1
00:00:00,420 --> 00:00:05,910
[Music]

2
00:00:10,719 --> 00:00:15,519
Hello. Welcome again to PyCon Day Three.

3
00:00:13,840 --> 00:00:17,279
Uh it's my pleasure to announce

4
00:00:15,519 --> 00:00:19,119
Christopher Nogabau with his talk, The

5
00:00:17,279 --> 00:00:19,840
Death of Consequences. Take it away,

6
00:00:19,119 --> 00:00:21,390
Chris.

7
00:00:19,840 --> 00:00:23,519
Thank you.

8
00:00:21,390 --> 00:00:26,640
[Applause]

9
00:00:23,519 --> 00:00:28,080
Hi. I you've come to my opinion talk.

10
00:00:26,640 --> 00:00:29,760
This is a talk I prepared for a

11
00:00:28,080 --> 00:00:32,320
different conference which is slightly

12
00:00:29,760 --> 00:00:33,840
less lax on timing. It is going to run

13
00:00:32,320 --> 00:00:35,680
long. It's not going to run so long that

14
00:00:33,840 --> 00:00:37,120
you'll miss the next talk. But if you're

15
00:00:35,680 --> 00:00:38,559
getting antsy, it's okay. You will get

16
00:00:37,120 --> 00:00:41,840
time to go to another room at the end of

17
00:00:38,559 --> 00:00:43,920
it. Um, so yeah, again, my name

18
00:00:41,840 --> 00:00:45,600
you may know me from the talk I gave

19
00:00:43,920 --> 00:00:47,120
yesterday or the fact that I host

20
00:00:45,600 --> 00:00:50,160
lightning talks or various other things

21
00:00:47,120 --> 00:00:51,920
I do in the Python community. Um, you

22
00:00:50,160 --> 00:00:53,199
may have seen over the last year a lot

23
00:00:51,920 --> 00:00:55,600
of talks about open source

24
00:00:53,199 --> 00:00:57,199
sustainability, the way things are in

25
00:00:55,600 --> 00:00:59,280
open source. And I'm here to set the

26
00:00:57,199 --> 00:01:01,760
stage for those talks that you've seen.

27
00:00:59,280 --> 00:01:04,320
This is a talk about why things are the

28
00:01:01,760 --> 00:01:05,840
way they are in open source. Now, how

29
00:01:04,320 --> 00:01:08,000
this talk is going to work is that I'm

30
00:01:05,840 --> 00:01:09,600
going to start with some background. I'm

31
00:01:08,000 --> 00:01:11,840
going to selectively present some

32
00:01:09,600 --> 00:01:13,920
evidence and then I'm going to present

33
00:01:11,840 --> 00:01:16,560
the conclusions that follow directly

34
00:01:13,920 --> 00:01:18,320
from that evidence.

35
00:01:16,560 --> 00:01:21,200
If you're into rhetoric, you will

36
00:01:18,320 --> 00:01:23,600
recognize this as reductionism.

37
00:01:21,200 --> 00:01:24,720
This is necessary because I have only 30

38
00:01:23,600 --> 00:01:26,560
minutes which I'm going to use up

39
00:01:24,720 --> 00:01:30,080
entirely to try to deliver a clear

40
00:01:26,560 --> 00:01:33,119
message. You have been warned.

41
00:01:30,080 --> 00:01:35,680
So the talk, this talk is about the

42
00:01:33,119 --> 00:01:37,280
first golden age of open source which I

43
00:01:35,680 --> 00:01:39,200
believe everyone here in this room has

44
00:01:37,280 --> 00:01:41,520
lived through

45
00:01:39,200 --> 00:01:44,960
and is now in my view firmly and

46
00:01:41,520 --> 00:01:46,880
indisputably at its end. Two key events

47
00:01:44,960 --> 00:01:49,680
happened last year that I think signal

48
00:01:46,880 --> 00:01:51,600
that downfall. The first, which I spoke

49
00:01:49,680 --> 00:01:53,759
about at Monktoberfest last year. You

50
00:01:51,600 --> 00:01:55,680
can look up the video, stems from us as

51
00:01:53,759 --> 00:01:57,920
a community being unwilling to address

52
00:01:55,680 --> 00:02:00,079
the false and destructive narrative that

53
00:01:57,920 --> 00:02:02,719
being a terrible person makes you a more

54
00:02:00,079 --> 00:02:04,960
effective contributor to open source.

55
00:02:02,719 --> 00:02:07,920
The result of this was almost an

56
00:02:04,960 --> 00:02:10,080
international security incident.

57
00:02:07,920 --> 00:02:11,760
The other is that IBM announced that it

58
00:02:10,080 --> 00:02:13,440
was to acquire Hashi Corp in a

59
00:02:11,760 --> 00:02:15,840
transaction that closed earlier this

60
00:02:13,440 --> 00:02:18,160
year. This acquisition followed the

61
00:02:15,840 --> 00:02:20,640
rejection first by industry and then by

62
00:02:18,160 --> 00:02:23,599
the market of Hashi Corp's decision to

63
00:02:20,640 --> 00:02:25,440
rellic their suite of open-source tools,

64
00:02:23,599 --> 00:02:28,000
notably including Terraform under

65
00:02:25,440 --> 00:02:30,319
non-opensource terms that prevented

66
00:02:28,000 --> 00:02:32,239
cloud providers that Terraform was

67
00:02:30,319 --> 00:02:35,599
designed to interoperate with from

68
00:02:32,239 --> 00:02:38,239
offering services based on those tools.

69
00:02:35,599 --> 00:02:40,400
Hashi Corp's actions were a reaction

70
00:02:38,239 --> 00:02:43,120
like countless other open core companies

71
00:02:40,400 --> 00:02:45,040
to better larger better capitalized

72
00:02:43,120 --> 00:02:47,040
companies launching services that

73
00:02:45,040 --> 00:02:49,680
competed with their revenue streams

74
00:02:47,040 --> 00:02:51,120
using code that they had written. And

75
00:02:49,680 --> 00:02:53,200
while it makes sense to fight back

76
00:02:51,120 --> 00:02:54,480
against such extractive behavior, the

77
00:02:53,200 --> 00:02:56,480
way in which they did so was

78
00:02:54,480 --> 00:02:59,440
comprehensively rejected by people who

79
00:02:56,480 --> 00:03:01,360
care about open source. That rejection

80
00:02:59,440 --> 00:03:03,840
likely caused the drop in share price

81
00:03:01,360 --> 00:03:06,159
that enabled that acquisition. A

82
00:03:03,840 --> 00:03:08,239
complete failure of an open-source pro

83
00:03:06,159 --> 00:03:10,560
uh product business to operate as an

84
00:03:08,239 --> 00:03:12,480
independent entity.

85
00:03:10,560 --> 00:03:14,560
And ultimately the root cause of both of

86
00:03:12,480 --> 00:03:16,480
these things are the same. We as a

87
00:03:14,560 --> 00:03:19,360
movement have created incentives for

88
00:03:16,480 --> 00:03:20,800
people to perform undesirable behaviors.

89
00:03:19,360 --> 00:03:22,560
And today I want to talk about

90
00:03:20,800 --> 00:03:25,200
extractive behavior that leads to

91
00:03:22,560 --> 00:03:27,120
smaller projects becoming unsustainable

92
00:03:25,200 --> 00:03:29,440
and how we as a movement have actively

93
00:03:27,120 --> 00:03:31,519
chosen to incentivize extractive

94
00:03:29,440 --> 00:03:33,840
behavior and how we've thrown away our

95
00:03:31,519 --> 00:03:35,680
ability to fight back.

96
00:03:33,840 --> 00:03:38,000
So extractive behavior is not unique to

97
00:03:35,680 --> 00:03:39,680
open source. Many creative industries

98
00:03:38,000 --> 00:03:40,799
suffer from extractive behavior from

99
00:03:39,680 --> 00:03:44,319
corporations that are not

100
00:03:40,799 --> 00:03:45,840
disincentivized to do so. In 2023, a

101
00:03:44,319 --> 00:03:47,840
group of workers realized that their

102
00:03:45,840 --> 00:03:50,560
industry was likely to be significantly

103
00:03:47,840 --> 00:03:52,879
impacted by the rise of AI. That their

104
00:03:50,560 --> 00:03:54,560
work could be used to train AI models,

105
00:03:52,879 --> 00:03:56,959
which could eventually be used to

106
00:03:54,560 --> 00:03:59,840
devalue their work or worse still,

107
00:03:56,959 --> 00:04:01,840
completely replace their jobs. A threat

108
00:03:59,840 --> 00:04:03,680
that we as software engineers feel all

109
00:04:01,840 --> 00:04:05,280
the time at the moment. And yet, it

110
00:04:03,680 --> 00:04:08,720
wasn't the tech industry that stepped

111
00:04:05,280 --> 00:04:10,799
up. It was the entertainment industry.

112
00:04:08,720 --> 00:04:13,519
Both the writers unions and the actors

113
00:04:10,799 --> 00:04:15,680
unions for film and television in the US

114
00:04:13,519 --> 00:04:18,400
withdrew their labor for more than six

115
00:04:15,680 --> 00:04:20,720
months in protest to changes affecting

116
00:04:18,400 --> 00:04:23,040
their industry and threats that they saw

117
00:04:20,720 --> 00:04:24,880
on the horizon. Entertainment

118
00:04:23,040 --> 00:04:27,120
productions halted for more than 6

119
00:04:24,880 --> 00:04:29,199
months. But as a result, both these

120
00:04:27,120 --> 00:04:31,600
unions achieved long-term protection

121
00:04:29,199 --> 00:04:33,600
against their work or likenesses being

122
00:04:31,600 --> 00:04:36,000
exploited for AI training without their

123
00:04:33,600 --> 00:04:38,080
consent.

124
00:04:36,000 --> 00:04:39,840
Collective action is a balance against

125
00:04:38,080 --> 00:04:42,479
the power dynamics inherent in

126
00:04:39,840 --> 00:04:44,720
employment situations. It's easy for an

127
00:04:42,479 --> 00:04:47,040
employer to compel a single employee to

128
00:04:44,720 --> 00:04:48,800
perform undesirable work because the

129
00:04:47,040 --> 00:04:50,960
employer can threaten that employees

130
00:04:48,800 --> 00:04:52,960
livelihood. By holding each other in

131
00:04:50,960 --> 00:04:56,320
solidarity and acting towards the same

132
00:04:52,960 --> 00:04:58,000
goals using the same tactics, workers

133
00:04:56,320 --> 00:05:00,000
can impose consequences on their

134
00:04:58,000 --> 00:05:01,680
employers for making unreasonable

135
00:05:00,000 --> 00:05:03,600
demands.

136
00:05:01,680 --> 00:05:06,000
The most drastic form of collective

137
00:05:03,600 --> 00:05:08,960
action is the threat to withdraw labor,

138
00:05:06,000 --> 00:05:11,199
a strike. The theory behind it in simple

139
00:05:08,960 --> 00:05:13,120
terms is that collectively workers

140
00:05:11,199 --> 00:05:15,440
demand that their employer materially

141
00:05:13,120 --> 00:05:17,520
improve their conditions. This is a form

142
00:05:15,440 --> 00:05:19,759
of sacrifice on the part of the

143
00:05:17,520 --> 00:05:21,600
employer. If the employer does not meet

144
00:05:19,759 --> 00:05:23,360
the demands of the workers, then the

145
00:05:21,600 --> 00:05:25,759
workers can instead show that there are

146
00:05:23,360 --> 00:05:27,520
potentially worse economic outcomes than

147
00:05:25,759 --> 00:05:30,880
the sacrifice that is being asked of

148
00:05:27,520 --> 00:05:32,400
them collectively. And and key to this

149
00:05:30,880 --> 00:05:34,000
is that the workers show that they are

150
00:05:32,400 --> 00:05:36,320
also willing to make collective

151
00:05:34,000 --> 00:05:38,320
sacrifice. Withdrawing labor means

152
00:05:36,320 --> 00:05:40,800
withdrawing the right to get paid while

153
00:05:38,320 --> 00:05:42,720
they aren't working. And while they are

154
00:05:40,800 --> 00:05:44,880
not direct parties to the dispute,

155
00:05:42,720 --> 00:05:46,240
strike action creates inconvenience for

156
00:05:44,880 --> 00:05:48,000
consumers of the business that is

157
00:05:46,240 --> 00:05:49,600
subject to strike action. And this is

158
00:05:48,000 --> 00:05:51,199
why workers need to educate the

159
00:05:49,600 --> 00:05:53,680
consumers about the reasons for their

160
00:05:51,199 --> 00:05:56,320
action and why demands are reasonable

161
00:05:53,680 --> 00:05:58,080
and necessary. This builds understanding

162
00:05:56,320 --> 00:06:00,960
and hopefully compels long-term

163
00:05:58,080 --> 00:06:03,039
behavioral change. Performing your own

164
00:06:00,960 --> 00:06:05,520
education is key because you can be

165
00:06:03,039 --> 00:06:07,280
certain that the impacted employers will

166
00:06:05,520 --> 00:06:10,720
spend money to make their point of view

167
00:06:07,280 --> 00:06:12,240
known and you have to compete with that.

168
00:06:10,720 --> 00:06:14,000
So it goes without saying that this sort

169
00:06:12,240 --> 00:06:17,360
of action is devastating for everyone

170
00:06:14,000 --> 00:06:19,600
involved. It demands sacrifice. It

171
00:06:17,360 --> 00:06:22,240
hurts. It should only be used when

172
00:06:19,600 --> 00:06:24,319
negotiations completely break down. But

173
00:06:22,240 --> 00:06:26,240
the mere possibility of workers walking

174
00:06:24,319 --> 00:06:28,560
off the job for their needs is a strong

175
00:06:26,240 --> 00:06:30,319
signal that collectively workers are

176
00:06:28,560 --> 00:06:33,360
willing to make sacrifices in order to

177
00:06:30,319 --> 00:06:35,360
create consequences for non-compliance.

178
00:06:33,360 --> 00:06:38,080
And so labor organizing is built around

179
00:06:35,360 --> 00:06:40,240
solidarity, shared sacrifice, and using

180
00:06:38,080 --> 00:06:42,319
inconvenience as moments to actively

181
00:06:40,240 --> 00:06:45,039
educate the public so that they can

182
00:06:42,319 --> 00:06:48,080
change their behavior to disincentivize

183
00:06:45,039 --> 00:06:50,000
poor behavior by employers.

184
00:06:48,080 --> 00:06:51,520
And so we have the Foss uh movement

185
00:06:50,000 --> 00:06:53,440
which is made up of the free software

186
00:06:51,520 --> 00:06:55,759
movement which believes that maintaining

187
00:06:53,440 --> 00:06:57,919
freedoms of users to use their computers

188
00:06:55,759 --> 00:06:59,520
without restrictions is key and the

189
00:06:57,919 --> 00:07:01,280
open- source movement on the other hand

190
00:06:59,520 --> 00:07:03,680
that believes that software produced

191
00:07:01,280 --> 00:07:05,360
collaboratively has a distinct technical

192
00:07:03,680 --> 00:07:07,840
advantage to software that is produced

193
00:07:05,360 --> 00:07:09,520
in isolation. Both movements are

194
00:07:07,840 --> 00:07:11,840
collective responses to power

195
00:07:09,520 --> 00:07:13,840
imbalances. For free software, it's

196
00:07:11,840 --> 00:07:16,160
against developers and distributors of

197
00:07:13,840 --> 00:07:18,400
software who prevent users from

198
00:07:16,160 --> 00:07:20,720
modifying or redistributing code. And

199
00:07:18,400 --> 00:07:22,720
for open-source, it's against management

200
00:07:20,720 --> 00:07:25,440
who prevent developers from releasing

201
00:07:22,720 --> 00:07:27,360
code to be collaborated upon.

202
00:07:25,440 --> 00:07:29,039
The success of both of these movements

203
00:07:27,360 --> 00:07:30,960
depends on how its participants

204
00:07:29,039 --> 00:07:32,240
organize. So, let's go through the

205
00:07:30,960 --> 00:07:34,960
history of these movements with an

206
00:07:32,240 --> 00:07:36,560
organizers lens. We'll start with the

207
00:07:34,960 --> 00:07:39,840
free software movement and the emergence

208
00:07:36,560 --> 00:07:42,000
of copy left. Free software is to put

209
00:07:39,840 --> 00:07:44,240
very simply the philosophy that software

210
00:07:42,000 --> 00:07:46,400
needs to be unencumbered so that users

211
00:07:44,240 --> 00:07:49,120
may freely realize the full utility of

212
00:07:46,400 --> 00:07:51,440
their computers. They argue that users

213
00:07:49,120 --> 00:07:53,520
should be free to run, distribute, and

214
00:07:51,440 --> 00:07:55,280
modify the programs on their computer so

215
00:07:53,520 --> 00:07:57,599
that if a program does not do what they

216
00:07:55,280 --> 00:07:59,440
want, they can modify it to do so and

217
00:07:57,599 --> 00:08:00,879
share those modifications to help their

218
00:07:59,440 --> 00:08:02,639
neighbor.

219
00:08:00,879 --> 00:08:05,039
The free software movement thought of

220
00:08:02,639 --> 00:08:07,919
the need to liberate software as a moral

221
00:08:05,039 --> 00:08:09,919
imperative with proprietary software or

222
00:08:07,919 --> 00:08:12,240
closed source software impeding the

223
00:08:09,919 --> 00:08:14,879
utility of our computers and impeding

224
00:08:12,240 --> 00:08:18,000
the ability of us to help our neighbors

225
00:08:14,879 --> 00:08:19,440
realize the utility of their computers.

226
00:08:18,000 --> 00:08:22,000
The early free software movement

227
00:08:19,440 --> 00:08:24,639
produced two achievements. A technical

228
00:08:22,000 --> 00:08:27,599
project called GNU whose goal was to

229
00:08:24,639 --> 00:08:29,199
create a free software operating system.

230
00:08:27,599 --> 00:08:30,879
That project largely succeeded in

231
00:08:29,199 --> 00:08:33,360
producing free software tools for use

232
00:08:30,879 --> 00:08:36,159
with the operating system but failed to

233
00:08:33,360 --> 00:08:38,399
deliver the operating system itself. And

234
00:08:36,159 --> 00:08:40,479
the more lasting achievement was the new

235
00:08:38,399 --> 00:08:43,039
general public license which is a copy

236
00:08:40,479 --> 00:08:45,040
leftft license. If you wish to modify or

237
00:08:43,039 --> 00:08:46,320
redistribute software under a copy left

238
00:08:45,040 --> 00:08:48,080
license, you would need to distribute

239
00:08:46,320 --> 00:08:51,360
the source code including any

240
00:08:48,080 --> 00:08:53,519
modifications that you make. A copy left

241
00:08:51,360 --> 00:08:55,760
serves as a control on the production

242
00:08:53,519 --> 00:08:58,080
and distribution side of software.

243
00:08:55,760 --> 00:09:00,880
Developers can only use copy left parts

244
00:08:58,080 --> 00:09:02,800
if their product is itself copy left and

245
00:09:00,880 --> 00:09:05,279
distributors can only distribute copy

246
00:09:02,800 --> 00:09:06,720
left software if they also commit to

247
00:09:05,279 --> 00:09:09,360
distributing the source code along with

248
00:09:06,720 --> 00:09:11,360
any binaries. Indeed, if the GNU project

249
00:09:09,360 --> 00:09:14,399
in its formative years did all their

250
00:09:11,360 --> 00:09:16,399
development behind closed doors, copy

251
00:09:14,399 --> 00:09:18,240
left was a distribution tool to them. If

252
00:09:16,399 --> 00:09:19,920
a distributor broke the license, then

253
00:09:18,240 --> 00:09:23,600
GNU had the tools they needed to

254
00:09:19,920 --> 00:09:26,399
penalize that distributor.

255
00:09:23,600 --> 00:09:28,720
In his talk at Linux Confu 2017, RML

256
00:09:26,399 --> 00:09:30,240
left given coined a term app freedom,

257
00:09:28,720 --> 00:09:32,399
which is the argument that the freedom

258
00:09:30,240 --> 00:09:34,399
to make your computer do what you want

259
00:09:32,399 --> 00:09:37,040
is no longer dependent on having access

260
00:09:34,399 --> 00:09:39,200
to the source code. As the means of

261
00:09:37,040 --> 00:09:41,519
distribution of complete apps has become

262
00:09:39,200 --> 00:09:44,240
more streamlined, it's just as easy for

263
00:09:41,519 --> 00:09:46,640
an end user to swap one app for another

264
00:09:44,240 --> 00:09:48,959
one that better meets their needs and

265
00:09:46,640 --> 00:09:50,880
often for significantly lower cost than

266
00:09:48,959 --> 00:09:54,000
finding someone to modify some free

267
00:09:50,880 --> 00:09:55,920
software for you. In the world of app

268
00:09:54,000 --> 00:09:57,360
freedom, that is the freedom to choose a

269
00:09:55,920 --> 00:09:59,839
different app that meets your needs

270
00:09:57,360 --> 00:10:02,480
better, free software becomes just

271
00:09:59,839 --> 00:10:04,480
another option for a user to pick.

272
00:10:02,480 --> 00:10:06,240
Indeed, people who decide to commit to

273
00:10:04,480 --> 00:10:08,399
free software and only using free

274
00:10:06,240 --> 00:10:10,560
software get less utility from their

275
00:10:08,399 --> 00:10:12,320
computers than people for whom free

276
00:10:10,560 --> 00:10:15,519
software is just one of a sea of

277
00:10:12,320 --> 00:10:18,079
options. Take me for example. This is a

278
00:10:15,519 --> 00:10:20,880
Mac. I could be using Libre Office to

279
00:10:18,079 --> 00:10:23,600
write my talks, but I'm using Keynote.

280
00:10:20,880 --> 00:10:26,720
If I bound myself to free software, I

281
00:10:23,600 --> 00:10:28,480
could not make that choice.

282
00:10:26,720 --> 00:10:30,320
And from an organizing perspective, this

283
00:10:28,480 --> 00:10:33,120
is actually quite problematic because in

284
00:10:30,320 --> 00:10:35,760
the labor model, workers, employees, and

285
00:10:33,120 --> 00:10:38,320
consumers are made all to contend with a

286
00:10:35,760 --> 00:10:40,560
moral imperative. In the copy left

287
00:10:38,320 --> 00:10:41,920
model, only developers and distributors

288
00:10:40,560 --> 00:10:44,079
are asked to reckon with the moral

289
00:10:41,920 --> 00:10:46,880
imperative of free software. There is

290
00:10:44,079 --> 00:10:48,640
nowhere in that model where users are

291
00:10:46,880 --> 00:10:50,560
directly presented with a chance to

292
00:10:48,640 --> 00:10:52,399
educate themselves as to why they should

293
00:10:50,560 --> 00:10:55,680
care despite the movement being

294
00:10:52,399 --> 00:10:57,519
ostensibly about user freedom.

295
00:10:55,680 --> 00:10:59,600
So the Linux kernel was the first

296
00:10:57,519 --> 00:11:02,000
massive open-source project to organize

297
00:10:59,600 --> 00:11:03,600
itself online. This is before the norms

298
00:11:02,000 --> 00:11:05,680
of sharing and communicating that we're

299
00:11:03,600 --> 00:11:07,760
familiar with today were established.

300
00:11:05,680 --> 00:11:10,640
Linux wanted developers to organize

301
00:11:07,760 --> 00:11:13,200
under the same project. And copy left in

302
00:11:10,640 --> 00:11:15,120
this example is a stick. It's a stick to

303
00:11:13,200 --> 00:11:16,800
put to put the need to release source

304
00:11:15,120 --> 00:11:20,000
code in terms that businesses

305
00:11:16,800 --> 00:11:22,320
understood. You do it or you get sued.

306
00:11:20,000 --> 00:11:24,880
But it also gives you a carrot. Most

307
00:11:22,320 --> 00:11:26,160
large and longived code bases exhibit a

308
00:11:24,880 --> 00:11:28,399
phenomenon that RML called

309
00:11:26,160 --> 00:11:30,720
lithification, which is the tendency for

310
00:11:28,399 --> 00:11:33,839
software to grow in layers upon layers

311
00:11:30,720 --> 00:11:35,760
built on stable interfaces. If your code

312
00:11:33,839 --> 00:11:38,480
targets an older and more stable

313
00:11:35,760 --> 00:11:41,040
interface, then you have no incentive to

314
00:11:38,480 --> 00:11:44,720
contribute it upstream. you can keep it

315
00:11:41,040 --> 00:11:46,880
to yourself. This happened with BSD.

316
00:11:44,720 --> 00:11:49,360
Linux in those days did not have a

317
00:11:46,880 --> 00:11:51,120
stable interface and so it resisted

318
00:11:49,360 --> 00:11:53,120
lithification.

319
00:11:51,120 --> 00:11:55,279
If you wanted your kernel driver to keep

320
00:11:53,120 --> 00:11:57,200
working, you really did need to have it

321
00:11:55,279 --> 00:11:59,839
merged into mainline to ensure that it

322
00:11:57,200 --> 00:12:02,079
did not suddenly break. So copy left

323
00:11:59,839 --> 00:12:04,160
provided useful guarantees as well. If

324
00:12:02,079 --> 00:12:06,560
you submit a patch to the kernel, you

325
00:12:04,160 --> 00:12:08,639
are protected against someone using your

326
00:12:06,560 --> 00:12:10,880
patch that you've contributed upstream

327
00:12:08,639 --> 00:12:13,440
in an extractive fashion. If they

328
00:12:10,880 --> 00:12:16,399
produce new software using your patch,

329
00:12:13,440 --> 00:12:17,920
they have to release it as copy.

330
00:12:16,399 --> 00:12:19,760
So the copy left side of the free

331
00:12:17,920 --> 00:12:22,000
software movement made a tool that was

332
00:12:19,760 --> 00:12:23,920
successful in compelling much of the

333
00:12:22,000 --> 00:12:26,160
development and distribution side of the

334
00:12:23,920 --> 00:12:28,320
industry to release software with its

335
00:12:26,160 --> 00:12:30,320
source. And the consequence of focusing

336
00:12:28,320 --> 00:12:32,240
on the supply side of the economy was

337
00:12:30,320 --> 00:12:35,200
that there was no opportunity to educate

338
00:12:32,240 --> 00:12:37,360
the user base. Despite the user focused

339
00:12:35,200 --> 00:12:40,959
rhetoric, free software success was a

340
00:12:37,360 --> 00:12:43,120
labor model for developers.

341
00:12:40,959 --> 00:12:44,800
And what's fascinating about Linux is

342
00:12:43,120 --> 00:12:48,160
that Linux Torvalds was a second

343
00:12:44,800 --> 00:12:50,399
generation of copy left users.

344
00:12:48,160 --> 00:12:52,959
He emerged independently of the GNU

345
00:12:50,399 --> 00:12:56,880
project and instead saw that copy left

346
00:12:52,959 --> 00:12:58,800
was a good developer focused tool to

347
00:12:56,880 --> 00:13:01,120
ensure that all contributors fairly

348
00:12:58,800 --> 00:13:04,079
benefited from each other's work. Now

349
00:13:01,120 --> 00:13:05,839
Linux emerged in 1991.

350
00:13:04,079 --> 00:13:07,839
But that year another notable thing

351
00:13:05,839 --> 00:13:09,680
happened. The GNU project had got to the

352
00:13:07,839 --> 00:13:11,680
point where they released enough user

353
00:13:09,680 --> 00:13:13,920
space programs but they were ready to

354
00:13:11,680 --> 00:13:16,399
release a C library which every Unix

355
00:13:13,920 --> 00:13:18,480
system needs. But there were already

356
00:13:16,399 --> 00:13:20,399
many C libraries out there and there was

357
00:13:18,480 --> 00:13:22,639
no real advantage to making it fully

358
00:13:20,399 --> 00:13:25,920
copy left. And so the result was a thing

359
00:13:22,639 --> 00:13:28,240
called the library GPL or the LGPL which

360
00:13:25,920 --> 00:13:30,000
is like the GPL in that if you modified

361
00:13:28,240 --> 00:13:32,720
covered code you would need to release

362
00:13:30,000 --> 00:13:34,160
the source code. But unlike the GPL you

363
00:13:32,720 --> 00:13:36,720
wouldn't have to release the source code

364
00:13:34,160 --> 00:13:38,880
of your own software that was otherwise

365
00:13:36,720 --> 00:13:40,639
closed source. And this makes a lot of

366
00:13:38,880 --> 00:13:44,480
sense for libraries that provide no

367
00:13:40,639 --> 00:13:46,959
competitive advantage to the author.

368
00:13:44,480 --> 00:13:48,720
In 1998, a group of people decided that

369
00:13:46,959 --> 00:13:51,440
we should refer to free software as

370
00:13:48,720 --> 00:13:52,880
opensource as a way to emphasize the

371
00:13:51,440 --> 00:13:55,440
technical advantage that source

372
00:13:52,880 --> 00:13:58,240
availability offered developers and

373
00:13:55,440 --> 00:14:00,880
deemphasize the moral imperative. So the

374
00:13:58,240 --> 00:14:03,120
free software foundation which remember

375
00:14:00,880 --> 00:14:05,440
successfully created a model to compel

376
00:14:03,120 --> 00:14:07,760
developers to produce free or now

377
00:14:05,440 --> 00:14:10,160
open-source software but had no real

378
00:14:07,760 --> 00:14:12,000
success in compelling user behavior

379
00:14:10,160 --> 00:14:13,600
decided that the moral imperative and

380
00:14:12,000 --> 00:14:15,519
user freedom should become their main

381
00:14:13,600 --> 00:14:17,440
focus

382
00:14:15,519 --> 00:14:19,360
and relatively soon after the rise of

383
00:14:17,440 --> 00:14:20,959
open source as an alternative

384
00:14:19,360 --> 00:14:23,279
justification for making source code

385
00:14:20,959 --> 00:14:24,880
available the free software foundation

386
00:14:23,279 --> 00:14:28,160
announced that they were renaming the

387
00:14:24,880 --> 00:14:30,480
library GPL to the lesser GPL. L to make

388
00:14:28,160 --> 00:14:33,120
it clear that not all libraries should

389
00:14:30,480 --> 00:14:34,880
be licensed under the LGPL.

390
00:14:33,120 --> 00:14:36,399
If you as an author want to encourage

391
00:14:34,880 --> 00:14:39,040
more developers to release free

392
00:14:36,399 --> 00:14:41,120
software, you should license your wares

393
00:14:39,040 --> 00:14:43,440
under the GPL so that they can't choose

394
00:14:41,120 --> 00:14:45,839
to use your software unless they use the

395
00:14:43,440 --> 00:14:47,920
GPL. It was a very highly nuanced read.

396
00:14:45,839 --> 00:14:51,040
It spelled out why you should actively

397
00:14:47,920 --> 00:14:52,959
think about your license choices,

398
00:14:51,040 --> 00:14:54,720
which was put under a headline, why you

399
00:14:52,959 --> 00:14:56,639
shouldn't use the library GPL for your

400
00:14:54,720 --> 00:14:58,320
next library.

401
00:14:56,639 --> 00:15:00,720
And the result of this headline was

402
00:14:58,320 --> 00:15:02,800
exactly what you would expect. Library

403
00:15:00,720 --> 00:15:05,360
developers stopped using the library

404
00:15:02,800 --> 00:15:07,040
GPL. However, the result was not the

405
00:15:05,360 --> 00:15:10,560
mass migration to copy left that the

406
00:15:07,040 --> 00:15:12,160
free software foundation recommended.

407
00:15:10,560 --> 00:15:14,399
Because following the rise of open

408
00:15:12,160 --> 00:15:16,639
source a few years earlier, we started

409
00:15:14,399 --> 00:15:18,880
to see a new breed of open source.

410
00:15:16,639 --> 00:15:20,639
Developers at product companies realized

411
00:15:18,880 --> 00:15:22,639
that they could take advantage of the

412
00:15:20,639 --> 00:15:24,800
technical benefits of open source and

413
00:15:22,639 --> 00:15:27,519
would release parts of their web apps

414
00:15:24,800 --> 00:15:29,440
under open- source licenses. This is how

415
00:15:27,519 --> 00:15:31,680
we got Ruby on Rails. It was a byproduct

416
00:15:29,440 --> 00:15:33,839
of a task tracking tool. This is how we

417
00:15:31,680 --> 00:15:36,160
got Django, which was the back end for a

418
00:15:33,839 --> 00:15:38,160
newspaper.

419
00:15:36,160 --> 00:15:39,920
Product companies where they knew a view

420
00:15:38,160 --> 00:15:42,079
of open-source as technically

421
00:15:39,920 --> 00:15:44,800
advantageous realized the trick was not

422
00:15:42,079 --> 00:15:46,399
to open source everything but to open

423
00:15:44,800 --> 00:15:49,120
source the parts of their application

424
00:15:46,399 --> 00:15:50,959
that offered them no differentiation.

425
00:15:49,120 --> 00:15:53,759
If every web app needs to talk to a

426
00:15:50,959 --> 00:15:56,079
database and produce forms then you

427
00:15:53,759 --> 00:15:57,839
commoditize that code leaving product

428
00:15:56,079 --> 00:15:59,519
developers to focus on what their

429
00:15:57,839 --> 00:16:01,360
business perceived as its competitive

430
00:15:59,519 --> 00:16:03,759
advantage.

431
00:16:01,360 --> 00:16:05,920
In this business model, strong copy left

432
00:16:03,759 --> 00:16:07,839
does not make sense as an option. And

433
00:16:05,920 --> 00:16:10,079
that also assumes that developers are

434
00:16:07,839 --> 00:16:11,759
even in a position to make the decision

435
00:16:10,079 --> 00:16:13,680
to release as copy left in the first

436
00:16:11,759 --> 00:16:15,440
place.

437
00:16:13,680 --> 00:16:17,600
And so despite focusing their rhetoric

438
00:16:15,440 --> 00:16:20,079
on user freedoms, free software's core

439
00:16:17,600 --> 00:16:22,000
organizing model that is copy left

440
00:16:20,079 --> 00:16:24,079
binding developers without that

441
00:16:22,000 --> 00:16:26,560
mechanism to educate users, it saw a

442
00:16:24,079 --> 00:16:28,399
real failure. These developers of of

443
00:16:26,560 --> 00:16:30,639
Django and Rails and things like that

444
00:16:28,399 --> 00:16:32,399
were users of open-source and they were

445
00:16:30,639 --> 00:16:35,040
embarking on their first large

446
00:16:32,399 --> 00:16:37,440
open-source projects. These developers

447
00:16:35,040 --> 00:16:39,279
were not predisposed to free software

448
00:16:37,440 --> 00:16:40,880
and free software had thrown away its

449
00:16:39,279 --> 00:16:42,720
tools to welcome people into the

450
00:16:40,880 --> 00:16:44,959
movement.

451
00:16:42,720 --> 00:16:47,120
Weak copy left like the LGPL was the

452
00:16:44,959 --> 00:16:49,360
exact copy left model that translated to

453
00:16:47,120 --> 00:16:52,320
these new developers views of the world.

454
00:16:49,360 --> 00:16:54,079
But with the official line being this,

455
00:16:52,320 --> 00:16:56,959
then it's no surprise that people looked

456
00:16:54,079 --> 00:16:58,720
elsewhere. Django, Rails, and basically

457
00:16:56,959 --> 00:17:01,440
every major library to emerge around

458
00:16:58,720 --> 00:17:04,000
that time used licenses that bind no

459
00:17:01,440 --> 00:17:06,160
behaviors on the lency other than giving

460
00:17:04,000 --> 00:17:07,600
attribution to the original developer.

461
00:17:06,160 --> 00:17:10,559
These are the so-called permissive

462
00:17:07,600 --> 00:17:13,120
licenses. And binding on licenses is the

463
00:17:10,559 --> 00:17:14,799
stick in Linux's organizing model. And

464
00:17:13,120 --> 00:17:16,559
these projects decided from the

465
00:17:14,799 --> 00:17:18,799
information available to them that it

466
00:17:16,559 --> 00:17:20,720
wasn't worth it.

467
00:17:18,799 --> 00:17:23,360
And as for the carrot, well, the the

468
00:17:20,720 --> 00:17:24,880
platform effects of GitHub replicated

469
00:17:23,360 --> 00:17:26,480
for small projects at least the

470
00:17:24,880 --> 00:17:28,880
collaborative effects of the Linux

471
00:17:26,480 --> 00:17:30,559
kernel. If you host on GitHub, when you

472
00:17:28,880 --> 00:17:32,880
fork a library, it's really easy to stay

473
00:17:30,559 --> 00:17:34,559
in sync with mainline. It's trivial to

474
00:17:32,880 --> 00:17:36,960
merge from external contributors who

475
00:17:34,559 --> 00:17:38,960
make bug fixes. There are mildly

476
00:17:36,960 --> 00:17:41,679
punitive behaviors that make it slightly

477
00:17:38,960 --> 00:17:43,520
painful to make your code closed source.

478
00:17:41,679 --> 00:17:45,600
GitHub charged a monthly fee to take

479
00:17:43,520 --> 00:17:47,440
things private.

480
00:17:45,600 --> 00:17:49,840
But the consequence of not participating

481
00:17:47,440 --> 00:17:52,559
in GitHub's commons was inconvenience.

482
00:17:49,840 --> 00:17:54,720
You either spent time or money. And this

483
00:17:52,559 --> 00:17:57,520
is generally speaking enough to keep

484
00:17:54,720 --> 00:17:59,679
most people in line. It does not stop

485
00:17:57,520 --> 00:18:03,120
people with means from maintaining their

486
00:17:59,679 --> 00:18:04,720
own closed source fork.

487
00:18:03,120 --> 00:18:06,400
So by the end of the first decade of

488
00:18:04,720 --> 00:18:08,000
this millennium, we really had the basis

489
00:18:06,400 --> 00:18:09,840
for what we think of as open source

490
00:18:08,000 --> 00:18:11,919
today.

491
00:18:09,840 --> 00:18:13,760
The things that generate value or

492
00:18:11,919 --> 00:18:16,000
represent competitive advantage should

493
00:18:13,760 --> 00:18:18,160
be kept proprietary so that they can be

494
00:18:16,000 --> 00:18:20,080
a source of revenue that others cannot

495
00:18:18,160 --> 00:18:21,919
exploit.

496
00:18:20,080 --> 00:18:23,760
Code that does not differentiate you

497
00:18:21,919 --> 00:18:25,679
from your competitor should be open

498
00:18:23,760 --> 00:18:27,760
sourced so that you can immortalize the

499
00:18:25,679 --> 00:18:29,919
maintenance costs across your entire

500
00:18:27,760 --> 00:18:32,000
industry and that you should use

501
00:18:29,919 --> 00:18:34,480
permissive licensing because copy left

502
00:18:32,000 --> 00:18:36,960
threatened your revenue and weak copy

503
00:18:34,480 --> 00:18:39,600
was somehow inappropriate.

504
00:18:36,960 --> 00:18:42,400
And so open- source code is not revenue

505
00:18:39,600 --> 00:18:45,200
generating code. It is a side effect

506
00:18:42,400 --> 00:18:47,280
with less need to be protected.

507
00:18:45,200 --> 00:18:49,760
And so the model of open source that won

508
00:18:47,280 --> 00:18:51,679
was simple. A commons was built out of

509
00:18:49,760 --> 00:18:54,720
the code that did not constitute a

510
00:18:51,679 --> 00:18:56,640
competitive advantage for its creators.

511
00:18:54,720 --> 00:18:58,240
The code you develop uh you derive

512
00:18:56,640 --> 00:19:00,880
economic benefit from would be

513
00:18:58,240 --> 00:19:03,840
proprietary. Anything that's open source

514
00:19:00,880 --> 00:19:05,840
is a side effect of production. And but

515
00:19:03,840 --> 00:19:08,000
even though it's a side effect, somebody

516
00:19:05,840 --> 00:19:10,880
still has to produce the code. And so

517
00:19:08,000 --> 00:19:12,799
that creates perverse incentives, the

518
00:19:10,880 --> 00:19:14,559
use of permissive licensing means that

519
00:19:12,799 --> 00:19:17,679
consumers are released from all

520
00:19:14,559 --> 00:19:20,400
obligation around the code. And so by

521
00:19:17,679 --> 00:19:22,720
definition, any consumer of a piece of

522
00:19:20,400 --> 00:19:25,200
open-source infrastructure is able to

523
00:19:22,720 --> 00:19:27,919
gain more economic advantage than its

524
00:19:25,200 --> 00:19:31,880
producer because they did not have to do

525
00:19:27,919 --> 00:19:31,880
the work to create it.

526
00:19:31,919 --> 00:19:36,080
And so when we emerged in the era where

527
00:19:33,760 --> 00:19:37,679
open source had won, we had a free

528
00:19:36,080 --> 00:19:40,080
software movement that was focused on

529
00:19:37,679 --> 00:19:43,840
users that created no compelling reason

530
00:19:40,080 --> 00:19:45,200
for for its users to participate.

531
00:19:43,840 --> 00:19:47,760
But it did create a burden for

532
00:19:45,200 --> 00:19:49,760
developers. And we also had an open-

533
00:19:47,760 --> 00:19:52,160
source movement where businesses that

534
00:19:49,760 --> 00:19:54,480
didn't write open-source infrastructure

535
00:19:52,160 --> 00:19:58,200
were at an economic advantage to the

536
00:19:54,480 --> 00:19:58,200
businesses that did.

537
00:20:00,640 --> 00:20:06,320
just let you get that photo.

538
00:20:03,120 --> 00:20:08,240
So that brings us to 2012 which is a

539
00:20:06,320 --> 00:20:10,240
notable year because a bunch of

540
00:20:08,240 --> 00:20:12,799
companies successfully sought venture

541
00:20:10,240 --> 00:20:15,360
funding to produce open-source tools as

542
00:20:12,799 --> 00:20:17,440
a product effort. This was a direction

543
00:20:15,360 --> 00:20:18,960
that ignored the successes in open

544
00:20:17,440 --> 00:20:20,640
source that had emerged the previous

545
00:20:18,960 --> 00:20:22,640
year.

546
00:20:20,640 --> 00:20:24,640
They didn't have the product revenue

547
00:20:22,640 --> 00:20:26,799
streams of the open- source as a side

548
00:20:24,640 --> 00:20:28,799
effect movement and they did this

549
00:20:26,799 --> 00:20:31,200
without the tools to organize against

550
00:20:28,799 --> 00:20:33,120
extractive behavior by organizations

551
00:20:31,200 --> 00:20:35,919
that are not yet motivated to play by

552
00:20:33,120 --> 00:20:37,600
the rules. The business model was to

553
00:20:35,919 --> 00:20:40,000
release open-source code under

554
00:20:37,600 --> 00:20:42,720
permissive licenses as their core

555
00:20:40,000 --> 00:20:45,280
activity and then attempt to build value

556
00:20:42,720 --> 00:20:46,960
added services on top of that.

557
00:20:45,280 --> 00:20:52,080
What those services were actually going

558
00:20:46,960 --> 00:20:54,400
to be was ambiguous at the time.

559
00:20:52,080 --> 00:20:56,080
Hashi Corp started trading publicly in

560
00:20:54,400 --> 00:20:58,720
2022.

561
00:20:56,080 --> 00:21:00,799
That is, they had reached the limit of

562
00:20:58,720 --> 00:21:03,360
what they could do with venture capital

563
00:21:00,799 --> 00:21:04,640
and instead sought market validation for

564
00:21:03,360 --> 00:21:06,480
their products and their business

565
00:21:04,640 --> 00:21:08,559
approach.

566
00:21:06,480 --> 00:21:10,400
And within a year, they decided that

567
00:21:08,559 --> 00:21:12,640
they could no longer meet their owner's

568
00:21:10,400 --> 00:21:15,440
metrics of success by releasing

569
00:21:12,640 --> 00:21:17,360
permissively licensed open-source tools.

570
00:21:15,440 --> 00:21:19,360
They relicensed all of their products

571
00:21:17,360 --> 00:21:22,640
under a source available license that

572
00:21:19,360 --> 00:21:25,360
locked out competitors. The industry and

573
00:21:22,640 --> 00:21:27,919
then the market rejected their choice.

574
00:21:25,360 --> 00:21:30,880
Their stock price precipitously dropped

575
00:21:27,919 --> 00:21:33,360
and within six months they accepted a

576
00:21:30,880 --> 00:21:37,159
buyout offer from IBM for less than

577
00:21:33,360 --> 00:21:37,159
their IPO value.

578
00:21:37,679 --> 00:21:42,480
And sure, businesses get acquired all

579
00:21:39,760 --> 00:21:44,720
the time. But in this case, this created

580
00:21:42,480 --> 00:21:46,320
n and a half years of noise, saying that

581
00:21:44,720 --> 00:21:48,640
they could go it alone and make this

582
00:21:46,320 --> 00:21:50,880
work, producing open-source software as

583
00:21:48,640 --> 00:21:53,840
a business activity that they projected

584
00:21:50,880 --> 00:21:55,760
as being sustainable before the investor

585
00:21:53,840 --> 00:21:58,640
demands came home to roost. And there

586
00:21:55,760 --> 00:22:01,520
are, of course, consequences to

587
00:21:58,640 --> 00:22:03,919
accepting strings attached money. They

588
00:22:01,520 --> 00:22:06,640
couldn't make it work. And as a result,

589
00:22:03,919 --> 00:22:08,480
we've lost 10 years of success. Uh we've

590
00:22:06,640 --> 00:22:11,360
lost 10 years seeing scores of

591
00:22:08,480 --> 00:22:13,919
phosphocused startups founded by a meme

592
00:22:11,360 --> 00:22:15,679
on the myth that merely receiving money

593
00:22:13,919 --> 00:22:18,480
is all it takes to make a project

594
00:22:15,679 --> 00:22:19,840
sustainable.

595
00:22:18,480 --> 00:22:21,840
And this is what I mean when I say that

596
00:22:19,840 --> 00:22:24,240
we're in a golden age. It's now clear

597
00:22:21,840 --> 00:22:28,080
that we can't just find money by saying

598
00:22:24,240 --> 00:22:30,240
that Foss is a product in and of itself.

599
00:22:28,080 --> 00:22:31,919
We have to rebuild with new assumptions.

600
00:22:30,240 --> 00:22:34,080
And it's still not completely clear to

601
00:22:31,919 --> 00:22:36,480
me that there even could be a successful

602
00:22:34,080 --> 00:22:39,280
business model built around open-source

603
00:22:36,480 --> 00:22:41,600
as the product. Because as we've seen

604
00:22:39,280 --> 00:22:43,360
over the course of decades, we as a

605
00:22:41,600 --> 00:22:45,200
movement have made choices to reject

606
00:22:43,360 --> 00:22:47,679
economic tools that give us a fighting

607
00:22:45,200 --> 00:22:49,520
chance of being sustainable. Indeed,

608
00:22:47,679 --> 00:22:52,720
we've seen that we've made a series of

609
00:22:49,520 --> 00:22:54,799
choices that every time results in

610
00:22:52,720 --> 00:22:56,880
producers of open source having to

611
00:22:54,799 --> 00:22:59,440
shoulder a burden that is greater than

612
00:22:56,880 --> 00:23:01,440
that of our consumers.

613
00:22:59,440 --> 00:23:03,600
If our current model of open source was

614
00:23:01,440 --> 00:23:05,360
a business, we would be spending more

615
00:23:03,600 --> 00:23:08,080
than we're taking in. We would be

616
00:23:05,360 --> 00:23:10,320
trading while insolvent.

617
00:23:08,080 --> 00:23:12,080
And because we choose to not bind on our

618
00:23:10,320 --> 00:23:14,559
users, we create a situation where our

619
00:23:12,080 --> 00:23:16,559
users become our competitors.

620
00:23:14,559 --> 00:23:18,320
Licensing without consequences makes us

621
00:23:16,559 --> 00:23:20,960
complicit in creating competitors with

622
00:23:18,320 --> 00:23:22,480
no external baggage. They don't have to

623
00:23:20,960 --> 00:23:24,480
pay to maintain the software that

624
00:23:22,480 --> 00:23:26,159
they're building upon. And so they can

625
00:23:24,480 --> 00:23:28,240
focus entirely on building revenue

626
00:23:26,159 --> 00:23:30,080
streams. You know, businesses, they do

627
00:23:28,240 --> 00:23:31,919
what they're allowed to do. And so, if

628
00:23:30,080 --> 00:23:33,600
we don't penalize behavior, they're

629
00:23:31,919 --> 00:23:35,600
going to do whatever is cheapest and

630
00:23:33,600 --> 00:23:38,159
makes them the most money. That's what

631
00:23:35,600 --> 00:23:39,760
businesses do. And this is a problem

632
00:23:38,159 --> 00:23:41,760
because as a movement, whenever we've

633
00:23:39,760 --> 00:23:43,760
been presented with tools to organize

634
00:23:41,760 --> 00:23:46,000
and ensure that extractive behavior is

635
00:23:43,760 --> 00:23:47,360
not possible, we've chosen the more

636
00:23:46,000 --> 00:23:49,120
convenient option without any

637
00:23:47,360 --> 00:23:51,679
consideration of the consequences of

638
00:23:49,120 --> 00:23:53,919
that choice. And the result is that open

639
00:23:51,679 --> 00:23:56,240
source has no inherent reason to be

640
00:23:53,919 --> 00:23:58,240
sustainable. And at least that's true

641
00:23:56,240 --> 00:24:00,880
for any independent project that hasn't

642
00:23:58,240 --> 00:24:03,120
chosen to organize.

643
00:24:00,880 --> 00:24:05,280
Which leads me to talk about Kubernetes,

644
00:24:03,120 --> 00:24:06,480
which is probably the biggest success in

645
00:24:05,280 --> 00:24:09,679
corporate level collaborative

646
00:24:06,480 --> 00:24:12,000
open-source since Linux. Unlike Linux,

647
00:24:09,679 --> 00:24:13,600
Kubernetes is permissively licensed,

648
00:24:12,000 --> 00:24:15,279
just like most open source tools that

649
00:24:13,600 --> 00:24:18,240
are produced today. So, it doesn't have

650
00:24:15,279 --> 00:24:20,960
the organizing benefits of copy. But

651
00:24:18,240 --> 00:24:23,440
Kubernetes as a project still organizes

652
00:24:20,960 --> 00:24:25,120
to ensure that businesses play fairly.

653
00:24:23,440 --> 00:24:27,360
It just does it in a different way that

654
00:24:25,120 --> 00:24:29,120
businesses understand.

655
00:24:27,360 --> 00:24:31,200
Kubernetes is maintained by the

656
00:24:29,120 --> 00:24:33,360
cloudnative computing foundation which

657
00:24:31,200 --> 00:24:36,159
is a trade organization. This is

658
00:24:33,360 --> 00:24:38,400
otherwise known as a 501c6 organization

659
00:24:36,159 --> 00:24:41,279
which is a sort of nonprofit under US

660
00:24:38,400 --> 00:24:43,520
tax law that allows member corporations

661
00:24:41,279 --> 00:24:45,600
to work together to do work that

662
00:24:43,520 --> 00:24:48,799
improves the conditions in which all of

663
00:24:45,600 --> 00:24:51,279
them trade. Now that sounds a lot like

664
00:24:48,799 --> 00:24:52,640
organizing.

665
00:24:51,279 --> 00:24:55,120
The main thing that the cloudnative

666
00:24:52,640 --> 00:24:57,039
computing foundation sells are seats on

667
00:24:55,120 --> 00:24:58,799
their governing board which appoints a

668
00:24:57,039 --> 00:25:00,880
voting majority of the foundation's

669
00:24:58,799 --> 00:25:02,400
technical oversight committee. And this

670
00:25:00,880 --> 00:25:04,559
committee sets the technical direction

671
00:25:02,400 --> 00:25:06,559
for its projects including Kubernetes.

672
00:25:04,559 --> 00:25:08,320
And those appointed members ensure that

673
00:25:06,559 --> 00:25:10,559
the decisions are made in the best

674
00:25:08,320 --> 00:25:13,039
interests of the corporations that

675
00:25:10,559 --> 00:25:15,120
appointed them. If you don't pay to join

676
00:25:13,039 --> 00:25:16,799
the foundation, you don't get a say in

677
00:25:15,120 --> 00:25:19,679
the technical direction of the project.

678
00:25:16,799 --> 00:25:23,440
and even worse, your competitors will be

679
00:25:19,679 --> 00:25:25,919
able to make decisions in your absence.

680
00:25:23,440 --> 00:25:27,679
Kubernetes has organized effectively.

681
00:25:25,919 --> 00:25:30,159
This governance model is sustainable

682
00:25:27,679 --> 00:25:32,320
because the stakes of not participating

683
00:25:30,159 --> 00:25:34,240
for each individual member organization

684
00:25:32,320 --> 00:25:36,559
is made in terms that these sponsors

685
00:25:34,240 --> 00:25:38,320
understand and the proceeds of

686
00:25:36,559 --> 00:25:40,480
sponsorship pay for things beyond just

687
00:25:38,320 --> 00:25:43,200
governance. Some of the maintenance is

688
00:25:40,480 --> 00:25:45,120
also paid for. The CloudNative Computing

689
00:25:43,200 --> 00:25:47,919
Foundation is a properly successful way

690
00:25:45,120 --> 00:25:50,000
for corporations to organize around a

691
00:25:47,919 --> 00:25:52,240
permissively licensed project with

692
00:25:50,000 --> 00:25:54,480
stakes far greater than we as small

693
00:25:52,240 --> 00:25:57,200
project maintainers are willing to even

694
00:25:54,480 --> 00:25:59,200
consider imposing.

695
00:25:57,200 --> 00:26:01,679
And that's the thing like big tech

696
00:25:59,200 --> 00:26:04,240
corporations that are extractive when it

697
00:26:01,679 --> 00:26:07,120
comes to small and independent projects

698
00:26:04,240 --> 00:26:08,640
organize when it's their own projects.

699
00:26:07,120 --> 00:26:11,120
There are consequences for not

700
00:26:08,640 --> 00:26:14,960
participating. the corporations that

701
00:26:11,120 --> 00:26:16,240
should participate generally do.

702
00:26:14,960 --> 00:26:17,840
And when you look at where corporate

703
00:26:16,240 --> 00:26:19,760
money and support for maintenance is

704
00:26:17,840 --> 00:26:21,760
going, it's not going to the projects

705
00:26:19,760 --> 00:26:23,440
that fail to organize. It's going to the

706
00:26:21,760 --> 00:26:25,360
ones where there is risk in not

707
00:26:23,440 --> 00:26:27,039
participating.

708
00:26:25,360 --> 00:26:29,840
Independent projects with permissive

709
00:26:27,039 --> 00:26:31,600
licensing generates no risk. And if

710
00:26:29,840 --> 00:26:34,480
there's no risk to extractive behavior,

711
00:26:31,600 --> 00:26:36,480
they're going to keep on doing it. And

712
00:26:34,480 --> 00:26:38,320
that leads us to the point we're at now.

713
00:26:36,480 --> 00:26:40,159
We have corporates who understand and

714
00:26:38,320 --> 00:26:41,760
have built structures around preventing

715
00:26:40,159 --> 00:26:43,840
extractive behavior for the projects

716
00:26:41,760 --> 00:26:46,080
that they contribute to. And because we

717
00:26:43,840 --> 00:26:48,880
don't stop them, many of them continue

718
00:26:46,080 --> 00:26:51,840
to apply extractive behavior.

719
00:26:48,880 --> 00:26:53,360
So what do we do about it?

720
00:26:51,840 --> 00:26:54,880
Firstly, we need to accept that the

721
00:26:53,360 --> 00:26:57,279
tools that formed the basis of

722
00:26:54,880 --> 00:26:59,919
organizing back when free software was

723
00:26:57,279 --> 00:27:02,240
new have now been completely defanged.

724
00:26:59,919 --> 00:27:05,360
The primary threat of copy was fear of

725
00:27:02,240 --> 00:27:07,360
being sued by bigname copyright holders.

726
00:27:05,360 --> 00:27:09,600
These days, the corporate stakeholders

727
00:27:07,360 --> 00:27:12,960
in Linux use the same trade association

728
00:27:09,600 --> 00:27:14,799
model as Kubernetes does.

729
00:27:12,960 --> 00:27:17,279
Corporate contributors now have at their

730
00:27:14,799 --> 00:27:20,159
disposal tools that are more useful to

731
00:27:17,279 --> 00:27:22,000
them than the fear of copy suits. And

732
00:27:20,159 --> 00:27:23,520
because we as a community of independent

733
00:27:22,000 --> 00:27:26,000
developers have stopped caring about

734
00:27:23,520 --> 00:27:28,240
copy theft, corporations aren't worried

735
00:27:26,000 --> 00:27:30,880
about being sued by us and routinely

736
00:27:28,240 --> 00:27:32,720
breach copy licenses.

737
00:27:30,880 --> 00:27:34,960
The single most important test case in

738
00:27:32,720 --> 00:27:37,279
copy left history is taking place in

739
00:27:34,960 --> 00:27:39,760
California at the moment. Vizio is a

740
00:27:37,279 --> 00:27:41,679
manufacturer that produces TVs that run

741
00:27:39,760 --> 00:27:44,559
Linux and they failed to produce source

742
00:27:41,679 --> 00:27:46,240
code when requested by users. The

743
00:27:44,559 --> 00:27:48,400
developers who bind distributors to the

744
00:27:46,240 --> 00:27:51,600
GPL have established standing in legal

745
00:27:48,400 --> 00:27:53,360
circles to sue to enforce copy theft,

746
00:27:51,600 --> 00:27:56,240
but it's not currently clear that third

747
00:27:53,360 --> 00:27:58,320
parties have that standing. SFC versus

748
00:27:56,240 --> 00:28:00,880
Vizio is a test case to establish

749
00:27:58,320 --> 00:28:03,360
standing for third parties. If it

750
00:28:00,880 --> 00:28:05,679
succeeds, then owners of devices with

751
00:28:03,360 --> 00:28:08,320
GPL software can demand that software

752
00:28:05,679 --> 00:28:10,399
like the licenses says say they can from

753
00:28:08,320 --> 00:28:12,640
the manufacturer and they can make a

754
00:28:10,399 --> 00:28:15,279
credible threat of legal action if the

755
00:28:12,640 --> 00:28:16,960
manufacturer does not comply.

756
00:28:15,279 --> 00:28:19,279
And from a right to repair perspective,

757
00:28:16,960 --> 00:28:21,200
this would be huge. We would be able to

758
00:28:19,279 --> 00:28:23,200
inspect and modify the software that our

759
00:28:21,200 --> 00:28:25,679
devices run on. But from the perspective

760
00:28:23,200 --> 00:28:28,000
of making open source more sustainable,

761
00:28:25,679 --> 00:28:30,080
this achieves basically nothing. It

762
00:28:28,000 --> 00:28:32,080
doesn't make a new funer of open- source

763
00:28:30,080 --> 00:28:33,600
projects. It doesn't contribute

764
00:28:32,080 --> 00:28:35,919
development time that would ease the

765
00:28:33,600 --> 00:28:38,240
maintenance burden on Linux. It just

766
00:28:35,919 --> 00:28:41,919
compels a TV manufacturer to do a bit

767
00:28:38,240 --> 00:28:44,080
more inconvenient work.

768
00:28:41,919 --> 00:28:47,279
And so while enforcement is entirely

769
00:28:44,080 --> 00:28:48,559
necessary and I donate to conservies so

770
00:28:47,279 --> 00:28:51,279
that they can keep doing this

771
00:28:48,559 --> 00:28:54,399
enforcement work, the fruits of copy

772
00:28:51,279 --> 00:28:56,640
enforcement are increasingly myopic. So

773
00:28:54,399 --> 00:28:59,440
let me revise my statement. Licensing as

774
00:28:56,640 --> 00:29:01,840
an organizing tool is dead. Licensing is

775
00:28:59,440 --> 00:29:04,480
a way to get what you want through the

776
00:29:01,840 --> 00:29:07,919
courts. So unless you have the means to

777
00:29:04,480 --> 00:29:09,760
sue every non-compliant actor, licensing

778
00:29:07,919 --> 00:29:12,559
will not bring about comprehensive

779
00:29:09,760 --> 00:29:14,399
behavioral change. Any movement that

780
00:29:12,559 --> 00:29:15,919
founds itself on license compliance is

781
00:29:14,399 --> 00:29:18,320
going to be doomed to failure because we

782
00:29:15,919 --> 00:29:20,880
as a movement gave up licensing as a

783
00:29:18,320 --> 00:29:23,440
socially acceptable way for us to compel

784
00:29:20,880 --> 00:29:25,600
non-extractive behavior. Extractive

785
00:29:23,440 --> 00:29:27,840
organizations will not take organ uh

786
00:29:25,600 --> 00:29:30,799
licensing seriously.

787
00:29:27,840 --> 00:29:32,559
So what remains for us? The strike

788
00:29:30,799 --> 00:29:34,799
action model asks us to consider what

789
00:29:32,559 --> 00:29:36,399
we're willing to sacrifice in order to

790
00:29:34,799 --> 00:29:39,840
show that we're serious about demands

791
00:29:36,399 --> 00:29:41,440
that we have. Um what could work as a

792
00:29:39,840 --> 00:29:43,600
movement? We've shown we're unwilling to

793
00:29:41,440 --> 00:29:45,440
impose consequences in favor of growth

794
00:29:43,600 --> 00:29:47,200
as a movement and we work in

795
00:29:45,440 --> 00:29:48,880
environments where there is risk in

796
00:29:47,200 --> 00:29:50,960
talking about impeding the revenue

797
00:29:48,880 --> 00:29:54,320
streams of our employers or of our

798
00:29:50,960 --> 00:29:56,240
sponsors. So let's look to labor.

799
00:29:54,320 --> 00:29:57,919
Starting with solidarity and community

800
00:29:56,240 --> 00:30:00,799
building would go a long way to making

801
00:29:57,919 --> 00:30:03,120
safe spaces spaces for us to even start

802
00:30:00,799 --> 00:30:04,480
talking about this. And we need to be

803
00:30:03,120 --> 00:30:06,880
willing to talk about what sort of

804
00:30:04,480 --> 00:30:08,640
consequences we can impose because not

805
00:30:06,880 --> 00:30:11,279
talking about consequences is what got

806
00:30:08,640 --> 00:30:13,120
us here in the first place. As a

807
00:30:11,279 --> 00:30:15,600
movement, we're unwilling to impose

808
00:30:13,120 --> 00:30:18,000
consequences for extracting behavior.

809
00:30:15,600 --> 00:30:20,000
But we need to work together. If just

810
00:30:18,000 --> 00:30:21,919
one project or one community support

811
00:30:20,000 --> 00:30:24,399
provider continues to fail to impose

812
00:30:21,919 --> 00:30:26,720
consequences for extractive behavior,

813
00:30:24,399 --> 00:30:29,440
then that extractive behavior is being

814
00:30:26,720 --> 00:30:31,440
allowed to continue.

815
00:30:29,440 --> 00:30:33,679
And we've seen that the industry and the

816
00:30:31,440 --> 00:30:36,240
market reacts really poorly to shutting

817
00:30:33,679 --> 00:30:38,480
out bad actors with no remedy. If we

818
00:30:36,240 --> 00:30:41,279
create a way to encourage organizations

819
00:30:38,480 --> 00:30:43,840
to stop engaging in extractive behavior,

820
00:30:41,279 --> 00:30:46,799
we absolutely need a way to bring those

821
00:30:43,840 --> 00:30:48,000
organizations back in to engage in a way

822
00:30:46,799 --> 00:30:50,559
that makes the projects more

823
00:30:48,000 --> 00:30:52,799
sustainable. And this is one hypothesis

824
00:30:50,559 --> 00:30:55,279
for why the market completely rejected

825
00:30:52,799 --> 00:30:57,520
these relicicensing.

826
00:30:55,279 --> 00:31:00,320
It locks out a potential revenue stream

827
00:30:57,520 --> 00:31:01,919
or source of collaboration. Terraform's

828
00:31:00,320 --> 00:31:05,120
selling point was that it is a better

829
00:31:01,919 --> 00:31:07,600
way to deploy to AWS. So removing all

830
00:31:05,120 --> 00:31:10,240
chances of them collaborating with AWS

831
00:31:07,600 --> 00:31:12,880
was a terrible idea. The technical

832
00:31:10,240 --> 00:31:14,399
outcomes of these relic licenses are bad

833
00:31:12,880 --> 00:31:16,320
as well as being bad for the business

834
00:31:14,399 --> 00:31:18,159
side of things.

835
00:31:16,320 --> 00:31:19,919
So I entitled this talk the death of

836
00:31:18,159 --> 00:31:22,320
consequences because free and open

837
00:31:19,919 --> 00:31:24,399
source software is a movement that is

838
00:31:22,320 --> 00:31:27,120
clearly not built on caring about the

839
00:31:24,399 --> 00:31:29,279
consequences of our decisions. But not

840
00:31:27,120 --> 00:31:31,039
caring about consequences comes with its

841
00:31:29,279 --> 00:31:32,960
own consequences.

842
00:31:31,039 --> 00:31:35,120
without questioning the basis on which

843
00:31:32,960 --> 00:31:37,279
we're willing to work. Then the tools

844
00:31:35,120 --> 00:31:39,120
for addressing sustainability

845
00:31:37,279 --> 00:31:41,679
operate entirely within the framework

846
00:31:39,120 --> 00:31:43,600
that we've laid out for ourselves.

847
00:31:41,679 --> 00:31:46,000
Feeling confident about any funding

848
00:31:43,600 --> 00:31:47,760
model requires us to ignore the

849
00:31:46,000 --> 00:31:50,640
consequences that we've built into our

850
00:31:47,760 --> 00:31:53,039
system. Every FOST developer that we pay

851
00:31:50,640 --> 00:31:55,360
to produce software in this system

852
00:31:53,039 --> 00:31:57,519
creates more software that can be

853
00:31:55,360 --> 00:31:58,960
extracted.

854
00:31:57,519 --> 00:32:01,840
And that's how I've reached the end of

855
00:31:58,960 --> 00:32:04,159
the first age of go of open source. If

856
00:32:01,840 --> 00:32:08,600
we want a second golden age, well, maybe

857
00:32:04,159 --> 00:32:08,600
let's start by not doing it alone.

858
00:32:12,240 --> 00:32:17,120
So, I don't have time for questions.

859
00:32:15,440 --> 00:32:18,960
If you do want to chat to me about this,

860
00:32:17,120 --> 00:32:20,240
have an argument, whatever, I will be

861
00:32:18,960 --> 00:32:22,559
hanging around the back of the room or

862
00:32:20,240 --> 00:32:25,120
in the hallway for the next few minutes.

863
00:32:22,559 --> 00:32:27,120
Thank you. So, is this one? Thank you so

864
00:32:25,120 --> 00:32:29,279
much, Christopher. What a compelling and

865
00:32:27,120 --> 00:32:30,960
engaging argument presented. One more

866
00:32:29,279 --> 00:32:36,689
time.

867
00:32:30,960 --> 00:32:36,689
[Applause]