1
00:00:01,760 --> 00:00:03,699
Join us as we gather around the hedge,

2
00:00:03,759 --> 00:00:05,139
where we dig into technology,

3
00:00:05,440 --> 00:00:08,160
business, and culture with the finest minds in

4
00:00:08,160 --> 00:00:09,139
computer networking.

5
00:00:21,054 --> 00:00:21,795
Well, hello,

6
00:00:22,814 --> 00:00:24,355
Jeff. How are you today?

7
00:00:24,974 --> 00:00:27,130
Hi, Russ. No. I'm good. I'm good. It's

8
00:00:27,130 --> 00:00:29,370
down here in the Southern Hemisphere. It's You're

9
00:00:29,370 --> 00:00:32,329
upside down. Summer. Yeah. I am I am

10
00:00:32,329 --> 00:00:34,250
upside down, and things are getting, you know,

11
00:00:34,250 --> 00:00:35,229
warm and green,

12
00:00:35,770 --> 00:00:37,149
which, I love.

13
00:00:37,690 --> 00:00:39,070
Here, it was.

14
00:00:39,935 --> 00:00:41,715
The high today was around

15
00:00:42,655 --> 00:00:45,395
10 degrees centigrade or a bit less.

16
00:00:45,935 --> 00:00:47,314
And I don't love that.

17
00:00:48,734 --> 00:00:49,475
And overnight,

18
00:00:49,774 --> 00:00:50,434
it was,

19
00:00:51,295 --> 00:00:52,994
I don't know, down into

20
00:00:54,500 --> 00:00:56,899
the twenties here, which is negative eight to

21
00:00:56,899 --> 00:01:00,119
negative 10 centigrade something along those lines.

22
00:01:00,659 --> 00:01:02,260
Not feeling the love for me on that.

23
00:01:02,260 --> 00:01:03,000
No. No.

24
00:01:03,460 --> 00:01:06,340
That's that's pretty actually, pretty common for Tennessee.

25
00:01:06,340 --> 00:01:08,645
This for Eastern Tennessee and being in the

26
00:01:08,645 --> 00:01:10,805
foothills of the mountains this time of year.

27
00:01:10,805 --> 00:01:12,344
We're on the West side of the mountains,

28
00:01:12,645 --> 00:01:14,484
and so we get all the weather that

29
00:01:14,484 --> 00:01:15,944
doesn't make it over the mountains.

30
00:01:16,564 --> 00:01:19,145
It all stops with us because it's So

31
00:01:20,564 --> 00:01:22,105
You're just lucky, I guess.

32
00:01:23,209 --> 00:01:24,969
Well, we don't get the nor'easters that come

33
00:01:24,969 --> 00:01:26,170
up the East side of the mountains. And,

34
00:01:26,170 --> 00:01:27,790
actually, the nor'easters are horrible.

35
00:01:28,250 --> 00:01:30,170
That's the worst weather in The US, I

36
00:01:30,170 --> 00:01:31,390
think, is the nor'easters

37
00:01:32,090 --> 00:01:33,530
that go up the East Coast, that come

38
00:01:33,530 --> 00:01:35,849
up the Florida Coast, hit North Carolina, South

39
00:01:35,849 --> 00:01:36,349
Carolina,

40
00:01:36,784 --> 00:01:38,704
and then work their way all the way

41
00:01:38,704 --> 00:01:39,525
up the coastline.

42
00:01:40,145 --> 00:01:42,625
Holy mackerel. Those things are those are some

43
00:01:42,625 --> 00:01:43,924
crazy, crazy storms.

44
00:01:44,305 --> 00:01:45,045
But, anyway,

45
00:01:45,424 --> 00:01:46,245
that said,

46
00:01:46,865 --> 00:01:47,765
we were talking

47
00:01:49,210 --> 00:01:50,650
by the way, Jeff, I mean, maybe you

48
00:01:50,650 --> 00:01:52,890
wanna say hi to people before Oh, hi,

49
00:01:52,890 --> 00:01:53,870
everyone. Yeah.

50
00:01:54,170 --> 00:01:56,569
I'm Jeff Houston. As might might be obvious,

51
00:01:56,569 --> 00:01:58,250
I I live in the Southern Hemisphere, in

52
00:01:58,250 --> 00:01:59,549
Australia to be precise,

53
00:01:59,930 --> 00:02:02,534
on the East Coast where right now it's

54
00:02:02,534 --> 00:02:05,854
all turning gloriously summer, which is highly enjoyable.

55
00:02:05,854 --> 00:02:06,134
And,

56
00:02:07,894 --> 00:02:09,574
yes. That's me. I work as the chief

57
00:02:09,574 --> 00:02:12,215
scientist at APNIC, which means I get paid

58
00:02:12,215 --> 00:02:13,194
to think thoughts.

59
00:02:13,974 --> 00:02:14,935
Oh, cool. Awesome.

60
00:02:15,735 --> 00:02:18,349
Do. Think thoughts. So before we started, we

61
00:02:18,349 --> 00:02:20,909
were gonna talk about Moore's law and the

62
00:02:20,909 --> 00:02:22,210
Internet and such.

63
00:02:22,510 --> 00:02:24,669
And you started off talking about telephone networks.

64
00:02:24,669 --> 00:02:26,349
Now if you've listened to me for a

65
00:02:26,349 --> 00:02:28,189
while, some people don't know this, but I

66
00:02:28,189 --> 00:02:30,224
started out in The US Air Force doing

67
00:02:30,224 --> 00:02:32,384
electronic. Actually, I started long before the US

68
00:02:32,384 --> 00:02:33,685
Air Force doing electronics.

69
00:02:34,064 --> 00:02:35,985
I got my immateriality license when I was

70
00:02:35,985 --> 00:02:36,485
12.

71
00:02:36,944 --> 00:02:38,944
So I was actually doing 40 words a

72
00:02:38,944 --> 00:02:40,625
minute Morse code when I was 12 years

73
00:02:40,625 --> 00:02:41,924
old, and I was doing

74
00:02:42,349 --> 00:02:45,229
climbing towers and hanging inverted v's and long

75
00:02:45,229 --> 00:02:45,729
wire

76
00:02:46,030 --> 00:02:48,990
40 meter and 10 meters and hand adjusting

77
00:02:48,990 --> 00:02:53,709
and soldering together YAGI antenna antennas and all

78
00:02:53,709 --> 00:02:56,189
those LPDs, log period dipoles, and all that

79
00:02:56,189 --> 00:02:56,930
other stuff.

80
00:02:57,294 --> 00:02:58,655
When I was in the air force, we

81
00:02:58,655 --> 00:02:59,155
had

82
00:02:59,534 --> 00:03:00,034
a

83
00:03:01,055 --> 00:03:03,694
an FPS 77, which was a storm detection

84
00:03:03,694 --> 00:03:04,194
radar,

85
00:03:04,574 --> 00:03:06,894
which was installed in 1964.

86
00:03:06,894 --> 00:03:08,435
It was serial number one.

87
00:03:08,974 --> 00:03:11,099
It was a weather radar, serial number one

88
00:03:11,900 --> 00:03:14,319
of the mid range. There was one earlier.

89
00:03:14,540 --> 00:03:16,639
There was a model before the 77,

90
00:03:16,860 --> 00:03:18,719
then they came out with the 77,

91
00:03:18,780 --> 00:03:20,780
and now they have Nexrad, which is all

92
00:03:20,780 --> 00:03:21,280
computerized

93
00:03:21,819 --> 00:03:22,639
stuff stuff.

94
00:03:23,340 --> 00:03:25,634
And this was all tube type. We even

95
00:03:25,634 --> 00:03:28,134
had tube type operational amplifiers,

96
00:03:30,034 --> 00:03:31,574
and it was all high voltage.

97
00:03:32,275 --> 00:03:34,435
Everything was high voltage in the whole thing.

98
00:03:34,435 --> 00:03:34,935
Absolutely.

99
00:03:35,314 --> 00:03:38,219
Yeah. Yeah. The magnetron was fed by a

100
00:03:38,460 --> 00:03:39,840
flywheel circuit that

101
00:03:40,139 --> 00:03:40,639
pushed,

102
00:03:41,340 --> 00:03:41,740
I,

103
00:03:42,139 --> 00:03:44,960
a 100 watts at, you know, one volt

104
00:03:45,099 --> 00:03:46,639
or 10 volts or something.

105
00:03:46,939 --> 00:03:48,540
But you said it was all same. It

106
00:03:48,540 --> 00:03:51,280
it was all tubes. Yeah? Yes. Magnetron,

107
00:03:51,580 --> 00:03:52,560
receive circuit,

108
00:03:52,875 --> 00:03:55,055
everything was all tubes. There were no

109
00:03:55,514 --> 00:03:57,375
digital electronics in it.

110
00:03:57,754 --> 00:03:59,594
And then on the other side of my

111
00:03:59,594 --> 00:04:01,435
world I mean, I worked on lots of

112
00:04:01,435 --> 00:04:03,435
other things that were like that, very old.

113
00:04:03,435 --> 00:04:05,354
I worked on the telco network from time

114
00:04:05,354 --> 00:04:07,775
to time. And k g 80 fours,

115
00:04:08,759 --> 00:04:09,959
which I don't know if you know much

116
00:04:09,959 --> 00:04:11,799
about 2 threes and k g 80 fours

117
00:04:11,799 --> 00:04:12,699
crypto gear.

118
00:04:13,159 --> 00:04:15,400
But the telco system was based on stroger

119
00:04:15,400 --> 00:04:15,900
switches.

120
00:04:16,360 --> 00:04:18,779
We had a 10,000 line frame

121
00:04:19,959 --> 00:04:22,300
that was based on paper wrapped cable

122
00:04:23,935 --> 00:04:25,154
and stroger switches.

123
00:04:26,574 --> 00:04:28,894
Right. Right. And, you know, that was the

124
00:04:28,894 --> 00:04:31,935
the early, you know, computer sorry, telephone networks

125
00:04:31,935 --> 00:04:33,634
were effectively electromechanical.

126
00:04:34,334 --> 00:04:37,475
The stroger switches actually really created an electronic

127
00:04:37,535 --> 00:04:38,035
circuit.

128
00:04:38,654 --> 00:04:40,419
And the issue was, you know, in some

129
00:04:40,419 --> 00:04:42,439
ways, technology was extensive.

130
00:04:42,899 --> 00:04:44,180
So all you could do at the other

131
00:04:44,180 --> 00:04:46,100
end of the wire out in people's houses

132
00:04:46,100 --> 00:04:46,839
and offices

133
00:04:47,220 --> 00:04:48,919
was to put a speaker and a microphone.

134
00:04:49,379 --> 00:04:52,339
Yep. And the power came from, you know,

135
00:04:52,339 --> 00:04:53,480
the thing in the middle,

136
00:04:53,944 --> 00:04:55,865
And all the switching was in the middle.

137
00:04:55,865 --> 00:04:58,524
And when they first started doing digital switching,

138
00:04:59,305 --> 00:05:00,925
then you only had valves.

139
00:05:01,464 --> 00:05:04,685
You only have really expensive power hungry things

140
00:05:05,224 --> 00:05:07,404
where you had to share that capacity

141
00:05:08,104 --> 00:05:11,139
across thousands of people to make it pay.

142
00:05:11,680 --> 00:05:13,860
And so the telephone network was

143
00:05:14,319 --> 00:05:17,439
incredibly network centric. You know? Yeah. It really

144
00:05:17,439 --> 00:05:20,100
was stashing all the money in the middle

145
00:05:20,479 --> 00:05:22,879
and then sharing that functionality to people at

146
00:05:22,879 --> 00:05:25,300
the edge, so highly network centric.

147
00:05:25,774 --> 00:05:27,615
Yeah. And a lot of the terminology we

148
00:05:27,615 --> 00:05:29,395
have today, by the way, comes from this.

149
00:05:29,455 --> 00:05:32,254
So the stroger frame on Maguire, we called

150
00:05:32,254 --> 00:05:33,475
it the mainframe

151
00:05:34,814 --> 00:05:35,714
because it was

152
00:05:36,014 --> 00:05:38,654
the frame that all of the tele telco

153
00:05:38,654 --> 00:05:39,955
circuits came into.

154
00:05:40,289 --> 00:05:42,930
Now there were IDFs and BDFs and, you

155
00:05:42,930 --> 00:05:46,310
know, subframes and and other things like that,

156
00:05:46,930 --> 00:05:50,129
but the frame there that was the stroger

157
00:05:50,129 --> 00:05:51,269
was the mainframe.

158
00:05:51,970 --> 00:05:53,669
And so it was called the mainframe.

159
00:05:54,194 --> 00:05:55,714
Right? And the Right. It was

160
00:05:57,714 --> 00:06:00,274
terminology. And and, you know, that that was

161
00:06:00,274 --> 00:06:02,435
the early design. But there was a couple

162
00:06:02,435 --> 00:06:04,214
of thoughts that were completely

163
00:06:04,594 --> 00:06:05,894
off the wall crazy.

164
00:06:07,259 --> 00:06:10,720
And because the telephone network was the concentration

165
00:06:10,939 --> 00:06:13,759
or one of the concentration points of technology

166
00:06:14,779 --> 00:06:16,699
because if you think about it, you know,

167
00:06:16,699 --> 00:06:18,699
there are a few giants around at the

168
00:06:18,699 --> 00:06:20,779
time, and one of them was RCA, the

169
00:06:20,779 --> 00:06:21,279
broadcaster.

170
00:06:21,894 --> 00:06:24,375
But there was also, you know, dear old

171
00:06:24,375 --> 00:06:27,574
Marbell, the telephone company. And they were desperately

172
00:06:27,574 --> 00:06:29,654
trying to make the telephone network bigger and

173
00:06:29,654 --> 00:06:30,154
cheaper.

174
00:06:30,535 --> 00:06:32,935
And so their research labs were trying

175
00:06:33,254 --> 00:06:35,274
I suppose, I guess, there was a department

176
00:06:35,334 --> 00:06:36,235
trying to make

177
00:06:36,539 --> 00:06:38,620
valves that were smaller and use less power

178
00:06:38,620 --> 00:06:40,939
because, you know, trying to perfect what you

179
00:06:40,939 --> 00:06:43,099
have is always a good idea. But down

180
00:06:43,099 --> 00:06:45,019
the end of the corridor, god knows where,

181
00:06:45,019 --> 00:06:46,079
but down the end

182
00:06:46,620 --> 00:06:48,560
was this strange sort of group

183
00:06:49,019 --> 00:06:51,039
trying to push current through

184
00:06:51,425 --> 00:06:53,745
this weird stuff of rare earths and and

185
00:06:53,745 --> 00:06:54,805
so called semiconductors.

186
00:06:55,904 --> 00:06:58,064
Where they weren't like copper. They didn't just

187
00:06:58,064 --> 00:06:58,964
always conduct.

188
00:06:59,824 --> 00:07:01,204
But it wasn't like

189
00:07:01,504 --> 00:07:03,504
stuff that doesn't conduct. It was sort of

190
00:07:03,504 --> 00:07:04,165
in between.

191
00:07:04,625 --> 00:07:05,845
It sort of conducted.

192
00:07:06,220 --> 00:07:07,439
The name goes semiconductors.

193
00:07:08,060 --> 00:07:10,079
And it was in 1947

194
00:07:10,379 --> 00:07:12,000
that a group at Bell Labs

195
00:07:12,540 --> 00:07:13,759
came up with an arrangement

196
00:07:14,699 --> 00:07:15,439
of silicon,

197
00:07:16,300 --> 00:07:18,300
which had been permeated with a certain amount.

198
00:07:18,300 --> 00:07:20,024
I think it was gallium at one point.

199
00:07:20,024 --> 00:07:21,884
I've forgotten, but it doesn't really matter.

200
00:07:22,264 --> 00:07:24,204
And instead of having a valve

201
00:07:24,584 --> 00:07:26,125
with power, vacuum,

202
00:07:26,504 --> 00:07:27,644
hideous unreliability,

203
00:07:28,745 --> 00:07:31,404
you had these strange semiconductor elements,

204
00:07:31,865 --> 00:07:34,044
which with a small push of electrons

205
00:07:34,904 --> 00:07:36,899
turned from conducting to not

206
00:07:37,279 --> 00:07:39,620
conducting or the absence of a small push,

207
00:07:39,839 --> 00:07:41,379
which was all a valve was.

208
00:07:41,759 --> 00:07:44,319
A valve was this large current. And if

209
00:07:44,319 --> 00:07:46,079
you put a small amount of voltage on

210
00:07:46,079 --> 00:07:48,079
the grid in the middle, it stopped it

211
00:07:48,079 --> 00:07:50,574
or it started it. So a little thing

212
00:07:50,574 --> 00:07:53,074
controlled a big thing. And in the semiconductor

213
00:07:53,375 --> 00:07:55,855
world, which they just sort of invented as

214
00:07:55,855 --> 00:07:56,514
a transistor,

215
00:07:56,975 --> 00:07:59,074
this little current controlled a big current.

216
00:08:00,095 --> 00:08:02,574
Amazing. Who would have thought? Which which, by

217
00:08:02,574 --> 00:08:04,834
the way, happens with tubes. This is something

218
00:08:05,770 --> 00:08:06,750
like a tube,

219
00:08:07,050 --> 00:08:07,550
essentially,

220
00:08:08,410 --> 00:08:10,509
a vacuum tube is essentially

221
00:08:11,050 --> 00:08:11,790
a cathode

222
00:08:12,889 --> 00:08:14,189
and a plate,

223
00:08:14,810 --> 00:08:16,490
and then you have a grid in the

224
00:08:16,490 --> 00:08:18,435
middle. And the whole thing about a tube

225
00:08:18,435 --> 00:08:20,834
is is you can have multiple grids. So

226
00:08:20,834 --> 00:08:22,615
basically, what you do is you shoot

227
00:08:23,074 --> 00:08:25,014
a stream of electrons

228
00:08:25,394 --> 00:08:27,574
through the vacuum in the vacuum tube

229
00:08:28,035 --> 00:08:28,935
at the plate

230
00:08:29,350 --> 00:08:31,509
and through the plate so that the receiver,

231
00:08:31,509 --> 00:08:31,910
the cap

232
00:08:32,470 --> 00:08:34,470
the to the plate, and the plate picks

233
00:08:34,470 --> 00:08:36,950
it up. And so what you do is

234
00:08:36,950 --> 00:08:39,429
you control the strength of the str of

235
00:08:39,429 --> 00:08:42,009
the stream of electrons, the amount of voltage

236
00:08:42,914 --> 00:08:45,875
that you're getting through the tube, by modifying

237
00:08:45,875 --> 00:08:47,794
the amount of voltage you drive on the

238
00:08:47,794 --> 00:08:49,335
on the on the grid.

239
00:08:49,715 --> 00:08:51,075
So you can have three or four grids

240
00:08:51,075 --> 00:08:52,434
and have them do things. So this is

241
00:08:52,434 --> 00:08:54,934
how you modulate signals, is you actually

242
00:08:55,315 --> 00:08:56,134
can adjust.

243
00:08:56,459 --> 00:08:58,379
You have a you have a straight pure

244
00:08:58,379 --> 00:09:00,779
sine wave going into the cathode side, and

245
00:09:00,779 --> 00:09:02,379
it is if if you don't have any

246
00:09:02,379 --> 00:09:04,139
if you don't have any grid, you just

247
00:09:04,139 --> 00:09:05,279
get the same straight,

248
00:09:05,899 --> 00:09:08,379
you know, out of the the the plate

249
00:09:08,379 --> 00:09:10,264
side. But you put that

250
00:09:10,644 --> 00:09:12,804
grid in the middle and you start messing

251
00:09:12,804 --> 00:09:15,524
with it, and you can actually modulate signals

252
00:09:15,524 --> 00:09:18,424
onto the sine wave. So it's actually exactly

253
00:09:19,044 --> 00:09:20,345
like a transistor,

254
00:09:21,365 --> 00:09:21,865
only

255
00:09:22,245 --> 00:09:23,784
it's not using these

256
00:09:24,320 --> 00:09:25,940
weird silicon things

257
00:09:26,559 --> 00:09:28,559
that you you know. It it it it

258
00:09:28,559 --> 00:09:30,879
was the other way. The the the silicon

259
00:09:30,879 --> 00:09:31,379
transistor

260
00:09:31,759 --> 00:09:35,299
was exactly like a simple valve. It wasn't

261
00:09:35,360 --> 00:09:38,534
multiple grids and multi unit and so on.

262
00:09:38,615 --> 00:09:41,414
It was really basic and really simple. Yeah.

263
00:09:41,414 --> 00:09:43,574
But, you know, when they invented it at

264
00:09:43,574 --> 00:09:46,214
Bell Labs, the answer from the telephone companies

265
00:09:46,214 --> 00:09:46,714
was,

266
00:09:47,334 --> 00:09:50,549
well, fascinating, but so what? Yeah. Because it

267
00:09:50,629 --> 00:09:53,589
wasn't immediately obvious that it had application there.

268
00:09:53,589 --> 00:09:55,769
And interestingly, in the next ten years,

269
00:09:56,470 --> 00:09:58,809
the silicon transistor industry

270
00:09:59,669 --> 00:10:00,570
went gangbusters

271
00:10:00,950 --> 00:10:02,330
in consumer devices.

272
00:10:02,789 --> 00:10:03,945
Transistor radios were a

273
00:10:07,865 --> 00:10:10,904
to actually differentiate between that big hulking thing

274
00:10:10,904 --> 00:10:12,745
with a whole bunch of valves connected to

275
00:10:12,745 --> 00:10:13,644
Maine's power

276
00:10:14,024 --> 00:10:16,120
and this little thing that ultimately, you you

277
00:10:16,120 --> 00:10:17,879
know, within a few years, could run off

278
00:10:17,879 --> 00:10:18,540
a battery.

279
00:10:19,000 --> 00:10:20,440
Yep. All of a sudden, it was a

280
00:10:20,440 --> 00:10:23,019
different world, but it didn't change telephony

281
00:10:23,559 --> 00:10:24,059
until

282
00:10:25,000 --> 00:10:27,340
and it was in 1958

283
00:10:27,800 --> 00:10:29,100
that Bell Labs

284
00:10:29,480 --> 00:10:32,485
came up with something else, which was almost

285
00:10:32,485 --> 00:10:35,204
like the multiple grids in a valve in

286
00:10:35,204 --> 00:10:36,184
a vacuum tube.

287
00:10:36,725 --> 00:10:39,625
But it was basically multiple transistors

288
00:10:40,485 --> 00:10:43,065
on the one substrate of silicon. Yep.

289
00:10:43,600 --> 00:10:45,139
The integrated circuit.

290
00:10:45,759 --> 00:10:48,179
And then all of a sudden, things turned

291
00:10:48,480 --> 00:10:48,980
bananas

292
00:10:49,360 --> 00:10:49,860
because

293
00:10:50,960 --> 00:10:53,360
we were starting to play with computers. We

294
00:10:53,360 --> 00:10:55,540
played with them a lot in the forties

295
00:10:55,759 --> 00:10:58,034
to try and do various feats of of

296
00:10:58,335 --> 00:10:59,554
encoding and decoding,

297
00:11:00,174 --> 00:11:02,914
various feats around calculations of projectiles

298
00:11:03,215 --> 00:11:04,674
and and orbits, etcetera.

299
00:11:05,455 --> 00:11:07,534
We were playing with this stuff, but all

300
00:11:07,534 --> 00:11:09,794
of a sudden, integrated circuits meant

301
00:11:10,190 --> 00:11:12,850
you could actually build a machine with

302
00:11:13,470 --> 00:11:13,970
thousands,

303
00:11:14,829 --> 00:11:17,730
tens of thousands of the equivalent of valves.

304
00:11:18,269 --> 00:11:20,209
And instead of taking up buildings,

305
00:11:20,909 --> 00:11:22,990
you could make computers that only took up

306
00:11:22,990 --> 00:11:23,490
rooms.

307
00:11:24,269 --> 00:11:24,769
Wow.

308
00:11:25,674 --> 00:11:28,815
And at that point, things started to impact

309
00:11:29,194 --> 00:11:29,934
back into

310
00:11:30,554 --> 00:11:32,414
this wonderful world of telephony.

311
00:11:33,274 --> 00:11:35,694
Because we were starting to uncover the fact

312
00:11:35,754 --> 00:11:38,714
that you can actually do this integrated circuits

313
00:11:38,714 --> 00:11:41,990
and integrate them with computers. And you can

314
00:11:41,990 --> 00:11:43,289
actually do switching.

315
00:11:44,470 --> 00:11:44,970
Now

316
00:11:45,350 --> 00:11:46,649
the switching in telephony,

317
00:11:47,269 --> 00:11:50,149
for some reason, the telephone network decided to

318
00:11:50,149 --> 00:11:52,649
switch the most difficult thing to switch.

319
00:11:53,190 --> 00:11:54,809
I'm like, you gotta be nuts.

320
00:11:55,535 --> 00:11:56,754
They switch time.

321
00:11:57,375 --> 00:11:59,455
Yeah. It's kind of, what are you talking

322
00:11:59,455 --> 00:12:01,315
about? That's really hard.

323
00:12:01,774 --> 00:12:04,014
You mean to say, I I sort of

324
00:12:04,014 --> 00:12:05,075
create a signal

325
00:12:05,774 --> 00:12:08,514
and I pass it through your digital switch.

326
00:12:09,100 --> 00:12:11,899
And out the other end becomes the same

327
00:12:11,899 --> 00:12:12,399
signal,

328
00:12:12,779 --> 00:12:14,000
the same voice

329
00:12:14,620 --> 00:12:15,519
without distortion,

330
00:12:16,139 --> 00:12:18,080
without sort of turning into unintelligible.

331
00:12:19,019 --> 00:12:21,500
And, mate, the telephone switch was again a

332
00:12:21,500 --> 00:12:22,335
modern miracle,

333
00:12:22,815 --> 00:12:24,975
and it was only really possible with with

334
00:12:24,975 --> 00:12:25,634
the invention

335
00:12:26,014 --> 00:12:27,875
of of the integrated circuit.

336
00:12:28,654 --> 00:12:30,815
And and, again, what was going on was

337
00:12:30,815 --> 00:12:32,595
this was still expensive stuff.

338
00:12:33,215 --> 00:12:36,200
Computers were big. They were kind of custom

339
00:12:36,339 --> 00:12:37,320
exotic things.

340
00:12:37,940 --> 00:12:38,440
But

341
00:12:38,899 --> 00:12:40,839
there was something else going on.

342
00:12:41,700 --> 00:12:44,580
Because once you'd figured out that what you

343
00:12:44,580 --> 00:12:47,240
did to build these integrated circuits

344
00:12:47,779 --> 00:12:49,860
was to take away for a pretty pure

345
00:12:49,860 --> 00:12:50,360
silicon,

346
00:12:51,085 --> 00:12:51,985
I. E. Sand,

347
00:12:52,764 --> 00:12:54,705
whack a photoresist on it,

348
00:12:55,245 --> 00:12:57,985
and then expose it to light and etch

349
00:12:58,044 --> 00:13:00,764
patterns into that chip. And you could apply

350
00:13:00,764 --> 00:13:01,264
various

351
00:13:01,725 --> 00:13:02,225
impregnated

352
00:13:02,524 --> 00:13:04,465
metals to create transistors

353
00:13:04,790 --> 00:13:06,889
on on this sort of silicon wafer.

354
00:13:07,429 --> 00:13:08,170
How many?

355
00:13:09,509 --> 00:13:11,850
Woah. Depends on how good your camera is,

356
00:13:11,910 --> 00:13:13,129
to be perfectly frank.

357
00:13:13,670 --> 00:13:15,370
Because two is easy.

358
00:13:15,750 --> 00:13:17,210
A thousand is easy.

359
00:13:17,754 --> 00:13:20,154
10,000? Well, if you've got a really good

360
00:13:20,154 --> 00:13:21,134
piece of optics,

361
00:13:21,434 --> 00:13:22,654
a fine lens,

362
00:13:23,355 --> 00:13:24,574
and and good positioning,

363
00:13:25,034 --> 00:13:27,294
yeah. You can put tens of thousands

364
00:13:27,674 --> 00:13:30,654
of individual circuits on one chip.

365
00:13:31,230 --> 00:13:32,669
And if you do it right, you can

366
00:13:32,669 --> 00:13:35,709
do this at a rate of thousands a

367
00:13:35,709 --> 00:13:36,209
second.

368
00:13:36,909 --> 00:13:38,269
So at the end of the other side

369
00:13:38,269 --> 00:13:40,429
of the production line, sand and a few

370
00:13:40,429 --> 00:13:42,589
rare animals get rare minerals go in one

371
00:13:42,589 --> 00:13:43,089
side.

372
00:13:43,629 --> 00:13:46,110
At the other end, basically, come to these

373
00:13:46,110 --> 00:13:48,414
fantastic sort of combinations

374
00:13:48,715 --> 00:13:51,054
of literally thousands of valves

375
00:13:51,434 --> 00:13:53,534
all hardwired together to create

376
00:13:53,835 --> 00:13:54,335
functionality.

377
00:13:55,914 --> 00:13:57,695
The computing industry was born.

378
00:13:58,634 --> 00:13:59,615
And Yeah.

379
00:14:00,309 --> 00:14:01,850
The first ones were big.

380
00:14:02,389 --> 00:14:04,629
They were really big. And what's interesting to

381
00:14:04,870 --> 00:14:06,649
a lot of people think about the computing

382
00:14:06,709 --> 00:14:08,870
side of this, but they don't what they

383
00:14:08,870 --> 00:14:11,190
don't think about so much is that without

384
00:14:11,190 --> 00:14:13,370
the network stuff that was done

385
00:14:13,824 --> 00:14:14,884
early in the days,

386
00:14:15,184 --> 00:14:17,584
computers themselves would have been a very limited

387
00:14:17,584 --> 00:14:18,084
value.

388
00:14:18,625 --> 00:14:20,945
And there would have probably been 20 or

389
00:14:20,945 --> 00:14:22,324
30 of them in the world

390
00:14:22,704 --> 00:14:24,164
at big research institutions

391
00:14:24,865 --> 00:14:26,164
that took on specialized

392
00:14:26,470 --> 00:14:26,970
tasks.

393
00:14:28,230 --> 00:14:29,990
So a number of things were going on.

394
00:14:29,990 --> 00:14:31,929
And and, you know, computers

395
00:14:32,549 --> 00:14:34,470
started to be used in business. And the

396
00:14:34,470 --> 00:14:36,629
whole issue was, I want them cheaper. I

397
00:14:36,629 --> 00:14:38,009
wanna build more of them.

398
00:14:38,389 --> 00:14:40,554
And and and the trick was, can I

399
00:14:40,554 --> 00:14:42,815
put more gates, more features

400
00:14:43,274 --> 00:14:45,434
on the one piece of silicon? And the

401
00:14:45,434 --> 00:14:47,294
answer was brighter lights,

402
00:14:47,834 --> 00:14:48,654
better lenses.

403
00:14:49,115 --> 00:14:51,695
Sure. Because the process could miniaturize

404
00:14:52,074 --> 00:14:52,894
really easily.

405
00:14:53,274 --> 00:14:54,495
And when it miniaturizes,

406
00:14:55,029 --> 00:14:57,929
it uses less power per unit of computing.

407
00:14:58,549 --> 00:15:01,190
So over the fifties and sixties sorry. The

408
00:15:01,190 --> 00:15:02,330
sixties and seventies,

409
00:15:02,710 --> 00:15:04,710
because the integrated circuit sort of appeared in

410
00:15:04,710 --> 00:15:06,409
the world in the late fifties,

411
00:15:06,710 --> 00:15:09,274
the sixties, The mainframe computer industry

412
00:15:09,654 --> 00:15:10,554
was constantly

413
00:15:10,855 --> 00:15:11,355
changing.

414
00:15:11,735 --> 00:15:14,394
They were getting smaller. They were getting cheaper.

415
00:15:15,575 --> 00:15:17,575
And we started to get not just the

416
00:15:17,575 --> 00:15:18,075
mainframe,

417
00:15:18,615 --> 00:15:20,455
which was the big, you know, multi tens

418
00:15:20,455 --> 00:15:21,720
of millions expenditure,

419
00:15:22,019 --> 00:15:24,519
we started to get mini mini mainframes.

420
00:15:25,379 --> 00:15:27,320
Digital equipment were keen on them.

421
00:15:28,100 --> 00:15:29,159
They're called PDPs.

422
00:15:30,259 --> 00:15:33,000
And in the late nineteen sixties AS 400.

423
00:15:33,754 --> 00:15:36,815
AS 400. Well, PDPs are interesting because,

424
00:15:37,514 --> 00:15:39,754
in the same Bell Labs, not the same

425
00:15:39,754 --> 00:15:40,254
office,

426
00:15:40,715 --> 00:15:41,115
but,

427
00:15:41,754 --> 00:15:43,455
Ken Thompson and Dennis Ritchie

428
00:15:44,075 --> 00:15:45,215
were mucking around

429
00:15:46,210 --> 00:15:49,409
looking at the entrails of a failed project

430
00:15:49,409 --> 00:15:52,370
called Multics, which was a grand experiment with

431
00:15:52,370 --> 00:15:55,029
MIT and a few others to produce the

432
00:15:55,250 --> 00:15:57,730
massive computer that everyone would have an account

433
00:15:57,730 --> 00:16:01,134
on, the shared multi user, you know, computing

434
00:16:01,195 --> 00:16:03,134
system that would, you know, do everything

435
00:16:03,514 --> 00:16:05,054
for everyone all at once.

436
00:16:05,514 --> 00:16:07,455
And they took this PDP 11,

437
00:16:08,154 --> 00:16:10,335
tiny machine, a single rack.

438
00:16:11,389 --> 00:16:13,710
That's a 19 inch equipment frame for those

439
00:16:13,710 --> 00:16:16,110
who haven't seen one. They're still around. They're

440
00:16:16,110 --> 00:16:18,509
still the same size. They took this this

441
00:16:18,509 --> 00:16:19,490
PDP 11

442
00:16:20,029 --> 00:16:23,570
and built their own little mini operating system.

443
00:16:23,710 --> 00:16:25,970
The control system that made the computer

444
00:16:26,315 --> 00:16:27,695
be able to do things.

445
00:16:28,075 --> 00:16:29,754
You know, it could run programs. It had

446
00:16:29,754 --> 00:16:31,834
a file store. It had a way you

447
00:16:31,834 --> 00:16:34,014
could type in commands. It had a display.

448
00:16:34,714 --> 00:16:36,315
And and they called it as a play

449
00:16:36,315 --> 00:16:37,855
on Multics, UNIX.

450
00:16:39,379 --> 00:16:41,879
So here was this operating system,

451
00:16:42,340 --> 00:16:43,480
small, portable,

452
00:16:44,580 --> 00:16:48,100
worked on this new generation of of machines

453
00:16:48,100 --> 00:16:50,179
that you could afford in a lab, not

454
00:16:50,179 --> 00:16:51,639
just afford in a company.

455
00:16:53,065 --> 00:16:55,304
And immediately, when they sort of did this

456
00:16:55,304 --> 00:16:56,904
and came to the attention of the the

457
00:16:56,904 --> 00:16:58,445
mothership AT and T,

458
00:16:59,705 --> 00:17:00,605
there was a problem.

459
00:17:03,065 --> 00:17:04,044
Under the terms

460
00:17:04,669 --> 00:17:07,150
of an agreement with the US administration and

461
00:17:07,150 --> 00:17:08,049
Department of Justice,

462
00:17:08,829 --> 00:17:09,809
AT and T

463
00:17:10,750 --> 00:17:12,990
was basically a phone company, and they were

464
00:17:12,990 --> 00:17:15,409
a supplier of technology to phone companies.

465
00:17:15,950 --> 00:17:16,690
They built

466
00:17:17,014 --> 00:17:19,595
five ESS switches. They built that technology.

467
00:17:20,294 --> 00:17:22,954
Now they were straying from the reserve

468
00:17:23,734 --> 00:17:26,375
if they were doing a business other than

469
00:17:26,375 --> 00:17:29,034
equipment for common carriers in the telephone world.

470
00:17:29,255 --> 00:17:30,554
That's not your job.

471
00:17:31,014 --> 00:17:31,914
I went, oh,

472
00:17:32,410 --> 00:17:34,410
but we have this really cool operating system.

473
00:17:34,410 --> 00:17:35,230
And they said,

474
00:17:35,690 --> 00:17:36,430
not your

475
00:17:36,970 --> 00:17:38,589
job. What can we do with it?

476
00:17:39,130 --> 00:17:40,029
Give it away.

477
00:17:41,529 --> 00:17:43,549
What? Give it away. And they did.

478
00:17:44,089 --> 00:17:45,684
So for the cost of shipping,

479
00:17:46,144 --> 00:17:47,444
anyone could have

480
00:17:47,744 --> 00:17:51,285
a really quite cool lean mean operating system

481
00:17:51,904 --> 00:17:55,025
that ran on not just PDP elevens, but

482
00:17:55,025 --> 00:17:56,625
because it was written in a high level

483
00:17:56,625 --> 00:17:58,944
source code, you can assemble it for almost

484
00:17:58,944 --> 00:17:59,444
anything.

485
00:18:00,410 --> 00:18:01,309
And and so

486
00:18:01,769 --> 00:18:04,589
by the late sixties, early seventies,

487
00:18:05,289 --> 00:18:07,390
this new operating system, UNIX,

488
00:18:07,849 --> 00:18:09,470
was certainly a thing.

489
00:18:10,329 --> 00:18:13,069
Every vendor had their own operating system, digital

490
00:18:13,130 --> 00:18:13,630
IBM,

491
00:18:14,009 --> 00:18:16,295
you know, Burrows, you name it. But here

492
00:18:16,295 --> 00:18:17,914
was this other one, this upstart

493
00:18:18,615 --> 00:18:21,335
that wasn't really owned by anyone. It was

494
00:18:21,335 --> 00:18:21,835
open

495
00:18:22,295 --> 00:18:23,674
and it was available

496
00:18:24,535 --> 00:18:27,255
as we evolved computers and made them smaller

497
00:18:27,255 --> 00:18:27,994
and faster.

498
00:18:28,455 --> 00:18:30,080
I want an operating system. Oh, I'll just

499
00:18:30,080 --> 00:18:30,900
put on UNIX.

500
00:18:31,440 --> 00:18:33,220
A morning's work and you're done.

501
00:18:33,600 --> 00:18:36,160
Not five months work of devising a new

502
00:18:36,160 --> 00:18:38,559
operating system for this new computer. It's going,

503
00:18:38,559 --> 00:18:40,480
nah. Just put on UNIX. I'm out of

504
00:18:40,480 --> 00:18:40,980
there.

505
00:18:41,944 --> 00:18:42,444
Revolutionary.

506
00:18:42,825 --> 00:18:43,884
What a thought.

507
00:18:44,265 --> 00:18:47,005
What a thought. And totally totally accidental.

508
00:18:48,184 --> 00:18:50,924
Totally accidental. But it made the software

509
00:18:51,305 --> 00:18:52,684
independent of the hardware.

510
00:18:53,224 --> 00:18:55,325
So that's the second of these serendipitous

511
00:18:55,785 --> 00:18:58,099
thoughts that kind of came from nowhere.

512
00:18:58,799 --> 00:19:00,799
Maybe it was a particularly strong coffee that

513
00:19:00,799 --> 00:19:03,299
day. I don't know. But, you know, kaboom.

514
00:19:03,519 --> 00:19:04,500
It just happened.

515
00:19:05,519 --> 00:19:07,200
And there was one more, and it was

516
00:19:07,200 --> 00:19:08,660
1973.

517
00:19:09,375 --> 00:19:11,775
Because one of the outcomes of this work

518
00:19:11,775 --> 00:19:14,494
in radar and so on was actually using

519
00:19:14,494 --> 00:19:15,634
radio for communications.

520
00:19:16,894 --> 00:19:18,035
That's pretty easy.

521
00:19:18,414 --> 00:19:19,875
Now you send a signal,

522
00:19:20,575 --> 00:19:23,714
in radio terms, and then you modulate that.

523
00:19:24,335 --> 00:19:26,409
So, So, you know, dot dot dash dash,

524
00:19:26,409 --> 00:19:28,329
all that kind of stuff is humming, not

525
00:19:28,329 --> 00:19:29,389
humming in radio.

526
00:19:30,089 --> 00:19:32,169
And over in the Hawaiian Islands, they built

527
00:19:32,169 --> 00:19:35,549
this radio based network, relatively primitive, pretty slow,

528
00:19:35,690 --> 00:19:36,750
called AlohaNet.

529
00:19:37,904 --> 00:19:39,445
And it was kind of neat.

530
00:19:41,265 --> 00:19:43,924
It was a simple system where

531
00:19:44,625 --> 00:19:46,305
you waited for a while to see if

532
00:19:46,305 --> 00:19:48,244
that carrier frequency was idle.

533
00:19:48,785 --> 00:19:51,105
And then you sent your your little packet

534
00:19:51,105 --> 00:19:51,765
of data

535
00:19:52,289 --> 00:19:54,390
over to the main hub point.

536
00:19:55,090 --> 00:19:56,769
And then it would send it to wherever

537
00:19:56,769 --> 00:19:58,450
you wanted to address it. It was a

538
00:19:58,450 --> 00:19:59,350
neat system

539
00:19:59,809 --> 00:20:02,549
because it just used radio. Nothing else.

540
00:20:03,090 --> 00:20:05,730
All the machines, the computers at the sort

541
00:20:05,730 --> 00:20:06,105
of connect

542
00:20:07,144 --> 00:20:10,765
to the network formed part of the functionality.

543
00:20:10,904 --> 00:20:12,184
One of the guys who worked on this

544
00:20:12,184 --> 00:20:13,244
was Bob Metcalfe,

545
00:20:13,944 --> 00:20:16,265
and he was then hired by that well

546
00:20:16,265 --> 00:20:17,964
known computing company, Xerox.

547
00:20:18,720 --> 00:20:19,220
Not

548
00:20:20,319 --> 00:20:22,819
the the only company that I think that

549
00:20:22,880 --> 00:20:26,240
made at least two, maybe three fundamental inventions

550
00:20:26,240 --> 00:20:28,500
in computing. And at the end of this,

551
00:20:28,960 --> 00:20:31,059
managed to stay a photocopier company.

552
00:20:32,684 --> 00:20:33,184
Yeah.

553
00:20:33,804 --> 00:20:35,244
So I sit there and think, how did

554
00:20:35,244 --> 00:20:36,944
you miss it? Every time.

555
00:20:37,484 --> 00:20:37,984
Missed.

556
00:20:38,365 --> 00:20:41,325
So this guy, Bob Bittcalf working at Xerox

557
00:20:41,325 --> 00:20:42,065
at the time,

558
00:20:42,524 --> 00:20:44,704
kind of came up with a wired equivalent

559
00:20:44,845 --> 00:20:45,505
of HelloAnet,

560
00:20:46,044 --> 00:20:47,105
a wired equivalent.

561
00:20:48,099 --> 00:20:50,259
And this is the beauty of it The

562
00:20:50,259 --> 00:20:51,880
network was a wire

563
00:20:52,500 --> 00:20:53,240
A wire

564
00:20:53,940 --> 00:20:54,839
No electronics

565
00:20:55,619 --> 00:20:56,119
Nothing

566
00:20:56,659 --> 00:20:58,759
It was the ultimate disappearing trick

567
00:20:59,619 --> 00:21:00,839
Because everything

568
00:21:01,299 --> 00:21:03,734
was actually in the computers attached to the

569
00:21:03,734 --> 00:21:04,234
wire.

570
00:21:05,335 --> 00:21:05,914
And architecturally,

571
00:21:06,375 --> 00:21:08,554
it was it was the opposite of telephony.

572
00:21:09,894 --> 00:21:12,855
Really smart concentration of technology in the middle.

573
00:21:12,855 --> 00:21:15,194
It's so expensive. We can only have one

574
00:21:15,335 --> 00:21:16,954
dumb devices at the edge.

575
00:21:18,830 --> 00:21:19,970
This new thing,

576
00:21:20,750 --> 00:21:21,250
Ethernet,

577
00:21:21,869 --> 00:21:23,009
is the exact opposite.

578
00:21:23,710 --> 00:21:25,650
Computing is now so affordable

579
00:21:26,350 --> 00:21:28,289
that every computer can participate

580
00:21:28,590 --> 00:21:30,384
in driving the network.

581
00:21:30,845 --> 00:21:32,285
And we can actually take it to the

582
00:21:32,285 --> 00:21:33,805
point, and it is a big point to

583
00:21:33,805 --> 00:21:36,684
take it to, where the network itself is

584
00:21:36,684 --> 00:21:39,025
just a piece of wire. It's it's nothing.

585
00:21:39,884 --> 00:21:41,884
It's the computers at the edge that are

586
00:21:41,884 --> 00:21:42,384
cooperatively

587
00:21:42,684 --> 00:21:44,545
talking to each other,

588
00:21:45,400 --> 00:21:47,259
driving this thing we call a network.

589
00:21:48,279 --> 00:21:48,779
Wow.

590
00:21:49,799 --> 00:21:51,720
And and when the industry saw this, it's

591
00:21:51,720 --> 00:21:53,960
kind of, you mean to say all I

592
00:21:53,960 --> 00:21:55,420
need to do is lay wire

593
00:21:55,799 --> 00:21:57,180
and I have a network?

594
00:21:57,765 --> 00:21:58,505
And the answer,

595
00:21:58,964 --> 00:22:00,984
yes. Or where do I order this stuff?

596
00:22:01,765 --> 00:22:03,144
And by the mid seventies,

597
00:22:03,525 --> 00:22:05,305
this 10 megabit system

598
00:22:05,845 --> 00:22:08,025
was being installed almost everywhere.

599
00:22:08,404 --> 00:22:10,505
It wasn't dependent on a single vendor.

600
00:22:10,805 --> 00:22:13,125
You know, Xerox did not make computers. And

601
00:22:13,125 --> 00:22:13,625
oddly

602
00:22:14,049 --> 00:22:15,430
enough, that was a good thing

603
00:22:15,809 --> 00:22:16,309
because

604
00:22:16,690 --> 00:22:18,470
it was openly available.

605
00:22:19,009 --> 00:22:21,170
It wasn't tied up with a particular brand

606
00:22:21,170 --> 00:22:21,830
of mainframe.

607
00:22:22,450 --> 00:22:25,430
And and literally, everyone who was looking on

608
00:22:25,490 --> 00:22:28,210
building their mini computers and then it got

609
00:22:28,210 --> 00:22:30,325
smaller and smaller, Connecting them up was the

610
00:22:30,325 --> 00:22:31,384
job of Ethernet.

611
00:22:32,484 --> 00:22:34,964
So we kind of had these three things

612
00:22:34,964 --> 00:22:36,345
happening all at once.

613
00:22:37,284 --> 00:22:39,684
Yeah. Moore's law was starting to take effect.

614
00:22:39,684 --> 00:22:42,904
Your integrated circuits were smaller, cheaper, and faster.

615
00:22:43,630 --> 00:22:44,130
Wow.

616
00:22:44,990 --> 00:22:46,369
We had software

617
00:22:46,829 --> 00:22:47,970
divorced from hardware.

618
00:22:48,669 --> 00:22:49,890
And oddly enough,

619
00:22:51,230 --> 00:22:52,669
no one knew how to value it. So

620
00:22:52,669 --> 00:22:54,990
we were giving it away. Just giving it

621
00:22:54,990 --> 00:22:55,490
away.

622
00:22:55,869 --> 00:22:56,369
Remarkable.

623
00:22:56,990 --> 00:22:58,724
And then we had networks which

624
00:22:59,605 --> 00:23:03,065
were effectively networks that weren't centralized. They were

625
00:23:03,125 --> 00:23:05,605
the ultimate of pushing it so far out

626
00:23:05,605 --> 00:23:07,204
to the edge, there was nothing left in

627
00:23:07,204 --> 00:23:07,785
the middle.

628
00:23:09,044 --> 00:23:11,865
These were very, very basic inventions, all three.

629
00:23:12,164 --> 00:23:14,890
Yeah. Yeah. And now when networks were first

630
00:23:14,890 --> 00:23:16,410
done, like, when we talk about mock mock

631
00:23:16,730 --> 00:23:19,769
Metcalf days, we actually aren't talking about integrated

632
00:23:19,769 --> 00:23:22,589
circuits. We're talking about discrete electronic circuits,

633
00:23:23,529 --> 00:23:26,089
which is a phase of electronic engine of

634
00:23:26,089 --> 00:23:28,490
engineering that people seem to miss. But I've

635
00:23:28,490 --> 00:23:29,390
worked on,

636
00:23:29,894 --> 00:23:30,475
you know,

637
00:23:32,055 --> 00:23:32,555
systems

638
00:23:33,015 --> 00:23:33,515
with

639
00:23:34,215 --> 00:23:36,715
a hundred, two hundred, 300 gates.

640
00:23:37,494 --> 00:23:39,355
There were literally two

641
00:23:40,215 --> 00:23:42,615
there were literally two capacitors and a resistor

642
00:23:42,615 --> 00:23:43,355
per gate,

643
00:23:43,679 --> 00:23:46,819
all discreetly soldered on to circuit boards.

644
00:23:47,440 --> 00:23:49,200
Because they they I mean, they were transistors,

645
00:23:49,200 --> 00:23:50,899
but they really I mean, they were mostly

646
00:23:50,960 --> 00:23:51,460
diodes

647
00:23:52,000 --> 00:23:52,899
and capacitors.

648
00:23:54,240 --> 00:23:55,299
And Right.

649
00:23:55,599 --> 00:23:58,159
And they were soldered onto physical boards to

650
00:23:58,159 --> 00:23:58,899
make gates.

651
00:23:59,975 --> 00:24:01,894
But the trick was you could do it

652
00:24:01,894 --> 00:24:02,715
on silicon

653
00:24:03,255 --> 00:24:05,654
Yeah. A lot smaller. At money level. And

654
00:24:05,654 --> 00:24:07,815
and it was cheaper to produce. So the

655
00:24:07,815 --> 00:24:09,735
cost per gate is a really good way

656
00:24:09,735 --> 00:24:11,035
of thinking about this.

657
00:24:11,609 --> 00:24:13,130
Whereas it used to be a couple of

658
00:24:13,130 --> 00:24:13,630
cents

659
00:24:14,009 --> 00:24:15,630
to actually build the transistor,

660
00:24:16,009 --> 00:24:17,930
solder it onto a circuit board, you know,

661
00:24:17,930 --> 00:24:19,769
put it with all of its mates, you

662
00:24:19,769 --> 00:24:22,490
know, create circuitry that supported the thing working

663
00:24:22,490 --> 00:24:24,970
in concert, and then doing the connectors at

664
00:24:24,970 --> 00:24:27,434
one end of the the PCB card to

665
00:24:27,434 --> 00:24:28,875
connect to all the rest of these, you

666
00:24:28,875 --> 00:24:31,515
know, little circuits. Yeah. You could shrink all

667
00:24:31,515 --> 00:24:34,494
of that to one integrated circuit

668
00:24:35,035 --> 00:24:36,795
with a bunch of of pins as its

669
00:24:36,795 --> 00:24:37,295
feet.

670
00:24:37,755 --> 00:24:39,674
And and this just kinda took over the

671
00:24:39,674 --> 00:24:40,494
world because

672
00:24:40,954 --> 00:24:42,950
the cost of production were now down at

673
00:24:42,950 --> 00:24:44,809
fractions of a cent per day.

674
00:24:45,110 --> 00:24:45,610
Minions.

675
00:24:46,230 --> 00:24:48,549
And and bizarrely and again, this is something

676
00:24:48,549 --> 00:24:50,490
that I think we failed to truly understand.

677
00:24:51,670 --> 00:24:52,890
Moore's law said,

678
00:24:53,350 --> 00:24:55,535
every two years as it turned out originally

679
00:24:55,535 --> 00:24:58,674
said every every single year, but very regularly,

680
00:24:59,134 --> 00:25:01,214
we were able to put twice the number

681
00:25:01,214 --> 00:25:01,954
of gates

682
00:25:02,335 --> 00:25:03,795
on a single circuit

683
00:25:04,174 --> 00:25:05,474
of the same size.

684
00:25:06,529 --> 00:25:09,509
So wow. This was the incredible shrinking world.

685
00:25:10,289 --> 00:25:12,230
Every two years, let's go two,

686
00:25:12,690 --> 00:25:14,789
I could buy state of the art technology

687
00:25:14,849 --> 00:25:16,630
that was twice as fast.

688
00:25:17,809 --> 00:25:19,109
Twice as cheap.

689
00:25:20,014 --> 00:25:22,595
And demanded twice as less power

690
00:25:23,214 --> 00:25:24,274
than my competitors.

691
00:25:26,095 --> 00:25:28,974
And so the whole reason why bell labs

692
00:25:28,974 --> 00:25:30,755
looked at the transistor and said

693
00:25:31,615 --> 00:25:32,115
Great.

694
00:25:33,500 --> 00:25:35,440
But we're not ready for it.

695
00:25:35,900 --> 00:25:37,420
Come back and tell us again in a

696
00:25:37,420 --> 00:25:38,400
few years' time.

697
00:25:38,940 --> 00:25:40,380
Was that they had a life cycle of

698
00:25:40,380 --> 00:25:42,400
equipment of twenty five years,

699
00:25:43,340 --> 00:25:45,180
you know, which is why the military, by

700
00:25:45,180 --> 00:25:47,580
the way, that that insisted on having that

701
00:25:47,580 --> 00:25:49,414
life cycle for decades longer than they should

702
00:25:49,414 --> 00:25:51,654
have were using twenty five year old equipment

703
00:25:51,654 --> 00:25:53,734
in the seventies because, you know, that was

704
00:25:53,734 --> 00:25:55,815
the mode of thinking. Yeah. I know. We

705
00:25:55,815 --> 00:25:57,894
were still using nineteen sixties equipment in the

706
00:25:57,894 --> 00:25:58,795
nineteen nineties.

707
00:25:59,654 --> 00:26:01,755
Twenty five year lifetime. And in

708
00:26:02,214 --> 00:26:03,755
in one extent, they were right.

709
00:26:04,210 --> 00:26:05,009
You know? Trans

710
00:26:05,730 --> 00:26:08,210
transistors, unlike valves, didn't have a half life

711
00:26:08,210 --> 00:26:10,850
of ours. Yeah. Transistors had a half life

712
00:26:10,850 --> 00:26:12,930
of decades. Yeah. And so the thing that

713
00:26:12,930 --> 00:26:14,850
you bought in the sixties still worked in

714
00:26:14,850 --> 00:26:15,509
the nineties.

715
00:26:15,970 --> 00:26:17,775
But if you're in the competitive world, the

716
00:26:17,934 --> 00:26:18,434
commercial

717
00:26:18,974 --> 00:26:22,035
world, everyone else was buying equipment and machinery

718
00:26:22,654 --> 00:26:24,195
that every two years

719
00:26:24,734 --> 00:26:27,315
was twice as good in almost every parameter,

720
00:26:27,375 --> 00:26:28,355
the one I have.

721
00:26:28,894 --> 00:26:29,394
So

722
00:26:29,775 --> 00:26:31,795
all of a sudden, I've got a problem.

723
00:26:32,240 --> 00:26:34,640
I can't run a twenty five year life

724
00:26:34,640 --> 00:26:35,140
cycle

725
00:26:35,440 --> 00:26:37,920
on my on my capital, on my equipment,

726
00:26:37,920 --> 00:26:38,820
on my asset.

727
00:26:39,599 --> 00:26:41,920
Because unlike the the military and the government,

728
00:26:41,920 --> 00:26:44,240
I'm in a competitive world. I need to

729
00:26:44,240 --> 00:26:47,619
change every year or so. Every year.

730
00:26:48,625 --> 00:26:51,365
And and that really did leave some companies

731
00:26:51,424 --> 00:26:51,924
behind

732
00:26:52,464 --> 00:26:54,724
and made giants out of the most amazing,

733
00:26:55,744 --> 00:26:56,724
almost unintended

734
00:26:57,184 --> 00:26:57,684
folk.

735
00:26:58,304 --> 00:27:00,005
Apple was an accident.

736
00:27:01,000 --> 00:27:02,539
Yeah. It really was.

737
00:27:03,000 --> 00:27:03,500
Because

738
00:27:04,039 --> 00:27:06,059
I think very few folk foresaw

739
00:27:06,759 --> 00:27:09,259
that when you scale and instead of producing

740
00:27:09,319 --> 00:27:10,940
a 100 computers a year,

741
00:27:11,559 --> 00:27:12,299
you produce

742
00:27:12,599 --> 00:27:15,259
a 100 integrated circuits per second,

743
00:27:15,715 --> 00:27:17,335
which turn into what?

744
00:27:18,115 --> 00:27:18,615
Mainframes?

745
00:27:19,075 --> 00:27:21,315
Don't be silly. Yeah. How do I sell

746
00:27:21,315 --> 00:27:23,174
that extraordinary production capacity?

747
00:27:23,795 --> 00:27:25,555
I need to convince more and more people

748
00:27:25,555 --> 00:27:26,375
to buy them.

749
00:27:26,674 --> 00:27:27,835
And so what you do is you get

750
00:27:27,955 --> 00:27:29,875
you you release Lotus one two three, and

751
00:27:29,875 --> 00:27:32,190
everybody puts one in their closet.

752
00:27:33,049 --> 00:27:35,150
Yeah. Right. Yeah. I build a closet computer,

753
00:27:35,369 --> 00:27:37,769
and I convince every single people that are

754
00:27:37,809 --> 00:27:39,210
you know, person who owns an apartment or

755
00:27:39,210 --> 00:27:39,789
a house,

756
00:27:40,090 --> 00:27:42,190
you need one of these in your closet.

757
00:27:42,330 --> 00:27:44,090
And all of a sudden, my market has

758
00:27:44,090 --> 00:27:46,349
expanded from a few 100,000

759
00:27:46,410 --> 00:27:47,505
enterprises and businesses

760
00:27:51,505 --> 00:27:53,045
games, not by business applications, by

761
00:27:53,664 --> 00:27:54,724
the way. Well,

762
00:27:55,105 --> 00:27:55,605
consumers.

763
00:27:56,065 --> 00:27:58,615
Yeah. Consumers. Because once you once you enter

764
00:27:58,615 --> 00:28:01,009
the consumer market, you know, most of us

765
00:28:01,009 --> 00:28:02,690
spend most of our time wanting to be

766
00:28:02,690 --> 00:28:05,329
entertained. And that's that's a fine human pursuit.

767
00:28:05,329 --> 00:28:06,309
I do it myself.

768
00:28:06,769 --> 00:28:08,769
But, you know, what we found then was

769
00:28:08,769 --> 00:28:09,990
that these things become

770
00:28:10,289 --> 00:28:12,309
a scaling consumer technology.

771
00:28:13,134 --> 00:28:15,134
Mhmm. Now, again, you look at sort of

772
00:28:15,134 --> 00:28:17,075
the the tools that we then had,

773
00:28:17,535 --> 00:28:18,275
cheap networks,

774
00:28:18,894 --> 00:28:20,595
really, really minimal networking,

775
00:28:21,694 --> 00:28:22,595
cheap software.

776
00:28:22,894 --> 00:28:24,974
Some of it was free like like, like

777
00:28:24,974 --> 00:28:25,849
a Unix.

778
00:28:26,309 --> 00:28:28,250
Hardware that was ever shrinking.

779
00:28:29,029 --> 00:28:31,349
And we had this kind of desire, if

780
00:28:31,349 --> 00:28:32,250
you will, to sign

781
00:28:32,789 --> 00:28:34,890
build more of them and link them together.

782
00:28:35,589 --> 00:28:37,690
And and out of that grew the Internet.

783
00:28:38,230 --> 00:28:40,150
And the Internet was kind of the same

784
00:28:40,150 --> 00:28:40,650
expression

785
00:28:41,744 --> 00:28:44,464
that it wasn't a network that was dense

786
00:28:44,464 --> 00:28:45,204
in the middle.

787
00:28:46,065 --> 00:28:47,045
Really wasn't.

788
00:28:47,664 --> 00:28:49,984
Because like Ethernet, it was all at the

789
00:28:49,984 --> 00:28:52,384
edge. Yeah. All the software that is the

790
00:28:52,384 --> 00:28:54,960
Internet, all all the protocols and applications and

791
00:28:54,960 --> 00:28:55,460
packages.

792
00:28:55,920 --> 00:28:58,080
I mean, the end to end yeah. The

793
00:28:58,080 --> 00:29:00,720
end to end principle. Right? Like, the the

794
00:29:00,720 --> 00:29:02,720
intelligence should be at the edge and the

795
00:29:02,720 --> 00:29:06,240
network should be dumb. Why? Because the network

796
00:29:06,240 --> 00:29:08,019
does has a longer life cycle.

797
00:29:08,384 --> 00:29:10,644
It's physical wires in the ground,

798
00:29:11,105 --> 00:29:13,505
and so, therefore, you want those physical wires

799
00:29:13,505 --> 00:29:16,464
to last for twenty, thirty years. And, you

800
00:29:16,464 --> 00:29:18,305
you know, you do all your upgrades on

801
00:29:18,305 --> 00:29:21,285
the edge to make the network more useful

802
00:29:21,424 --> 00:29:24,220
over time. And because Right. The network itself

803
00:29:24,220 --> 00:29:25,200
is a big investment.

804
00:29:25,579 --> 00:29:26,079
Well,

805
00:29:26,700 --> 00:29:28,799
boy, have we reversed all of that.

806
00:29:29,180 --> 00:29:31,519
Well, that then leads to the next point.

807
00:29:32,059 --> 00:29:34,079
Because you kinda think, oh, it's just computers,

808
00:29:34,940 --> 00:29:36,859
and the stuff in the network's kind of

809
00:29:37,019 --> 00:29:39,275
well, it's expensive and fixed, isn't it?

810
00:29:39,975 --> 00:29:40,875
Yeah. No.

811
00:29:41,335 --> 00:29:41,835
No.

812
00:29:42,295 --> 00:29:44,215
And and we first saw it with the

813
00:29:44,215 --> 00:29:45,994
use of the telephone network

814
00:29:46,455 --> 00:29:48,555
for dial up modems. Remember them?

815
00:29:49,255 --> 00:29:50,934
Yeah. Remember when you used to make the

816
00:29:50,934 --> 00:29:53,355
call and these things would start screechy?

817
00:29:54,140 --> 00:29:56,220
You've heard for the first ten seconds or

818
00:29:56,220 --> 00:29:59,019
so as it kind of had this screech

819
00:29:59,019 --> 00:30:00,940
down the wire. What was it trying to

820
00:30:00,940 --> 00:30:03,360
do by screeching, you know, down the wire?

821
00:30:03,740 --> 00:30:05,360
Well, we were getting smarter

822
00:30:05,980 --> 00:30:09,180
because the original idea of putting signals on

823
00:30:09,180 --> 00:30:09,759
a wire

824
00:30:10,275 --> 00:30:12,914
was to take a carrier tone and turn

825
00:30:12,914 --> 00:30:14,455
it off and turn it on again.

826
00:30:15,394 --> 00:30:17,234
The absence of a carrier tone was a

827
00:30:17,234 --> 00:30:19,414
zero and its presence was a one.

828
00:30:19,875 --> 00:30:21,795
And in essence, you could put a microphone

829
00:30:21,795 --> 00:30:23,795
at the other end and reassemble the bits.

830
00:30:23,795 --> 00:30:25,680
And that was about as sophisticated as it

831
00:30:25,759 --> 00:30:26,259
was.

832
00:30:26,960 --> 00:30:28,960
But if you try and go as fast

833
00:30:28,960 --> 00:30:30,420
as the carrier tone,

834
00:30:31,039 --> 00:30:32,960
you start to think about, well, how do

835
00:30:32,960 --> 00:30:33,860
I encode

836
00:30:34,799 --> 00:30:36,420
a one zero sequence

837
00:30:37,519 --> 00:30:38,019
into

838
00:30:38,724 --> 00:30:41,605
a frequency signal that's the same frequency as

839
00:30:41,605 --> 00:30:42,264
my bits.

840
00:30:42,804 --> 00:30:43,304
Oh.

841
00:30:43,684 --> 00:30:45,524
So I have to kind of modulate the

842
00:30:45,524 --> 00:30:46,024
bits.

843
00:30:46,644 --> 00:30:48,325
I have to kind of set it so

844
00:30:48,325 --> 00:30:48,825
that

845
00:30:49,365 --> 00:30:51,365
a high voltage is a zero, a low

846
00:30:51,365 --> 00:30:53,819
voltage is a one, and somehow manage to

847
00:30:53,980 --> 00:30:55,279
to encode that.

848
00:30:55,819 --> 00:30:56,319
Interesting.

849
00:30:56,779 --> 00:30:59,259
It's called amplitude encoding. It's kind of on

850
00:30:59,259 --> 00:31:01,519
off. Or you can try to be cleverer.

851
00:31:01,980 --> 00:31:03,279
I have four volumes.

852
00:31:04,220 --> 00:31:06,319
No volume at all. That's a zero.

853
00:31:07,265 --> 00:31:09,205
Full volume. That's one one.

854
00:31:09,585 --> 00:31:11,664
And there's two intermediate states. Well, I can

855
00:31:11,664 --> 00:31:14,245
encode that as zero one and one zero.

856
00:31:14,384 --> 00:31:14,884
Oh.

857
00:31:15,505 --> 00:31:17,745
Out of the same carrier tone, I've got

858
00:31:17,745 --> 00:31:18,725
double the bits.

859
00:31:19,740 --> 00:31:20,480
How clever.

860
00:31:20,940 --> 00:31:22,960
All I need is four amplitude levels.

861
00:31:23,339 --> 00:31:24,320
How about 16?

862
00:31:25,019 --> 00:31:27,920
Well, I can get more. How about 32?

863
00:31:28,539 --> 00:31:30,140
You know, the human ear is pretty good

864
00:31:30,140 --> 00:31:32,700
at this. It can distinguish about 256

865
00:31:32,700 --> 00:31:34,545
or so discrete levels of

866
00:31:35,424 --> 00:31:38,164
volume. Well, computers can do better because

867
00:31:38,625 --> 00:31:41,125
we're using digital signal processes.

868
00:31:41,825 --> 00:31:43,825
And so you start to do really clever

869
00:31:43,825 --> 00:31:44,325
encoding.

870
00:31:45,105 --> 00:31:46,565
But is it just all volume?

871
00:31:47,279 --> 00:31:47,779
No.

872
00:31:48,240 --> 00:31:50,740
I can change the phase of the carrier.

873
00:31:51,600 --> 00:31:52,100
Woah.

874
00:31:52,720 --> 00:31:53,460
Big thought.

875
00:31:53,920 --> 00:31:55,859
All of a sudden, by the

876
00:31:56,480 --> 00:31:57,380
late eighties,

877
00:31:57,920 --> 00:32:00,654
we were getting out of a carrier voice

878
00:32:00,654 --> 00:32:02,894
call, which should have been capable of doing

879
00:32:02,894 --> 00:32:04,595
300 bits per second.

880
00:32:05,134 --> 00:32:08,275
By the screeching and the remarkably clever modulation,

881
00:32:08,975 --> 00:32:10,914
we were getting 56,000

882
00:32:10,975 --> 00:32:12,035
bits per second.

883
00:32:13,160 --> 00:32:15,579
The carrier tone was actually only

884
00:32:16,119 --> 00:32:17,019
eight kilohertz.

885
00:32:18,039 --> 00:32:18,839
8,000

886
00:32:18,839 --> 00:32:21,559
cycles a second. But we were managing by

887
00:32:21,559 --> 00:32:23,740
really clever modulation of that carrier

888
00:32:24,119 --> 00:32:26,279
getting seven times the signal out of it.

889
00:32:26,279 --> 00:32:27,339
56 kilobits.

890
00:32:28,214 --> 00:32:31,015
You go, well that's really clever. You analog

891
00:32:31,015 --> 00:32:31,994
guys are amazing.

892
00:32:32,855 --> 00:32:34,234
And then there was a fascinating

893
00:32:34,615 --> 00:32:35,115
accident.

894
00:32:36,214 --> 00:32:37,914
Over in Motorola land,

895
00:32:38,615 --> 00:32:40,454
they were running out of money. God knows

896
00:32:40,454 --> 00:32:43,500
why. And they were really clever radio folk.

897
00:32:43,720 --> 00:32:46,460
And they were really clever fiber optic folk.

898
00:32:47,080 --> 00:32:47,980
Really clever.

899
00:32:48,600 --> 00:32:50,460
But as a budget saving measure

900
00:32:50,840 --> 00:32:52,539
they put them in the same building.

901
00:32:54,494 --> 00:32:56,355
Now over in the carrier world,

902
00:32:56,735 --> 00:32:59,475
we've long since run away from copper conductors

903
00:32:59,535 --> 00:33:00,674
for our big circuitry.

904
00:33:01,215 --> 00:33:02,914
We were using fiber optic.

905
00:33:03,455 --> 00:33:04,755
It was, I suppose,

906
00:33:05,055 --> 00:33:05,555
cheaper,

907
00:33:06,095 --> 00:33:07,075
far more resilient,

908
00:33:07,859 --> 00:33:10,500
lasted for years longer than copper, didn't rust.

909
00:33:10,500 --> 00:33:11,880
You know, it was great stuff.

910
00:33:12,500 --> 00:33:14,819
And the technique was you have a torch

911
00:33:14,819 --> 00:33:15,640
at one end

912
00:33:16,819 --> 00:33:18,420
and the equivalent of an eye at the

913
00:33:18,420 --> 00:33:20,660
other end, a digital eye, and you go

914
00:33:20,660 --> 00:33:22,954
on or off, On off key

915
00:33:23,654 --> 00:33:26,294
And you can control the torch with some

916
00:33:26,294 --> 00:33:26,794
circuitry

917
00:33:27,414 --> 00:33:29,095
And you can start doing the torch a

918
00:33:29,095 --> 00:33:31,414
few million times a second On off on

919
00:33:31,414 --> 00:33:33,914
off on off And it kind of works

920
00:33:34,294 --> 00:33:35,674
up until a few million

921
00:33:36,359 --> 00:33:37,980
But how do you make it go faster?

922
00:33:39,160 --> 00:33:41,660
Well, switching lasers is difficult.

923
00:33:42,599 --> 00:33:45,019
On off keying really does strike a limit.

924
00:33:45,240 --> 00:33:47,240
And and the best they're able to do

925
00:33:47,240 --> 00:33:50,299
was just below 10 mill ten ten megahertz,

926
00:33:50,359 --> 00:33:51,555
ten minutes per second,

927
00:33:52,115 --> 00:33:53,875
which is Ethernet have been doing that for

928
00:33:53,875 --> 00:33:55,875
a decade. It was hardly rocket science in

929
00:33:55,875 --> 00:33:56,375
fiber.

930
00:33:57,515 --> 00:33:59,235
I'm just sorry. I I tell a lie.

931
00:33:59,235 --> 00:33:59,975
I got confused.

932
00:34:00,515 --> 00:34:03,555
They're up to a 10

933
00:34:03,555 --> 00:34:05,634
gigabits per second in on off keying. That's

934
00:34:05,634 --> 00:34:06,934
sort of where it got to.

935
00:34:07,440 --> 00:34:08,820
Right? Which is okay.

936
00:34:09,599 --> 00:34:11,920
But because they're in the same lab as

937
00:34:11,920 --> 00:34:15,280
the modem people, the modem folks said, well,

938
00:34:15,280 --> 00:34:16,980
you know, photons, electrons,

939
00:34:17,760 --> 00:34:20,639
same thing really, isn't it? Quantum mechanically, you

940
00:34:20,639 --> 00:34:23,855
know, just waves, dude. It's all waves.

941
00:34:24,554 --> 00:34:26,735
Why don't we phase modulate

942
00:34:27,355 --> 00:34:28,094
the light?

943
00:34:29,275 --> 00:34:30,574
Right. Can you?

944
00:34:31,275 --> 00:34:32,414
Yes. Sure.

945
00:34:33,514 --> 00:34:35,375
And amplitude modulate it.

946
00:34:36,440 --> 00:34:37,579
And all of a sudden

947
00:34:38,119 --> 00:34:38,619
with

948
00:34:39,079 --> 00:34:40,940
better digital signal processes,

949
00:34:41,799 --> 00:34:44,139
you can push more and more capacity

950
00:34:44,920 --> 00:34:46,460
through the same piece of fiber.

951
00:34:47,159 --> 00:34:48,859
Nothing's changed on the glass,

952
00:34:49,355 --> 00:34:52,014
but everything's changed electronics at either end.

953
00:34:52,474 --> 00:34:52,974
Everything.

954
00:34:54,074 --> 00:34:56,315
So instead of having a carrier network that

955
00:34:56,315 --> 00:34:58,635
ran at a few, well, 100 megabits per

956
00:34:58,635 --> 00:35:00,414
second, a few gigabits per second,

957
00:35:01,355 --> 00:35:03,135
I'm talking 20 gigs,

958
00:35:03,674 --> 00:35:04,654
a 100 gigs,

959
00:35:05,239 --> 00:35:06,299
200 gigs.

960
00:35:06,679 --> 00:35:08,699
And all I did was change the ends.

961
00:35:09,960 --> 00:35:10,460
Yeah.

962
00:35:10,760 --> 00:35:11,239
So

963
00:35:11,880 --> 00:35:13,719
yeah. I mean, it's a bit more complex

964
00:35:13,719 --> 00:35:15,980
than that because you're actually in in QSQ

965
00:35:16,440 --> 00:35:18,139
or quad quad keying.

966
00:35:18,440 --> 00:35:20,859
QSQ. Yeah. Yeah. You're you're on

967
00:35:21,255 --> 00:35:21,755
sidebands.

968
00:35:22,135 --> 00:35:23,914
You're dealing with multiple sidebands.

969
00:35:24,934 --> 00:35:26,295
You know, like, I used to run a

970
00:35:26,295 --> 00:35:27,914
radio that was a single sideband,

971
00:35:28,295 --> 00:35:30,474
but they were running all eight sidebands.

972
00:35:30,855 --> 00:35:33,355
And there's they were modulating onto eight sidebands

973
00:35:33,890 --> 00:35:36,289
or four side bands with more cymbal depth.

974
00:35:36,289 --> 00:35:37,809
So, I mean, there's a lot more going

975
00:35:37,809 --> 00:35:39,269
on there. But, yeah, basically,

976
00:35:40,289 --> 00:35:42,210
we, you know, got to the point where

977
00:35:42,210 --> 00:35:43,110
we were modulating

978
00:35:43,969 --> 00:35:45,750
the same way on fiber

979
00:35:46,210 --> 00:35:47,190
as we were

980
00:35:47,570 --> 00:35:48,070
on,

981
00:35:49,244 --> 00:35:52,204
on copper. And, you know, there were there

982
00:35:52,204 --> 00:35:52,704
were

983
00:35:54,764 --> 00:35:57,885
light based modulation schemes before. They just weren't

984
00:35:57,885 --> 00:35:59,505
ever put inside of a cable.

985
00:36:00,045 --> 00:36:02,710
Well, that work that it didn't work cost

986
00:36:02,710 --> 00:36:03,210
effectively,

987
00:36:03,510 --> 00:36:05,769
Moore's law. But if you wait long enough,

988
00:36:06,309 --> 00:36:08,710
you know, the the the gates get cheaper.

989
00:36:08,710 --> 00:36:10,570
You can put more on a single chip.

990
00:36:10,789 --> 00:36:13,269
I can do more on a single integrated

991
00:36:13,269 --> 00:36:14,650
circuit, so we did.

992
00:36:15,184 --> 00:36:16,784
Now I'm gonna swing back a bit a

993
00:36:16,784 --> 00:36:18,644
little bit and talk a little bit about

994
00:36:18,864 --> 00:36:21,525
the way the telephone networks actually worked.

995
00:36:21,905 --> 00:36:24,144
Because if everyone lifted up their handset at

996
00:36:24,144 --> 00:36:26,304
the same time and tried to dial, nothing

997
00:36:26,304 --> 00:36:27,045
would happen.

998
00:36:27,664 --> 00:36:27,989
Yep.

999
00:36:28,550 --> 00:36:30,389
There wasn't enough network to go around. It's

1000
00:36:30,469 --> 00:36:31,690
they're non blocking,

1001
00:36:32,070 --> 00:36:34,710
but they're non blocking because you have access

1002
00:36:34,710 --> 00:36:35,210
control.

1003
00:36:35,989 --> 00:36:38,150
You can say no. You can say no.

1004
00:36:38,150 --> 00:36:40,010
You're not allowed to make a phone call.

1005
00:36:40,335 --> 00:36:41,875
And you looked at some of the telephone

1006
00:36:41,934 --> 00:36:43,775
control rooms, and I remember visiting one here

1007
00:36:43,775 --> 00:36:46,175
in Australia, and they had the TV running

1008
00:36:46,175 --> 00:36:47,614
on massive screens at the front of the

1009
00:36:47,614 --> 00:36:50,114
room. Why? Because if there was an earthquake

1010
00:36:50,175 --> 00:36:53,234
in in Italy, some natural disaster in Japan,

1011
00:36:53,614 --> 00:36:55,554
they would block the calls manually.

1012
00:36:56,049 --> 00:36:57,809
I'm sorry. No one's ringing up Italy right

1013
00:36:57,809 --> 00:37:00,289
now because, you know, everyone's trying. It won't

1014
00:37:00,289 --> 00:37:00,789
fit.

1015
00:37:01,250 --> 00:37:03,250
But you think again about what they were

1016
00:37:03,250 --> 00:37:04,929
trying to do. And what they were trying

1017
00:37:04,929 --> 00:37:05,989
to do was ration

1018
00:37:06,609 --> 00:37:07,109
inadequacy,

1019
00:37:07,809 --> 00:37:08,309
scarcity.

1020
00:37:08,849 --> 00:37:10,949
There wasn't enough to go around.

1021
00:37:11,755 --> 00:37:14,335
And indeed, the whole margins of why telephony

1022
00:37:14,394 --> 00:37:15,775
made us so much money

1023
00:37:16,155 --> 00:37:18,974
was because there wasn't enough to go around.

1024
00:37:19,355 --> 00:37:21,514
So as well as the cost of I

1025
00:37:21,514 --> 00:37:24,094
run a telephone network, there was the opportunity

1026
00:37:24,554 --> 00:37:27,089
to charge what the economists call a scarcity

1027
00:37:27,309 --> 00:37:27,809
premium.

1028
00:37:28,589 --> 00:37:30,289
To actually damp demand,

1029
00:37:30,670 --> 00:37:32,670
you charge a whole lot more than it

1030
00:37:32,670 --> 00:37:33,170
costs

1031
00:37:33,630 --> 00:37:35,630
just to dissuade everyone from picking up the

1032
00:37:35,630 --> 00:37:37,089
handset at the same time.

1033
00:37:37,389 --> 00:37:39,661
It's it's the economics of scarcity, and the

1034
00:37:39,661 --> 00:37:41,054
telephone companies became the

1035
00:37:41,614 --> 00:37:44,355
became the rationalist and became very rich.

1036
00:37:45,054 --> 00:37:46,275
But Moore's law

1037
00:37:46,894 --> 00:37:49,315
was the insidious worm

1038
00:37:49,694 --> 00:37:53,074
that was undereating the entire foundation of this.

1039
00:37:53,599 --> 00:37:55,840
Because when you applied Moore's law to these

1040
00:37:55,840 --> 00:37:58,420
digital signal processes and the carrier's

1041
00:37:58,719 --> 00:37:59,699
own networks,

1042
00:38:00,719 --> 00:38:02,559
all of a sudden, they didn't have not

1043
00:38:02,559 --> 00:38:03,699
enough to go around.

1044
00:38:04,640 --> 00:38:06,579
They had so much. It wasn't funny.

1045
00:38:07,119 --> 00:38:08,019
It was abundance.

1046
00:38:09,035 --> 00:38:10,894
And no matter how they tried,

1047
00:38:11,355 --> 00:38:13,934
they couldn't put a damper on that abundant

1048
00:38:14,155 --> 00:38:14,655
capacity.

1049
00:38:15,914 --> 00:38:17,135
So all of a sudden,

1050
00:38:17,594 --> 00:38:20,015
there was this suspicion that the telephone companies

1051
00:38:20,235 --> 00:38:20,735
were

1052
00:38:21,309 --> 00:38:24,269
eating everyone. They they were monopolistic. They were

1053
00:38:24,269 --> 00:38:27,069
gouging. They were they were working against you

1054
00:38:27,069 --> 00:38:27,809
and I.

1055
00:38:28,269 --> 00:38:30,449
And to hang the the era of deregulation,

1056
00:38:31,389 --> 00:38:33,389
which was actually the era of entering a

1057
00:38:33,389 --> 00:38:35,195
different phase, the phase of abundance,

1058
00:38:35,655 --> 00:38:37,755
the utopia of Moore's law.

1059
00:38:38,775 --> 00:38:41,755
So now we're designing networks about abundance.

1060
00:38:43,175 --> 00:38:43,994
Where transmission

1061
00:38:44,454 --> 00:38:46,555
is so cheap, it's worthless.

1062
00:38:48,019 --> 00:38:50,980
Didn't someone sell Sprint's carriage network to someone

1063
00:38:50,980 --> 00:38:53,539
else for a dollar? Yeah. Pretty much. Yeah.

1064
00:38:53,539 --> 00:38:55,000
It's not cheap. Much.

1065
00:38:55,300 --> 00:38:56,920
They paid way too much because

1066
00:38:57,460 --> 00:38:59,159
it's not worth anything anymore.

1067
00:38:59,539 --> 00:39:00,519
It's abundance.

1068
00:39:01,405 --> 00:39:03,244
And you think again about what the job

1069
00:39:03,244 --> 00:39:05,485
of the Internet was. It was to take

1070
00:39:05,485 --> 00:39:07,025
you and your computer

1071
00:39:07,405 --> 00:39:09,724
over to that server a long, long way

1072
00:39:09,724 --> 00:39:10,224
away

1073
00:39:10,525 --> 00:39:12,125
to get the product or the service you

1074
00:39:12,125 --> 00:39:12,625
wanted.

1075
00:39:13,085 --> 00:39:15,440
Dear old Microsoft was having this room full

1076
00:39:15,440 --> 00:39:18,559
of servers to distribute the Windows operating system

1077
00:39:18,559 --> 00:39:20,719
in Seattle. And all of a sudden Seattle

1078
00:39:20,719 --> 00:39:23,440
was a networking black hole. It just couldn't

1079
00:39:23,440 --> 00:39:25,219
stand it. What about you with abundance?

1080
00:39:26,324 --> 00:39:29,904
Well, hang on a second. Computers are cheap

1081
00:39:29,940 --> 00:39:33,844
now. Storage is cheap now. Processing is cheap

1082
00:39:33,844 --> 00:39:34,344
now.

1083
00:39:34,964 --> 00:39:36,804
Why don't I take what you're trying to

1084
00:39:36,804 --> 00:39:37,304
distribute

1085
00:39:38,005 --> 00:39:40,025
and distribute a copy everywhere

1086
00:39:40,565 --> 00:39:42,025
right beside the user?

1087
00:39:42,860 --> 00:39:45,500
The content distribution networks. And it's all about

1088
00:39:45,500 --> 00:39:46,000
cost.

1089
00:39:46,780 --> 00:39:48,860
It's cheaper to do it that way than

1090
00:39:48,860 --> 00:39:51,099
to carry the bits. Even though we've made

1091
00:39:51,099 --> 00:39:54,460
transmission cheaper, computing got even cheaper. So did

1092
00:39:54,460 --> 00:39:54,960
storage.

1093
00:39:55,474 --> 00:39:57,875
And so you find companies like Netflix is

1094
00:39:57,875 --> 00:39:58,695
a good example,

1095
00:39:59,235 --> 00:40:00,215
Google's YouTube,

1096
00:40:00,914 --> 00:40:03,394
where they take the entire inventory of what

1097
00:40:03,394 --> 00:40:05,015
they have, that digital inventory,

1098
00:40:05,394 --> 00:40:06,215
and reproduce

1099
00:40:07,074 --> 00:40:08,375
it 10,000 times

1100
00:40:09,369 --> 00:40:10,829
close to everybody.

1101
00:40:11,449 --> 00:40:11,949
Mhmm.

1102
00:40:12,570 --> 00:40:15,309
It's the great vanishing network all over again.

1103
00:40:15,769 --> 00:40:18,010
Yeah. Because all of a sudden, the number

1104
00:40:18,010 --> 00:40:20,030
of miles that packets travel

1105
00:40:20,489 --> 00:40:22,730
from where they were stored in a storage

1106
00:40:22,730 --> 00:40:23,230
device

1107
00:40:23,804 --> 00:40:25,724
to my computer on, you know, in my

1108
00:40:25,724 --> 00:40:28,125
hand or on my desk, they've shrunk to

1109
00:40:28,125 --> 00:40:29,264
close to nothing.

1110
00:40:30,684 --> 00:40:31,184
Wow.

1111
00:40:31,804 --> 00:40:33,905
We've taken distance out of networking.

1112
00:40:35,724 --> 00:40:36,464
You see,

1113
00:40:36,844 --> 00:40:39,264
networking always had the distance square rule.

1114
00:40:40,739 --> 00:40:42,980
You know, it comes from physics that force

1115
00:40:42,980 --> 00:40:44,599
over distance, force reduces,

1116
00:40:45,300 --> 00:40:47,780
by a factor of distance squared because, you

1117
00:40:47,780 --> 00:40:50,280
know, that's the way geometry works.

1118
00:40:51,380 --> 00:40:53,614
The circumference of a circle. Sorry. The area

1119
00:40:53,614 --> 00:40:56,175
of a circle. Distance squared. So cost is

1120
00:40:56,175 --> 00:40:59,074
distance squared. Performance is distance squared.

1121
00:40:59,855 --> 00:41:02,434
Everything is distance squared. If you reduce distance,

1122
00:41:02,894 --> 00:41:03,714
things become

1123
00:41:04,014 --> 00:41:04,514
phenomenally

1124
00:41:04,815 --> 00:41:05,315
cheap.

1125
00:41:06,739 --> 00:41:09,460
Really so. So we took distance out of

1126
00:41:09,460 --> 00:41:09,960
networking.

1127
00:41:10,579 --> 00:41:12,599
When we took the networking out of networking.

1128
00:41:13,460 --> 00:41:15,940
And all of a sudden, today's networks are

1129
00:41:15,940 --> 00:41:16,440
remarkable.

1130
00:41:17,299 --> 00:41:19,699
We've gone from a provisioning model of, well,

1131
00:41:19,699 --> 00:41:21,319
it's over there if you need it.

1132
00:41:22,505 --> 00:41:24,364
To, look, I bought it to your doorstep

1133
00:41:24,825 --> 00:41:27,385
along with a thousand, a million other artifacts.

1134
00:41:27,385 --> 00:41:28,905
Yeah. If you want it, just open the

1135
00:41:28,905 --> 00:41:30,344
door and pick it up. It's just sitting

1136
00:41:30,344 --> 00:41:32,825
there waiting for you. It's just in case

1137
00:41:32,825 --> 00:41:33,325
delivery.

1138
00:41:33,890 --> 00:41:35,250
Just in case you want it, I've already

1139
00:41:35,250 --> 00:41:37,730
done it for you. I have pre networked

1140
00:41:37,730 --> 00:41:40,070
it for you. It's just waiting there.

1141
00:41:40,449 --> 00:41:42,530
And so we've managed to actually achieve the

1142
00:41:42,530 --> 00:41:43,829
one thing that this industry

1143
00:41:44,289 --> 00:41:46,369
always said. And I learned that when I

1144
00:41:46,369 --> 00:41:48,390
was a child sort of starting to think

1145
00:41:48,664 --> 00:41:51,144
bigger, faster, cheaper. Pick two. I can make

1146
00:41:51,144 --> 00:41:51,724
it bigger.

1147
00:41:52,425 --> 00:41:53,945
I can make it faster. But if I

1148
00:41:53,945 --> 00:41:55,405
do so, I can't make it cheaper.

1149
00:41:55,784 --> 00:41:57,304
I can make it bigger and cheaper, but

1150
00:41:57,304 --> 00:41:59,144
it won't be fast. Now there was always

1151
00:41:59,144 --> 00:42:00,925
a compromise. But all of a sudden,

1152
00:42:01,619 --> 00:42:02,839
bigger, faster, cheaper,

1153
00:42:03,299 --> 00:42:04,199
well, I'm good.

1154
00:42:04,500 --> 00:42:05,319
I'm good.

1155
00:42:05,940 --> 00:42:08,099
Well, it's and it still is to some

1156
00:42:08,099 --> 00:42:08,599
degree.

1157
00:42:08,900 --> 00:42:10,739
I mean, because you still have to store

1158
00:42:10,739 --> 00:42:13,319
the data at a thousand, a million locations.

1159
00:42:13,460 --> 00:42:15,779
You still have to like, it still costs

1160
00:42:15,779 --> 00:42:16,279
something.

1161
00:42:16,775 --> 00:42:17,655
And so you could

1162
00:42:18,295 --> 00:42:20,454
If it costs less than a millionth of

1163
00:42:20,454 --> 00:42:22,454
what it used to cost Yeah. It actually

1164
00:42:22,454 --> 00:42:23,195
costs less.

1165
00:42:23,735 --> 00:42:25,894
You can replicate it and still pay less.

1166
00:42:25,894 --> 00:42:27,494
I mean, you yeah. You but you could

1167
00:42:27,494 --> 00:42:29,735
actually still not store the data and still

1168
00:42:29,735 --> 00:42:31,909
pay less. But economically,

1169
00:42:32,210 --> 00:42:34,130
the trade off is such that it's better

1170
00:42:34,130 --> 00:42:34,789
not to.

1171
00:42:35,250 --> 00:42:37,730
It's better said yeah. Moore's Law said buy

1172
00:42:37,730 --> 00:42:38,469
more computers.

1173
00:42:38,929 --> 00:42:39,670
They're cheaper.

1174
00:42:40,130 --> 00:42:41,969
Keep up with your competitive. You know? It's

1175
00:42:41,969 --> 00:42:43,329
a race. Come on. Come on. Come on.

1176
00:42:43,329 --> 00:42:45,914
Change. Change. Change. Get on with it. And

1177
00:42:45,914 --> 00:42:47,434
so now I think we're looking at at

1178
00:42:47,434 --> 00:42:48,815
the next revolution.

1179
00:42:49,675 --> 00:42:52,015
And it's interesting because we've lived through

1180
00:42:52,394 --> 00:42:53,215
Moore's Law,

1181
00:42:54,715 --> 00:42:56,414
but things are looking hairy.

1182
00:42:57,035 --> 00:42:58,574
And this is, I suppose, the

1183
00:42:58,875 --> 00:43:01,640
the portent, which is a useful place to

1184
00:43:01,640 --> 00:43:02,300
end this.

1185
00:43:03,079 --> 00:43:05,579
The feature size on today's silicon

1186
00:43:06,200 --> 00:43:07,660
is down around two nanometers.

1187
00:43:09,000 --> 00:43:09,980
It's remarkable

1188
00:43:10,519 --> 00:43:12,780
to get there. We have to vaporize

1189
00:43:13,480 --> 00:43:16,700
Indian tin using laser pulsing

1190
00:43:17,375 --> 00:43:20,034
to get a light signal of nine nanometers

1191
00:43:20,655 --> 00:43:22,335
in order to focus it down onto a

1192
00:43:22,335 --> 00:43:25,394
chip to etch features that are two nanometers

1193
00:43:25,534 --> 00:43:26,034
wide.

1194
00:43:26,335 --> 00:43:27,315
Yeah. Wow.

1195
00:43:28,574 --> 00:43:30,494
What do we do next? Oh, let's make

1196
00:43:30,494 --> 00:43:31,714
it down to one nanometer.

1197
00:43:32,730 --> 00:43:33,230
Woo!

1198
00:43:34,010 --> 00:43:35,469
How big is a silicon atom?

1199
00:43:36,090 --> 00:43:37,869
Well, a fifth of a nanometer

1200
00:43:39,210 --> 00:43:41,530
So I get five of them across a

1201
00:43:41,530 --> 00:43:42,909
one nanometer track

1202
00:43:43,210 --> 00:43:44,030
Just five

1203
00:43:45,434 --> 00:43:47,135
Now the whole idea of semiconductors

1204
00:43:47,914 --> 00:43:49,775
was that electrons are well behaved.

1205
00:43:50,235 --> 00:43:52,655
And if I put a slightly excess amount,

1206
00:43:52,795 --> 00:43:54,635
it can change a conductor to a non

1207
00:43:54,635 --> 00:43:55,135
conductor.

1208
00:43:55,515 --> 00:43:56,015
Right?

1209
00:43:56,980 --> 00:43:58,599
But this kind of relies on

1210
00:43:58,900 --> 00:44:00,920
the polite behavior of electrons.

1211
00:44:01,460 --> 00:44:03,300
But when you get down to five atoms,

1212
00:44:03,300 --> 00:44:04,440
they're not very polite.

1213
00:44:04,980 --> 00:44:06,980
And in fact, they're not even electrons. They're

1214
00:44:06,980 --> 00:44:07,800
a wave function

1215
00:44:08,180 --> 00:44:09,160
and they leak.

1216
00:44:10,275 --> 00:44:11,255
Can we make

1217
00:44:12,755 --> 00:44:14,135
gates at one nanometer?

1218
00:44:15,074 --> 00:44:15,734
Well, no.

1219
00:44:17,074 --> 00:44:18,054
It just can't.

1220
00:44:18,514 --> 00:44:20,214
They don't behave well enough.

1221
00:44:20,514 --> 00:44:22,034
So at some point, you've got to be

1222
00:44:22,034 --> 00:44:22,934
a lot cleverer

1223
00:44:23,554 --> 00:44:24,855
to get the next doubling.

1224
00:44:25,320 --> 00:44:27,480
I can't just make it smaller anymore. We've

1225
00:44:27,480 --> 00:44:30,380
been making it smaller since 1957.

1226
00:44:31,400 --> 00:44:31,900
Wow.

1227
00:44:32,680 --> 00:44:34,840
You know? So we've done sixty five years

1228
00:44:34,840 --> 00:44:37,000
of making it smaller. And we've now got

1229
00:44:37,000 --> 00:44:38,519
to the size of an atom. And it's

1230
00:44:38,519 --> 00:44:39,660
kind of, dudes,

1231
00:44:40,554 --> 00:44:42,735
you're done. In this this universe,

1232
00:44:43,114 --> 00:44:45,114
in this model of physics, you know, physics

1233
00:44:45,114 --> 00:44:47,054
version one, that's it.

1234
00:44:47,594 --> 00:44:49,375
You you've got to go elsewhere.

1235
00:44:49,675 --> 00:44:51,355
And, you know, you've go into the fuzzy

1236
00:44:51,355 --> 00:44:53,940
world of quantum computing, and my brain just

1237
00:44:53,940 --> 00:44:56,019
exploded a second ago because I don't understand

1238
00:44:56,019 --> 00:44:56,519
it.

1239
00:44:56,820 --> 00:44:58,739
Or you can try and build three d

1240
00:44:58,739 --> 00:44:59,719
silicon lattices

1241
00:45:00,179 --> 00:45:02,119
at two nanometer feature size.

1242
00:45:02,659 --> 00:45:03,960
And and quite frankly,

1243
00:45:04,260 --> 00:45:04,920
you know,

1244
00:45:05,219 --> 00:45:06,039
that's a challenge.

1245
00:45:06,755 --> 00:45:10,135
In anybody's language, that's a challenge. Yeah. Yeah.

1246
00:45:10,355 --> 00:45:11,494
Or you can say,

1247
00:45:11,875 --> 00:45:13,815
have we reached the end of this road?

1248
00:45:14,515 --> 00:45:16,775
And and it's sort of a plausible question.

1249
00:45:18,400 --> 00:45:19,760
And and that's what I'd like to leave

1250
00:45:19,760 --> 00:45:20,659
you with, because

1251
00:45:20,960 --> 00:45:22,420
if you look at the

1252
00:45:23,039 --> 00:45:27,059
feature size and processing capacity of cores lately,

1253
00:45:28,000 --> 00:45:30,559
instead of making it, if you will, smaller

1254
00:45:30,559 --> 00:45:32,655
and smaller and smaller, what we've actually been

1255
00:45:32,655 --> 00:45:35,135
doing is adding to the number of cores

1256
00:45:35,135 --> 00:45:38,015
at each circuit. So it's integrated circuit process.

1257
00:45:38,015 --> 00:45:39,775
It now has 256

1258
00:45:39,775 --> 00:45:40,275
cores,

1259
00:45:40,894 --> 00:45:43,715
but it probably won't have 512.

1260
00:45:44,255 --> 00:45:46,094
And and even if it does but even

1261
00:45:46,094 --> 00:45:46,914
if it does,

1262
00:45:47,469 --> 00:45:50,690
you're still looking at now a chip die

1263
00:45:51,390 --> 00:45:53,630
that is Bigger. One and a half to

1264
00:45:53,630 --> 00:45:54,849
two times larger

1265
00:45:55,469 --> 00:45:57,469
that consumes one and a half to two

1266
00:45:57,469 --> 00:45:58,849
times the amount of power.

1267
00:45:59,150 --> 00:46:01,764
So you're really not gaining a lot. And

1268
00:46:01,764 --> 00:46:04,005
it has 10 times the failure rate in

1269
00:46:04,005 --> 00:46:06,585
production Yeah. Etcetera. So all of a sudden,

1270
00:46:06,644 --> 00:46:08,344
this is this next generation

1271
00:46:08,724 --> 00:46:09,385
big chip

1272
00:46:09,764 --> 00:46:12,264
is actually more expensive than today's generation

1273
00:46:12,724 --> 00:46:13,465
small chip.

1274
00:46:13,960 --> 00:46:15,320
And and it's kind of what have you

1275
00:46:15,320 --> 00:46:17,079
won? And the other thing is too, the

1276
00:46:17,079 --> 00:46:19,239
physically bigger chip, it takes more time to

1277
00:46:19,239 --> 00:46:21,559
get from one side to another. Yeah. Again,

1278
00:46:21,559 --> 00:46:22,380
it's slower.

1279
00:46:22,920 --> 00:46:24,460
So we're topping out.

1280
00:46:25,480 --> 00:46:28,599
CPU processing, three gigahertz clocks, not really getting

1281
00:46:28,599 --> 00:46:29,724
faster in production.

1282
00:46:30,344 --> 00:46:31,324
Feature size,

1283
00:46:31,625 --> 00:46:34,125
we might hit a trillion features per chip.

1284
00:46:34,344 --> 00:46:36,525
But beyond that, there's no clear indicator.

1285
00:46:36,905 --> 00:46:39,085
Because unless someone thinks of something

1286
00:46:39,545 --> 00:46:41,704
as out of the box, so much as

1287
00:46:41,704 --> 00:46:44,765
the transistor and the integrated circuit, something else,

1288
00:46:45,059 --> 00:46:47,460
Unless something else walks in the room, we've

1289
00:46:47,460 --> 00:46:49,320
kind of milked this for all it's worth.

1290
00:46:50,180 --> 00:46:52,420
And and my closing thought is, well, what

1291
00:46:52,420 --> 00:46:53,320
does that mean?

1292
00:46:54,180 --> 00:46:56,660
This means that the future is no longer

1293
00:46:56,660 --> 00:46:57,320
a competition

1294
00:46:57,619 --> 00:46:58,440
to the current.

1295
00:46:59,555 --> 00:47:01,315
I can go back to twenty five year

1296
00:47:01,315 --> 00:47:02,855
life cycles on my equipment.

1297
00:47:03,474 --> 00:47:03,974
Yep.

1298
00:47:04,994 --> 00:47:05,894
I can be

1299
00:47:06,355 --> 00:47:08,835
as big as I want without worrying about

1300
00:47:08,835 --> 00:47:09,494
my competitors.

1301
00:47:10,195 --> 00:47:12,515
And this is even happening down to the

1302
00:47:12,515 --> 00:47:13,559
cell phone market

1303
00:47:13,880 --> 00:47:16,739
and even to the laptop and the the

1304
00:47:16,760 --> 00:47:20,440
home computer market. Right? Like, okay. Used to

1305
00:47:20,440 --> 00:47:22,360
be I replaced my computer every two years.

1306
00:47:22,360 --> 00:47:23,180
Why? Because

1307
00:47:23,640 --> 00:47:25,500
as I upgraded software, it gets slower.

1308
00:47:26,094 --> 00:47:28,894
Yep. Now It's it's well, no. You have

1309
00:47:28,894 --> 00:47:30,655
to upgrade. No. You have to upgrade because

1310
00:47:30,655 --> 00:47:32,195
the current model comes in black.

1311
00:47:32,494 --> 00:47:35,135
Yes. That's that's exactly right. That's what's happening.

1312
00:47:35,135 --> 00:47:37,055
Next year next year's will come in gray.

1313
00:47:37,055 --> 00:47:39,389
Yeah. You know? And that's the the Yeah.

1314
00:47:39,549 --> 00:47:40,769
We've moved from capability

1315
00:47:41,069 --> 00:47:42,449
yeah. We've come from capability

1316
00:47:42,750 --> 00:47:43,489
to fashion.

1317
00:47:44,190 --> 00:47:46,429
And whenever you get to fashion, like

1318
00:47:47,309 --> 00:47:48,909
Well, the people who are big stay big,

1319
00:47:48,909 --> 00:47:50,849
and the people who are small can't compete

1320
00:47:51,230 --> 00:47:52,289
because the technological

1321
00:47:52,670 --> 00:47:53,170
edge

1322
00:47:53,585 --> 00:47:56,244
that gave small its sort of cut through

1323
00:47:56,464 --> 00:47:57,764
that actually made Google.

1324
00:47:58,304 --> 00:48:00,545
Let's index all the worlds of information. The

1325
00:48:00,545 --> 00:48:03,105
incumbents at the time were going, laugh point,

1326
00:48:03,105 --> 00:48:04,884
giggle hysterically, you're nuts.

1327
00:48:05,184 --> 00:48:07,349
Now Google's the behemoth because it did it.

1328
00:48:07,750 --> 00:48:10,230
And we're gonna strike this right now as

1329
00:48:10,230 --> 00:48:12,010
a closing thought about AI.

1330
00:48:12,869 --> 00:48:15,130
Because the current thought about AI is,

1331
00:48:15,430 --> 00:48:17,510
as long as it gets cheaper and bigger

1332
00:48:17,510 --> 00:48:18,890
and cheaper and bigger,

1333
00:48:19,190 --> 00:48:20,090
this will work.

1334
00:48:20,465 --> 00:48:22,224
It doesn't work at the moment. It's funded

1335
00:48:22,224 --> 00:48:25,585
by venture capital. It'll work tomorrow. Trust me,

1336
00:48:25,585 --> 00:48:26,405
Moore's law.

1337
00:48:26,704 --> 00:48:28,565
But what if Moore's law isn't?

1338
00:48:29,184 --> 00:48:31,344
What if it doesn't? Yeah. How long does

1339
00:48:31,344 --> 00:48:34,420
the venture capital last? Yeah. People say, just

1340
00:48:34,420 --> 00:48:36,500
remember about AI that it's the worst it'll

1341
00:48:36,500 --> 00:48:37,239
ever be.

1342
00:48:37,780 --> 00:48:39,539
And you know what? You should remember the

1343
00:48:39,539 --> 00:48:42,179
opposite of that, that it may not actually

1344
00:48:42,179 --> 00:48:43,640
ever get much better.

1345
00:48:44,260 --> 00:48:46,440
Well, you're relying you're relying on

1346
00:48:46,855 --> 00:48:49,015
a universe whose physics are not the physics

1347
00:48:49,015 --> 00:48:50,474
you live in. Yeah. You're

1348
00:48:50,934 --> 00:48:53,335
relying on something else that we don't understand.

1349
00:48:53,335 --> 00:48:55,414
So Yeah. It's an interesting thought that we've

1350
00:48:55,414 --> 00:48:58,555
lived out sixty five years of phenomenal,

1351
00:48:59,335 --> 00:49:00,474
prodigious growth

1352
00:49:00,880 --> 00:49:03,760
coming from those subtle inventions of the transistor

1353
00:49:03,760 --> 00:49:06,000
and the integrated circuit. And we've milked them

1354
00:49:06,000 --> 00:49:07,460
for all they're worth.

1355
00:49:08,079 --> 00:49:10,340
And and the intrusive little thought is,

1356
00:49:10,800 --> 00:49:12,400
is that all there is? Have we milked

1357
00:49:12,400 --> 00:49:14,079
it as far as it will go? And

1358
00:49:14,079 --> 00:49:15,460
if the answer is yes,

1359
00:49:16,965 --> 00:49:19,545
start to think differently about tomorrow's world.

1360
00:49:20,164 --> 00:49:22,585
Yeah. It might look a lot more like

1361
00:49:23,364 --> 00:49:26,485
the nineteen twenties, the nineteen thirties. Thereafter, there

1362
00:49:26,485 --> 00:49:28,405
was a huge burst of innovation in the

1363
00:49:28,405 --> 00:49:29,545
mechanical world,

1364
00:49:29,989 --> 00:49:32,630
and then everything kinda settled down for a

1365
00:49:32,630 --> 00:49:34,409
couple of decades and,

1366
00:49:35,109 --> 00:49:37,589
you know. Yes. It's it's the same kind

1367
00:49:37,589 --> 00:49:39,670
of of of issues flying around here. And

1368
00:49:39,670 --> 00:49:41,989
and, you know, that's the transition we're going

1369
00:49:41,989 --> 00:49:43,109
through at the moment. So,

1370
00:49:44,315 --> 00:49:46,474
I leave I leave our listeners with with

1371
00:49:46,474 --> 00:49:48,255
that thought that, you know,

1372
00:49:48,875 --> 00:49:49,535
this particular

1373
00:49:49,914 --> 00:49:51,755
oh, it's a happy thought. Oh, but the

1374
00:49:51,755 --> 00:49:54,235
opportunities of tomorrow, we're not just making today

1375
00:49:54,235 --> 00:49:54,735
smaller.

1376
00:49:55,275 --> 00:49:57,275
Tomorrow will not be like, you know, will

1377
00:49:57,275 --> 00:49:57,775
be

1378
00:49:58,679 --> 00:50:00,599
it's a different kind of world when you

1379
00:50:00,599 --> 00:50:02,920
just can't rely on the prodigious nature of

1380
00:50:02,920 --> 00:50:05,019
Moore's Law to fix everything up,

1381
00:50:05,320 --> 00:50:07,800
including AI. You've gotta do something else with

1382
00:50:07,800 --> 00:50:09,420
it. Yeah. Yeah. Exactly.

1383
00:50:10,519 --> 00:50:11,019
Well,

1384
00:50:11,559 --> 00:50:13,099
thanks, Jeff, for that,

1385
00:50:13,905 --> 00:50:16,144
long or short or however you wanna put

1386
00:50:16,144 --> 00:50:18,085
it, that trip through history.

1387
00:50:18,625 --> 00:50:19,125
And,

1388
00:50:20,945 --> 00:50:22,545
I don't know. I don't have much else

1389
00:50:22,545 --> 00:50:24,945
to add to that. So do you have

1390
00:50:24,945 --> 00:50:26,785
a place where people can follow you or

1391
00:50:26,785 --> 00:50:28,065
get in touch with you if they want

1392
00:50:28,065 --> 00:50:28,565
to?

1393
00:50:29,139 --> 00:50:33,000
All my ramblings are at dubdubdub.potaroo.net.

1394
00:50:33,619 --> 00:50:35,460
You might need a spare hour or two,

1395
00:50:35,460 --> 00:50:36,980
but I promise I try and make it

1396
00:50:36,980 --> 00:50:37,960
varied and entertaining

1397
00:50:38,340 --> 00:50:40,019
with the kind of stuff I put up

1398
00:50:40,019 --> 00:50:42,420
there. But they ain't tweets. Oh my god.

1399
00:50:42,420 --> 00:50:44,280
No. They're not tweets. I'm sorry.

1400
00:50:44,775 --> 00:50:46,795
This is actually the curse of the writer.

1401
00:50:47,014 --> 00:50:49,275
Even when I write short post on LinkedIn,

1402
00:50:49,335 --> 00:50:51,655
they're, like, 300 words. And, like, I just

1403
00:50:51,655 --> 00:50:54,454
can't I just I just I'm sorry. I

1404
00:50:54,454 --> 00:50:54,954
can't.

1405
00:50:55,894 --> 00:50:57,815
Right. So, you know, if you want the

1406
00:50:57,815 --> 00:51:00,329
slightly condensed version, you've got it in one

1407
00:51:00,329 --> 00:51:02,570
hour on this podcast here and there. That's

1408
00:51:02,570 --> 00:51:04,510
as short as it gets. Exactly.

1409
00:51:05,210 --> 00:51:08,010
So I try. Alright. Cool. And I'm Russ

1410
00:51:08,010 --> 00:51:09,210
White. You can always find me here at

1411
00:51:09,210 --> 00:51:11,230
the hedge at rule eleven dot tech

1412
00:51:11,690 --> 00:51:12,829
on x occasionally,

1413
00:51:13,130 --> 00:51:15,789
whenever I feel like logging into that account,

1414
00:51:16,144 --> 00:51:17,204
and LinkedIn.

1415
00:51:17,744 --> 00:51:19,184
Thanks for listening to this episode of The

1416
00:51:19,184 --> 00:51:21,585
Hedge. We know your attention is we live

1417
00:51:21,585 --> 00:51:24,224
in an attention driven economy, and your attention

1418
00:51:24,224 --> 00:51:26,464
is high value at this point. So we

1419
00:51:26,464 --> 00:51:28,144
thank you for listening all the way to

1420
00:51:28,144 --> 00:51:29,920
this bitter end of the hedge, and we

1421
00:51:29,920 --> 00:51:31,300
will catch you next time.