1469
01:23:37,560 --> 01:23:41,990
He is a firm believer
in Murphy's law.

1470
01:23:41,990 --> 01:23:45,090
Everything that could
go wrong will go on.

1471
01:23:45,090 --> 01:23:48,350
So he always says,
when this will--

1472
01:23:48,350 --> 01:23:51,780
when this happens,
what will I do?

1473
01:23:51,780 --> 01:23:55,010
And that is the secret of
the time-sharing system

1474
01:23:55,010 --> 01:23:57,440
that he has developed.

1475
01:23:57,440 --> 01:24:01,160
He always knows what
to do with that.

1476
01:24:01,160 --> 01:24:04,790
Now, let me make a couple
concluding remarks,

1477
01:24:04,790 --> 01:24:06,440
because time is really flying.

1478
01:24:06,440 --> 01:24:12,190


1479
01:24:12,190 --> 01:24:16,210
We are going through a
microprocessor revolution.

1480
01:24:16,210 --> 01:24:18,910
Now, essentially,
microprocessors

1481
01:24:18,910 --> 01:24:24,250
having invalidated what used
to be called Grosch's law.

1482
01:24:24,250 --> 01:24:27,400
Now, Grosch's law,
way back, said

1483
01:24:27,400 --> 01:24:32,920
that you always get a
bigger bang per buck

1484
01:24:32,920 --> 01:24:37,540
with a bigger machine, right?

1485
01:24:37,540 --> 01:24:38,790
That's what it said.

1486
01:24:38,790 --> 01:24:43,470
And that was, to a good
part, the rationale

1487
01:24:43,470 --> 01:24:46,860
for time-sharing a large
machine rather than

1488
01:24:46,860 --> 01:24:49,560
having separate computers.

1489
01:24:49,560 --> 01:24:51,630
That was the rationale.

1490
01:24:51,630 --> 01:24:54,820
Now, clearly, Grosch's
law is no longer valid.

1491
01:24:54,820 --> 01:24:58,050
So this is just changing
the entire technological

1492
01:24:58,050 --> 01:25:00,840
environment in which
time-sharing grew up.

1493
01:25:00,840 --> 01:25:04,920
But the technology
of implementation

1494
01:25:04,920 --> 01:25:05,890
may be different.

1495
01:25:05,890 --> 01:25:08,700
But the goal is still valid--

1496
01:25:08,700 --> 01:25:14,370
communication between people,
sharing, online computation.

1497
01:25:14,370 --> 01:25:15,810
That is still valid.

1498
01:25:15,810 --> 01:25:18,990
But you will probably
want to implement

1499
01:25:18,990 --> 01:25:21,660
in the future in
a different way.

1500
01:25:21,660 --> 01:25:25,590
You may want to have
the computational power

1501
01:25:25,590 --> 01:25:28,260
at the terminal
rather than central.

1502
01:25:28,260 --> 01:25:30,120
But you still need mass memory.

1503
01:25:30,120 --> 01:25:31,740
You still need communication.

1504
01:25:31,740 --> 01:25:35,010


1505
01:25:35,010 --> 01:25:39,600
Well, finally, I think
I ought to give you

1506
01:25:39,600 --> 01:25:44,310
my view on the
successes and failures

1507
01:25:44,310 --> 01:25:50,370
of the time-sharing movement
and our effort at MIT.

1508
01:25:50,370 --> 01:25:54,570
I think, from a
research point of view,

1509
01:25:54,570 --> 01:25:58,320
I think the
time-sharing work at MIT

1510
01:25:58,320 --> 01:26:03,570
was a major, major
contribution, in many ways.

1511
01:26:03,570 --> 01:26:06,840
The variety of techniques
that were developed just

1512
01:26:06,840 --> 01:26:08,100
went out in the field.

1513
01:26:08,100 --> 01:26:10,860
And you find them today
all over the place.

1514
01:26:10,860 --> 01:26:13,855
Virtual memory, in
some sense, was started

1515
01:26:13,855 --> 01:26:15,510
in those time-sharing systems.

1516
01:26:15,510 --> 01:26:18,450
Now everybody uses
virtual memory.

1517
01:26:18,450 --> 01:26:23,670
And a lot of techniques went
out and are in actual use.

1518
01:26:23,670 --> 01:26:26,970
It certainly fostered
research, online research,

1519
01:26:26,970 --> 01:26:30,060
in universities,
certainly did at MIT.

1520
01:26:30,060 --> 01:26:34,640
There was really a sudden growth
of computer-based research

1521
01:26:34,640 --> 01:26:36,440
at MIT in the '60s.

1522
01:26:36,440 --> 01:26:39,520
There is no question about it.

1523
01:26:39,520 --> 01:26:45,300
It did provide people
with physical access

1524
01:26:45,300 --> 01:26:49,830
to a computer from
wherever they might be.

1525
01:26:49,830 --> 01:26:52,140
Certainly, it did that.

1526
01:26:52,140 --> 01:26:56,910
It failed to provide the
corresponding intellectual

1527
01:26:56,910 --> 01:26:59,850
access from wherever
people might be.

1528
01:26:59,850 --> 01:27:03,720
The reason is that when
you work on this system,

1529
01:27:03,720 --> 01:27:06,420
you're always in
a learning mode.

1530
01:27:06,420 --> 01:27:08,790
And it makes a great
deal of difference

1531
01:27:08,790 --> 01:27:11,250
whether sitting
in the next office

1532
01:27:11,250 --> 01:27:13,680
is somebody that has used
the program that you're

1533
01:27:13,680 --> 01:27:17,700
going to use next, or you
are isolated miles away

1534
01:27:17,700 --> 01:27:22,200
from anybody that knows
what you would like to do.

1535
01:27:22,200 --> 01:27:25,830
Well, at one point, we tried
to do something about that.

1536
01:27:25,830 --> 01:27:28,420
But it never came out.

1537
01:27:28,420 --> 01:27:31,440
So the intellectual access
is still a problem today.

1538
01:27:31,440 --> 01:27:36,830


1539
01:27:36,830 --> 01:27:39,485
While it was successful,
the research community

1540
01:27:39,485 --> 01:27:42,890
at MIT, online
community, I think

1541
01:27:42,890 --> 01:27:47,640
that the movement
failed to create

1542
01:27:47,640 --> 01:27:53,730
online communities,
certainly outside university,

1543
01:27:53,730 --> 01:27:56,730
and even within universities.

1544
01:27:56,730 --> 01:28:01,440
That is, for instance, we
didn't have an online student

1545
01:28:01,440 --> 01:28:04,250
community at MIT.

1546
01:28:04,250 --> 01:28:07,940
We are just beginning
now to have some.

1547
01:28:07,940 --> 01:28:09,210
It was just [INAUDIBLE].

1548
01:28:09,210 --> 01:28:12,420
The major reason is cost--

1549
01:28:12,420 --> 01:28:15,320
has been too
expensive until now.

1550
01:28:15,320 --> 01:28:18,510
We are just getting now to the
time in which it's cheap enough

1551
01:28:18,510 --> 01:28:19,560
to do it.

1552
01:28:19,560 --> 01:28:23,550
But there is another reason if
you go outside the university.

1553
01:28:23,550 --> 01:28:26,550
You see, the kind of
time-sharing system that

1554
01:28:26,550 --> 01:28:31,590
I've been talking about are
really designed with one

1555
01:28:31,590 --> 01:28:33,570
objective in mind--

1556
01:28:33,570 --> 01:28:40,150
to create decentralized
communication and control.

1557
01:28:40,150 --> 01:28:44,160
That is, it's a
decentralizing system.

1558
01:28:44,160 --> 01:28:45,390
It's inherently so.

1559
01:28:45,390 --> 01:28:46,720
That was the intent.

1560
01:28:46,720 --> 01:28:51,120
And then, of course, that
runs against the grain

1561
01:28:51,120 --> 01:28:53,055
of traditional management.

1562
01:28:53,055 --> 01:28:55,850


1563
01:28:55,850 --> 01:28:57,470
That's also quite clear.

1564
01:28:57,470 --> 01:29:00,800
And I can give you examples.

1565
01:29:00,800 --> 01:29:04,550
Now, the situation, I
believe, is changing.

1566
01:29:04,550 --> 01:29:08,420
Well, let me give
you an example.

1567
01:29:08,420 --> 01:29:14,720
You remember, there
was the celebration

1568
01:29:14,720 --> 01:29:17,540
of the telephone
a few years ago.

1569
01:29:17,540 --> 01:29:19,790
Among other things,
there was a series

1570
01:29:19,790 --> 01:29:23,090
of seminars with
speakers coming from all

1571
01:29:23,090 --> 01:29:27,170
over the place, organized
by the late '80s boom,

1572
01:29:27,170 --> 01:29:30,590
on the sociology of telephony.

1573
01:29:30,590 --> 01:29:33,800
Now, you are all aware
that in this country,

1574
01:29:33,800 --> 01:29:39,380
telephony has been a public for
the public from the beginning,

1575
01:29:39,380 --> 01:29:41,900
because that was Bell's idea.

1576
01:29:41,900 --> 01:29:42,770
He had a choice.

1577
01:29:42,770 --> 01:29:44,030
But that's what he decided.

1578
01:29:44,030 --> 01:29:48,710
It was very interesting that
a French sociologist came over

1579
01:29:48,710 --> 01:29:49,580
to speak.

1580
01:29:49,580 --> 01:29:53,960
And I never realized before
that the situation in France

1581
01:29:53,960 --> 01:29:56,570
was and is still very different.

1582
01:29:56,570 --> 01:30:02,360
Telephony was used as a tool of
government and of the military.

1583
01:30:02,360 --> 01:30:07,460
The reason was-- and it remained
that way, in fact, still--

1584
01:30:07,460 --> 01:30:08,570
well, 10 years ago.

1585
01:30:08,570 --> 01:30:10,070
I can't say today.

1586
01:30:10,070 --> 01:30:15,410
If you try-- I tried many times
to call Paris from Switzerland.

1587
01:30:15,410 --> 01:30:18,320
And the communication was lousy.

1588
01:30:18,320 --> 01:30:23,310
The telephone system was behind
other European countries,

1589
01:30:23,310 --> 01:30:26,540
and certainly the United States,
because it was government

1590
01:30:26,540 --> 01:30:30,020
policy not to push it.

1591
01:30:30,020 --> 01:30:34,430
France is run in a very
centralized way from Paris.

1592
01:30:34,430 --> 01:30:37,580
And the government never
wanted to provide the tool

1593
01:30:37,580 --> 01:30:38,735
to bypass Paris.

1594
01:30:38,735 --> 01:30:41,880


1595
01:30:41,880 --> 01:30:44,520
That comes from a
French psychology--

1596
01:30:44,520 --> 01:30:46,500
I mean, social psychology.

1597
01:30:46,500 --> 01:30:49,110
I'm not inventing this, you see.

1598
01:30:49,110 --> 01:30:51,300
Now, you see the
similar thing relative

1599
01:30:51,300 --> 01:30:53,760
to time-sharing systems
in organizations.

1600
01:30:53,760 --> 01:30:56,250
I think the situation
is changing,

1601
01:30:56,250 --> 01:30:59,550
because there is a
tremendous pressure now

1602
01:30:59,550 --> 01:31:03,490
on individual productivity,
improving individual quality,

1603
01:31:03,490 --> 01:31:09,000
that this is what online use
of computers does for you.

1604
01:31:09,000 --> 01:31:12,610
The second is that the cost of
computers is going way down.

1605
01:31:12,610 --> 01:31:15,060
So I think that we are
going to see a change.

1606
01:31:15,060 --> 01:31:20,520
The technology will lead to
a different types of systems.

1607
01:31:20,520 --> 01:31:22,860
But the goal will be the same.

1608
01:31:22,860 --> 01:31:24,960
Well, I talked way too long.

1609
01:31:24,960 --> 01:31:25,800
Thank you.

1610
01:31:25,800 --> 01:31:29,750
[APPLAUSE]

1611
01:31:29,750 --> 01:31:47,000