1
00:00:00,350 --> 00:00:04,700
So in the last section we ended by
showing how large professional mining

2
00:00:04,700 --> 00:00:07,990
data centers have taken over
the business of Bitcoin mining.

3
00:00:08,990 --> 00:00:13,440
And we showed the parallel to traditional
mining with those open pit mines on

4
00:00:13,440 --> 00:00:16,720
the bottom, which you may know have
been a huge source of concern for

5
00:00:16,720 --> 00:00:19,090
environmentalists over the years asking.

6
00:00:19,090 --> 00:00:21,640
How much damage are these pit
miners doing to the environment?

7
00:00:22,910 --> 00:00:25,310
Now, Bitcoin is not
quite at that level yet.

8
00:00:25,310 --> 00:00:29,710
But as I said,
it is a very difficult computation

9
00:00:29,710 --> 00:00:33,950
with a lot of large players
interested in competing now.

10
00:00:33,950 --> 00:00:37,630
And it is starting to use
a significant amount of energy,

11
00:00:37,630 --> 00:00:39,320
which has become a topic of discussion.

12
00:00:40,800 --> 00:00:44,129
So in this section we'll talk about
how much energy Bitcoin is using and

13
00:00:44,129 --> 00:00:46,803
what that might mean for
the currency and for the planet.

14
00:00:50,421 --> 00:00:51,630
So to start with,

15
00:00:51,630 --> 00:00:56,400
we'll talk about why computation
inherently requires some energy.

16
00:00:57,730 --> 00:01:02,326
So there's a principle developed
by Rolph Landauer in the 1960s.

17
00:01:02,326 --> 00:01:09,650
That any non-reversible computation
must use as minimum amount of energy.

18
00:01:09,650 --> 00:01:12,028
So this is derived from basic physics.

19
00:01:12,028 --> 00:01:14,791
We're not going to go
through the derivation here,

20
00:01:14,791 --> 00:01:19,196
except to say that every time you flip
on bit in a non-reversible computation,

21
00:01:19,196 --> 00:01:22,375
there's a minimum number of
joules that you have to use.

22
00:01:25,320 --> 00:01:29,940
And of course, if you remember some
fundamental theorems of physics,

23
00:01:29,940 --> 00:01:32,060
energy is never destroyed.

24
00:01:32,060 --> 00:01:35,544
It's only converted from
one form into another.

25
00:01:35,544 --> 00:01:39,530
In the case of computation, it's
mostly that energy is transformed from

26
00:01:39,530 --> 00:01:42,808
electricity, which is very
high grade energy, Into heat,

27
00:01:42,808 --> 00:01:45,204
which is dissipated into the environment.

28
00:01:47,783 --> 00:01:51,270
Now of course,
SHA-256 being a hash function,

29
00:01:51,270 --> 00:01:56,480
which is the basis of Bitcoin,
is not a reversible computation.

30
00:01:56,480 --> 00:01:58,330
And if you remember all
the way back to lecture one,

31
00:01:58,330 --> 00:02:01,799
this is a basic requirement of hash
functions, that they're not reversible.

32
00:02:03,370 --> 00:02:08,685
Since I told you that any non-reversible
computation has to use some energy and

33
00:02:08,685 --> 00:02:11,190
SHA-256 is not reversible,

34
00:02:11,190 --> 00:02:14,390
energy consumption is an inevitable
factor doing Bitcoin mining.

35
00:02:15,590 --> 00:02:19,740
What it's saying is that the limits
provided by Landauer's principle

36
00:02:21,280 --> 00:02:26,550
are far, far below the amount of
electricity that's being used today by

37
00:02:26,550 --> 00:02:28,220
over a factor of a thousand, so

38
00:02:28,220 --> 00:02:32,100
we're nowhere close to theoretical
optimum efficiency of computing.

39
00:02:34,040 --> 00:02:36,570
But even if we did get to
the theoretical optimum,

40
00:02:36,570 --> 00:02:39,540
we would still be using some
energy to perform Bitcoin mining.

41
00:02:41,280 --> 00:02:43,980
So why does Bitcoin mining require energy?

42
00:02:43,980 --> 00:02:46,470
There are three steps in
the process that require energy.

43
00:02:47,610 --> 00:02:51,110
First, you have to manufacture
your Bitcoin mining equipment.

44
00:02:51,110 --> 00:02:55,330
So that requires both physical mining,
digging up things out of the ground,

45
00:02:55,330 --> 00:02:59,500
especially rare earth metals and
copper that go into integrated circuits.

46
00:02:59,500 --> 00:03:03,010
And then you have to manufacture
it into a Bitcoin mining ASIC.

47
00:03:03,010 --> 00:03:07,940
So it takes a lot of energy to
run fabs and to tune out chips.

48
00:03:07,940 --> 00:03:11,050
So all of that energy is
called the embodied energy.

49
00:03:11,050 --> 00:03:14,960
As soon as you receive a bit coin
mining ASIC in the mail that you order,

50
00:03:14,960 --> 00:03:19,490
you've already consumed a lot of energy,
including the shipping energy of course,

51
00:03:19,490 --> 00:03:23,340
just to get that to you before you've even
turned it on and tried to mine Bitcoins.

52
00:03:24,740 --> 00:03:26,540
Then you'll plug it into the wall and
turn it on,

53
00:03:26,540 --> 00:03:30,520
and of course it will be drawing
electricity constantly while it's on,

54
00:03:30,520 --> 00:03:33,050
and that's the electrical
energy consumed in mining.

55
00:03:33,050 --> 00:03:36,930
And that's the step that
no matter what happens

56
00:03:37,940 --> 00:03:41,130
has to be consumed because
of Landauer's principle.

57
00:03:41,130 --> 00:03:44,880
So hopefully over time,
the embodied energy will go down.

58
00:03:44,880 --> 00:03:49,100
As less and
less new capacity comes online,

59
00:03:49,100 --> 00:03:52,180
fewer people are going out
to buy new mining ASICs.

60
00:03:52,180 --> 00:03:54,960
They're being obsoleted less quickly.

61
00:03:54,960 --> 00:03:57,480
The existing fleet can last a long time.

62
00:03:57,480 --> 00:04:01,360
The embodied energy will be amortized
over years and years of mining.

63
00:04:02,530 --> 00:04:06,470
But the electricity consumption,
even though it will go down a little bit

64
00:04:06,470 --> 00:04:10,620
because rigs will get more efficient,
that will be a fact of life forever.

65
00:04:12,290 --> 00:04:14,633
The other thing about both electricity and

66
00:04:16,099 --> 00:04:20,840
energy is that both are probably less
if you're operating at a large scale.

67
00:04:20,840 --> 00:04:23,270
If you're running a huge
mining data center.

68
00:04:23,270 --> 00:04:25,240
You can do it more efficiently.

69
00:04:25,240 --> 00:04:29,550
It's cheaper to build chips that are
designed to run in a large data center and

70
00:04:29,550 --> 00:04:31,340
you can deliver the power
more efficiently.

71
00:04:31,340 --> 00:04:33,880
Because you don't need
as many power supplies,

72
00:04:33,880 --> 00:04:36,660
you can deliver all the electricity
to one place and so on.

73
00:04:38,010 --> 00:04:39,820
But there's a third important component,

74
00:04:39,820 --> 00:04:43,330
which is cooling off your equipment to
make sure that it doesn't malfunction.

75
00:04:45,320 --> 00:04:47,650
So if you're operating your
equipment in Antarctica,

76
00:04:47,650 --> 00:04:49,780
maybe your cooling budget is very small.

77
00:04:49,780 --> 00:04:55,090
But almost anywhere else, you're going to
have to pay extra, usually electricity,

78
00:04:55,090 --> 00:04:58,790
to cool off your equipment from all
the waste heat that it's generating.

79
00:05:00,100 --> 00:05:05,030
An interesting aspect about cooling
Is that cooling actually costs more

80
00:05:05,030 --> 00:05:09,560
the bigger your scale is, so if you want
to run a very large operation and have

81
00:05:09,560 --> 00:05:14,140
a lot of Bitcoin mining equipment all in
one place, you're cooling budget is going

82
00:05:14,140 --> 00:05:19,320
to increase, because cooling that big
mass is going to be much more difficult.

83
00:05:19,320 --> 00:05:22,973
There's less air for the heat to, to
dissipate into surrounding your equipment.

84
00:05:26,021 --> 00:05:29,921
So how much energy is the entire
Bitcoin network using?

85
00:05:29,921 --> 00:05:34,056
There are two basic approaches to trying
to estimate how much energy the Bitcoin

86
00:05:34,056 --> 00:05:37,882
network is using of course we can't
compute this precisely because it's

87
00:05:37,882 --> 00:05:41,277
a decentralized network with miners
operating all over the place

88
00:05:41,277 --> 00:05:43,953
without documenting exactly
what they are doing.

89
00:05:43,953 --> 00:05:47,706
But we'll start with a really
simple approximation strategy,

90
00:05:47,706 --> 00:05:52,643
which is to take the fact that about
$15,000 and again, that's 25 Bitcoin

91
00:05:52,643 --> 00:05:56,900
of reward are created with every block
which is found every ten minutes.

92
00:05:57,940 --> 00:06:02,190
So if we convert that to revenue per
second, we get about $25 US dollars per

93
00:06:02,190 --> 00:06:06,430
second that are being minted and
given to the mining community.

94
00:06:08,040 --> 00:06:13,330
Now, if the miners are turning all of
those $25 per second into electricity,

95
00:06:13,330 --> 00:06:14,046
how much can they get?

96
00:06:14,046 --> 00:06:19,370
Well, at US industrial electricity prices,
and

97
00:06:19,370 --> 00:06:23,470
this will vary from state to state or
certainly from country to country, but

98
00:06:23,470 --> 00:06:26,550
we'll go with about $0.10
US per kilowatt hour.

99
00:06:27,960 --> 00:06:30,900
And kilowatt hour is kind of
a funny marketing unit, so

100
00:06:30,900 --> 00:06:33,920
we'll go to the more standard
scientific unit of the megajoule.

101
00:06:35,810 --> 00:06:40,880
So, if the miner took those $25 that they
earned every second and converted it

102
00:06:40,880 --> 00:06:46,180
purely into electricity, they would
get about 900 megajoules every second.

103
00:06:47,880 --> 00:06:50,340
And of course joules per
second are just watts.

104
00:06:50,340 --> 00:06:55,210
So those 900 megajoules per second
are 900 megawatts or 900 million watts.

105
00:06:58,920 --> 00:07:02,850
A second way to estimate the same
figure is to do a bottom up approach.

106
00:07:02,850 --> 00:07:07,330
And to say let's look at how many hashes
the miners are actually computing,

107
00:07:07,330 --> 00:07:11,110
which we know by observing
the difficulty of each block.

108
00:07:11,110 --> 00:07:13,630
And what is the best hardware
that miners might be using?

109
00:07:15,850 --> 00:07:20,480
So if you look online at mining rigs
that are being commercially sold today,

110
00:07:20,480 --> 00:07:25,110
one of the best performance figures that
you'll see Is rigs that are able to turn

111
00:07:25,110 --> 00:07:29,480
one watt of electricity into
about one gigahertz of hashing.

112
00:07:29,480 --> 00:07:33,770
So they perform one billion hashes
per second while consuming about one

113
00:07:33,770 --> 00:07:34,550
watt of power.

114
00:07:35,890 --> 00:07:42,560
And the total network hash rate is about
150 million gigahertz or 150 petahertz.

115
00:07:46,670 --> 00:07:48,790
Of course that excludes all
of the cooling energy and

116
00:07:48,790 --> 00:07:50,880
all of the embodied energy
that's in those chips.

117
00:07:50,880 --> 00:07:53,020
But we're doing an optimal
calculation here.

118
00:07:54,500 --> 00:07:57,620
So if the entire network was
running at about the efficiency of

119
00:07:57,620 --> 00:08:00,650
generally the better chips on the market,
what would we get?

120
00:08:01,940 --> 00:08:06,740
If we just multiply these two
together We would get about 150 MW to

121
00:08:06,740 --> 00:08:09,850
produce that many hashes per second,
at that efficiency.

122
00:08:11,960 --> 00:08:15,710
So again, last slide I said, at a high
end, using the top down approach,

123
00:08:15,710 --> 00:08:18,330
we estimated about 900 MW.

124
00:08:18,330 --> 00:08:22,920
And using the bottom up approach,
this is a lower bound, about 150 MW.

125
00:08:24,140 --> 00:08:27,580
So maybe for the whole network today
somewhere between 100 megawatts and

126
00:08:27,580 --> 00:08:30,070
a gigawatt of electricity
are being consumed.

127
00:08:31,390 --> 00:08:33,740
In reality, of course,
it's probably somewhere in the middle and

128
00:08:33,740 --> 00:08:35,760
it's going to evolve over time.

129
00:08:35,760 --> 00:08:38,240
But that's a useful ballpark
to think about right now.

130
00:08:40,780 --> 00:08:41,960
So how much is a megawatt?

131
00:08:43,110 --> 00:08:46,370
Well, we can look at what
big power plants produce.

132
00:08:46,370 --> 00:08:48,800
So one of the largest
power plants in the world,

133
00:08:48,800 --> 00:08:51,960
the Three Gorges Dam in China,
is a 10,000 megawatt power plant.

134
00:08:51,960 --> 00:08:56,550
It actually has slightly
higher capacity than that, but

135
00:08:56,550 --> 00:08:59,960
that's the average rate of
power that's being produced,

136
00:08:59,960 --> 00:09:02,950
whereas a typical large hydro plant
is more like 1,000 megawatts.

137
00:09:05,790 --> 00:09:10,311
If you're interested in nuclear power, we
can look at the largest nuclear plant in

138
00:09:10,311 --> 00:09:13,024
Japan, and that's about a 7,000 MW plant,

139
00:09:13,024 --> 00:09:16,790
whereas the average nuclear power
plant is more like 4,000 MW.

140
00:09:19,432 --> 00:09:24,465
Or back in the carbon-heavy way of
producing electricity, we can look at

141
00:09:24,465 --> 00:09:30,580
a major large coal fired plants, and
you might get 1,000 to 2,000 megawatts.

142
00:09:30,580 --> 00:09:35,080
So again, our high end estimate was still
that Bitcoin was consuming less than 1,000

143
00:09:35,080 --> 00:09:39,402
megawatts, so
the whole Bitcoin network is consuming

144
00:09:39,402 --> 00:09:42,380
less than a large power
plants worth of electricity.

145
00:09:43,540 --> 00:09:44,840
So it's not nothing.

146
00:09:44,840 --> 00:09:49,530
It still means that we have to essentially
to run a large power plant purely to power

147
00:09:49,530 --> 00:09:53,570
Bitcoin and not any of the other things
that we need electricity for in the world.

148
00:09:54,610 --> 00:09:58,750
But it's not yet to the point where it's
a large amount of electricity compared to

149
00:09:58,750 --> 00:10:01,609
all the other things that people
are using electricity for on the planet.

150
00:10:03,210 --> 00:10:06,240
And, it's certainly worth pointing out
that any payment system is going to

151
00:10:06,240 --> 00:10:09,040
require energy and electricity.

152
00:10:09,040 --> 00:10:11,030
So, when we look at traditional currency,

153
00:10:12,080 --> 00:10:17,440
a lot of energy is consumed moving gold
bullion around, guarding the gold bullion,

154
00:10:17,440 --> 00:10:22,390
running ATM machines, running coin
sorting machines, running cash registers.

155
00:10:22,390 --> 00:10:25,080
Transporting the money
around in armored cars.

156
00:10:25,080 --> 00:10:28,030
All of that is energy consumed
by the traditional money system.

157
00:10:29,430 --> 00:10:34,300
So sometimes people have a tendency
to think Bitcoin is wasting energy,

158
00:10:34,300 --> 00:10:39,820
because the energy is being expended
in this SHA-256 computation

159
00:10:39,820 --> 00:10:42,590
that doesn't serve any apparent purpose.

160
00:10:42,590 --> 00:10:46,370
But you could also look at all of the
energies in a traditional currency system

161
00:10:46,370 --> 00:10:48,180
and say that it's also wasted and

162
00:10:48,180 --> 00:10:51,940
that it doesn't serve any other purpose
besides maintaining the currency system.

163
00:10:54,110 --> 00:10:56,220
So that' s a really important disclaimer,
I think,

164
00:10:56,220 --> 00:11:00,630
that just because Bitcoin uses
electricity it's not necessarily wasted.

165
00:11:00,630 --> 00:11:02,950
If Bitcoin is a useful currency system,

166
00:11:02,950 --> 00:11:05,879
then the electricity is essentially
being used for that purpose.

167
00:11:08,350 --> 00:11:11,220
But we still might think is
there something better that

168
00:11:11,220 --> 00:11:13,330
we could do with this electricity.

169
00:11:13,330 --> 00:11:16,330
Rather than just heating up air which
is sent off into the atmosphere.

170
00:11:17,440 --> 00:11:21,010
And one pretty interesting idea is what if
we tried to capture the heat that we're

171
00:11:21,010 --> 00:11:27,210
turning that electricity used in Bitcoin
mining and use it for practical purposes.

172
00:11:27,210 --> 00:11:30,630
So, this is called
the Data Furnaces Model, and

173
00:11:30,630 --> 00:11:34,620
the basic idea is that you would go
down to your local hardware store, and

174
00:11:34,620 --> 00:11:37,920
instead of buying a traditional
electric heater to heat your home, or

175
00:11:37,920 --> 00:11:42,800
to heat water in your home, you would
buy a Bitcoin mining rig that you would

176
00:11:42,800 --> 00:11:47,050
plug in both to the electricity outlet and
also to your Internet connection.

177
00:11:48,280 --> 00:11:51,650
And your heater would essentially
be doing Bitcoin mining, and

178
00:11:51,650 --> 00:11:56,110
using the heat produced as a byproduct of
that computation to heat your water or

179
00:11:56,110 --> 00:11:59,090
to heat your home,
which is hopefully useful.

180
00:12:01,250 --> 00:12:03,800
And it turns out that
the efficiency of doing this isn't

181
00:12:03,800 --> 00:12:07,530
actually that much worse that
just buying an electric heater.

182
00:12:07,530 --> 00:12:09,000
So this seems to be a great idea,

183
00:12:09,000 --> 00:12:12,590
and maybe it's a promising avenue
to explore for the future.

184
00:12:13,600 --> 00:12:15,970
There's a couple of challenges here.

185
00:12:15,970 --> 00:12:20,910
For one, electric heaters are still
much less efficient than gas heaters, so

186
00:12:20,910 --> 00:12:24,170
if you're in a really cold climate,
where people really need heat,

187
00:12:24,170 --> 00:12:26,779
hopefully they'll have gas
heating in their home anyway.

188
00:12:28,870 --> 00:12:31,350
It's also not clear what
the ownership model is here.

189
00:12:32,360 --> 00:12:36,290
If you buy the Bitcoin data furnace,
do you own the Bitcoin mining rewards that

190
00:12:36,290 --> 00:12:39,690
you get, or
does the company that sold them to you?

191
00:12:39,690 --> 00:12:43,560
Most people don't have any interest in
Bitcoin mining and probably never will, so

192
00:12:43,560 --> 00:12:46,470
it might make more sense to
buy this as an appliance.

193
00:12:46,470 --> 00:12:48,970
And have the company that sold
it to you keep the rewards.

194
00:12:50,390 --> 00:12:54,090
And then there's the really basic question
of what happens if everybody turns off

195
00:12:54,090 --> 00:12:55,970
their Bitcoin mining rig in the summer?

196
00:12:57,880 --> 00:13:01,750
Will the capacity of the Bitcoin network
go way down seasonally based on how much

197
00:13:01,750 --> 00:13:02,850
heat people need?

198
00:13:02,850 --> 00:13:06,507
Will it go way down on days that
happen to be warmer, than average.

199
00:13:08,250 --> 00:13:11,440
This would be really interesting if the
data for this model actually caught on.

200
00:13:14,250 --> 00:13:17,690
So a couple of open questions related
to Bitcoin's energy consumption.

201
00:13:18,780 --> 00:13:23,130
Does the fact that Bitcoin provides such
a good way to turn electricity into cash

202
00:13:23,130 --> 00:13:26,790
mean that countries that have strong
electricity subsidies will have to rethink

203
00:13:26,790 --> 00:13:27,410
that model?

204
00:13:28,970 --> 00:13:33,800
So right now in many countries around the
world, the government actually subsidizes

205
00:13:33,800 --> 00:13:36,640
electricity, particularly
industrial electricity.

206
00:13:37,650 --> 00:13:39,502
And one of the reasons they do so
is to try and

207
00:13:39,502 --> 00:13:43,174
encourage industry to be located in their
country, as opposed to other countries.

208
00:13:45,572 --> 00:13:49,354
Now, if one of the main things that
Bitcoin miners need to be successful is

209
00:13:49,354 --> 00:13:53,441
cheap energy and you can mine Bitcoins
basically anywhere, it may not be stable

210
00:13:53,441 --> 00:13:57,711
to have countries subsidizing electricity
heavily, because all that will mean is

211
00:13:57,711 --> 00:14:01,499
that you're paying for a lot of Bitcoin
miners to move into your country.

212
00:14:03,600 --> 00:14:05,820
There's also the interesting question of,

213
00:14:05,820 --> 00:14:09,420
will the fact that you can turn
electricity into money easily with Bitcoin

214
00:14:09,420 --> 00:14:12,390
mean that people have to start
guarding their power outlets?

215
00:14:12,390 --> 00:14:14,010
Particularly around universities and

216
00:14:14,010 --> 00:14:17,530
corporations, large buildings
with a lot of power outlets.

217
00:14:17,530 --> 00:14:21,440
Will they need security cameras to
make sure that employees or students

218
00:14:21,440 --> 00:14:25,020
aren't trying to mine Bitcoins by plugging
into unmonitored power outlets and

219
00:14:25,020 --> 00:14:25,810
just letting them run.

220
00:14:27,610 --> 00:14:30,720
And you might ask would we be better
off if we didn't have this electricity

221
00:14:30,720 --> 00:14:32,140
consumption.

222
00:14:32,140 --> 00:14:35,630
Could we make a currency that didn't have
proof of work and didn't have to use so

223
00:14:35,630 --> 00:14:36,430
much electricity.

224
00:14:36,430 --> 00:14:40,340
And again I'm not going to talk
about that directly today, but

225
00:14:40,340 --> 00:14:43,190
that's going to be a topic that
we talk about quite a bit.

226
00:14:43,190 --> 00:14:45,441
In our future lecture
on alternative mining.

