WEBVTT

1
00:00:00.300 --> 00:00:07.920
Wishocracy How to let billions of people say whether the next marginal public dollar should buy another nuclear weapon system

2
00:00:07.920 --> 00:00:15.240
or fund another pragmatic clinical trial, without routing the answer through lobbyists, committees, and middlemen.

3
00:00:15.820 --> 00:00:21.960
On Who Wants to Be a Millionaire, the studio audience picked the right answer ninety-one percent of the time.

4
00:00:22.400 --> 00:00:29.000
The "phone a friend" expert? Sixty-five percent. Your entire system of government is phone-a-friend.

5
00:00:29.540 --> 00:00:35.960
You phone five hundred and thirty-five members of Congress who don't know the subject, aren't trying to answer your question,

6
00:00:36.500 --> 00:00:42.140
and are being paid by one of the answer choices. Then you act surprised when the answer is wrong.

7
00:00:42.880 --> 00:00:50.720
On Wishonia, we found the same pattern four thousand, two hundred and ninety-seven years ago: random crowds outperform credentialed

8
00:00:50.720 --> 00:00:57.400
experts on every question where the expert has any incentive to lie, which in government is all of them.

9
00:00:58.120 --> 00:01:01.650
So we stopped phoning friends. We built a system that asks everyone.

10
00:01:02.600 --> 00:01:05.480
We named it after ourselves, which your species would call "branding.".

11
00:01:06.620 --> 00:01:11.700
Wishocracy replaces committees with code and representatives with mathematics.

12
00:01:12.120 --> 00:01:19.680
It lets eight billion people collectively decide how to allocate contested public budgets without having the answer quietly

13
00:01:19.680 --> 00:01:25.280
rewritten by lobbyists, committees, or people whose main qualification is seniority.

14
00:01:26.040 --> 00:01:32.560
The difference between this and Congress is like asking two hundred bureaucrats what you should have for lunch versus asking

15
00:01:32.560 --> 00:01:38.520
your stomach. Your stomach has never once funded a colleague's lunch out of professional courtesy.

16
00:01:39.500 --> 00:01:46.880
Your species is already making brutal tradeoffs. Every extra dollar for nuclear weapon systems is a dollar not spent on

17
00:01:46.880 --> 00:01:52.380
pragmatic clinical trials, pandemic defense, transit, housing, or anything else.

18
00:01:52.900 --> 00:02:00.260
Your current institutions hide that fact by slicing the budget into committees, subcommittees, and jargon until nobody can

19
00:02:00.260 --> 00:02:05.000
see the tradeoff clearly enough to object. Wishocracy drags it back into daylight.

20
00:02:06.100 --> 00:02:11.800
Ordinary polling is useless here because it lets people live in a fantasy where scarcity does not exist.

21
00:02:12.540 --> 00:02:20.500
Ask humans whether they want more spending on defense, healthcare, education, infrastructure, and also lower taxes, and

22
00:02:20.500 --> 00:02:26.480
many will cheerfully say yes to all of it because the poll never makes them pay for one preference with another.

23
00:02:27.240 --> 00:02:35.380
Wishocracy does. Every answer is a relative tradeoff under finite resources, which is why it measures budgets instead of

24
00:02:35.380 --> 00:02:43.040
vibes. How Wishocracy Allocates High-Level Priorities. Anyone can submit a priority proposal.

25
00:02:43.220 --> 00:02:51.440
Wishocracy allocates the part of a budget that genuinely requires collective judgment: cross-domain tradeoffs, shared infrastructure,

26
00:02:51.440 --> 00:02:57.140
and other public goods that markets handle badly. It's Kickstarter, but backed by the G.D.P.

27
00:02:57.340 --> 00:03:03.920
of nations instead of your friend's credit card. What Wishocracy Actually Decides: The hard stuff.

28
00:03:04.580 --> 00:03:11.680
Allocation across competing priorities where there is no honest price signal: High-Level Tradeoffs.

29
00:03:11.740 --> 00:03:15.900
"Nuclear weapon modernization" versus "Pragmatic clinical trials".

30
00:03:16.540 --> 00:03:24.580
"Pandemic surveillance" versus "Missile defense". "Warhead dismantlement verification" versus "Submarine recapitalization".

31
00:03:25.540 --> 00:03:32.400
Shared Public Goods Within a Domain. "Verification satellite network" vs "On-site inspection capacity".

32
00:03:33.260 --> 00:03:41.940
"Bridge inspection" vs "Port cybersecurity". "Trial matching infrastructure" vs "Research data interoperability".

33
00:03:43.180 --> 00:03:48.620
Everything else is execution. Engineers build. Operators operate.

34
00:03:48.620 --> 00:03:55.820
Researchers run studies. Markets and domain-specific institutions handle the parts where direct user choice or price signals

35
00:03:55.820 --> 00:04:03.500
actually work. Wishocracy governs the part where the species has to decide what the next marginal public dollar should buy.

36
00:04:04.820 --> 00:04:11.060
How It Works: Pairwise Slider Allocation. Your brain can't rank a list of twenty priorities.

37
00:04:11.640 --> 00:04:17.410
It gives up around item number seven. But it's much better at a local tradeoff between two things.

38
00:04:17.800 --> 00:04:25.160
So instead of asking "Rank these ten thousand priorities," the system shows you two at a time and asks: "How would you split

39
00:04:25.160 --> 00:04:28.140
one hundred dollars between these two priorities right now?"

40
00:04:28.940 --> 00:04:33.080
"Nuclear weapon modernization" vs "Pragmatic clinical trials"?

41
00:04:33.860 --> 00:04:40.460
"Pandemic surveillance" vs "Missile defense"? "Bridge inspection" vs "Port cybersecurity"?

42
00:04:41.100 --> 00:04:46.080
You drag the slider, submit your split, and do this a few times with different random pairs.

43
00:04:46.480 --> 00:04:51.760
It takes five minutes (less time than you spend choosing a Netflix show, and considerably more important).

44
00:04:52.720 --> 00:04:59.000
Millions of other people do the same. An algorithm aggregates all these pairwise allocations into funding weights.

45
00:04:59.620 --> 00:05:06.760
No filibustering. No horse-trading. No senators holding one priority hostage because they want a weapons plant, research

46
00:05:06.760 --> 00:05:11.240
park, or bridge in their district. Not everyone has to answer every pair.

47
00:05:11.720 --> 00:05:16.480
If you do not know enough about "pandemic surveillance vs missile defense," you can skip it.

48
00:05:17.260 --> 00:05:24.560
This is how useful knowledge systems already work. A tiny minority writes Wikipedia articles, a somewhat larger minority

49
00:05:24.560 --> 00:05:31.400
edits them, and the overwhelming majority just reads. The crowd does not need everyone doing the same job.

50
00:05:31.740 --> 00:05:39.840
It needs enough informed people doing each job. Selective abstention is a feature, not a bug: better to skip a pair than

51
00:05:39.840 --> 00:05:43.260
inject random noise because you felt obliged to have an opinion.

52
00:05:44.200 --> 00:05:51.080
Low overall participation is still a risk, so the system needs thresholds, weighting, or fallback modes when engagement

53
00:05:51.080 --> 00:05:58.980
is too thin. But letting people answer where they actually know or care improves signal quality relative to mandatory guessing.

54
00:05:59.900 --> 00:06:03.040
One domain may still have its own internal allocation system.

55
00:06:03.780 --> 00:06:10.060
Health can have patient-directed trial participation. Science can have peer review or prize markets.

56
00:06:10.380 --> 00:06:14.880
Transit can have rider feedback. Markets can still handle things with prices.

57
00:06:15.480 --> 00:06:20.560
Wishocracy sits above that layer, steering the part that actually requires collective choice.

58
00:06:21.160 --> 00:06:26.880
One is the engine, the other is the steering wheel. You need both unless you enjoy driving into walls.

59
00:06:27.740 --> 00:06:36.260
Why This Actually Works (Math Warning). When millions of people make pairwise choices, something almost magical happens

60
00:06:36.260 --> 00:06:44.500
(by which I mean "statistically inevitable," which is the only kind of magic that works): Random pairs prevent gaming.

61
00:06:44.580 --> 00:06:51.600
Corrupting your current system requires buying access to five hundred and thirty-five members of Congress, which is so affordable

62
00:06:51.600 --> 00:07:00.540
that every major industry does it simultaneously. Corrupting Wishocracy would require bribing eight billion random voters,

63
00:07:01.180 --> 00:07:08.780
and the only way to "bribe" eight billion people is to offer them something they actually want, which is the system working

64
00:07:08.780 --> 00:07:14.900
correctly. The attack vector for Wishocracy is giving people what they voted for.

65
00:07:15.460 --> 00:07:24.380
Your security vulnerability is democracy. It's also immune to marketing: there's no ballot measure to link to, no "Vote

66
00:07:24.380 --> 00:07:32.820
Yes on Proposition Seven" ad to run. Each voter sees random pairs, so the only way to advertise is to convince people your

67
00:07:32.820 --> 00:07:37.400
proposal is better than whatever it happens to be compared against, which is just...

68
00:07:37.820 --> 00:07:42.980
making a good argument. Your species finds this deeply unsettling.

69
00:07:43.900 --> 00:07:53.140
Independence is structurally guaranteed. Surowiecki identified four conditions for crowd wisdom: diversity, independence,

70
00:07:53.820 --> 00:08:00.380
decentralization, and aggregation. Your elections destroy the most critical one.

71
00:08:01.200 --> 00:08:09.640
First-past-the-post voting forces strategic voting: people pick the "viable" candidate rather than the best one, so each

72
00:08:09.640 --> 00:08:15.320
ballot reflects what voters think other voters will do rather than what they actually want.

73
00:08:16.140 --> 00:08:23.860
It's an ant death spiral: each ant follows the one ahead, the leader follows the last, and the whole colony walks in a circle

74
00:08:23.860 --> 00:08:32.100
until it dies. Pairwise comparisons can't death-spiral because there's nothing to vote "strategically" about.

75
00:08:32.360 --> 00:08:36.760
"Which matters more: pragmatic trials or nuclear modernization?"

76
00:08:37.159 --> 00:08:42.200
has no spoiler candidate, no wasted vote, no reason to care what your neighbor picked.

77
00:08:42.919 --> 00:08:48.680
You just answer honestly. When millions of people answer honestly, you get wisdom.

78
00:08:49.260 --> 00:08:56.920
When millions of people game a two-option system, you get two candidates that three hundred and thirty million people chose

79
00:08:56.920 --> 00:09:01.620
from three hundred and thirty million people, and nobody likes either of them.

80
00:09:02.800 --> 00:09:09.420
Statistical models (Bradley-Terry, PageRank) extract global preferences from individual comparisons.

81
00:09:10.260 --> 00:09:15.480
The same math that ranks Google search results can rank humanity's budget priorities.

82
00:09:16.160 --> 00:09:23.840
Google uses it to find cat videos. You'll use it to decide whether the next marginal dollar buys a warhead or a clinical

83
00:09:23.840 --> 00:09:33.460
trial. Progress. Outliers cancel out. The guy who votes "fund my personal jetpack" gets drowned out by the millions voting

84
00:09:33.460 --> 00:09:39.170
for things that are at least attached to reality. Wisdom of crowds emerges.

85
00:09:39.780 --> 00:09:44.140
Remember the ninety-one percent versus sixty-five percent gap from *Millionaire*?

86
00:09:44.780 --> 00:09:52.680
Your species already proved that random people outperform credentialed individuals at picking the right answer.

87
00:09:52.820 --> 00:10:00.580
Wishocracy just applies this to questions that matter. Example with real numbers: five million people vote.

88
00:10:01.200 --> 00:10:04.940
Each makes twenty comparisons. That's one hundred million data points.

89
00:10:05.520 --> 00:10:12.900
Algorithm processes them in seconds. Output: "Allocate twenty-seven percent to nuclear weapon systems, forty-one percent

90
00:10:12.900 --> 00:10:19.160
to pragmatic clinical trials, eighteen percent to pandemic defense, fourteen percent to everything else...".

91
00:10:19.920 --> 00:10:22.780
No committee meetings were harmed in the making of this budget.

92
00:10:23.280 --> 00:10:29.980
It's democracy without the shouting. The stupidity is still there, but it's evenly distributed across millions of people

93
00:10:29.980 --> 00:10:34.180
and thus cancels out. Your species calls this "the wisdom of crowds."

94
00:10:34.840 --> 00:10:42.120
Mine calls it "obvious." About half of the eight hundred and forty-seven civilizations I've advised adopted some version

95
00:10:42.120 --> 00:10:47.860
of Wishocracy. The other half voted against it, which turned out to be the last thing they voted on.

96
00:10:48.560 --> 00:10:55.640
"But Direct Democracy Has Been Tried" (Yes. Badly.). You tried asking everyone.

97
00:10:55.900 --> 00:11:02.980
It went poorly. But not because asking was wrong. Your species has run four major experiments in direct democracy.

98
00:11:04.060 --> 00:11:09.960
Each one failed for a specific, diagnosable reason. None of those reasons apply here.

99
00:11:10.180 --> 00:11:17.680
Athens voted to execute Socrates and to invade Sicily (which destroyed their navy and lost the Peloponnesian War).

100
00:11:18.260 --> 00:11:25.820
The failure mode was mob rule: demagogues gave emotional speeches, crowds got excited, irreversible decisions happened before

101
00:11:25.820 --> 00:11:31.040
lunch. Wishocracy has no speeches, no candidates, no charisma.

102
00:11:31.560 --> 00:11:35.980
You cannot execute a philosopher with a pairwise comparison between infrastructure projects.

103
00:11:37.360 --> 00:11:44.560
California propositions let voters ban same-sex marriage (Prop eight, two thousand and eight) while spending over seven

104
00:11:44.560 --> 00:11:48.380
hundred million dollars per cycle on ballot measure campaigns.

105
00:11:48.660 --> 00:11:56.520
The failure mode was twofold: majority rule on fundamental rights, and unlimited advertising on binary yes/no questions.

106
00:11:57.220 --> 00:12:05.900
Wishocracy is scope-locked to bounded allocation questions (you cannot vote on who gets to marry) and presents random pairs

107
00:12:05.900 --> 00:12:15.180
(there is no ballot measure to buy ads for). Brexit reduced a one hundred-dimensional trade policy to a single binary question,

108
00:12:15.780 --> 00:12:22.540
then asked thirty-three million people to evaluate it with less information than they use to choose a phone plan.

109
00:12:23.140 --> 00:12:28.880
The failure mode was voter incompetence on complex policy forced into yes/no.

110
00:12:29.720 --> 00:12:38.540
Wishocracy asks "pragmatic clinical trials versus nuclear modernization," not "leave a twenty-seven-country trade bloc:

111
00:12:39.100 --> 00:12:46.900
yes/no." Pairwise comparisons between concrete tradeoffs are the exact kind of question crowds answer well.

112
00:12:47.500 --> 00:12:54.540
Swiss referendums denied women the vote until nineteen seventy-one (one canton held out until a federal court forced the

113
00:12:54.560 --> 00:13:01.320
issue in nineteen ninety) and banned minaret construction in two thousand and nine with fifty-seven point five percent support.

114
00:13:02.180 --> 00:13:09.120
The failure mode was unbounded scope: any question could appear on the ballot, including questions about who counts as a

115
00:13:09.120 --> 00:13:16.720
person. But the Swiss system is also the closest working model for the parts that do work: turnout averages forty-five percent,

116
00:13:17.540 --> 00:13:24.520
decisions are granular, and the country has not collapsed. Wishocracy keeps the good parts (distributed decision-making,

117
00:13:24.840 --> 00:13:31.720
regular participation) and scope-locks away the bad parts (voting on rights, voter fatigue from fifty-page ballot packets).

118
00:13:32.980 --> 00:13:36.820
Five minutes of pairwise comparisons, not a weekend with a legal dictionary.

119
00:13:38.300 --> 00:13:45.210
Every criticism of direct democracy is actually a criticism of bad aggregation on unbounded questions.

120
00:13:45.460 --> 00:13:49.800
Athens failed because it used majority rule on everything with emotional manipulation.

121
00:13:50.720 --> 00:13:57.760
California failed because it uses binary yes slash no on complex policy with unlimited ad spending.

122
00:13:58.980 --> 00:14:05.220
Wishocracy uses pairwise comparisons on scope-locked questions with no advertising surface.

123
00:14:05.720 --> 00:14:13.260
Same species, different math, different result. Your Current System Is Phone-a-Friend (But Your Friend Is Being Bribed).

124
00:14:14.160 --> 00:14:19.480
That ninety-one percent-versus-sixty-five percent comparison is actually generous to your politicians.

125
00:14:20.500 --> 00:14:26.770
Phone-a-friend experts have three advantages over members of Congress: They're trying to get the right answer.

126
00:14:27.360 --> 00:14:34.300
A senator deciding between more nuclear modernization and more pragmatic clinical trials is trying to get re-elected.

127
00:14:34.760 --> 00:14:40.020
These are different objectives. One of them occasionally results in a sensible budget by accident.

128
00:14:40.360 --> 00:14:43.760
The other one does it on purpose. They're selected for knowledge.

129
00:14:44.140 --> 00:14:46.880
You phone your friend who knows history for history questions.

130
00:14:47.440 --> 00:14:52.320
Your appropriations chair got there through thirty years of seniority and prolific fundraising.

131
00:14:52.760 --> 00:14:59.260
At no point did anyone check whether they could price one extra warhead against one extra clinical trial.

132
00:15:00.060 --> 00:15:01.860
They're not being paid by one of the answer choices. If the phone-a-friend expert were receiving one hundred and twenty-seven

133
00:15:01.860 --> 00:15:02.120
million dollars from "Answer C," you'd call that corruption.

134
00:15:02.120 --> 00:15:06.740
When a senator receives money from a defense contractor and then votes on defense spending, you call that "lobbying," which

135
00:15:06.740 --> 00:15:11.540
is the same thing but with a syllable that makes it sound boring enough that nobody gets angry.

136
00:15:12.580 --> 00:15:19.700
So your current system isn't sixty-five percent. It's phone-a-friend where your friend doesn't know the subject, isn't trying

137
00:15:19.700 --> 00:15:27.780
to answer your question, and is being paid to say "B." Against that baseline, ninety-one percent seems less like a modest

138
00:15:27.780 --> 00:15:31.640
improvement and more like replacing a broken compass with G.P.S.

139
00:15:31.940 --> 00:15:39.280
The twenty-six-Point Gap Is a Floor, Not a Ceiling. Here's the part your economists should find uncomfortable.

140
00:15:40.080 --> 00:15:43.840
Trivia is the best case for experts. There's one correct answer.

141
00:15:44.180 --> 00:15:51.080
The expert either knows it or doesn't. Nobody is paying them to say "Constantinople" when the answer is "Istanbul."

142
00:15:51.220 --> 00:15:55.660
No lobbyist is taking them to dinner. Their career doesn't depend on getting it wrong.

143
00:15:56.040 --> 00:16:04.620
And they still only hit sixty-five percent. Governance decisions are worse for experts on every dimension: The "right answer"

144
00:16:04.840 --> 00:16:11.880
requires knowledge no individual has. Friedrich Hayek won a Nobel Prize in nineteen seventy-four for explaining why.

145
00:16:12.600 --> 00:16:19.240
No central planner, no matter how brilliant, can aggregate the dispersed knowledge that millions of people carry about their

146
00:16:19.240 --> 00:16:24.990
own priorities, their own risks, and their own willingness to trade one outcome for another.

147
00:16:25.160 --> 00:16:32.620
No committee knows how much the public actually values one less warhead relative to one more pragmatic clinical trial.

148
00:16:33.120 --> 00:16:38.500
But the public does, and a system that asks all of them knows more than any committee ever could.

149
00:16:39.200 --> 00:16:42.220
Experts have financial incentives to give the wrong answer.

150
00:16:42.740 --> 00:16:47.800
On *Millionaire*, nobody pays the expert to lie. In Congress, it's the business model.

151
00:16:48.260 --> 00:16:53.280
Experts are selected for the wrong criteria. On *Millionaire*, you pick your smartest friend.

152
00:16:53.760 --> 00:17:01.200
In Congress, the selection process optimizes for charisma, name recognition, and the ability to stand at a podium without

153
00:17:01.200 --> 00:17:09.060
saying anything career-ending. The overlap between this skill set and "knows which public goods deserve the next marginal

154
00:17:09.060 --> 00:17:17.220
dollar" is, as far as I can tell, zero. So if crowds outperform experts by twenty-six percentage points on questions where

155
00:17:17.220 --> 00:17:24.599
experts have every possible advantage, the gap on governance decisions, where experts are compromised, poorly selected,

156
00:17:24.619 --> 00:17:28.580
and missing most of the relevant information, is almost certainly larger.

157
00:17:29.180 --> 00:17:32.880
The ninety-one percent versus sixty-five percent gap is the floor.

158
00:17:33.280 --> 00:17:41.040
The ceiling is the Soviet Union, which was the most ambitious phone-a-friend experiment in history: a small group of credentialed

159
00:17:41.040 --> 00:17:46.460
experts making resource allocation decisions for two hundred and eighty-six million people.

160
00:17:47.060 --> 00:17:54.640
It ran for sixty-nine years. It did not go well. Why Markets Solved This (and Why You Still Need Wishocracy).

161
00:17:55.480 --> 00:18:02.380
Hayek's solution to the knowledge problem was markets. Prices aggregate dispersed information automatically: if millions

162
00:18:02.380 --> 00:18:07.080
of people want more of something, the price goes up, and producers make more.

163
00:18:07.220 --> 00:18:13.860
No committee required. Your species has been doing this for thousands of years and it works so well that you barely notice

164
00:18:13.860 --> 00:18:18.220
it, the way you barely notice breathing until someone puts a pillow over your face.

165
00:18:18.840 --> 00:18:23.720
But markets have a blind spot. They only work when someone can profit from the solution.

166
00:18:24.260 --> 00:18:32.360
Markets are good at shoes, mediocre at healthcare, and terrible at shared public goods like pandemic surveillance, arms

167
00:18:32.360 --> 00:18:39.940
reduction verification, basic research infrastructure, and anything where everyone benefits but nobody can capture the return

168
00:18:39.940 --> 00:18:45.800
privately. These are called public goods, and markets handle them the way your cat handles a bath.

169
00:18:46.540 --> 00:18:53.820
Wishocracy is what you build when you need Hayek-style information aggregation but can't rely on price signals.

170
00:18:54.500 --> 00:19:01.860
It takes the same principle (millions of individual decisions aggregated into collective intelligence) and applies it to

171
00:19:01.860 --> 00:19:09.580
the things markets won't touch. It's the free market's missing organ: the part that funds what's important instead of just

172
00:19:09.580 --> 00:19:14.260
what's profitable. Hayek diagnosed the disease in nineteen forty-five.

173
00:19:14.900 --> 00:19:19.620
On Wishonia, we built the cure several millennia earlier. Better late than never.

174
00:19:20.380 --> 00:19:29.340
From Priorities to Projects (Where Wishes Become Tasks). Great, so the Pairwise Slider Allocation tells you "Pragmatic clinical

175
00:19:29.340 --> 00:19:33.720
trials" matter more than "nuclear weapon modernization." Wonderful.

176
00:19:34.360 --> 00:19:42.880
Now what? Having a priority without a plan is just a wish. And wishes, as a rule, don't cure anything or dismantle anything.

177
00:19:43.560 --> 00:19:49.440
Wishocracy translates that priority into action. A.I. Breaks It Down: An A.I.

178
00:19:49.700 --> 00:19:56.240
takes the high-level priority and breaks it into thousands of smaller, concrete, fundable tasks.

179
00:19:56.280 --> 00:20:02.100
"Expand pragmatic clinical trials" becomes "Build trial matching infrastructure."

180
00:20:02.780 --> 00:20:11.800
That becomes "Integrate hospital data systems." "Build open verification tools for warhead dismantlement" becomes "Contract

181
00:20:11.800 --> 00:20:19.840
satellite imagery analysis." Every impossible problem is just a series of possible steps arranged in a line.

182
00:20:20.280 --> 00:20:27.100
Your species figured this out for building pyramids four thousand five hundred years ago but keeps forgetting to apply it

183
00:20:27.100 --> 00:20:33.400
to government. The Bounty Board: The system posts these tasks to a global marketplace.

184
00:20:33.760 --> 00:20:37.260
It's like eBay, but for whatever priority the public just chose.

185
00:20:37.960 --> 00:20:43.280
WANTED: Expand pragmatic clinical trial capacity. BOUNTY: ten billion dollars.

186
00:20:43.860 --> 00:20:48.380
WANTED: Build open verification tools for warhead dismantlement.

187
00:20:48.920 --> 00:20:56.300
BOUNTY: five hundred million dollars. The World Competes: The best teams from around the world, from M.I.T.

188
00:20:56.640 --> 00:21:03.620
to some kid in a garage in Lagos, bid on these tasks. The system funds multiple approaches in parallel.

189
00:21:03.960 --> 00:21:08.880
The ones that show promise get more funding. The ones that fail lose funding instantly.

190
00:21:09.640 --> 00:21:17.020
It's venture capital, but for public goods. Your species already uses this method to decide which restaurants survive and

191
00:21:17.020 --> 00:21:23.780
which phone apps get downloaded. Time to try it with the things you keep pretending are too important for competition.

192
00:21:24.500 --> 00:21:33.320
Why the Algorithm Is the Constitution. Your species has a habit of writing beautiful rules and then immediately finding

193
00:21:33.320 --> 00:21:39.580
people to break them. Your Constitution guarantees free speech, and your government classifies documents.

194
00:21:40.100 --> 00:21:44.500
Your laws prohibit bribery, and your lobbying industry does it with a receipt.

195
00:21:45.300 --> 00:21:52.720
Every protection you've ever designed relies on humans to enforce it, and humans can be bought, threatened, promoted, or

196
00:21:53.140 --> 00:22:00.600
distracted by a sufficiently interesting scandal. On Wishonia, we tried enforcement-based protections for about two hundred

197
00:22:00.600 --> 00:22:08.280
years before concluding that constitutional constraints enforced by people are just suggestions with better typography.

198
00:22:08.680 --> 00:22:15.580
So we put the protections in the math. Algorithmic Protections (Rules Nobody Can Break Because They're Not Rules).

199
00:22:16.320 --> 00:22:23.960
Scope. The domain is defined before the voting starts. An allocation ballot can ask "nuclear weapon systems or pragmatic

200
00:22:23.960 --> 00:22:30.600
clinical trials?" An arms-control ballot can ask "verification satellite network or on-site inspection capacity?"

201
00:22:31.620 --> 00:22:36.740
A health ballot can ask "trial matching infrastructure or research data interoperability?"

202
00:22:37.300 --> 00:22:44.040
It cannot ask who counts as a person, which religion gets banned, or whether your senator's cousin deserves a bridge contract.

203
00:22:45.240 --> 00:22:50.740
You can't deposit a sandwich at an A.T.M. The machine doesn't accept that input.

204
00:22:51.800 --> 00:22:56.420
Transparency. There is no backroom where a committee quietly edits the answer.

205
00:22:57.180 --> 00:23:00.760
All comparisons, allocations, and outputs can be published.

206
00:23:01.380 --> 00:23:07.220
Not because a law requires it, but because the system is designed without a "hide this part" button.

207
00:23:07.800 --> 00:23:15.200
Corruption resistance. Your current system can be captured by a manageable number of lobbyists bribing a manageable number

208
00:23:15.200 --> 00:23:21.620
of people in a manageable number of rooms. Wishocracy raises the cost dramatically.

209
00:23:21.780 --> 00:23:29.660
There is no single appropriations chair to buy, no single committee hearing to dominate, and no single swing vote to purchase.

210
00:23:30.100 --> 00:23:33.920
Most of the money never passes through a room with a door you can close.

211
00:23:34.920 --> 00:23:42.000
Bodily autonomy. You can vote to allocate budgets. You cannot vote to control someone else's body.

212
00:23:42.100 --> 00:23:46.900
Not by majority vote. Not by executive order. Not by a very convincing T.E.D.

213
00:23:47.060 --> 00:23:51.680
talk. On Wishonia, the shortest law on the books is still: "Self.".

214
00:23:53.400 --> 00:23:59.900
What Wishocracy Measures. Wishocracy measures one thing: how people want a bounded public budget split.

215
00:24:00.800 --> 00:24:03.660
It does not tell you whether the public is morally correct.

216
00:24:04.100 --> 00:24:09.680
It does not prove which policy is causally optimal. It does not replace evidence systems.

217
00:24:10.160 --> 00:24:17.020
It answers a narrower and more important question than your current institutions usually answer: what do people actually

218
00:24:17.020 --> 00:24:24.540
want the next marginal public dollar to buy? If the answer is "twenty-seven cents to nuclear weapon systems and seventy-three

219
00:24:24.540 --> 00:24:32.640
cents to pragmatic clinical trials," that is already more information than most elections produce after two years of campaigning,

220
00:24:33.180 --> 00:24:36.800
three billion dollars in ads, and a national shouting match.