1
00:00:00,000 --> 00:00:06,448


2
00:00:06,448 --> 00:00:09,920
[MUSIC PLAYING]

3
00:00:09,920 --> 00:00:59,070


4
00:00:59,070 --> 00:01:03,900
We are beginning now the first
session of the afternoon.

5
00:01:03,900 --> 00:01:09,630
And the first and last and
only speaker in this session

6
00:01:09,630 --> 00:01:12,930
is Professor Michael
Dertouzos, who clearly doesn't

7
00:01:12,930 --> 00:01:17,410
need any introduction, because
he's your host, first of all,

8
00:01:17,410 --> 00:01:23,730
and second has been director of
Project MAC and the Laboratory

9
00:01:23,730 --> 00:01:26,800
for computer science
for some 14 years--

10
00:01:26,800 --> 00:01:30,370
more than half the existence
of the whole thing.

11
00:01:30,370 --> 00:01:35,380
But I will say just
a couple of things

12
00:01:35,380 --> 00:01:38,420
because they're pertinent
to his presentation.

13
00:01:38,420 --> 00:01:46,820
What you will hear originates
is largely from two,

14
00:01:46,820 --> 00:01:48,710
let me call them,
extra-curricular

15
00:01:48,710 --> 00:01:53,900
or above the call
of duty activities

16
00:01:53,900 --> 00:01:56,400
in which he has been involved.

17
00:01:56,400 --> 00:01:59,780
He has been chairman
of the MIT Commission

18
00:01:59,780 --> 00:02:02,360
on Industrial
Productivity, which

19
00:02:02,360 --> 00:02:06,110
is a very serious effort
on the part of MIT to try

20
00:02:06,110 --> 00:02:08,479
to understand what's
going on and make

21
00:02:08,479 --> 00:02:11,960
some sensible recommendations
for the future.

22
00:02:11,960 --> 00:02:17,490
The preliminary report of
the commission was issued,

23
00:02:17,490 --> 00:02:20,240
I believe, this
spring at some time,

24
00:02:20,240 --> 00:02:23,000
and are still working
on the final report.

25
00:02:23,000 --> 00:02:25,340
The other very
pertinent activity,

26
00:02:25,340 --> 00:02:29,720
he a member of the Computer
Science and Technology Board

27
00:02:29,720 --> 00:02:32,030
of the National
Research Council,

28
00:02:32,030 --> 00:02:37,610
which has issued very
recently an excellent report--

29
00:02:37,610 --> 00:02:41,990
I read it, called the National
Challenge in Computer Science

30
00:02:41,990 --> 00:02:43,470
and Technology.

31
00:02:43,470 --> 00:02:49,910
And I hope you will get a copy
at some point and read it.

32
00:02:49,910 --> 00:02:54,650
I think that the topic
of his presentation today

33
00:02:54,650 --> 00:02:59,350
emerges quite a bit from his
experience in these two tasks.

34
00:02:59,350 --> 00:03:01,427
Professor Dertouzos.

35
00:03:01,427 --> 00:03:04,850
[MUSIC PLAYING]

36
00:03:04,850 --> 00:03:10,718


37
00:03:10,718 --> 00:03:13,620
Good afternoon,
ladies and gentlemen.

38
00:03:13,620 --> 00:03:15,050
I'm going to start
by telling you

39
00:03:15,050 --> 00:03:18,530
how I increased my own
productivity by doing

40
00:03:18,530 --> 00:03:20,030
these slides on my Macintosh.

41
00:03:20,030 --> 00:03:21,832
It took 200 hours.

42
00:03:21,832 --> 00:03:25,206
[LAUGHTER]

43
00:03:25,206 --> 00:03:31,472


44
00:03:31,472 --> 00:03:34,160
The first talk is a little
different than the ones

45
00:03:34,160 --> 00:03:35,690
you heard this morning.

46
00:03:35,690 --> 00:03:37,880
So I don't I don't have
to rise to the challenge

47
00:03:37,880 --> 00:03:39,290
of giving a better talk.

48
00:03:39,290 --> 00:03:42,650
[LAUGHTER]

49
00:03:42,650 --> 00:03:49,910
This talk, first of all, is
going to try to change your pH.

50
00:03:49,910 --> 00:03:53,540
It's going to try to change
your mindset, our mindset.

51
00:03:53,540 --> 00:03:54,518
I'm very ambitious.

52
00:03:54,518 --> 00:03:56,060
I would like to
suggest that we ought

53
00:03:56,060 --> 00:03:59,990
to begin thinking differently
about the next 20, 30, 40--

54
00:03:59,990 --> 00:04:03,680
pick up a grandiose
enough scale years.

55
00:04:03,680 --> 00:04:07,130
Now, all of us know
about computers.

56
00:04:07,130 --> 00:04:10,940
But I'd like to remind
you about productivity.

57
00:04:10,940 --> 00:04:13,460
Productivity is
defined as the output,

58
00:04:13,460 --> 00:04:17,329
usually in some form of money,
of the goods that are produced

59
00:04:17,329 --> 00:04:19,190
or the services
that are delivered

60
00:04:19,190 --> 00:04:23,270
divided by the input, which
is either the labor that goes

61
00:04:23,270 --> 00:04:26,450
in it-- again expressed
as money or as time,

62
00:04:26,450 --> 00:04:29,450
or the capital that goes
into capital equipment

63
00:04:29,450 --> 00:04:33,620
to produce those goods, or
some combination of both.

64
00:04:33,620 --> 00:04:38,480
Now, let's just pause here
and ask a very basic question.

65
00:04:38,480 --> 00:04:44,750
What is this number in the large
for the entire United States?

66
00:04:44,750 --> 00:04:48,080
Take the gross national product
divided by all the people

67
00:04:48,080 --> 00:04:50,060
who are working, and
come up with a figure

68
00:04:50,060 --> 00:04:55,370
of productivity of dollars
per worker in one year.

69
00:04:55,370 --> 00:04:59,270
That number for the
United States is $38,000.

70
00:04:59,270 --> 00:05:03,230
Now, that same number
for Japan is $37,000.

71
00:05:03,230 --> 00:05:07,050
And I'm giving you 1987 data,
so it's just a little below it.

72
00:05:07,050 --> 00:05:07,550
By now.

73
00:05:07,550 --> 00:05:10,490
I'm sure it's crossed.

74
00:05:10,490 --> 00:05:13,370
West Germany is a little
bit below that in the $20s,

75
00:05:13,370 --> 00:05:16,130
and then you go to Korea,
a little bit further below.

76
00:05:16,130 --> 00:05:19,700
Now, that in itself perhaps
is not terribly significant.

77
00:05:19,700 --> 00:05:24,690
But ask, what is the slope
at which this figure has

78
00:05:24,690 --> 00:05:26,400
been growing, adjusted
for inflation,

79
00:05:26,400 --> 00:05:30,580
adjusted for everything else,
through the last 10 years?

80
00:05:30,580 --> 00:05:38,260
The Japanese slope is
35% in the last 10 years.

81
00:05:38,260 --> 00:05:42,160
The West German slope is 21%.

82
00:05:42,160 --> 00:05:44,710
The Korean slope is 60%-some.

83
00:05:44,710 --> 00:05:51,980
The United States United
States slope is 7%.

84
00:05:51,980 --> 00:05:54,320
So in the last 10
years, the productivity

85
00:05:54,320 --> 00:05:58,070
adjusted for inflation, and for
everything else, in real terms,

86
00:05:58,070 --> 00:05:59,540
has increased by 7%--

87
00:05:59,540 --> 00:06:03,980
if you want to think of
divided by 10, 0.7% per year

88
00:06:03,980 --> 00:06:07,010
that's an important
number to remember.

89
00:06:07,010 --> 00:06:12,753
What I'd like to do today is try
to answer the question of what

90
00:06:12,753 --> 00:06:15,170
are some of the weaknesses
that have contributed to the US

91
00:06:15,170 --> 00:06:16,870
productivity being low.

92
00:06:16,870 --> 00:06:19,910
Then I'd like to move away
from the US productivity,

93
00:06:19,910 --> 00:06:22,580
look at the world at large,
and say what other things could

94
00:06:22,580 --> 00:06:26,630
we do with computers to enlarge
and increase the productivity

95
00:06:26,630 --> 00:06:28,370
worldwide.

96
00:06:28,370 --> 00:06:32,120
And then I'd like to bring
in the technological supply

97
00:06:32,120 --> 00:06:36,050
and beat the two together in
some form of brokered activity

98
00:06:36,050 --> 00:06:38,630
to see what we might
get out of them.

99
00:06:38,630 --> 00:06:41,390
I admit to you at the
outset that I'm biased.

100
00:06:41,390 --> 00:06:43,790
I think we've been supply
siders for too long,

101
00:06:43,790 --> 00:06:44,990
worrying about this.

102
00:06:44,990 --> 00:06:46,610
And the time has
come now to begin

103
00:06:46,610 --> 00:06:48,500
worrying about the demand side.

104
00:06:48,500 --> 00:06:51,530
Finally, I'd like to take
this to some limiting argument

105
00:06:51,530 --> 00:06:53,900
of how far we might
be able to go,

106
00:06:53,900 --> 00:06:56,210
how we might measure
productivity, and then

107
00:06:56,210 --> 00:07:02,480
conclude with some actions
of what we ought to do.

108
00:07:02,480 --> 00:07:08,540
Now, as Bob Fano told you, MIT
put together this major effort

109
00:07:08,540 --> 00:07:11,220
of 18 commissioners who
brought in another 30,

110
00:07:11,220 --> 00:07:13,100
40 people from MIT.

111
00:07:13,100 --> 00:07:16,910
And this group, the
commission, asked the question

112
00:07:16,910 --> 00:07:19,940
what went wrong
with US productivity

113
00:07:19,940 --> 00:07:23,750
and what might be done
to fix it at places

114
00:07:23,750 --> 00:07:25,390
like MIT and elsewhere?

115
00:07:25,390 --> 00:07:27,140
Now, I'm not going to
tell you everything.

116
00:07:27,140 --> 00:07:29,720
In fact, I only have one
slide from that commission.

117
00:07:29,720 --> 00:07:32,030
If you want the rest,
there is a book coming out

118
00:07:32,030 --> 00:07:33,590
called Made in America.

119
00:07:33,590 --> 00:07:35,100
And that's going
to be available,

120
00:07:35,100 --> 00:07:36,170
I think, around March.

121
00:07:36,170 --> 00:07:38,580
And that's our report.

122
00:07:38,580 --> 00:07:42,680
The commission is a little
different from other groups.

123
00:07:42,680 --> 00:07:44,630
We focused on the
production system.

124
00:07:44,630 --> 00:07:47,030
We focused inside the company--

125
00:07:47,030 --> 00:07:50,390
how people work, how they use
equipment, how they're managed,

126
00:07:50,390 --> 00:07:51,410
how they're organized.

127
00:07:51,410 --> 00:07:54,770
Most studies on productivity
are macroeconomic or fiscal,

128
00:07:54,770 --> 00:07:56,690
and they look at
everything like a black box

129
00:07:56,690 --> 00:07:59,045
and adjust interest rates
and things of that sort.

130
00:07:59,045 --> 00:08:03,860
The second thing is we looked
in detail from the micro level

131
00:08:03,860 --> 00:08:05,390
at eight industries.

132
00:08:05,390 --> 00:08:08,240
We looked at consumer
electronics, textiles,

133
00:08:08,240 --> 00:08:12,620
chemicals, automobiles, steel,
machine tools, aircraft,

134
00:08:12,620 --> 00:08:14,540
and semiconductors.

135
00:08:14,540 --> 00:08:16,790
A group of 10 or
12 people attacked

136
00:08:16,790 --> 00:08:18,860
each group,
conducted interviews,

137
00:08:18,860 --> 00:08:21,740
put together big
reports, and then we

138
00:08:21,740 --> 00:08:25,940
beat everything together to
find out common weaknesses.

139
00:08:25,940 --> 00:08:28,760
And the final thing
is that we did use

140
00:08:28,760 --> 00:08:30,470
an interdisciplinary approach.

141
00:08:30,470 --> 00:08:32,299
This is not a
group of economists

142
00:08:32,299 --> 00:08:33,799
or a group of technologists.

143
00:08:33,799 --> 00:08:35,059
It's a mixed group.

144
00:08:35,059 --> 00:08:36,980
And it took us six
months to learn

145
00:08:36,980 --> 00:08:40,309
how to talk to each
other, but we managed.

146
00:08:40,309 --> 00:08:45,110
I'm reporting now on the
six dominant weaknesses that

147
00:08:45,110 --> 00:08:49,710
have caused productivity
weakness as determined

148
00:08:49,710 --> 00:08:52,930
by looking at these
eight industrial groups.

149
00:08:52,930 --> 00:08:55,530
And I will explain.

150
00:08:55,530 --> 00:08:58,860
First, short time horizons.

151
00:08:58,860 --> 00:09:01,290
I think all of us
understand what this means.

152
00:09:01,290 --> 00:09:05,790
At one company, the bonus
is based on the next six

153
00:09:05,790 --> 00:09:08,580
months of profitability.

154
00:09:08,580 --> 00:09:10,400
We know that our
financial system

155
00:09:10,400 --> 00:09:16,370
operates with a flywheel
that is very small.

156
00:09:16,370 --> 00:09:19,370
In another company, a very
large electronics company

157
00:09:19,370 --> 00:09:22,880
that was dominating our
industry for many years,

158
00:09:22,880 --> 00:09:27,170
we find that it abandons
its area of expertise,

159
00:09:27,170 --> 00:09:32,540
goes out and buys a rental
car agencies and finance

160
00:09:32,540 --> 00:09:34,280
institutions.

161
00:09:34,280 --> 00:09:38,600
Now, this lack of staying
power, this moving away,

162
00:09:38,600 --> 00:09:41,000
going to an area that appears
to be in the short term

163
00:09:41,000 --> 00:09:44,330
profitable, and this
constant preoccupation

164
00:09:44,330 --> 00:09:47,120
with the short term, from
managing within the company

165
00:09:47,120 --> 00:09:50,210
to the financial
side, is one recurrent

166
00:09:50,210 --> 00:09:52,610
and dominant weakness.

167
00:09:52,610 --> 00:09:56,670
Now, as I go down my list, I
want you to start thinking,

168
00:09:56,670 --> 00:10:00,740
how can we use computers
and computer technology

169
00:10:00,740 --> 00:10:02,010
to help in these areas.

170
00:10:02,010 --> 00:10:04,760
Now, we can't help all of them,
nor is computer technology

171
00:10:04,760 --> 00:10:06,680
the only panacea,
but how can we?

172
00:10:06,680 --> 00:10:09,580
Please start thinking.

173
00:10:09,580 --> 00:10:13,960
Outdated strategies-- two
major outdated strategies

174
00:10:13,960 --> 00:10:16,600
mark our demise.

175
00:10:16,600 --> 00:10:18,340
And the more we
got into trouble,

176
00:10:18,340 --> 00:10:21,280
the more we went back
to these old favorites,

177
00:10:21,280 --> 00:10:23,290
the old standbys.

178
00:10:23,290 --> 00:10:26,300
First one is mass production.

179
00:10:26,300 --> 00:10:28,190
We have achieved so much
with mass production

180
00:10:28,190 --> 00:10:30,080
that we think it's the answer.

181
00:10:30,080 --> 00:10:32,780
But in fact, the world has
moved, and is moving away,

182
00:10:32,780 --> 00:10:37,490
from a mass production system to
a niche system where consumers

183
00:10:37,490 --> 00:10:40,100
are much more sophisticated and
where production is much more

184
00:10:40,100 --> 00:10:41,760
flexible.

185
00:10:41,760 --> 00:10:44,720
The second one is parochialism.

186
00:10:44,720 --> 00:10:48,170
This country, for the
longest possible time,

187
00:10:48,170 --> 00:10:52,070
has been dealing only with
its own internal borders.

188
00:10:52,070 --> 00:10:56,270
We did not look outside
when the Europeans started

189
00:10:56,270 --> 00:10:58,370
making small cars, when
the Japanese started

190
00:10:58,370 --> 00:11:01,370
making small cars, and
using them and so forth.

191
00:11:01,370 --> 00:11:03,620
And in the textiles
area, we again

192
00:11:03,620 --> 00:11:08,390
dominantly pursued the
manufacture of apparel,

193
00:11:08,390 --> 00:11:12,260
pretty much all of the same
mass production quality,

194
00:11:12,260 --> 00:11:14,060
whereas others
concentrated on niches.

195
00:11:14,060 --> 00:11:17,040


196
00:11:17,040 --> 00:11:18,890
The pursuit of
outdated strategies,

197
00:11:18,890 --> 00:11:20,990
the strategic inertia
in our companies,

198
00:11:20,990 --> 00:11:24,750
is another factor we have found.

199
00:11:24,750 --> 00:11:28,070
The third factor hurts
in a place like MIT,

200
00:11:28,070 --> 00:11:31,670
because while we're great
at inventing products,

201
00:11:31,670 --> 00:11:34,070
we don't seem to be
so good at carrying

202
00:11:34,070 --> 00:11:36,770
the products from the invention
to the production stage

203
00:11:36,770 --> 00:11:38,570
to market.

204
00:11:38,570 --> 00:11:42,270
And a lot of the weaknesses
there seem to be technological.

205
00:11:42,270 --> 00:11:46,840
We do not design for
manufacturability.

206
00:11:46,840 --> 00:11:51,241
We design for glory--

207
00:11:51,241 --> 00:11:54,230
[LAUGHTER]

208
00:11:54,230 --> 00:11:56,930
--or for greed if
we have a startup.

209
00:11:56,930 --> 00:12:00,610


210
00:12:00,610 --> 00:12:03,730
If you look around at the
examples, they abound.

211
00:12:03,730 --> 00:12:07,300
The VCR, which we
invented, we couldn't

212
00:12:07,300 --> 00:12:10,060
produce at the tolerance
and manufacturing

213
00:12:10,060 --> 00:12:11,195
cost that was necessary.

214
00:12:11,195 --> 00:12:12,820
Also, we didn't have
the vision that it

215
00:12:12,820 --> 00:12:15,910
would be a worldwide market,
but leave that aside.

216
00:12:15,910 --> 00:12:19,130
We had trouble producing it.

217
00:12:19,130 --> 00:12:22,990
So this is a very
bad third factor--

218
00:12:22,990 --> 00:12:27,250
third weakness if you wish,
in our productivity, which

219
00:12:27,250 --> 00:12:30,760
also is characterized by
over-innovation on products,

220
00:12:30,760 --> 00:12:33,250
another way of saying
it, and under-innovation

221
00:12:33,250 --> 00:12:35,740
on the processes that
take these products

222
00:12:35,740 --> 00:12:39,830
all the way to the marketplace.

223
00:12:39,830 --> 00:12:44,840
If you want another measure
of that, the automobiles,

224
00:12:44,840 --> 00:12:48,230
it takes us 50% longer
to produce a car

225
00:12:48,230 --> 00:12:51,380
from the time we conceive
it than it takes in Japan.

226
00:12:51,380 --> 00:12:56,900
And after the car is out,
we have two to three times--

227
00:12:56,900 --> 00:13:01,010
that's 200% to 300% as
many, reported faults

228
00:13:01,010 --> 00:13:02,590
in the first six months of use.

229
00:13:02,590 --> 00:13:06,200


230
00:13:06,200 --> 00:13:09,800
The fourth factor is neglect
of our education and human

231
00:13:09,800 --> 00:13:10,910
resources.

232
00:13:10,910 --> 00:13:13,580
A recent survey in
metro Philadelphia

233
00:13:13,580 --> 00:13:17,780
asked high school
graduates giving them

234
00:13:17,780 --> 00:13:20,840
a table of Amtrak,
which train they

235
00:13:20,840 --> 00:13:26,130
should take to arrive in
Washington DC just before noon?

236
00:13:26,130 --> 00:13:29,650
How many of them do you think
were able to answer correctly?

237
00:13:29,650 --> 00:13:34,750
5%-- 4.9% to be precise.

238
00:13:34,750 --> 00:13:38,080
We're having serious
problems and a trend

239
00:13:38,080 --> 00:13:40,870
downward in the area
of basic education

240
00:13:40,870 --> 00:13:43,010
and in the area of
technological literacy.

241
00:13:43,010 --> 00:13:45,160
This at a time
when our technology

242
00:13:45,160 --> 00:13:50,050
is moving toward more use of
scientific and technological

243
00:13:50,050 --> 00:13:52,820
methods of production.

244
00:13:52,820 --> 00:13:58,840
And in many of our companies,
we find an attitude

245
00:13:58,840 --> 00:14:01,360
which treats labor--

246
00:14:01,360 --> 00:14:04,660
and by labor, I mean everybody,
professionals and labor

247
00:14:04,660 --> 00:14:06,760
people in a company,
white and blue collar,

248
00:14:06,760 --> 00:14:09,250
treat labor as a
cost factor to be

249
00:14:09,250 --> 00:14:14,580
minimized rather than a
precious asset to be nurtured.

250
00:14:14,580 --> 00:14:16,040
There is a lot of room there.

251
00:14:16,040 --> 00:14:18,800
Ask yourselves,
can computers help?

252
00:14:18,800 --> 00:14:21,140
Lack of cooperation
is a dominant

253
00:14:21,140 --> 00:14:23,180
and recurring weakness.

254
00:14:23,180 --> 00:14:25,190
We see that within a
company when the design

255
00:14:25,190 --> 00:14:28,160
department designs something,
tosses it over to production.

256
00:14:28,160 --> 00:14:30,860
Production makes
it, finds a bug.

257
00:14:30,860 --> 00:14:33,530
They toss it over the
wall again, fix the bug.

258
00:14:33,530 --> 00:14:37,280
And all these tosses account
for these 50% longer time delays

259
00:14:37,280 --> 00:14:40,790
that I mentioned earlier
in getting products out.

260
00:14:40,790 --> 00:14:44,780
But we also found this
problem between companies.

261
00:14:44,780 --> 00:14:49,220
Companies treat their suppliers
as if they were adversaries,

262
00:14:49,220 --> 00:14:51,830
with a legal contract
as an instrument

263
00:14:51,830 --> 00:14:55,650
instead of as partners
toward the common goal.

264
00:14:55,650 --> 00:14:57,800
And we find this,
even in the area

265
00:14:57,800 --> 00:15:00,530
that Bob [? Scheiffler ?]
discussed this morning,

266
00:15:00,530 --> 00:15:02,330
among competitors.

267
00:15:02,330 --> 00:15:04,430
Competitors must get
together every now and then

268
00:15:04,430 --> 00:15:06,980
for example to set standards.

269
00:15:06,980 --> 00:15:09,980
And we have been unable
to do this, as well

270
00:15:09,980 --> 00:15:11,780
as some other people.

271
00:15:11,780 --> 00:15:15,920
In particular, where we lost
the entire machine tool business

272
00:15:15,920 --> 00:15:19,320
to Japan because they
did set standards.

273
00:15:19,320 --> 00:15:23,450
Finally, when one
talks to executives

274
00:15:23,450 --> 00:15:25,610
especially-- chief
executives of companies,

275
00:15:25,610 --> 00:15:27,650
they always complain that
our productivity ills

276
00:15:27,650 --> 00:15:29,720
are caused by the government.

277
00:15:29,720 --> 00:15:33,770
The commission found that
this is, in fact, a factor.

278
00:15:33,770 --> 00:15:36,140
But we found no
dominant pattern.

279
00:15:36,140 --> 00:15:39,110
We did not see, for
example, antitrust

280
00:15:39,110 --> 00:15:43,580
as being a dominant factor
or excessive environmental

281
00:15:43,580 --> 00:15:45,860
regulations, or many
of these factors

282
00:15:45,860 --> 00:15:49,040
that come up in
regulatory policy.

283
00:15:49,040 --> 00:15:51,740
What we did find is a mismatch.

284
00:15:51,740 --> 00:15:54,260
It's as if the
government was proceeding

285
00:15:54,260 --> 00:15:58,160
on a different agenda
from that of the nation.

286
00:15:58,160 --> 00:16:05,270
Now, let me put up what
by now you must recognize

287
00:16:05,270 --> 00:16:09,920
is the central strategy
of the Laboratory

288
00:16:09,920 --> 00:16:10,860
for Computer Science.

289
00:16:10,860 --> 00:16:12,620
I alluded to this
in my introduction

290
00:16:12,620 --> 00:16:15,470
and you heard about
it in the talks.

291
00:16:15,470 --> 00:16:17,540
But as we look at
the future, if we

292
00:16:17,540 --> 00:16:20,360
want to assess what
do we have to offer

293
00:16:20,360 --> 00:16:24,560
against these weaknesses first,
and then against other things

294
00:16:24,560 --> 00:16:26,690
beyond the weaknesses,
we ought to look

295
00:16:26,690 --> 00:16:29,210
at what our technology is
going to have available to us.

296
00:16:29,210 --> 00:16:33,170


297
00:16:33,170 --> 00:16:36,650
Now, up here at the top
are the basic things.

298
00:16:36,650 --> 00:16:41,063
The theory which is going
to keep growing, I hope.

299
00:16:41,063 --> 00:16:42,480
We can't predict
where it's going.

300
00:16:42,480 --> 00:16:45,870
But it's been doing great,
so it'll keep doing great.

301
00:16:45,870 --> 00:16:48,380
Our technology base,
where most of us

302
00:16:48,380 --> 00:16:52,190
expect two to three orders
of magnitude improvement

303
00:16:52,190 --> 00:16:54,420
in the next 20 years.

304
00:16:54,420 --> 00:16:56,600
And that means that
whatever we do now,

305
00:16:56,600 --> 00:16:59,090
we'll be able to do faster,
better, and cheaper,

306
00:16:59,090 --> 00:17:04,599
as has been the case in the last
25 years in terms of trends.

307
00:17:04,599 --> 00:17:07,960
In the area of software
with John Guttag,

308
00:17:07,960 --> 00:17:11,089
there will be progress, but
I think it will be slow.

309
00:17:11,089 --> 00:17:13,240
And the reason is
the one you heard.

310
00:17:13,240 --> 00:17:17,290
We inextricably mix the
design, the production,

311
00:17:17,290 --> 00:17:20,859
and the maintenance of
software all in one activity.

312
00:17:20,859 --> 00:17:23,170
And until we get past
that, I don't think

313
00:17:23,170 --> 00:17:25,359
we'll see much change here.

314
00:17:25,359 --> 00:17:30,100
These are the two big engines
of innovation, in my opinion.

315
00:17:30,100 --> 00:17:32,800
The multiprocessors of
harnessing hundreds, thousands,

316
00:17:32,800 --> 00:17:35,740
millions of processors all
together, like lots of horses

317
00:17:35,740 --> 00:17:39,880
pulling a cart, is
the new kid in town.

318
00:17:39,880 --> 00:17:43,180
That's the one big new
thing around the corner.

319
00:17:43,180 --> 00:17:46,870
And the other one is
the distributed systems,

320
00:17:46,870 --> 00:17:49,870
hooking together all these
machines from personal

321
00:17:49,870 --> 00:17:52,750
to big machines together.

322
00:17:52,750 --> 00:17:54,640
Now, you'll notice some
arrows on this graph.

323
00:17:54,640 --> 00:17:57,550
And the arrows mean
that A is used in B.

324
00:17:57,550 --> 00:17:59,740
But that's not
terribly significant.

325
00:17:59,740 --> 00:18:03,310
You may want to play with it
later and disagree with me.

326
00:18:03,310 --> 00:18:06,070
Let me move on to the
bottom of the chart.

327
00:18:06,070 --> 00:18:08,980
These are the
cognitive, the sensory,

328
00:18:08,980 --> 00:18:10,960
and the robotic systems,
which I consider

329
00:18:10,960 --> 00:18:13,840
as the intelligent
systems of the future.

330
00:18:13,840 --> 00:18:16,715
And I think here in this
area, it's tough to predict.

331
00:18:16,715 --> 00:18:19,090
There's a lot of hoopla, a
lot of hype, about neural nets

332
00:18:19,090 --> 00:18:20,080
and this and that.

333
00:18:20,080 --> 00:18:23,510
Learning research is beginning
again, learning from practice.

334
00:18:23,510 --> 00:18:25,640
But I think we'll
have to wait and see.

335
00:18:25,640 --> 00:18:28,070
It's perhaps a little
early to call it.

336
00:18:28,070 --> 00:18:30,970
Whereas here, I think the
progress is going to be big.

337
00:18:30,970 --> 00:18:33,880
In the sensory area,
I expect progress

338
00:18:33,880 --> 00:18:35,830
simply because there
will be progress here,

339
00:18:35,830 --> 00:18:38,170
I think, especially
in the special purpose

340
00:18:38,170 --> 00:18:40,900
multiprocessors, and
because a lot of what we're

341
00:18:40,900 --> 00:18:43,240
learning from biology is
going to be used together

342
00:18:43,240 --> 00:18:45,740
with that progress in equipment.

343
00:18:45,740 --> 00:18:47,350
And finally, in
robotics systems,

344
00:18:47,350 --> 00:18:51,130
I think a lot will depend on how
cognitive these robots will be,

345
00:18:51,130 --> 00:18:55,520
so we can reduce that problem
to a large extent to the others.

346
00:18:55,520 --> 00:19:00,040
Now, let's ask, in
what way, or what ways,

347
00:19:00,040 --> 00:19:05,150
might computers
address the weaknesses?

348
00:19:05,150 --> 00:19:10,390
Let me take an area that we have
called computer-aided design.

349
00:19:10,390 --> 00:19:11,920
But let me broaden it.

350
00:19:11,920 --> 00:19:13,975
Let me broaden it
to include much more

351
00:19:13,975 --> 00:19:15,100
than computer-aided design.

352
00:19:15,100 --> 00:19:16,760
I want to include simulation.

353
00:19:16,760 --> 00:19:20,080
I wanted to include every
and any application where

354
00:19:20,080 --> 00:19:21,790
a lot of computing clicks--

355
00:19:21,790 --> 00:19:24,640
a lot of compute times
as Sussman calls them,

356
00:19:24,640 --> 00:19:30,870
are used to tell us something
about what we're working on.

357
00:19:30,870 --> 00:19:35,890
Now, take the first weakness--
outdated strategies.

358
00:19:35,890 --> 00:19:40,930
And there, I think we
could use simulation

359
00:19:40,930 --> 00:19:44,740
in ways we don't use it today
to gain a better understanding

360
00:19:44,740 --> 00:19:47,680
of consumers and of markets.

361
00:19:47,680 --> 00:19:50,470
Today, this great unknown
of what market to hit

362
00:19:50,470 --> 00:19:54,220
and how much to hit it by is
decided on judgemental terms,

363
00:19:54,220 --> 00:19:55,485
and that's fine.

364
00:19:55,485 --> 00:19:57,610
But with information that
we have, with the ability

365
00:19:57,610 --> 00:20:00,820
to profile world
markets, we might

366
00:20:00,820 --> 00:20:05,860
be able to try simulation and
have a new kind of activity

367
00:20:05,860 --> 00:20:09,910
called computer-aided marketing,
or computer-aided consumer

368
00:20:09,910 --> 00:20:12,260
simulation.

369
00:20:12,260 --> 00:20:15,930
The second area is perhaps
one of the most promising,

370
00:20:15,930 --> 00:20:19,760
especially since our products
have so many troubles.

371
00:20:19,760 --> 00:20:23,970
This is where we torture the
product at the design stage.

372
00:20:23,970 --> 00:20:25,850
Through excessive
stimulation, try

373
00:20:25,850 --> 00:20:28,340
to anticipate every
possible mishap.

374
00:20:28,340 --> 00:20:31,850
And this seems an area where we
could apply computer technology

375
00:20:31,850 --> 00:20:32,964
productively.

376
00:20:32,964 --> 00:20:35,690


377
00:20:35,690 --> 00:20:41,030
The third area is one that
is interesting, again,

378
00:20:41,030 --> 00:20:43,760
because we've tried
computer-aided education

379
00:20:43,760 --> 00:20:44,720
for a long time.

380
00:20:44,720 --> 00:20:46,740
And while we've
made some progress,

381
00:20:46,740 --> 00:20:50,150
I think the jury is still
out as to the bottom line

382
00:20:50,150 --> 00:20:51,960
of this activity.

383
00:20:51,960 --> 00:20:53,810
But we might want to
try something else.

384
00:20:53,810 --> 00:20:58,370
I mean, we put our
lives in aircraft

385
00:20:58,370 --> 00:21:01,340
flown for the first
time by pilots

386
00:21:01,340 --> 00:21:06,100
who have just but 30 or
40 hours on a simulator.

387
00:21:06,100 --> 00:21:10,630
Why couldn't we think of
some of our managers, some

388
00:21:10,630 --> 00:21:13,700
of our technologists,
of the future,

389
00:21:13,700 --> 00:21:18,940
putting 40 or 50 hours on a
simulator which simulates how

390
00:21:18,940 --> 00:21:22,330
their jobs are done, how
they ought to be done,

391
00:21:22,330 --> 00:21:24,910
what they encounter, and how
they could react with it--

392
00:21:24,910 --> 00:21:26,980
and get feedback before
they have a chance

393
00:21:26,980 --> 00:21:29,500
to practice on real people?

394
00:21:29,500 --> 00:21:32,740
[LAUGHTER]

395
00:21:32,740 --> 00:21:35,140
But we need not
stop at weaknesses.

396
00:21:35,140 --> 00:21:36,560
We could go beyond.

397
00:21:36,560 --> 00:21:41,050
Raj Reddy speaks of
thinking aids-- computers

398
00:21:41,050 --> 00:21:42,430
as thinking aids.

399
00:21:42,430 --> 00:21:46,270
I think we can envision
this notion of simulation,

400
00:21:46,270 --> 00:21:49,210
rapid prototyping,
computer-aided,

401
00:21:49,210 --> 00:21:52,240
whatever you want to call
it, as an entity that

402
00:21:52,240 --> 00:21:55,060
increases the thinking ability
of all kinds of people,

403
00:21:55,060 --> 00:21:58,510
from the professional to
the non-professional ones.

404
00:21:58,510 --> 00:22:03,640
This area is another one that I
have a great expectation from.

405
00:22:03,640 --> 00:22:05,590
We are at the point
where we can take

406
00:22:05,590 --> 00:22:08,440
massive amounts of
computing and apply them

407
00:22:08,440 --> 00:22:11,080
to very small portions
of our universe--

408
00:22:11,080 --> 00:22:14,170
perhaps not physically,
but structurally small.

409
00:22:14,170 --> 00:22:17,560
And we can understand
and see things

410
00:22:17,560 --> 00:22:19,510
that we couldn't even
see with the largest

411
00:22:19,510 --> 00:22:22,345
telescopes or the most
powerful microscopes.

412
00:22:22,345 --> 00:22:24,220
And it seems to me that
this could strengthen

413
00:22:24,220 --> 00:22:25,930
the area of scientific
productivity

414
00:22:25,930 --> 00:22:28,060
in ways we have
not thought about.

415
00:22:28,060 --> 00:22:30,790
And then of course, a lot of
the things we've talked about--

416
00:22:30,790 --> 00:22:34,870
storage of active procedural
knowledge, how we do things.

417
00:22:34,870 --> 00:22:37,030
Still, our knowledge
is largely stored

418
00:22:37,030 --> 00:22:40,030
as books of dead diagrams.

419
00:22:40,030 --> 00:22:43,780
Yet, all the human activity
out there is active.

420
00:22:43,780 --> 00:22:46,060
It involves time, and
I do this, I do this.

421
00:22:46,060 --> 00:22:47,230
I go under it.

422
00:22:47,230 --> 00:22:50,350
And those kinds of
activities we have not

423
00:22:50,350 --> 00:22:51,980
been able yet to capture.

424
00:22:51,980 --> 00:22:53,200
So here is one way.

425
00:22:53,200 --> 00:22:56,740
I'm not pretending that these
are the only ways, but one

426
00:22:56,740 --> 00:23:01,780
way in which if you ask how can
I bring supply to help demand,

427
00:23:01,780 --> 00:23:03,250
you get some answers.

428
00:23:03,250 --> 00:23:05,260
Now, let me give you
another one, where

429
00:23:05,260 --> 00:23:12,640
I think we might have something
perhaps even more important.

430
00:23:12,640 --> 00:23:16,990
And as before, while I call
this the National Information

431
00:23:16,990 --> 00:23:19,450
Network, it is much more.

432
00:23:19,450 --> 00:23:21,640
It is a global network.

433
00:23:21,640 --> 00:23:23,080
And it's not a network.

434
00:23:23,080 --> 00:23:25,070
It is an infrastructure.

435
00:23:25,070 --> 00:23:27,340
Now, the vision here--
and you heard this morning

436
00:23:27,340 --> 00:23:32,740
from Dave Clark about the
architecture of such a network.

437
00:23:32,740 --> 00:23:36,040
And in fact, it is precisely
the statement that we want.

438
00:23:36,040 --> 00:23:39,520
We, MIT, want to be instrumental
in the national information

439
00:23:39,520 --> 00:23:43,030
network business that
is driving the work

440
00:23:43,030 --> 00:23:46,540
that Dr. Clark and his
colleagues are doing.

441
00:23:46,540 --> 00:23:48,710
So you heard from him
about the architecture

442
00:23:48,710 --> 00:23:49,960
and how we might structure it.

443
00:23:49,960 --> 00:23:54,040
Now, the way I envision
this infrastructure

444
00:23:54,040 --> 00:23:58,840
is that it's as ubiquitous
as the telephone and as easy

445
00:23:58,840 --> 00:24:00,400
to use as the telephone.

446
00:24:00,400 --> 00:24:01,450
No big hassle.

447
00:24:01,450 --> 00:24:04,930
You plug your machine on it,
whether you're a company,

448
00:24:04,930 --> 00:24:06,570
or whether you're an individual.

449
00:24:06,570 --> 00:24:08,400
And somehow, it works.

450
00:24:08,400 --> 00:24:10,860
There is a yellow
pages, white pages.

451
00:24:10,860 --> 00:24:12,150
All that is taken care of.

452
00:24:12,150 --> 00:24:14,700
And more importantly,
there is some understanding

453
00:24:14,700 --> 00:24:19,360
between the machines and what
you want what you want to do.

454
00:24:19,360 --> 00:24:21,390
Now we have a hard
time doing this

455
00:24:21,390 --> 00:24:24,840
because the telephone
system had it easy.

456
00:24:24,840 --> 00:24:29,570
It had a fixed activity
on either side--

457
00:24:29,570 --> 00:24:34,160
the vocal box here,
the human ear here.

458
00:24:34,160 --> 00:24:39,490
And all it had to do was
carry faithfully the voice

459
00:24:39,490 --> 00:24:41,750
from the mouth to the ear.

460
00:24:41,750 --> 00:24:44,090
But what is the vocal
box of a computer

461
00:24:44,090 --> 00:24:46,610
and what is the ear
of the computer?

462
00:24:46,610 --> 00:24:50,570
It can go from bits per second
to megabits, and soon gigabits

463
00:24:50,570 --> 00:24:52,220
per second.

464
00:24:52,220 --> 00:24:55,940
It doesn't have the beautiful
property of a narrow spectrum.

465
00:24:55,940 --> 00:24:57,940
So there, you see
some of the reasons

466
00:24:57,940 --> 00:25:01,480
for building these networks
with very variable bandwidth

467
00:25:01,480 --> 00:25:04,360
that you pay for, variable
delay that you pay for,

468
00:25:04,360 --> 00:25:09,300
perhaps variable reliability
that Clark was talking about.

469
00:25:09,300 --> 00:25:13,260
Bring this kind of a vision
of this infrastructure

470
00:25:13,260 --> 00:25:16,500
to a country like America
and ask what might it do.

471
00:25:16,500 --> 00:25:20,970
First in terms of weaknesses,
it could bring knowledge

472
00:25:20,970 --> 00:25:25,342
about what's happening out
in the world to the companies

473
00:25:25,342 --> 00:25:27,550
throughout, because it would
be easy to connect to it

474
00:25:27,550 --> 00:25:30,400
and find out what is happening
in France, what's happening

475
00:25:30,400 --> 00:25:33,700
in England, in Japan, what
is needed by the market,

476
00:25:33,700 --> 00:25:36,640
where are the
tastes shifting to.

477
00:25:36,640 --> 00:25:40,120
It could address some of
the techno weaknesses.

478
00:25:40,120 --> 00:25:42,340
As we grow more
geographically distributed,

479
00:25:42,340 --> 00:25:44,230
it could bring
together the designer

480
00:25:44,230 --> 00:25:46,870
with a production person
with a marketeer, wherever

481
00:25:46,870 --> 00:25:49,930
they may reside, and have
them work cooperatively

482
00:25:49,930 --> 00:25:53,350
on the same kind of
activity, the same goal--

483
00:25:53,350 --> 00:25:57,310
to produce a reliable
product fast and so forth.

484
00:25:57,310 --> 00:25:59,170
In the area of lack
of cooperation,

485
00:25:59,170 --> 00:26:02,380
no computer is going to
force people to cooperate.

486
00:26:02,380 --> 00:26:05,230
And remember, these are
not exclusive panaceas.

487
00:26:05,230 --> 00:26:07,930
These are tools that
are supposed to aid.

488
00:26:07,930 --> 00:26:13,360
But if we do have a network
that helps communication

489
00:26:13,360 --> 00:26:16,330
by exchanges of voice,
by exchange of mail,

490
00:26:16,330 --> 00:26:18,550
or by exchange of
designs, by exchange

491
00:26:18,550 --> 00:26:22,250
of software, of
opinion and so forth,

492
00:26:22,250 --> 00:26:24,080
we are a lot better
off than if we're

493
00:26:24,080 --> 00:26:26,330
isolated in our little
cubicles through all

494
00:26:26,330 --> 00:26:28,040
these hierarchical
trees that form

495
00:26:28,040 --> 00:26:30,450
our current
organizational structures.

496
00:26:30,450 --> 00:26:32,240
So here is yet another
way that we might

497
00:26:32,240 --> 00:26:33,990
increase the communication.

498
00:26:33,990 --> 00:26:36,210
Certainly in the
ARPANET community,

499
00:26:36,210 --> 00:26:39,920
we saw this as this
culture grew that

500
00:26:39,920 --> 00:26:42,770
was able to communicate
rapidly, notwithstanding

501
00:26:42,770 --> 00:26:47,740
who stood on top of
these various pyramids.

502
00:26:47,740 --> 00:26:49,390
The same goes for
government/industry

503
00:26:49,390 --> 00:26:52,570
communication, which certainly
has not been at its best

504
00:26:52,570 --> 00:26:53,720
in the past.

505
00:26:53,720 --> 00:26:56,380
So there are things,
again, that computers

506
00:26:56,380 --> 00:27:01,000
could do if we looked
at these weaknesses

507
00:27:01,000 --> 00:27:02,710
where they might
increase productivity.

508
00:27:02,710 --> 00:27:06,250
But going beyond
those, I think we

509
00:27:06,250 --> 00:27:09,520
could see a whole
new way to increase

510
00:27:09,520 --> 00:27:10,720
white-collar productivity.

511
00:27:10,720 --> 00:27:12,678
And I'll be coming back
to this because I think

512
00:27:12,678 --> 00:27:16,010
this is a major target for us.

513
00:27:16,010 --> 00:27:20,250
I think the purchase and
sale of services today,

514
00:27:20,250 --> 00:27:23,810
which involves an immense
amount of paperwork,

515
00:27:23,810 --> 00:27:27,710
could take place in
seconds instead of in days.

516
00:27:27,710 --> 00:27:30,800
Just one little
nugget, business mail,

517
00:27:30,800 --> 00:27:32,300
which accounts for
tens of billions

518
00:27:32,300 --> 00:27:35,092
of in the United States,
and takes five days

519
00:27:35,092 --> 00:27:36,800
to be delivered today,
could be delivered

520
00:27:36,800 --> 00:27:38,927
in five seconds or less.

521
00:27:38,927 --> 00:27:41,260
What would that mean to the
productivity of the country?

522
00:27:41,260 --> 00:27:44,470


523
00:27:44,470 --> 00:27:47,020
Access to unique
common resources.

524
00:27:47,020 --> 00:27:51,310
Again, these could be
offered by entrepreneurs

525
00:27:51,310 --> 00:27:53,470
or they could be owned
by the government.

526
00:27:53,470 --> 00:27:56,050
We've often talked
about a digital library.

527
00:27:56,050 --> 00:27:57,850
There are many
other such examples

528
00:27:57,850 --> 00:28:02,110
that I'm sure you've discussed
that could fulfill this role.

529
00:28:02,110 --> 00:28:06,940
A network would facilitate that,
and we talked about the mail.

530
00:28:06,940 --> 00:28:12,070
So again, here is another way of
saying I have these weaknesses

531
00:28:12,070 --> 00:28:12,610
here.

532
00:28:12,610 --> 00:28:14,710
I have this supply here.

533
00:28:14,710 --> 00:28:17,380
How can I marry them together?

534
00:28:17,380 --> 00:28:19,670
Now, I could play
this game for hours.

535
00:28:19,670 --> 00:28:21,100
But I'm not.

536
00:28:21,100 --> 00:28:23,690
I'm going to try
to go to the limit.

537
00:28:23,690 --> 00:28:26,080
So if you'd like, you
can play this game.

538
00:28:26,080 --> 00:28:29,290
You can tick the boxes and
ask yourselves how could I

539
00:28:29,290 --> 00:28:32,650
use effectively these
kinds of technologies

540
00:28:32,650 --> 00:28:35,920
for other activities
in intelligent systems,

541
00:28:35,920 --> 00:28:37,630
or what have you, on
the factory floor.

542
00:28:37,630 --> 00:28:41,190


543
00:28:41,190 --> 00:28:43,960
I want to go to the limit.

544
00:28:43,960 --> 00:28:48,010
Who is going to use computers?

545
00:28:48,010 --> 00:28:51,940
Well, so far, mostly people.

546
00:28:51,940 --> 00:28:55,380
And what will people
do with computers?

547
00:28:55,380 --> 00:28:57,510
Well, what they do.

548
00:28:57,510 --> 00:29:00,190
They just use computers
to help them do it better.

549
00:29:00,190 --> 00:29:02,290
So what do people do?

550
00:29:02,290 --> 00:29:07,680
Well, let's look at what the
Department of Labor says.

551
00:29:07,680 --> 00:29:11,070
This is what people do in 1987.

552
00:29:11,070 --> 00:29:13,980
What I'm plotting here
is millions of workers,

553
00:29:13,980 --> 00:29:16,170
with 40 on the
peak of the scale.

554
00:29:16,170 --> 00:29:19,920
And I'm showing you in each
bar each of the major sectors

555
00:29:19,920 --> 00:29:21,240
of the United States economy.

556
00:29:21,240 --> 00:29:24,690
And I'll read them to
you-- agriculture, mining,

557
00:29:24,690 --> 00:29:29,580
construction, manufacturing,
transportation and utilities,

558
00:29:29,580 --> 00:29:34,500
trade, finance, insurance, and
real estate, and government.

559
00:29:34,500 --> 00:29:36,060
Now, what I've
marked here is I've

560
00:29:36,060 --> 00:29:40,530
marked as red the number
of white-collar workers.

561
00:29:40,530 --> 00:29:44,550
Now, white-collar workers are
the executives, the managers,

562
00:29:44,550 --> 00:29:47,730
the professional specialties,
the doctors, the technicians,

563
00:29:47,730 --> 00:29:50,760
the sales, the administrative
support people.

564
00:29:50,760 --> 00:29:54,060
I've marked with this
grid of green the service

565
00:29:54,060 --> 00:29:58,560
people, who are the household
helpers, the firefighters,

566
00:29:58,560 --> 00:30:02,820
the police, the food
providers, the cleaning

567
00:30:02,820 --> 00:30:06,900
people, and a lot of personal
services and handlers.

568
00:30:06,900 --> 00:30:11,340
Then I've marked with blue
the blue-collar workers,

569
00:30:11,340 --> 00:30:15,000
who are the precision
craftsmen, the repair

570
00:30:15,000 --> 00:30:20,160
people, the construction trades,
the operators, the fabricators,

571
00:30:20,160 --> 00:30:23,130
the laborers, and the
transportation people.

572
00:30:23,130 --> 00:30:27,930
And then the farm people
are sitting up there at 3%.

573
00:30:27,930 --> 00:30:32,620
Now, the red, which is
the white-collar workers,

574
00:30:32,620 --> 00:30:37,020
and I think that is our
dominant target in the future,

575
00:30:37,020 --> 00:30:40,140
comprises a rather
large percentage--

576
00:30:40,140 --> 00:30:45,000
55% or 63 million of
the working force.

577
00:30:45,000 --> 00:30:49,020
The blue-collar workers are 28%.

578
00:30:49,020 --> 00:30:51,270
And the so-called service--

579
00:30:51,270 --> 00:30:52,890
I say so-called,
because normally when

580
00:30:52,890 --> 00:30:55,860
we speak of services, we include
a lot of these white-collar

581
00:30:55,860 --> 00:31:02,110
into so-called services by this
definition are the order of 13%

582
00:31:02,110 --> 00:31:04,920
So if you want to think about
this as a limiting argument,

583
00:31:04,920 --> 00:31:09,120
there are 120 million people
out there in this country,

584
00:31:09,120 --> 00:31:13,052
and 55% of them, 60
million, used computers.

585
00:31:13,052 --> 00:31:15,510
And you can decide how many
computers you want each of them

586
00:31:15,510 --> 00:31:18,600
to use for white-collar.

587
00:31:18,600 --> 00:31:20,770
And then you can decide
for the blue-collar ones

588
00:31:20,770 --> 00:31:22,410
how many computers
each should have.

589
00:31:22,410 --> 00:31:24,690
And maybe you
arrive at, depending

590
00:31:24,690 --> 00:31:26,460
on your level of
optimism or pessimism,

591
00:31:26,460 --> 00:31:30,330
at 200, 300, 400, maybe
half a billion computers.

592
00:31:30,330 --> 00:31:31,140
I'm probably wrong.

593
00:31:31,140 --> 00:31:32,473
All these predictions are wrong.

594
00:31:32,473 --> 00:31:35,592
But ballpark, this is the kind
of number we're looking at.

595
00:31:35,592 --> 00:31:37,050
And if you get even
more grandiose,

596
00:31:37,050 --> 00:31:41,760
multiply it by 3 or 4 to
get world economy figures.

597
00:31:41,760 --> 00:31:44,550
But that's not quite the
limit that interests me.

598
00:31:44,550 --> 00:31:45,930
Let me proceed.

599
00:31:45,930 --> 00:31:50,640
I've done a calculation here,
which I find interesting.

600
00:31:50,640 --> 00:31:55,500
I multiply the fraction
of white-collar workers

601
00:31:55,500 --> 00:31:59,440
by the GNP in that industry.

602
00:31:59,440 --> 00:32:04,050
So if an industry like
trade has $800 billion,

603
00:32:04,050 --> 00:32:07,890
and the white-collar workers
in that industry are 60%,

604
00:32:07,890 --> 00:32:11,130
I multiply 60% by the $800
billion to get the number

605
00:32:11,130 --> 00:32:12,340
that I plot here.

606
00:32:12,340 --> 00:32:15,870
So here is exactly the same
industries, the same sectors.

607
00:32:15,870 --> 00:32:19,140
And the scale here
is in dollars.

608
00:32:19,140 --> 00:32:20,940
And you're looking
at $1 trillion

609
00:32:20,940 --> 00:32:22,680
at the top of the scale.

610
00:32:22,680 --> 00:32:26,250
And of course, everything has
to add up to $4.5 trillion,

611
00:32:26,250 --> 00:32:29,280
which is the GNP in 1987.

612
00:32:29,280 --> 00:32:33,510
What happens is that this
kind of limiting factor,

613
00:32:33,510 --> 00:32:37,530
which I call
white-collar leverage,

614
00:32:37,530 --> 00:32:41,260
is even higher than
the other percentage.

615
00:32:41,260 --> 00:32:45,420
So it exceeds 60% of the
economy. $2.7 trillion

616
00:32:45,420 --> 00:32:50,100
in the United States are
leveraged-- are controlled

617
00:32:50,100 --> 00:32:53,880
if you wish, by
white-collar people.

618
00:32:53,880 --> 00:32:56,250
Now let's just
pause for a second

619
00:32:56,250 --> 00:32:58,170
and go back to
the way I started.

620
00:32:58,170 --> 00:33:03,860
The US productivity was
7/10 of 1% per year growth.

621
00:33:03,860 --> 00:33:08,780
Suppose that we, with our
great and wondrous technology,

622
00:33:08,780 --> 00:33:13,010
could induce a 3% increase
in the productivity

623
00:33:13,010 --> 00:33:14,540
of the white-collar workers--

624
00:33:14,540 --> 00:33:17,430
3%.

625
00:33:17,430 --> 00:33:19,620
Reflecting this
by the 60% ratios

626
00:33:19,620 --> 00:33:23,100
and so forth, that would mean
2% more per year for the US.

627
00:33:23,100 --> 00:33:25,470
It would take us up
to 2.7%, suddenly

628
00:33:25,470 --> 00:33:30,540
we're up there with Japan
at the same rate of growth.

629
00:33:30,540 --> 00:33:36,270
I'm trying to impress on you
the great sensitivity between,

630
00:33:36,270 --> 00:33:38,940
on one hand, doing something at
the white-collar productivity

631
00:33:38,940 --> 00:33:42,600
level, and on the other hand,
effecting a major change

632
00:33:42,600 --> 00:33:44,760
in the productivity of the US.

633
00:33:44,760 --> 00:33:48,570
And the argument extends
to every nation on Earth

634
00:33:48,570 --> 00:33:52,470
that is on an advanced economy.

635
00:33:52,470 --> 00:33:57,180
Well, so it's sensitive,
and there are lots of people

636
00:33:57,180 --> 00:33:57,750
out there.

637
00:33:57,750 --> 00:34:02,360
But how far could we go?

638
00:34:02,360 --> 00:34:06,650
Now, this is an interesting
little chart it's $4.6 trillion

639
00:34:06,650 --> 00:34:09,139
of United States activity.

640
00:34:09,139 --> 00:34:14,030
Think of it as a little
kit where you can play.

641
00:34:14,030 --> 00:34:15,620
Here is one of your pawns.

642
00:34:15,620 --> 00:34:18,060
It's $100 billion,
that little rectangle.

643
00:34:18,060 --> 00:34:20,330
So you can move it around.

644
00:34:20,330 --> 00:34:22,219
Makes you feel good.

645
00:34:22,219 --> 00:34:28,219
And what I've done is
I fairly arbitrarily--

646
00:34:28,219 --> 00:34:32,389
not quite arbitrarily, but
I divided by my own taste

647
00:34:32,389 --> 00:34:36,380
in colors which sector, as
you go one notch deeper,

648
00:34:36,380 --> 00:34:39,920
would use blue-collar versus
white-collar computers.

649
00:34:39,920 --> 00:34:43,850
And then I put in black the
ones where I wasn't sure.

650
00:34:43,850 --> 00:34:45,830
And I'll tell you about that.

651
00:34:45,830 --> 00:34:47,389
Let's start with it.

652
00:34:47,389 --> 00:34:51,199
Real estate-- now
there is $500 billion

653
00:34:51,199 --> 00:34:53,449
of real estate changing hands.

654
00:34:53,449 --> 00:34:56,659
And that's counted in the
GNP as part of the GNP.

655
00:34:56,659 --> 00:35:00,740
And I think we could just
make computers go to GIPS,

656
00:35:00,740 --> 00:35:03,720
and that wouldn't change much
the value of the real estate.

657
00:35:03,720 --> 00:35:06,290
So we can change the value
of the paperwork that

658
00:35:06,290 --> 00:35:06,930
goes with it.

659
00:35:06,930 --> 00:35:09,890
So here, I think we
will be less effective.

660
00:35:09,890 --> 00:35:15,200
Trade at $800 billion of
wholesale and retail trade.

661
00:35:15,200 --> 00:35:18,120
And again, you're dealing
with the movement of goods.

662
00:35:18,120 --> 00:35:20,780
Now here, I think we can have
a bigger effect because we

663
00:35:20,780 --> 00:35:25,040
can locate the goods at
the best possible prices

664
00:35:25,040 --> 00:35:27,290
at the right places with
the use of computers--

665
00:35:27,290 --> 00:35:29,825
better than we do
today with humans.

666
00:35:29,825 --> 00:35:31,950
So here is an area where
we might have some impact.

667
00:35:31,950 --> 00:35:33,440
But again, I don't
know how much.

668
00:35:33,440 --> 00:35:35,960
Now, as you move it
away from these two

669
00:35:35,960 --> 00:35:39,800
things like insurance, banking,
the whole finance enterprise,

670
00:35:39,800 --> 00:35:45,670
and then, goodness,
government, huge amount

671
00:35:45,670 --> 00:35:48,550
of federal and state
government where I'm sure

672
00:35:48,550 --> 00:35:52,990
we can see ways of increasing
the productivity by 2% or 3%

673
00:35:52,990 --> 00:35:55,350
a year.

674
00:35:55,350 --> 00:35:57,210
And then on to the services.

675
00:35:57,210 --> 00:35:59,730
The legal profession that
we scream so much about

676
00:35:59,730 --> 00:36:01,740
is a very small sliver here.

677
00:36:01,740 --> 00:36:05,460
And, well, we are even smaller.

678
00:36:05,460 --> 00:36:10,590
We are in machinery, and you got
to go one notch below to find

679
00:36:10,590 --> 00:36:14,310
machinery, non electrical--

680
00:36:14,310 --> 00:36:15,630
electronic machinery.

681
00:36:15,630 --> 00:36:19,560
And then in there, we're
about $100 billion.

682
00:36:19,560 --> 00:36:21,570
I think you can
play your own games.

683
00:36:21,570 --> 00:36:24,150
I don't want to
belabor this more.

684
00:36:24,150 --> 00:36:26,040
The point I want
to make right now

685
00:36:26,040 --> 00:36:29,700
is that there is a
lot of room to go.

686
00:36:29,700 --> 00:36:32,310
There is a lot of room
with white-collar, as well

687
00:36:32,310 --> 00:36:35,040
as with blue-collar, computers.

688
00:36:35,040 --> 00:36:37,500
But I want to focus
on the white-collar,

689
00:36:37,500 --> 00:36:39,810
because of the great
sensitivity it has

690
00:36:39,810 --> 00:36:42,990
and because of another factor.

691
00:36:42,990 --> 00:36:44,910
Now, what is that factor?

692
00:36:44,910 --> 00:36:48,870
I told you that the United
States total productivity

693
00:36:48,870 --> 00:36:53,270
went up 7% in 10 years.

694
00:36:53,270 --> 00:36:55,570
Now, what happened to the
blue-collar productivity?

695
00:36:55,570 --> 00:36:58,690
What happened to the factory
workers, the precision workers,

696
00:36:58,690 --> 00:36:59,530
and all that?

697
00:36:59,530 --> 00:37:00,250
You can count.

698
00:37:00,250 --> 00:37:03,880
You can measure their output
and you can measure their input.

699
00:37:03,880 --> 00:37:04,720
It went up 20%.

700
00:37:04,720 --> 00:37:07,480


701
00:37:07,480 --> 00:37:09,610
Let us subtract.

702
00:37:09,610 --> 00:37:13,530
If blue-collar
productivity went up 20%,

703
00:37:13,530 --> 00:37:21,200
and total productivity went
up 7%, something went down.

704
00:37:21,200 --> 00:37:24,230
Now, the evil people out there
who do not love computers,

705
00:37:24,230 --> 00:37:28,430
and do not stay up until 3:00
AM hacking on their machines,

706
00:37:28,430 --> 00:37:30,830
tell me you did it.

707
00:37:30,830 --> 00:37:33,020
You did it with your computers.

708
00:37:33,020 --> 00:37:35,507
That's what's taking
up and dropping

709
00:37:35,507 --> 00:37:36,590
white-collar productivity.

710
00:37:36,590 --> 00:37:37,350
I don't think so.

711
00:37:37,350 --> 00:37:39,600
I think in some instances,
we've done it incidentally.

712
00:37:39,600 --> 00:37:42,710
But by and large there have
been a lot of other factors.

713
00:37:42,710 --> 00:37:44,210
And there's a very
interesting paper

714
00:37:44,210 --> 00:37:46,280
on that by Les Thoreau,
which brings up

715
00:37:46,280 --> 00:37:50,270
how all these increases of
benefits and the legal aspect

716
00:37:50,270 --> 00:37:52,430
have really caused a
tremendous decrease

717
00:37:52,430 --> 00:37:54,810
in white-collar productivity.

718
00:37:54,810 --> 00:37:59,120
But the point I'm making now is,
all right, so we didn't do it.

719
00:37:59,120 --> 00:38:01,980
Or maybe we did
some of this mishap.

720
00:38:01,980 --> 00:38:06,420
If computers are that great,
why haven't we already

721
00:38:06,420 --> 00:38:08,578
improved white-collar
productivity.

722
00:38:08,578 --> 00:38:11,390


723
00:38:11,390 --> 00:38:12,960
Well, I have some thoughts.

724
00:38:12,960 --> 00:38:14,760
I'm sure you have your own.

725
00:38:14,760 --> 00:38:18,590
Let me give you my
first set of reasons.

726
00:38:18,590 --> 00:38:21,110
I think we've been derelict.

727
00:38:21,110 --> 00:38:22,940
I think we've been remiss.

728
00:38:22,940 --> 00:38:26,090
I don't think we've
worried about productivity.

729
00:38:26,090 --> 00:38:28,700
I think we've had too much fun.

730
00:38:28,700 --> 00:38:31,970
Those of us who make the
machines and the programs

731
00:38:31,970 --> 00:38:35,030
have thrown them at those
of you who use them,

732
00:38:35,030 --> 00:38:39,020
and we've been seduced by
the rapid growth of the field

733
00:38:39,020 --> 00:38:41,930
on both sides of this divide.

734
00:38:41,930 --> 00:38:45,350
And I don't think we've
asked too many times

735
00:38:45,350 --> 00:38:47,540
how we are increasing
productivity

736
00:38:47,540 --> 00:38:49,640
through use of computers.

737
00:38:49,640 --> 00:38:54,470
I identify three pitfall types,
and there are probably more.

738
00:38:54,470 --> 00:38:59,330
First, adding the computer to
whatever was going on before.

739
00:38:59,330 --> 00:39:01,610
I think you've seen
that throughout.

740
00:39:01,610 --> 00:39:02,970
A little example here--

741
00:39:02,970 --> 00:39:05,610
I've seen it just a
couple of weeks ago.

742
00:39:05,610 --> 00:39:06,985
A waiter takes
the order and then

743
00:39:06,985 --> 00:39:08,693
he goes and keeps some
things over there,

744
00:39:08,693 --> 00:39:10,820
and does everything he used
to do plus some keying.

745
00:39:10,820 --> 00:39:12,320
I don't know what
the keying is for.

746
00:39:12,320 --> 00:39:15,920


747
00:39:15,920 --> 00:39:17,690
This one really burns me up.

748
00:39:17,690 --> 00:39:20,270
And it burns me up every
time I get a new program,

749
00:39:20,270 --> 00:39:23,360
and I get a lot of new
programs and try them.

750
00:39:23,360 --> 00:39:26,630
And I can't believe that the
manual for a word processor

751
00:39:26,630 --> 00:39:28,910
is this fat, and all
it teaches is how

752
00:39:28,910 --> 00:39:31,760
to put ink markings on paper.

753
00:39:31,760 --> 00:39:36,770
Sure you have power,
but six inches?

754
00:39:36,770 --> 00:39:40,597
I mean, at the limit, I could go
to medical school for 12 years

755
00:39:40,597 --> 00:39:42,930
and become a brain surgeon
and increase my productivity.

756
00:39:42,930 --> 00:39:45,230
But there must be a
limit to the amount

757
00:39:45,230 --> 00:39:47,390
of learning that
we ask people to do

758
00:39:47,390 --> 00:39:49,010
before they use computers.

759
00:39:49,010 --> 00:39:51,680
So I think we ought
to put a moratorium

760
00:39:51,680 --> 00:39:55,175
on excessive learning
for computers.

761
00:39:55,175 --> 00:39:58,178
[LAUGHTER]

762
00:39:58,178 --> 00:39:59,080


763
00:39:59,080 --> 00:40:03,730
The third one is that the
presence of the computer

764
00:40:03,730 --> 00:40:06,840
somehow induces us to work
more and increases the input

765
00:40:06,840 --> 00:40:08,710
in the denominator here.

766
00:40:08,710 --> 00:40:10,180
And of course, I did this one.

767
00:40:10,180 --> 00:40:11,620
I prepared these slides.

768
00:40:11,620 --> 00:40:12,550
But we tried to--

769
00:40:12,550 --> 00:40:15,070
[LAUGHTER]

770
00:40:15,070 --> 00:40:17,180
I think you know what
I'm talking about.

771
00:40:17,180 --> 00:40:21,280
So in one sense, we haven't
been keeping our eye focused

772
00:40:21,280 --> 00:40:22,600
on productivity.

773
00:40:22,600 --> 00:40:24,410
And it's not difficult
to focus our eye.

774
00:40:24,410 --> 00:40:26,300
It doesn't take
higher mathematics.

775
00:40:26,300 --> 00:40:28,690
Let me give you
a simple example.

776
00:40:28,690 --> 00:40:31,090
Take a situation where
you have something

777
00:40:31,090 --> 00:40:34,270
before the computer and
something after the computer.

778
00:40:34,270 --> 00:40:41,180
And you ask yourself, what gain
in post-computer productivity

779
00:40:41,180 --> 00:40:41,740
have you had.

780
00:40:41,740 --> 00:40:44,890
This numerator divided
by this labor cost only,

781
00:40:44,890 --> 00:40:48,190
relative to this, has
with the same labor--

782
00:40:48,190 --> 00:40:51,580
have you raised your
ability to produce 50% more

783
00:40:51,580 --> 00:40:53,170
if you consider
that you now have

784
00:40:53,170 --> 00:40:54,820
to also pay not
only for the labor

785
00:40:54,820 --> 00:40:56,510
but for the computer expense?

786
00:40:56,510 --> 00:40:57,760
And I've taken a simple thing.

787
00:40:57,760 --> 00:40:59,980
I've taken the computer
cost and divided it by 3.

788
00:40:59,980 --> 00:41:01,930
I say we amortize in 3 years.

789
00:41:01,930 --> 00:41:04,210
And to compare it
with a labor input,

790
00:41:04,210 --> 00:41:07,750
I've taken a direct input
around $30,000 to $40,000,

791
00:41:07,750 --> 00:41:08,920
put some maintenance.

792
00:41:08,920 --> 00:41:10,700
I did not put in
direct overhead and all

793
00:41:10,700 --> 00:41:12,700
that because that's the
white-collar stuff we're

794
00:41:12,700 --> 00:41:14,080
going to solve differently.

795
00:41:14,080 --> 00:41:17,650
I just took the direct stuff,
and I drew basically a tradeoff

796
00:41:17,650 --> 00:41:18,310
curve.

797
00:41:18,310 --> 00:41:20,500
And the curve says
that at this threshold

798
00:41:20,500 --> 00:41:23,800
here, you are not
gaining and you're not

799
00:41:23,800 --> 00:41:26,008
losing anything in this
multi-factor productivity.

800
00:41:26,008 --> 00:41:27,550
You're exactly where
you were before.

801
00:41:27,550 --> 00:41:30,490
You get as much output
divided by input as you used

802
00:41:30,490 --> 00:41:33,020
to get before in labor terms.

803
00:41:33,020 --> 00:41:36,700
Now, let's look at
office automation.

804
00:41:36,700 --> 00:41:40,180
And in particular, let's look
at the PCs and word processing.

805
00:41:40,180 --> 00:41:42,850
There, we have
reported gains of 30%.

806
00:41:42,850 --> 00:41:44,500
We know what the costs are.

807
00:41:44,500 --> 00:41:47,800
So we can plot a
little blob and see

808
00:41:47,800 --> 00:41:49,090
that it's in the good region.

809
00:41:49,090 --> 00:41:51,460
We are improving
our productivity

810
00:41:51,460 --> 00:41:53,710
relative to the cost that
we're paying for the capital

811
00:41:53,710 --> 00:41:55,600
equipment.

812
00:41:55,600 --> 00:41:57,790
Now, you can choose to
put your favorite activity

813
00:41:57,790 --> 00:41:58,570
on this chart.

814
00:41:58,570 --> 00:42:01,340


815
00:42:01,340 --> 00:42:03,350
And let me try one--

816
00:42:03,350 --> 00:42:05,180
robotics.

817
00:42:05,180 --> 00:42:09,140
The average robot today
costs over $100,000.

818
00:42:09,140 --> 00:42:11,810
In fact, it's around $150,000.

819
00:42:11,810 --> 00:42:14,060
And except for very
rare instances,

820
00:42:14,060 --> 00:42:16,580
where it does a lot of
painting or welding,

821
00:42:16,580 --> 00:42:21,410
its productivity barely
increases above the 50% level.

822
00:42:21,410 --> 00:42:23,960
So it's sitting somewhere
up here on the curve,

823
00:42:23,960 --> 00:42:26,210
on the bad region.

824
00:42:26,210 --> 00:42:30,230
Computer-aided drafting
increases the output product

825
00:42:30,230 --> 00:42:31,940
of drafting by a factor of 2--

826
00:42:31,940 --> 00:42:35,010
200%, and the capital
costs are somewhere here.

827
00:42:35,010 --> 00:42:37,250
So that one is over
in this region.

828
00:42:37,250 --> 00:42:40,700
No wonder that it works, as does
computer-aided design for chip

829
00:42:40,700 --> 00:42:42,390
manufacture and so forth.

830
00:42:42,390 --> 00:42:44,600
So indeed, by using
simple arguments

831
00:42:44,600 --> 00:42:48,140
like this, if we start
changing our mindset,

832
00:42:48,140 --> 00:42:54,680
we might be able to worry about
productive uses of computers.

833
00:42:54,680 --> 00:42:57,470
But this, I think, has been
only one of the reasons

834
00:42:57,470 --> 00:43:01,520
that we were unable, up to now,
to increase our productivity

835
00:43:01,520 --> 00:43:02,210
with computers.

836
00:43:02,210 --> 00:43:06,380
The other one is, I think
that we don't really

837
00:43:06,380 --> 00:43:09,785
know how to measure the
value of information.

838
00:43:09,785 --> 00:43:11,160
I mean, it's nice
and easy to say

839
00:43:11,160 --> 00:43:12,810
that the value of the
output of this machine

840
00:43:12,810 --> 00:43:13,740
is so many widgets.

841
00:43:13,740 --> 00:43:16,590
At $10 a widget, bang,
you get the output.

842
00:43:16,590 --> 00:43:19,740
But what is the value if you're
looking at the output of a word

843
00:43:19,740 --> 00:43:23,340
processor, or if you're looking
at the output of a CAD--

844
00:43:23,340 --> 00:43:26,520
a computer-aided design
program that produces a design?

845
00:43:26,520 --> 00:43:28,620
What's the value of that design?

846
00:43:28,620 --> 00:43:31,610
A lot has been
said on the topic.

847
00:43:31,610 --> 00:43:35,970
Let me just say that
the traditional notion

848
00:43:35,970 --> 00:43:37,620
of the value of
information being

849
00:43:37,620 --> 00:43:41,282
a signal in a perfect
marketplace doesn't work.

850
00:43:41,282 --> 00:43:43,740
This is the traditional economic
definition of information.

851
00:43:43,740 --> 00:43:48,870
They say it helps to
decrease the entropy.

852
00:43:48,870 --> 00:43:50,730
And so you can
calculate its value

853
00:43:50,730 --> 00:43:52,860
by the amount by which
it decreases the entropy.

854
00:43:52,860 --> 00:43:56,400
You learn that GM is
going to buy Ford,

855
00:43:56,400 --> 00:43:59,010
and that decreases
the uncertainty,

856
00:43:59,010 --> 00:44:01,230
and that, you can
calculate the value.

857
00:44:01,230 --> 00:44:03,390
I think that's not
very helpful when

858
00:44:03,390 --> 00:44:06,090
I have a computer-aided
design program

859
00:44:06,090 --> 00:44:08,250
and I'm producing a
final design for a chip.

860
00:44:08,250 --> 00:44:11,430
I just don't understand how I'm
decreasing the entropy over all

861
00:44:11,430 --> 00:44:13,980
the designs for a chip.

862
00:44:13,980 --> 00:44:17,190
So let me give you
an alternative here.

863
00:44:17,190 --> 00:44:18,640
You can think of your own.

864
00:44:18,640 --> 00:44:22,900
But this is my proposal
I give you a definition

865
00:44:22,900 --> 00:44:25,960
that information has
economic value if and only

866
00:44:25,960 --> 00:44:29,290
if it can lead to the
acquisition of goods

867
00:44:29,290 --> 00:44:31,160
and services--

868
00:44:31,160 --> 00:44:33,440
and I couldn't resist
the recursive definition,

869
00:44:33,440 --> 00:44:36,587
part or of other economic
valuable information.

870
00:44:36,587 --> 00:44:39,170
Of course, because you process
information to get information.

871
00:44:39,170 --> 00:44:43,610
For example, suppose that
what you're trying to do

872
00:44:43,610 --> 00:44:48,260
is produce chips and make money
by producing silicon chips.

873
00:44:48,260 --> 00:44:50,752
So all the factors of
production are here.

874
00:44:50,752 --> 00:44:51,710
You need so much labor.

875
00:44:51,710 --> 00:44:53,585
You need so much capital
equipment, this kind

876
00:44:53,585 --> 00:44:55,130
of devices, this kind of plant.

877
00:44:55,130 --> 00:44:57,800
You've got to make
this kind of profit.

878
00:44:57,800 --> 00:44:59,570
All I'm saying is
that to the extent

879
00:44:59,570 --> 00:45:02,720
that a piece of
information, like the design

880
00:45:02,720 --> 00:45:04,670
of a particular chip--

881
00:45:04,670 --> 00:45:07,220
four-function calculator
or what have you,

882
00:45:07,220 --> 00:45:10,100
the value of that
piece of information

883
00:45:10,100 --> 00:45:12,770
has to also go in there
like a legitimate factor

884
00:45:12,770 --> 00:45:13,490
of production.

885
00:45:13,490 --> 00:45:15,770
And then you add all of
these up and they make up

886
00:45:15,770 --> 00:45:17,600
the value of the
total, or you subtract

887
00:45:17,600 --> 00:45:20,150
this from that to get
the value of that.

888
00:45:20,150 --> 00:45:22,520
Now, if you need a couple
of pieces of information--

889
00:45:22,520 --> 00:45:26,810
perhaps a library of cells
for your design program

890
00:45:26,810 --> 00:45:30,890
and the output of your
logic gate analysis,

891
00:45:30,890 --> 00:45:33,560
then the value of these
pieces of information,

892
00:45:33,560 --> 00:45:37,490
added to the cost of
information processing amortized

893
00:45:37,490 --> 00:45:39,400
over all its users,
should give you

894
00:45:39,400 --> 00:45:40,650
the value of that information.

895
00:45:40,650 --> 00:45:42,883
Now, don't worry
about determinism

896
00:45:42,883 --> 00:45:44,300
of how you figure
the value of one

897
00:45:44,300 --> 00:45:45,570
from the value of the other.

898
00:45:45,570 --> 00:45:47,420
This is where business
decisions come in

899
00:45:47,420 --> 00:45:52,040
as they do in figuring out how
much are things worth here.

900
00:45:52,040 --> 00:45:55,370
But the point I'm making
is that information,

901
00:45:55,370 --> 00:45:57,830
the value of information,
and information

902
00:45:57,830 --> 00:46:01,730
itself have to be treated
like first-class factors

903
00:46:01,730 --> 00:46:04,400
of production, not like
second-class citizens

904
00:46:04,400 --> 00:46:06,830
because you did them
on a piece of paper.

905
00:46:06,830 --> 00:46:07,640
That's the message.

906
00:46:07,640 --> 00:46:09,533
Now, there are
differences, and I don't

907
00:46:09,533 --> 00:46:10,700
have the time to go in them.

908
00:46:10,700 --> 00:46:16,695
But unlike a brick here, which
has the same roughly value,

909
00:46:16,695 --> 00:46:18,320
if you use it for
this kind of building

910
00:46:18,320 --> 00:46:21,530
or another kind of
building, this information

911
00:46:21,530 --> 00:46:24,230
changes its value, depending
on what it leads to.

912
00:46:24,230 --> 00:46:28,340
So like specialized labor,
you must always keep in mind

913
00:46:28,340 --> 00:46:29,510
where it's leading.

914
00:46:29,510 --> 00:46:31,400
Be that as it may,
the point I want

915
00:46:31,400 --> 00:46:34,250
to leave you with
from this slide

916
00:46:34,250 --> 00:46:37,010
is that we need
measures of information

917
00:46:37,010 --> 00:46:38,390
and the value of
information that

918
00:46:38,390 --> 00:46:43,250
are far more realistic than
what we've had in the past.

919
00:46:43,250 --> 00:46:47,030
Now I'd like to bring
this talk to an end,

920
00:46:47,030 --> 00:46:49,400
and I would like to
make suggestions as

921
00:46:49,400 --> 00:46:53,038
to what it is that we might do.

922
00:46:53,038 --> 00:46:54,455
Now, I want to ask
you a question.

923
00:46:54,455 --> 00:46:57,290


924
00:46:57,290 --> 00:47:01,000
Have you ever seen
architects throwing

925
00:47:01,000 --> 00:47:06,010
windows and doors
and bricks at people

926
00:47:06,010 --> 00:47:08,080
and saying, here, build a house?

927
00:47:08,080 --> 00:47:10,640


928
00:47:10,640 --> 00:47:13,080
Sounds familiar?

929
00:47:13,080 --> 00:47:15,340
We've been doing a little
bit of that, haven't we?

930
00:47:15,340 --> 00:47:18,050
We have been throwing at
people "this great machine that

931
00:47:18,050 --> 00:47:21,230
has 20% more MIPS," or
this great program here

932
00:47:21,230 --> 00:47:23,192
that, if you only found
the right application,

933
00:47:23,192 --> 00:47:24,400
you could do wonders with it.

934
00:47:24,400 --> 00:47:27,879
[LAUGHTER]

935
00:47:27,879 --> 00:47:31,092


936
00:47:31,092 --> 00:47:32,550
Ladies and gentlemen,
I think we've

937
00:47:32,550 --> 00:47:35,100
got to begin behaving
like architects.

938
00:47:35,100 --> 00:47:36,990
That's my first message.

939
00:47:36,990 --> 00:47:40,510
Have you ever heard architects
say, I'm going out today.

940
00:47:40,510 --> 00:47:43,950
I'm going to be working on
applications of architecture.

941
00:47:43,950 --> 00:47:45,570
Have you ever heard
them say that?

942
00:47:45,570 --> 00:47:46,650
Of course not.

943
00:47:46,650 --> 00:47:48,990
No architect in his
right mind thinks of what

944
00:47:48,990 --> 00:47:52,680
he's doing as applications
of his premier discipline

945
00:47:52,680 --> 00:47:54,210
called architecture.

946
00:47:54,210 --> 00:47:58,590
Yet, we, cocky as we should
be from the great discoveries

947
00:47:58,590 --> 00:48:01,470
we've made, think of the
world as applications

948
00:48:01,470 --> 00:48:02,940
of computer science.

949
00:48:02,940 --> 00:48:05,970
Wrong, rotate.

950
00:48:05,970 --> 00:48:11,640
We are supposed to go out
there and build libraries,

951
00:48:11,640 --> 00:48:13,170
just like the
architects, or maybe

952
00:48:13,170 --> 00:48:14,580
they're digital libraries.

953
00:48:14,580 --> 00:48:20,270
We're supposed to go out
there and do projects thinking

954
00:48:20,270 --> 00:48:23,180
of the project as the
primary objective,

955
00:48:23,180 --> 00:48:25,670
not of them as being
applications of our field.

956
00:48:25,670 --> 00:48:30,110
We have to use the tools of
our trade to make that happen.

957
00:48:30,110 --> 00:48:33,800
That is, I think, my big
message about the mindset

958
00:48:33,800 --> 00:48:36,480
change that I'd
like to see happen.

959
00:48:36,480 --> 00:48:37,700
Now, there are many others.

960
00:48:37,700 --> 00:48:40,880
General Motors threw the
equipment at the people

961
00:48:40,880 --> 00:48:42,920
and said, we automate it.

962
00:48:42,920 --> 00:48:45,380
And then they lost.

963
00:48:45,380 --> 00:48:48,110
We cannot just throw
computers at people.

964
00:48:48,110 --> 00:48:51,830
We must very seriously
consider how to integrate them

965
00:48:51,830 --> 00:48:54,380
with people's work,
people's objectives,

966
00:48:54,380 --> 00:48:56,540
and what people
really like to do.

967
00:48:56,540 --> 00:48:59,180
And this is the second
important message.

968
00:48:59,180 --> 00:49:01,880
As I said earlier,
we've got to start using

969
00:49:01,880 --> 00:49:04,190
computer productivity measures.

970
00:49:04,190 --> 00:49:06,020
In the old days,
we had horsepower.

971
00:49:06,020 --> 00:49:07,730
If somebody said,
here's a diesel,

972
00:49:07,730 --> 00:49:13,170
it's got 100 horsepower, and it
costs no more than four horses,

973
00:49:13,170 --> 00:49:14,450
you said, all right.

974
00:49:14,450 --> 00:49:17,780
But now you give me a machine,
and what's the horsepower?

975
00:49:17,780 --> 00:49:19,413
It's MIPS.

976
00:49:19,413 --> 00:49:21,080
You know it's very
hard for me to relate

977
00:49:21,080 --> 00:49:26,210
MIPS to increasing the
productivity in the office

978
00:49:26,210 --> 00:49:27,690
or on the factory floor.

979
00:49:27,690 --> 00:49:30,020
So we need to develop
the horsepower measures

980
00:49:30,020 --> 00:49:32,600
for those activities as
technicians and scientists

981
00:49:32,600 --> 00:49:34,700
in this business.

982
00:49:34,700 --> 00:49:37,190
We have to focus on the
discovery of algorithms,

983
00:49:37,190 --> 00:49:40,980
software, and hardware that
can increase our productivity.

984
00:49:40,980 --> 00:49:45,680
Now, the network, I think, is
a big, big boon to all this.

985
00:49:45,680 --> 00:49:47,630
Imagine in the
United States, we're

986
00:49:47,630 --> 00:49:49,760
number one in the world
in computer science

987
00:49:49,760 --> 00:49:50,720
and technology.

988
00:49:50,720 --> 00:49:53,870
Why can't we use that great
science and technology

989
00:49:53,870 --> 00:49:56,660
to improve our productivity
across the board?

990
00:49:56,660 --> 00:49:59,990
3% per year, 2%
per year, increase

991
00:49:59,990 --> 00:50:03,380
in white-collar productivity,
which means that over 20 years

992
00:50:03,380 --> 00:50:08,480
we set as a goal to get 40%
increase in productivity.

993
00:50:08,480 --> 00:50:11,450
That doesn't sound unrealistic
for the white-collar work.

994
00:50:11,450 --> 00:50:14,772
That amount of activity could
make a tremendous difference

995
00:50:14,772 --> 00:50:15,480
for this country.

996
00:50:15,480 --> 00:50:16,850
This is big stakes.

997
00:50:16,850 --> 00:50:20,810
We, our profession, can
really help tremendously

998
00:50:20,810 --> 00:50:25,170
in the economy both of this
country and of the world.

999
00:50:25,170 --> 00:50:28,490
All we have to do is
get it through our heads

1000
00:50:28,490 --> 00:50:31,920
that we ought to be
thinking of productivity.

1001
00:50:31,920 --> 00:50:35,060
Now, I'm not asking that
people stop their great work.

1002
00:50:35,060 --> 00:50:38,150
I'm not asking that
people stop having fun.

1003
00:50:38,150 --> 00:50:39,680
You can still
invent great things.

1004
00:50:39,680 --> 00:50:43,100
You can have tremendous
amount of fun as a producer.

1005
00:50:43,100 --> 00:50:46,130
But think a little bit
about how it might be used

1006
00:50:46,130 --> 00:50:48,360
and how to measure what
you would gain for it.

1007
00:50:48,360 --> 00:50:50,690
And on the other side
of the fence as users,

1008
00:50:50,690 --> 00:50:53,810
let's not get arbitrarily
the next fad and use it.

1009
00:50:53,810 --> 00:50:55,580
Let's also apply
the calculations.

1010
00:50:55,580 --> 00:50:57,770
Let's get out the
time study chart

1011
00:50:57,770 --> 00:51:00,420
and see how can we
benefit from this.

1012
00:51:00,420 --> 00:51:03,860
I think working together, we
can achieve a great deal here

1013
00:51:03,860 --> 00:51:04,943
if we change our attitude.

1014
00:51:04,943 --> 00:51:06,693
And it is this thought
I want to leave you

1015
00:51:06,693 --> 00:51:07,880
with for the next 30 years.

1016
00:51:07,880 --> 00:51:08,750
Thank you.

1017
00:51:08,750 --> 00:51:12,100
[MUSIC PLAYING]

1018
00:51:12,100 --> 00:51:45,174