ABOUT THE SPEAKER
Randall Munroe - Cartoonist
Randall Munroe sketches elegant and illuminating explanations of the weird science and math questions that keep geeks awake at night.

Why you should listen

One of a small group of professional web cartoonists, math obsessive and chronic explainer Randall Munroe dazzles the online world (and racks up millions of monthly page views) with the meaninglessly-named (and occasionally heartbreaking) webcomic xkcd.
 
Munroe’s blog What If? specializes in cunning answers to, as the Atlantic put it, "the kinds of of wonderful and fanciful hypotheticals that might arise when the nerdily inclined get together in bars," like “How fast can you hit a speed bump while driving and live?” or “What would happen if a hair dryer with continuous power was turned on and put in an airtight 1x1x1 meter box?” As he told Math Horizons, I really enjoy solving these kinds of things, and it’s a bonus if I realize that I can put boxes around it and make it a comic."

He is also the author of the book Thing Explainer: Complicated Stuff in Simple Words

More profile about the speaker
Randall Munroe | Speaker | TED.com
TED2014

Randall Munroe: Comics that ask "what if?"

Filmed:
3,496,101 views

Web cartoonist Randall Munroe answers simple what-if questions ("what if you hit a baseball moving at the speed of light?") using math, physics, logic and deadpan humor. In this charming talk, a reader's question about Google's data warehouse leads Munroe down a circuitous path to a hilariously over-detailed answer — in which, shhh, you might actually learn something.
- Cartoonist
Randall Munroe sketches elegant and illuminating explanations of the weird science and math questions that keep geeks awake at night. Full bio

Double-click the English transcript below to play the video.

00:12
So, I have a feature on my website where every week
0
882
2509
00:15
people submit hypothetical questions
1
3391
2000
00:17
for me to answer,
2
5391
1588
00:18
and I try to answer them using math, science
3
6979
2381
00:21
and comics.
4
9360
1460
00:22
So for example, one person asked,
5
10820
2408
00:25
what would happen if you tried to hit a baseball
6
13228
1694
00:26
pitched at 90 percent of the speed of light?
7
14922
2406
00:29
So I did some calculations.
8
17328
2952
00:32
Now, normally, when an object flies through the air,
9
20280
2267
00:34
the air will flow around the object,
10
22547
1474
00:36
but in this case, the ball would be going so fast
11
24021
1929
00:37
that the air molecules wouldn't have time
12
25950
1451
00:39
to move out of the way.
13
27401
1619
00:41
The ball would smash right into and through them,
14
29020
2987
00:44
and the collisions with these air molecules
15
32007
1661
00:45
would knock away the nitrogen,
16
33668
1665
00:47
carbon and hydrogen from the ball,
17
35333
1958
00:49
fragmenting it off into tiny particles,
18
37291
1837
00:51
and also triggering waves of thermonuclear fusion
19
39128
2484
00:53
in the air around it.
20
41612
1528
00:55
This would result in a flood of x-rays
21
43140
2165
00:57
that would spread out in a bubble
22
45305
1751
00:59
along with exotic particles,
23
47056
1426
01:00
plasma inside, centered on the pitcher's mound,
24
48482
3120
01:03
and that would move away from the pitcher's mound
25
51602
3542
01:07
slightly faster than the ball.
26
55144
2485
01:09
Now at this point, about 30 nanoseconds in,
27
57629
2682
01:12
the home plate is far enough away
28
60311
1903
01:14
that light hasn't had time to reach it,
29
62214
1906
01:16
which means the batter
30
64120
1451
01:17
still sees the pitcher about to throw
31
65571
2764
01:20
and has no idea that anything is wrong.
32
68335
2017
01:22
(Laughter)
33
70352
2272
01:24
Now, after 70 nanoseconds,
34
72624
2426
01:27
the ball will reach home plate,
35
75050
1454
01:28
or at least the cloud of expanding plasma
36
76504
2737
01:31
that used to be the ball,
37
79241
2129
01:33
and it will engulf the bat and the batter
38
81370
3765
01:37
and the plate and the catcher and the umpire
39
85135
3461
01:40
and start disintegrating them all
40
88596
2805
01:43
as it also starts to carry them backward
41
91401
3166
01:46
through the backstop, which
also starts to disintegrate.
42
94567
3464
01:50
So if you were watching this whole thing
43
98031
1770
01:51
from a hill,
44
99801
1307
01:53
ideally, far away,
45
101108
3373
01:56
what you'd see is a bright flash of light
46
104481
1455
01:57
that would fade over a few seconds,
47
105936
1506
01:59
followed by a blast wave spreading out,
48
107442
2128
02:01
shredding trees and houses
49
109570
2460
02:04
as it moves away from the stadium,
50
112030
2119
02:06
and then eventually a mushroom cloud
51
114149
2932
02:09
rising up over the ruined city. (Laughter)
52
117081
3761
02:12
So the Major League Baseball rules
53
120842
1156
02:13
are a little bit hazy,
54
121998
2104
02:16
but — (Laughter) — under rule 6.02 and 5.09,
55
124102
4885
02:20
I think that in this situation,
56
128987
1609
02:22
the batter would be considered hit by pitch
57
130596
2453
02:25
and would be eligible to take first base,
58
133049
2634
02:27
if it still existed.
59
135683
3203
02:30
So this is the kind of question I answer,
60
138886
1796
02:32
and I get people writing in with
61
140682
2139
02:34
a lot of other strange questions.
62
142821
1874
02:36
I've had someone write and say,
63
144695
2547
02:39
scientifically speaking, what is the best
64
147242
1919
02:41
and fastest way to hide a body?
65
149161
2070
02:43
Can you do this one soon?
66
151231
2222
02:45
And I had someone write in,
67
153453
1860
02:47
I've had people write in about,
68
155313
2015
02:49
can you prove whether or not you can find love again
69
157328
2204
02:51
after your heart's broken?
70
159532
1672
02:53
And I've had people send in
71
161204
1775
02:54
what are clearly homework questions
72
162979
1991
02:56
they're trying to get me to do for them.
73
164970
3143
03:00
But one week, a couple months ago,
74
168113
2905
03:03
I got a question that was actually about Google.
75
171018
3661
03:06
If all digital data in the world
were stored on punch cards,
76
174679
2966
03:09
how big would Google's data warehouse be?
77
177645
2679
03:12
Now, Google's pretty secretive
about their operations,
78
180324
2722
03:15
so no one really knows how much data Google has,
79
183046
3376
03:18
and in fact, no one really knows
how many data centers Google has,
80
186422
2515
03:20
except people at Google itself.
81
188937
2491
03:23
And I've tried, I've met them a few times,
82
191428
1234
03:24
tried asking them, and they
aren't revealing anything.
83
192662
5223
03:29
So I decided to try to figure this out myself.
84
197885
2535
03:32
There are a few things that I looked at here.
85
200420
2159
03:34
I started with money.
86
202579
1906
03:36
Google has to reveal how much they spend,
87
204485
2142
03:38
in general, and that lets you put some caps
88
206627
1990
03:40
on how many data centers could they be building,
89
208617
3786
03:44
because a big data center costs
a certain amount of money.
90
212403
2300
03:46
And you can also then put a cap on
91
214703
2432
03:49
how much of the world hard
drive market are they taking up,
92
217135
2331
03:51
which turns out, it's pretty sizable.
93
219466
1952
03:53
I read a calculation at one point,
94
221418
1937
03:55
I think Google has a drive failure
95
223355
1465
03:56
about every minute or two,
96
224820
3254
04:00
and they just throw out the hard drive
97
228074
1350
04:01
and swap in a new one.
98
229424
1386
04:02
So they go through a huge number of them.
99
230810
2200
04:05
And so by looking at money,
100
233010
1111
04:06
you can get an idea of how
many of these centers they have.
101
234121
2849
04:08
You can also look at power.
102
236970
1860
04:10
You can look at how much electricity they need,
103
238830
3897
04:14
because you need a certain amount
of electricity to run the servers,
104
242727
1807
04:16
and Google is more efficient than most,
105
244534
2039
04:18
but they still have some basic requirements,
106
246573
3097
04:21
and that lets you put a limit
107
249670
1916
04:23
on the number of servers that they have.
108
251586
1947
04:25
You can also look at square footage and see
109
253533
4191
04:29
of the data centers that you know,
110
257724
1276
04:31
how big are they?
111
259000
1353
04:32
How much room is that?
112
260353
922
04:33
How many server racks could you fit in there?
113
261275
2360
04:35
And for some data centers,
114
263635
1703
04:37
you might get two of these pieces of information.
115
265338
1989
04:39
You know how much they spent,
116
267327
1696
04:41
and they also, say, because they had to contract
117
269023
2357
04:43
with the local government
118
271380
1419
04:44
to get the power provided,
119
272799
1680
04:46
you might know what they made a deal to buy,
120
274479
2657
04:49
so you know how much power it takes.
121
277136
1754
04:50
Then you can look at the ratios of those numbers,
122
278890
2294
04:53
and figure out for a data center
123
281184
1529
04:54
where you don't have that information,
124
282713
1636
04:56
you can figure out,
125
284349
1092
04:57
but maybe you only have one of those,
126
285441
2187
04:59
you know the square footage,
then you could figure out
127
287628
1679
05:01
well, maybe the power is proportional.
128
289307
2323
05:03
And you can do this same thing
with a lot of different quantities,
129
291630
2320
05:05
you know, with guesses about
the total amount of storage,
130
293950
2076
05:08
the number of servers, the
number of drives per server,
131
296026
2593
05:10
and in each case using what you know
132
298619
3047
05:13
to come up with a model that narrows down
133
301666
2527
05:16
your guesses for the things that you don't know.
134
304193
2130
05:18
It's sort of circling around the
number you're trying to get.
135
306323
2265
05:20
And this is a lot of fun.
136
308588
2555
05:23
The math is not all that advanced,
137
311143
2182
05:25
and really it's like nothing more than
138
313325
3135
05:28
solving a sudoku puzzle.
139
316460
2012
05:30
So what I did, I went through all of this information,
140
318472
5309
05:35
spent a day or two researching.
141
323781
1759
05:37
And there are some things I didn't look at.
142
325540
2039
05:39
You could always look at the Google
143
327579
3043
05:42
recruitment messages that they post.
144
330622
1817
05:44
That gives you an idea of where they have people.
145
332439
1938
05:46
Sometimes, when people visit a data center,
146
334377
1611
05:47
they'll take a cell-cam photo and post it,
147
335988
1845
05:49
and they aren't supposed to,
148
337833
1490
05:51
but you can learn things about
their hardware that way.
149
339323
2813
05:54
And in fact, you can just
look at pizza delivery drivers.
150
342136
2553
05:56
Turns out, they know where all
the Google data centers are,
151
344689
2655
05:59
at least the ones that have people in them.
152
347344
2876
06:02
But I came up with my estimate,
153
350220
2185
06:04
which I felt pretty good about,
154
352405
1864
06:06
that was about 10 exabytes of data
155
354269
3249
06:09
across all of Google's operations,
156
357518
2422
06:11
and then another maybe five exabytes or so
157
359940
3432
06:15
of offline storage in tape drives,
158
363372
2045
06:17
which it turns out Google is
159
365417
1575
06:18
about the world's largest consumer of.
160
366992
2510
06:21
So I came up with this estimate, and this is
161
369502
3561
06:25
a staggering amount of data.
162
373063
1243
06:26
It's quite a bit more than any other organization
163
374306
2874
06:29
in the world has, as far as we know.
164
377180
1747
06:30
There's a couple of other contenders,
165
378927
1746
06:32
especially everyone always thinks of the NSA.
166
380673
2344
06:35
But using some of these same methods,
167
383017
1429
06:36
we can look at the NSA's data centers,
168
384446
2023
06:38
and figure out, you know, we
don't know what's going on there,
169
386469
2482
06:40
but it's pretty clear that their operation
170
388951
2090
06:43
is not the size of Google's.
171
391041
1597
06:44
Adding all of this up, I came up with
172
392638
1739
06:46
the other thing that we can answer, which is,
173
394377
1896
06:48
how many punch cards would this take?
174
396273
1985
06:50
And so a punch card can hold
175
398258
2958
06:53
about 80 characters,
176
401216
1991
06:55
and you can fit about 2,000 or so cards into a box,
177
403207
3773
06:58
and you put them in, say,
178
406980
1274
07:00
my home region of New England,
179
408254
1952
07:02
it would cover the entire region
180
410206
2340
07:04
up to a depth of a little less than five kilometers,
181
412546
3474
07:08
which is about three times deeper
182
416020
1443
07:09
than the glaciers during the last ice age
183
417463
2399
07:11
about 20,000 years ago.
184
419862
2168
07:14
So this is impractical, but I think
185
422030
2423
07:16
that's about the best answer I could come up with.
186
424453
2951
07:19
And I posted it on my website. I wrote it up.
187
427404
2138
07:21
And I didn't expect to get an answer from Google,
188
429542
3690
07:25
because of course they've been so secretive,
189
433232
1631
07:26
they didn't answer of my questions,
190
434863
1404
07:28
and so I just put it up and said,
191
436267
1507
07:29
well, I guess we'll never know.
192
437774
1523
07:31
But then a little while later
193
439297
2024
07:33
I got a message, a couple weeks later, from Google,
194
441321
2449
07:35
saying, hey, someone here has an envelope for you.
195
443770
3442
07:39
So I go and get it, open it up,
196
447212
2801
07:42
and it's punch cards. (Laughter)
197
450013
2527
07:44
Google-branded punch cards.
198
452540
3082
07:47
And on these punch cards,
there are a bunch of holes,
199
455622
3248
07:50
and I said, thank you, thank you,
200
458870
1301
07:52
okay, so what's on here?
201
460171
1519
07:53
So I get some software and start reading it,
202
461690
1630
07:55
and scan them, and it turns out
203
463320
2000
07:57
it's a puzzle.
204
465320
1369
07:58
There's a bunch of code,
205
466689
1638
08:00
and I get some friends to help,
206
468327
1254
08:01
and we crack the code, and then
inside that is another code,
207
469581
2552
08:04
and then there are some equations,
208
472133
1295
08:05
and then we solve those equations,
209
473428
1171
08:06
and then finally out pops a message from Google
210
474599
3471
08:10
which is their official answer to my article,
211
478070
3189
08:13
and it said, "No comment."
212
481259
2452
08:15
(Laughter) (Applause)
213
483711
6174
08:26
And I love calculating these kinds of things,
214
494292
3217
08:29
and it's not that I love doing the math.
215
497509
1715
08:31
I do a lot of math,
216
499224
1565
08:32
but I don't really like math for its own sake.
217
500789
3076
08:35
What I love is that it lets you take
218
503865
1950
08:37
some things that you know,
219
505815
1400
08:39
and just by moving symbols
around on a piece of paper,
220
507215
4778
08:43
find out something that you didn't know
221
511993
1697
08:45
that's very surprising.
222
513690
2180
08:47
And I have a lot of stupid questions,
223
515870
1676
08:49
and I love that math gives the power
224
517546
1803
08:51
to answer them sometimes.
225
519349
2757
08:54
And sometimes not.
226
522106
1527
08:55
This is a question I got from a reader,
227
523633
2924
08:58
an anonymous reader,
228
526557
1173
08:59
and the subject line just said, "Urgent,"
229
527730
2045
09:01
and this was the entire email:
230
529775
1929
09:03
"If people had wheels and could fly,
231
531704
2762
09:06
how would we differentiate them from airplanes?"
232
534466
2696
09:09
Urgent. (Laughter)
233
537162
2675
09:11
And I think there are some questions
234
539837
3353
09:15
that math just cannot answer.
235
543190
2670
09:17
Thank you.
236
545860
2471
09:20
(Applause)
237
548331
4350

▲Back to top

ABOUT THE SPEAKER
Randall Munroe - Cartoonist
Randall Munroe sketches elegant and illuminating explanations of the weird science and math questions that keep geeks awake at night.

Why you should listen

One of a small group of professional web cartoonists, math obsessive and chronic explainer Randall Munroe dazzles the online world (and racks up millions of monthly page views) with the meaninglessly-named (and occasionally heartbreaking) webcomic xkcd.
 
Munroe’s blog What If? specializes in cunning answers to, as the Atlantic put it, "the kinds of of wonderful and fanciful hypotheticals that might arise when the nerdily inclined get together in bars," like “How fast can you hit a speed bump while driving and live?” or “What would happen if a hair dryer with continuous power was turned on and put in an airtight 1x1x1 meter box?” As he told Math Horizons, I really enjoy solving these kinds of things, and it’s a bonus if I realize that I can put boxes around it and make it a comic."

He is also the author of the book Thing Explainer: Complicated Stuff in Simple Words

More profile about the speaker
Randall Munroe | Speaker | TED.com