counterpart in Computer Network Offense. Even intelligence and military
organizations, considered to be the best positioned to defend their own
infrastructures, struggle to keep the constant onslaught of attackers with varying
motives, skills and resources at bay.” 135 The long list of failed US government
security attempts express a strange digital logic: The more essential it is that an
organization keep a secret, the less it seems able to do so. A decade behind? That is
the gap between a flip phone and an iPhone. In the hyperspeed world of technology
it is like confronting a laser weapon with a hoplite. The losing race slips easily
enough into Donald Rumsfeld’s aheader-behinder dynamic, the one that haunts the
paradoxes of national power we face now. Are we killing more terrorists than the
madrassas are producing? Rumsfeld wondered. We can ask: Are we plugging more
machines with more layers, software and applications than we can protect? Are we
making more bugs than we’re patching? (Yes and yes.) “Attackers are not like
natural catastrophes,” Lindner and Gaycken write. “They can analyze their targets.”
Bratus, a math genius who turned to computer science out of curiosity and now
teaches at Dartmouth, has spent a fair amount of time trying to understand just
what happens when a computer or a network is exploited by a hacker, or “pwned” in
the funny idiom of Warez Dudes language. (The phrase means to take control, or to
“own” a system. The spelling is an artifact of an overenthusiastic video-game death
match gloat, in which one player killed another and in his rush to celebrate typed
something along the lines of “I pwned you!” The mis-typing lives today: The highest
award in hacking is known as The Pwnie.) Bratus calls the resulting, pwned device a
134 “Exploit engineers”:Sergei Bratus, et al. “Chapter 13: ‘Weird Machine’ Patterns”
in C. Blackwell and H. Zhu (eds.), Cyberpatterns, Springer International Publishing
Switzerland 2014, p. 13
135 Even intelligence: Felix “F.X” Lidner and Sandro Gaycken, “Back to Basics:
Beyond Network Hygiene”, in Best Practices in Computer Network Defense: Incident
Detection and Response, M.E. Hathaway (Ed.) IOS Press, 2014
98
“Weird Machine”: Computers or sensors or network webs silently made to do what
is not intended. Made weird. Hacking is, after all, a kind of perverse programming. It
involves slipping inside a target machine, and then driving it to do things it wasn’t
intended to do, by giving it instructions its designers never knew it might receive. 136
The process of developing and using computer bugs, Bratus found, is not unlike the
most sophisticated software research. Hacker follow careful patterns. The best of
them really conceive of whole systems in the way the finest data architects might.
They look for particular designs, weaponize their code with a delicate elegance and
aim relentlessly at total control. A normal machine does what you tell it. A weird
machine does what someone else commands it to do.
How is such a system born? Well, a potential software hole of the sort that produces
a “Weird Machine” might be as simple as a failure to secure computer code after it is
compiled – sort of like not locking the door on your house after you leave – or a
programming oversight that means a machine can’t handle unexpected inputs. Take
the technique of “fuzzing”, for example, a famously effective way to turn a normal
machine into a weird one. The process involves confusing a digital security system
by throwing unexpected data into normal, apparently safe-looking procedures like
logging into a mail system or transferring money by wire. Think of all the
“username” and “password” forms you see when you’re on the Internet. In a fuzzing
attack, instead of placing a legitimate user name or email address in a registration
field, hackers might add some unexpected characters known to cause a system to
cough up a confused response. If you type in, joe@user.com!’ instead of
joe@user.com as the machine expects, the !’ at the end of the address can baffle and
stall a mis-programmed device. In some cases, that hiccup opens a vulnerability. A
proficient programmer can then order the dazed computer, for instance, to open a
door to the root of the system. It’s as if you could walk up to the teller at your local
bank and shout “Xhsuhgnnsh!!” at her when she asks how you are doing – and in her
confusion she lets you into the safe. You’ve made a weird machine of your bank.
System designers in later generations have become much more sophisticated in
trying to avoid such problems, not least because they’ve so often fingered the
embarrassing or costly aftermath of these kinds of holes in their own code. “You do
not understand how your program really works until it has been exploited,” Bratus
has said, a sentiment that hints at the stomach-lurching moment many coders and
their suddenly victimized users have now had. 137 You don’t understand yourself until
you’ve been pwned. The odds that the endless possible glitches can ever be
completely patched is honestly zero. Hackers continue to use classic exploits like
136 It involves: Julian Bangert, Bratus et al. p2 “The Page-Fault Weird Machine:
Lessons in Instruction-less Computing”, Presented as part of the 7th USENIX
Workshop on Offensive Technologies, (Washington, D.C., 2013) available on
www.usenix.org
137 “You do not really understand”: Rebecca Shapiro, Sergey Bratus, Sean W.
Smith, “’Weird Machines’ in ELF: A Spotlight on the Underappreciated Metadata”
paper delivered 7th USENIX Workshop on Offensive Technologies, (Washington, D.C.,
2013) available on www.usenix.org
99
fuzzing, back-doors, or rootkits – and to develop new, more intricate ways to steal a
machine’s mind. This race for mastery is a sprint, one that heads ever deeper into
the depths of a vulnerable system. The closer you are to the very core of a computer
program or a network, the more control you have. Real mastery of the heart of a
system would be like having a spy win the Presidency, turning the whole US
government into a “Weird Machine”. That prize of immediate, high-level and totally
“trusted” access remains the Warez Dude gold standard. 138
The most dangerous – and therefore the most alluringly valuable – of these sorts of
attacks, are known as “zero-day” exploits. The danger they represent only becomes
apparent at some awful instant, “Day Zero”, when they are revealed to have been
running wild inside some hapless network or machine. That first moment in the
knowledge of the bug is like day zero in a cancer diagnosis, and it begins an
immediate race to find and deliver a cure. Such completely unknown, secret
vulnerabilities are fissures in the walls of computers that their manufacturers,
system engineers, and security experts usually don’t realize are there. The dream of
hackers and spies and greedy Warez Dudes is a version of this trick called an
“Advanced Persistent Threat” – hidden, back-door access to a machine that endures
even for years, through upgrades and security checks and system cleanups, all the
while forcing the now-weird computer to do things its user won’t even be aware of:
Send a copy of every keystroke to another machine, for instance, or serve as a
robotic launching pad for attacks on other machines. All while acting like a perfectly
normal machine.
The best of the zero day exploits are based not on the idea of sneaking malicious
software onto machines so much as on taking existing, trusted code and finding tiny
holes that can be blasted into giant, insecure data tunnels. These tools for
manipulation are digital opiates, in a sense, for connected nervous systems. Such
attacks often rely on errors already knit into computer systems, or innocent
seeming features that can be made dangerous. Every computer and software
designer knows their systems are vulnerable. Mathematicians have proven you can
never be absolutely sure a connected machine is safe. A mobile phone, for instance,
contains nearly 30 million lines of code. The systems that run massive cloud
computing basins like those at Google or Amazon are even larger, updated every
day, and have to cope with unprecedented flows of data at very high speeds. Even
expert programmers will leave four or five errors in every million lines of code. 139
138 That prize: Sergey Bratus, Julian Bangert, Alexandar Gabrovsky, Anna Shubina,
Daniel Bilar, and Michael E. Locasto, “Composition Patterns of Hacking”, Proceedings
of Cyberpatterns (2012)
139 Even expert progammers: For a good explanation of how a machine’s code can
be turned against itself, see Sergey Bratus, Michael E. Locasto, Meredith l. Patterson,
Len Sassaman, and Anna Shubina, “Exploit Programming: From Buffer Overflows to
‘Weird Machines’ and Theory of Computation.” ;login December 2011, p. 13-21. The
piece is co-authored by and in honor of Len Sassman, one of the leading thinkers of
the Language Security or LangSec movement who died in 2011. Many of his talks,
still available online, reflect an unusually powerful mix of philosophical and
100
Software and hardware manufacturers usually struggle to keep such exploits secret
until they can deliver a fix, but this doesn’t always work. Secrets get out. And,
anyhow, even once a patch is developed, it can take weeks or months before it’s
widely installed. It’s not uncommon, therefore, that within hours of the
announcement of a newly found zero day, attacks using that method explode around
the net. Thousands of hackers try to take advantage of the vulnerability, to kick at
the defensive corners of systems while they are down for repair or restart – or
simply left vulnerable by slower-witted system administrators who don’t yet know
that it is now open hunting season on a particular bit of code. Heartbleed, a “zero
day” that permitted hackers to slip into your computer via holes in website and
browser security, was disclosed to the world on April 7, 2014 – more than two years
after it had apparently been put in place because of a programming error.
Accidentally? By an overworked engineer? Deliberately? By some state security
agency? In any event, two days after it was announced and long before it had been
fully patched, attacks using the method grew from a few dozen per hour to millions
as hackers tried to suck data from unsecured networks. 140 The exponential power of
a connected system is as apparent in sickness as in health.
4.
In recent years, hacking has moved deeper still, beyond the level of software and
USB drives and into the very atomic level of computers, the places where the
electrons that make up bits and bytes float. The technical elegance of these microlevel
hacks has been, often, breathtaking – exploits that look like Wagnerian operas
compared to the Cap’n Crunch’s thin, reedy weird-machine whistle. As companies
like Intel and AMD began packing more memory cells on silicon wafers, for example,
they noticed magnetic interference flowing across the surface of their chips like
waves. Electrical signals, recall, have a magnetic element, so more tiny digital cells,
closer together, is like a bowl of interacting magnets. Physics would have predicted
such a result. In 2014 security researchers Mark Seaborn and Thomas Dullien, who
worked at Google, discovered that they could use the magnetic vibrations on two
parallel rows of memory chips to flip the electrical state of a third row – sort of like
using a magnet under a table to move a paperclip around – in a way that the system
might never notice 141 . This permitted them to reach “off limits”, super secure areas
of the machine’s memory where they could do what they wanted. They called the
break, “Rowhammer” and it represented an ideal and essentially unfixable hole that
affected nearly every small chipset made for a half a decade. They published the
technical consideration about modern computing systems. See also Rebecca
Shapiro, Sergey Bratus and Sean W. Smith, ““Weird Machines” in ELF: A Spotlight on
the Underappreciated Metadata”, paper published online by Bratus.
140 In any event: Leyla Bilge, Tudor Dumitras, “Before We Knew It: An Empirical
Study of Zero-Day Attacks In the Real World,“ Paper presented at ACM CCS ’12, Oct
16-18, 2012, p 10
141 In 2014: Mark Seaborn and Thomas Dullien, “Exploiting the DRAM rowhammer
bug to gain kernel privileges”, March 18. 2015, Google Project Zero blog
101
result immediately as a warning to possibly affected victims, but the exploit
hummed at such a basic level of the system that it proved impossible to fully patch.
It was like trying to patch physics.
Connection changes the nature of an object. This has a particular impact on the
security of our systems, and by extension, our own safety. Computer researcher
Nathaniel Husted has described a world of “emergent vulnerabilities” – wormholes
in software and hardware, communications or finance, that pop up in the connected
universe unbidden and unplanned. “The fundamental aspect of emergent
vulnerabilities and attacks,” Husted writes, “are that they appear benign until
certain criticality conditions are met, at which point they become malignant.” 142
These risks we don’t want sit right alongside all the things we do want from
connectivity. In fact – and this is Hulsted’s point – they are the things we want from
connectivity, perverted into danger.
Paul Baran would have been impressed to see just how right he was, how
connections between us now are like irresistible gravity waves – and how gravity
always wins. In 2015, for instance, Israeli security developed an astonishing hack
that proved that nearly spiritual claim that all objects were linked by connection–
and demonstrated the way that slippery, hungry attacks can breach even the safestlooking
arrangements. “It has been assumed that the physical separation of
computers provides a reliable level of security,” Mordechai Guri and his team wrote
in a paper describing how they had used one isolated machine to infect another.
Physical separation is, in fact, one of the cardinal rules of safe computing, a kind of
lemma to join Robert Morris Sr,’s “don’t connect” rule of network safety: Two
machines, unconnected by a network, should not be able to affect each other.
Imagine I put one kid with a flu in one classroom and a schoolmate of his in another
building. The second kid should remain healthy. The Tel Aviv research team wanted
to challenge this. They first placed two computers side by side on a desk,
unconnected to each other by any wire or network. One machine was connected to
the Internet. The other was completely isolated – it was “air gapped”, like the
healthy kid in the distant building, a sort of digital “boy in the bubble,” in touch with
only the air around it. Then, the researchers began their Houdini trick: Look! Watch
us corrupt this completely unconnected machine! Running a set of programs on the
network-connected machine, the Israeli team was able to warm the processor board
of that computer like a revving car engine, eventually making it hot enough that the
temperature changes were detected by sensors inside the secure, allegedly
impregnable “boy in the bubble” machine sitting nearby. The heat wave triggered a
fan system inside the clean machine, which in turn activated a piece of pre-installed
malware that let the hot machine pwn the bubble machine through temperature
variations. In a video demonstration of the exploit, you can watch the infecting
machine glow ever hotter, issuing “thermal pings” as it sweats and then infects its
142 “The fundamental aspect”: Nathaniel Husted “Analysis techniques for exploring
emergent vulnerabilities and attacks on mobile devices” PhD. Thesis (available
online from Indiana University, 2013) p. v
102
safe, “unconnected” neighbor. 143 The heat transfer had a simple message: Nothing is
safe.
Why put such effort, worthy of the deepest physics problems, into the challenge of
sneaking into a cellphone undetected? Well, for Seaborn and Dullien, the drive was
part of a “discover and publish” effort to keep the overall system clean. It is better to
hack, discover and patch than to be hacked, and remain undiscovered. But these
“good guy” engineers are racing against different, equivalently sophisticated, lessdecently
inspired teams. The development and sale of zero-day bugs is, after all, a
business. Modern versions of Cap’n Crunch whistles crack access to some of the
most essential financial, political and security data stores on the planet. As the
power and value of hacking targets has increased, so has the price of the exploits.
Public “zero day markets” sponsored by companies like Google and Microsoft pay
hundreds of thousands of dollars to researchers who discover holes in their
systems. “Better to find them ourselves,” the thinking goes. Though that does not
always make the embarrassment less acute when holes are spotted. At one of the
most carefully watched public hacking competitions in early 2015, for instance, a
skinny, smiling South Korean named Jung Hoon Lee took home $225,000 in prize
money by pwning a series of some of the most important, common programs on the
planet, Apple’s web browser Safari and Google’s Chrome among them. These
systems had been constructed at the cost of hundreds of millions of dollars. They‘d
been assembled under the gaze of some of the best PhD-led computer scientists in
the world. Jung Hoon Lee’s exploits ran through their complete defenses in less than
a minute. 144
As good and fast as someone like Lee might be, he’s nothing compared to what the
best hackers do. They don’t work in public or compete in hotel ballrooms. They
don’t brag. And they develop ideas that make $225,000 look like a bargain. These
successors to the Warez Dudes work for cybercriminal billionaires, for intelligence
agencies, and even (often) just for themselves. They help find and deploy the sorts of
really deep system exploits that enable brazen cyber thefts of millions of pieces of
personal data or attacks like the Stuxnet virus, which caused thousands of Iranian
nuclear centrifuges to vibrate themselves apart. And they do still more: Most of the
attacks we’ve talked about so far occur in installed, running boxes. But the
companies that make those boxes oversee a whole, vulnerable process of building
and testing and designing and installing them. And it’s here, with billion dollar
budgets at work, that some exploit teams make and leave vulnerabilities that they
can later, ruthlessly exploit. Every step of that gestation – from sneaking secrets into
early code bases to intercepting and rewiring routers as they ship overseas – is now
an opportunity for secret control. Or for unanticipated risk, for “emergent
143 In a video: For a description of this exploit see Mordechai Guri, Matan Monitz,
Yisroel Mirski, Yuval Elovici, “BitWhisper: Covert Signaling Channel between Air-
Gapped Computers using Thermal Manipulations” (2015) available on
arXiv:1503.07919 [cs.CR]
144 Jung Hoon Lee’s exploits: “Chrome, Firefox, Explorer, Safari Were All hacked at
Pwn2Own Contest”, PC World via IDG News Service Mar 20, 2015
103
misbehavior” that defies simple analysis. You can’t predict where you might be
attacked by merely looking at the possible holes in each piece. Rather, it’s the whole
system that breeds risks. It acts in ways that the designer could not have predicted
in advance. “Clearly the system itself is misbehaving,” the researcher Jeffrey Mogul
writes of his study of various cases where networks are cracked in this fashion.
“However, none of the components have failed per-se.” 145
The complexity of the systems themselves has been, not surprisingly, mimicked in
the design of hacking attacks. What was once done by a single Warez Dude is now
handled with division of labor, technical specialization and intensive pre-attack
research. Every innovation in “righteous malware” is quickly copied and used in
dangerous attack tools. The clever modular design of Stuxnet, for instance, was
studied by criminals and was found years later still echoing in weapons aimed at
banks, credit card companies and health insurance firms. “We are not experts in
military history, doctrine, or philosophy,” cybersecurity researchers Stephen Cobb
and Andrew Lee have written, “so we are unaware of the correct word for the
following category of weapons: the ones you deliver to your enemies in re-usable
form.” Cyberattack systems can be dangerous not least because they boomerang.
They are delivered intact, primed for re-use to enemies who may choose to bounce
them back at your banks, hospitals and electrical grids. “Righteous malware is
unique,” Cobb and Lee conclude. “You are giving away your weapons, tactics and
designs simply by using them.” 146
It’s not only American services hunting and using such backdoor keys and battering
rams, of course; not only the NSA that sees its viruses retooled and reused.
Computer security researchers describe opening up the laptops of unwary business
travelers and finding the machines blasted inside by malware and other technical
cancers, carefully planted by a half-dozen intelligence agencies and criminal
organizations. It’s like discovering a closet full of spies in your house, each being
careful not to step on the other’s toes as they watch and listen to your life. Why is my
computer so slow, a government official in a Eurasian capital might ask. It is because
it has been simultaneously pwned by Americans, Russians, Israelis, Chinese, and
maybe a local Mafioso or two – and their code is not running smoothly.
A couple of years ago I had a naïve moment when I thought, perhaps, it would be
possible and in everyone’s interest to go back to those simpler, innocent Hacktic
days, when information about vulnerabilities was widely shared and easily
discussed – and holes were quickly patched as a result. I was thinking about the
problem of cybertension between the US and China and suggested applying an
145 “However”: Jeffrey C. Mogul, “Emergent (Mis)behavior vs. Complex Software
Systems”, HP Labs Research Papers, 2006, HPL-2006-2
146 “We are not experts”: Stephen Cobb and Andrew Lee, “Malware is Called
Malicious for a Reason: The Risks of Weaponizing Code” in P. Brangetto, M.
Maybaum, J. Stinissen eds., 6 th Annual Conference on Cyber Conflict (NATO
Publications, 2014) 71-82
104
important idea of the scientist Dan Geer to the ever more fraught relations. 147
Perhaps the US and China could work together to buy up and then publish all the
zero days as they emerged, I thought. 148 Instead of the dangerous code falling into
the hands of cybercriminals, Mafiosi and terrorists, the two countries could lead an
effort to jointly buy any new exploit for five times what anyone else would pay – and
then immediately publish what they had bought and the necessary patch. This
would make the network safer for everyone. I should have known better.
The US and China and other nations were (and are) buying world-class zero days.
But they were never going to publish them. They were buying them to use.
Sometimes against each other. Sometimes, unnervingly, against their own citizens.
And they needed to keep on buying and developing and hiding such tools on an
exhausting, never-ending security treadmill because, unlike traditional weapons,
which could be stockpiled to use whenever they were needed in the future, the holes
of the most valuable zero-days might be patched at any moment, making a once
devastating bit of malware instantly useless. And, as the systems matured and
accelerated, this meant that they had to run ever faster to keep up. Which further
reduced their incentive to “buy and publish” what they did have. Little surprise then
that all around the IT universe in recent years, the incidence of reporting dangerous
bugs has been declining, even as we know the number of known security holes is
certainly growing. 149
Grab the five nearest electronic devices near you and you can be pretty sure each is
vulnerable; which of course means you are vulnerable too. Not merely to the loss of
your secrets, but also to perversion and control. This is the cold truth: that old
hacker ethos, the one spread out so warmly on Amsterdam grass 20 years ago, the
be liberal in what you accept frontier society instinct, is dead. Weird machines and
normal machines, weird networks and normal ones, people made weird by
technological manipulation and those who have note – they will all, inevitably, live
side by side. The more essential machines become to our connected lives, the more
avidly weird and hacked the networks will become. The incentive to whistle up
control of the systems, and control of those of enmeshed in them, is the only thing
that seems to grow faster than the system itself.
This vulnerability of connected systems is an important mile marker in our route
towards understanding some principles of power in a connected age. “Read over
and over again the campaigns of Alexander, Hannibal, Caesar, Gustavus, Turenne,
Eugene and Frederick,” Napoleon wrote once. “Make them your models. This is the
5.
147 I was thinking: Dan Geer, “Cybersecurity as Realpolitik”, speech delivered at
BlackHat, August 6,2014
148 Perhaps: Joshua Cooper Ramo “Talking Cyberthreat with China”, International
Herald Tribune, July 10, 2013
149 Little surprise: Linder Gayten Back to Basics p. 58
105
only way to become a great general and master the secrets of war.” 150 I sometimes
feel the same reading over stories of zero-day attacks, clever hacks like
“rowhammer” or the Tel Aviv heat hack. You can distill from each tale of a broken,
once-secure systems an essential principle: The hackers rush always, relentlessly at
the central core of a system. They aim to make it weird, to manipulate it madly from
the inside out. Network power doesn’t merely come from that 10 million device-perday
spread of global connectivity, after all, it also comes from incredible
concentration of power inside certain systems we all rely on: Chips, data bases,
centralized and gatekept platforms. Control of such hubs and roots of our world can
influence everything; little wonder they are such an appealing target. “The
conventional belief that all nuclear systems are ‘air gapped’ is a myth,” the Russian
security researcher Eugene Kaspersky has warned. The result: “There are three
types of people: Scared to death. Opportunists. Don’t care.” 151
This sense that the systems are so vulnerable if you can get to their hearts is what
lures hackers ever deeper, into the code kernels where the most basic instructions
are decided. That they can often make machines weird by using the device’s own
code against them, like some sort of autoimmune disease, is only a marker of the
particular perversity of the problem here. Security researchers call such holes
“vulnerabilities” in a system, but of course they are much more than weak spots.
They are potentially fatal. In a way, the hot rush to touch and tickle and maliciously
use these already waiting cancers reveals to us the essential Seventh Sense secret of
the Warez Dudes: Connection makes an object vulnerable, yes; but it can also reveal
the possibility of total control, of the fundamental root mastery of a connected
system. Such a hole, when it is exposed by connection and then corruption, can be
complete in the scope of its damage, devastating. Lord Acton’s famous line that
“Absolute power corrupts absolutely,” twists in this age to something like “Absolute
access corrupts absolutely.” Connection makes total exploitation, total control,
possible.
Every new generation of connected technologies is breeding essential black boxes,
complex (not merely complicated!) containers filled with algorithmic levers and
code tools for digital work that can be understood by only a few people, and
exploited and used effectively – for good or ill – by a still smaller group. “The greater
the dependence on a technology the greater the need to study and expose its inner
workings,” one group of radical digital activists has argued in The Critical
Engineering Manifesto. 152 They mean that as we turn our safety, freedom, and health
over to a world of devices and their makers, we must know what goes on inside the
very heart of such systems. It’s not merely that everything is connected now; it’s
150 “Read over and over”: Napoleon, “Maxims” from Thomas Raphael Phillip, ed.,
Roots of Strategy: The 5 Greatest Military Classics of All Time, (Stackpole Books 1985)
p 432
151 “There are three types:” Eugene Kaspersky Talk at the Press Club in Canberra,
Australia (2013)
152 “The greater the dependence”: “The Critical Engineering Manifesto,” The
Critical Engineering Working Group, Berlin, October 2011-2014. Available online
106
also that everything is monitored. Remebered. Studied. The Warez Dudes’ drive to
get ever closer to the core, to perform even that atomic level hacking, tells us
something about just how much power is locked up in those central cores where
this information accumulates. The 2600 hz whistle was, it seems, only the first of an
endless series of battles for control of the roots and trunklines of modern power.
All around us today, huge power accumulates to certain irreplaceable cores. We
know this is a problem of connected age design: Giant search engines, certain
algorithms, database or communications protocols overmaster us because they can
gather so much data, so fast, and process it with unique fidelity. What makes a city?
urban scholars often ask. We might wonder: What makes a platform for network
power? The answer to both questions is the same: Density. 153 If the first cities of
Aztecs or Mesopotamiaman civilization differed from early tribal clusters because of
their density, the same is true for our first platforms of instant connection. Facebook
is denser than AOL ever was. More people, more data, thicker connections. Future
platforms will be denser still. And if cities and density were once sadly
unanticipated accelerants to plague, poverty and revolution, we should be aware of
the risks of our own tight-clustered centers of dense connection. The security of
these cores that link us to each other and our essential data – when jacked by
hackers, by companies, or even by fast algorithms we don’t understand – is
important not merely because of the possibility of total control a breach might
represent, but because they show us the very fact of such totalizing control exists.
To infect, surprise, sicken – all this is alluringly possible and dreamable for anyone
with a hunger for mastery. Imagine if you knew your government could be switched
instantly and invisibly to malice. (Or, to effectiveness!) Or picture a nation of
connected citizens wired for flash-started nationalism and hate. Such a possibility
exists on linked systems. This potential for total, weird control of the cores – and
thus total control of anyone connected to them – should force us to wonder a bit.
Every evil thing beats in those central nodes: The power to manipulate, to master, to
destroy even. With such access I can change what you know about the world, how
you vote, where your money sits, what you remember, how soon we spot (or don’t)
a slipped knot in your DNA.
“Just like every drinking binge ends at vodka, so every hacking session ends at
kernel.org” Thomas Dullien, the mathematician and good-guy hacker who won a
Pwnie in 2015 for lifetime security achievement, observed once 154 . So much power
in a connected system lies at its root: Kernel.org, for instance, is the reference copy
for LINUX computer code that powers most digital machines on earth, sort of like
the original DNA of the net. To manipulate Kernel.org would be to reach into the
very spine of the Internet. If the aim is control, if it is to find and exploit the most
fundamental of cracks in the surface of the black boxes, to get even deeper inside,
153 What makes a city?: Colin McFarlane, “The geographies of urban density:
topology, politics and the city.” Progress in Human Geography (2015) p. 2
154 “Just like every drinking binge”: Halvar Flake/Thomas Dullien, “Why Johnny
Can’t Tell If He Is Compromised,” speech delievered at Area41 Conference, (Zurich,
June 2014)
107
then inevitably kernel.org or it’s equivalent in any system is the ultimate target.
Hackers might start with beer (your phone), a few glasses of wine (your office email
system), but what they really want is full blotto (kernel.org). Such central black
boxes exist in any linked system and they represent, at once, both the greatest
accomplishments of our most masterful systems designers and the point at which
other, as masterful digital machinists direct their most relentless attacks.
Dullien saw something else too as he considered the work of hackers. Comparing
system cracking to drinking wasn’t an accidental, funny aside for him. Hacking was
almost a kind of addiction. It became a chase after a bigger and bigger high, which in
computing terms meant a race to compromise as many machines as possible. Rapid
escalation, a loss of self-control, the need for more and more – all these are the
hallmarks of the best widespread attacks, which aim to expand the “compromise
boundary” until every machine has been made sick. This is why stealing source
code, the original instructions that lay behind any computer program, are such a
prize for Warez Dudes. Source code is the DNA of the black box, in a sense, it can be
used break into other machines to, well, steal still more source code. This looked, to
Dullien, an awful lot like addiction. And it’s not just lone teen hackers looking for a
dopamine jump who were chasing machines with a blind addicts urgency. It infects
governments too. “Surprising realization (at least for me) after the Snowden leaks,”
Dullien observed. “Hacking is so addictive that entire organizations can become
addicts and show addict-style behavior.” 155 This was the NSA or GHSQ or who knew
what other intelligence service, for instance, in a feral hunger for more and more
and eventually all data. And the ultimate high? Imagine if you could crack the CA, the
Certificate Authority that provides proof that anyone working on a network is
trustworthy. CA “trustworthiness guarantees” are keys that permit access to any
computer, phone and network. The whole connected world depends on CAs to know
if software is safe. To control the CA would be a dream for an addicted organization,
like a set of keys to the local pharmacy. No lock would really serve much of a
purpose. You could touch the beating heart of any machine on earth. The ultimate
black box. We should all hope the CA, at least, always remains pristine in this
dangerous world.
It was hacked in the summer of 2011.
6.
Dullien’s observation that “every hacking binge ends at Kernel.org” touches
something far deeper than just the world of hackers. Sure, that urgent drive to get to
the very core of systems is a very specific, unquenchable digital thirst. It’s where the
most power is, as we’ve seen. There’s a rush in getting there: When brain
researchers found a spike in dopamine levels of university students engaged in
password hacking, that was hardly a surprise to anyone who’s ever written or used
155 ”Surprising realization”: Thomas Dullien, “Offensive Work and Addiction”,
keynote presentation delivered at ISACA Nordic Conference (2014)
108
an exploit. 156 But the real drive to get inside is about more than an adrenaline rush.
Remember that network power, the power that we’re trying to figure out how to
bend and shape for our own security, exists in a kind of dynamic tension. It’s like a
stretched, taut fabric spread between concentrated cores and billions of connected
users or devices. The logic of spreading and distributed power, the force that makes
the network bigger, is driven by Baran’s principle of open design, and by our own
hunger for communications and connection and cool new devices. But there is
another side to this tension.
In a sense, over the years, a whole set of hot, infectious pressures descended on the
network of values and friendships and easy cooperation of the Hack-Tic days. “Be
generous in what you receive,” had let the networks of our age grow at an incredible
pace, but at the price of vulnerability, of commercial ambition, and of an eerie
technological lemma that what made the systems faster and stronger might also kill
them. A change in culture of the digital elite, naturally, followed. The brutal,
inarguable, profitable demands of power and politics had cracked apart the unique
social webs of the HacTic era. I did not like watching this sad evolution; none of us
did, but anyhow it has produced the world in which our new sensibility will have to
operate. The openness that we loved and craved in so many areas of life, from our
minds to our markets, had now become a liability. “I remember what the Internet
was like before it was being watched and there had never been anything in the
history of man that is like it,” Edward Snowden once observed, nostalgic for the
datascape he saw melt away during his time at the NSA. 157 I realize now that there
is a whole new generation of young programmers that won’t ever know that original
generous ethos of a place like Hacktic, a fresh cohort of the digital age that operates
at levels of technical mastery far beyond anything we might have imagined in the
Citicorp Tower basement 20 years ago. They will confront endless battles to get
inside and exploit and make “weird” the cores of network power. They will know
and design and manage instead a world of gates, built for protection. Their instincts
will be for opacity and control, not openness and generosity. Invariably this shift
will affect the design of the systems this new generation builds which will, in turn,
affect all of us.
The black boxes and the crackers are in a dance, now, a sort of dangerous
evolutionary waltz that offers a foretaste of what you and I will face as we consider
the problems of attack and defense and strategy in a networked age 158 . It reminds
156 There’s a rush: Wael Khalifa, Kenneth Revett and Abdel-Badeeh Salem “In the
Hacker’s Eye: The Neurophysiology of a Computer Hacker”, in Global Security, Safety
and Sustainability & e-Democracy Volume 99 of the series Lecture Notes of the
Institute for Computer Sciences, Social Informatics and Telecommunications
Engineering, (Berlin: Springer), 112-119
157 “I remember what the Internet was like”: Snowden interview in CITIZENFOUR,
Laura Poitras (2015)
158 The black boxes: Nikos Virvilis, Dimitris Gritzalis, Theodoros Apostolopoulos,
“Trusted Computing vs. Advanced Persistent Threats: Can a defender win this
game?” 2013 IEEE 10th International Conference on Ubiquitous Intelligence &
109
me of an old puzzle of Chinese history: Why were the very strongest dynasties – the
Han, the Ming, the Tang – always confronted by the best organized, most deadly
rebels? The answer was rooted in a violent, greedy tango of the development of each
side. The better a dynasty defended its farmers or its trade or its annual rice harvest,
the more the rebels from the steppe had to become strategic and unified and
powerful. When it was easy to pick off single farmers, then the rebels had no need to
be well organized. They could be a bit lazy. Could snatch a year’s undefended
harvest in an afternoon. But the stronger the empire, the stronger the rebels had to
become 159 . Amsterdam in 1994 was a bunch of lone, weak, unprepared farmers.
Twenty years later we have a digital empire, and the attackers it deserves are
appearing, refined and evolved by this same competitive logic. They are better
organized, smart, intent on getting to the heart of the black boxes of power. And the
more we defend against them? The better they will become.
That our most essential systems are vulnerable to loss of control is a chilly feeling. It
is a reminder of the power of the people who know how to crash and manipulate –
or build and operate – parts of our world most of us barely understand. It’s like
discovering someone could take over your lungs or your heart. We don’t understand
those devices either, most of us. But we depend on them entirely. Some of these
hackers are moved to mischief by technical beauty. Some by the giddy smashing
rush of breaking in, of touching the core. Others by greed or patriotism or by secret,
zealous, unlawful obsessions. What the technically best of this group share,
however, is a pressing desire to get as close as possible to the kernels where
inarguable and even invisible code decisions are made, where digital DNA is printed
in a sense, and where a total mastery of the binary guts of the system is possible.
That Cap’n Crunch thrill, the dream of whistling up control over the thick, helpless
trunk of a network, that remains the dream. Remember Conway’s Law: The design
and activity and control of a network redounds on, even determines the real world.
If the whole network is, in a sense, filled with holes, if it contains inherently the
possibility of being turned into a “weird machine” – what does that mean for the real
world?
The Seventh Sense insight of this chapter, the key feeling, is that all the systems we
rely on, that we think we control – financial or political or digital – can all be made
weird and pwned by forces we cannot see and struggle to stop. Our markets, our
elections, our knowledge – all of these, dependent on linked systems themselves,
can become weird. We can no longer regard them as certain and harmless. The
hacker’s drive to get to the kernels is not merely an information technology
problem. It’s a larger statement about networks, about the way that power and
danger are still and always one and the same. The desire to hack the cores of our
world is a marker of just how profoundly, even secretly influential the kernels have
become – in ways we’re just now beginning to understand. Bratus was right, that we
don’t really understand any system until it’s been exploited, pwned. This is as true
Computing and 2013 IEEE 10th International Conference on Autonomic & Trusted
Computing
159 But the stronger the empire: See Turchin, p. 3
110
for our minds as it is for our markets. The Seventh Sense feeds a bit on the trail of
hackers; it is a way to tune our protective instincts. The cracks they find and use do
tell us an awful lot. But not everything. For that we need to look beyond the people
attacking. What we really need is a feel for the temprament of the fresh group of
innovators and investors and technologists who run and own the dark cores of
power now – and who, before long, may also run and pwn each of us as a result.
111
Chapter Seven:
The New Caste
In which we meet a powerful group defined, enabled and enriched by their mastery of
the Seventh Sense.
1.
Looking back over several hundred years of European history, the Oxford professor
David Priestland found that the movement of power might be scored by reviewing
the alliances and hatreds and hopes of three distinct, interacting groups. He called
them “castes”: merchants, soldiers and sages. By merchants, Priestland meant the
bankers, traders and industrialists whose capital, goods and political power bent
Europe’s once-feudal economy into something modern and industrial. 160 The Medici,
Dutch coffee traders, Scottish cotton barons. By sages, he had in mind the
churchmen and later the technocrats of various empires, the men who helped birth
and then manage the problems of an Enlightened, urban social order: Locke,
Bismarck, Disraeli. And by soldiers he had in mind both the great aristocratic
warrior classes of Europe and upstart, genius figures like Napoleon or Wellington –
men who handled martial force with the fresh, surpasssing brilliance of a paintbrush
or chisel, not an instrument of mass murder.
The aligned, shifting interests of these three castes, Priestland wrote, were like
gears of sorts, each offering special leverage, meshing together to drive nations to
great power. Mix the influence of France’s sage-bureaucrats with her artful soldiers
and you get the French Imperial period. Marry the interests of Britain’s shrewd 17 th
Century trading bankers with her martially inclined sailors and globe-spanning
Victorian dominance results. Today, of course, the merchants and soldiers and sages
of our era are also at work. They sit in sovereign wealth funds, wired situation
rooms and madrasas, churches and research labs. The force of America’s merchants
and financiers, bolstered by Washington’s security caste, defines much of American
power. No other nation could, at the moment, comprehensively replace what the
country does with such breadth, intensity and speed. And now, all around the globe,
we’re seeing the emergence of what we might think of as a New Caste joining the
merchants, soldiers and sages. This is the infotech caste.
What was going on in that field in Amsterdam back in the summer of 1993 was
nothing less than the birth of the first figures of a connected technological era, of a
new elite. This caste I have in mind is defined by their personal proximity and
fingertip feel for the linked machines that drive so much of our world. They
represent a tiny fraction of our population, but operate with non-linear, massive
levels of influence. This New Caste clusters, in ever tighter circles of intimacy,
around the systems and networks we depend on. They are building new connected
160 Looking back: David Priestland, Merchant, Soldier, Sage: A New History of Power,
(New York: Penguin, 2013)
112
financial systems. Fresh AI designs. Protocols and platforms for secure
communication. Maybe one million people can write object-oriented code at a high
level. A hundred thousand of them can shape that code into some sort of innovative
data structure. A few thousand might be able to use it to access and manage a large
data center. But get down to the couple of dozen who know how Google or Intel or
bitcoin really work, the group who can make machines seem to think, who know
and use backdoors at that atomic level of hacking – well, then you have a tight,
powerful elite. They are called system designers, algorithmic traders, growth
hackers or any of a dozen hazy and somewhat unnamable lines of work that fuse
network mastery with economic, political or social power. If connection changes the
nature of an object it also elevates, to a level of rare power and influence, those who
control that connection.
Through the networks and black boxes they control, the group of transcendent
talents working on search algorithms, data management and machine learning
touch, at any one instant, more parts of our lives than any group of elites ever has.
That many of them are billionaires as a result should hardly be a surprise. This is a
caste marked by constant one upsmanship, by endless and compulsive innovation,
by a cold and ceaseless fear of obsolescence. They are marked, of course, by a full
expression of the Seventh Sense. Their every instinct for starting, financing, growing
and using new firms or technologies or data-spinning biological tools reflects
complete confidence in network power. If the leading figures of the Enlightenment
shared a certainty near to faith that reason would unlock nearly any puzzle of mind
or politics, the New Caste shares a certainty too, that connection can produce a new,
better order. This caste battles with each other. They struggle for dominance over
markets and – more profoundly – for control of the lucrative systems they have
built.
My point in this chapter is simple. It is to introduce this new and powerful caste and
see what their successes and failures might tell us about the world we now all share.
Even now, however, I think we can make this sobering judgment: Whatever nation
cultivates and trains legions of this New Caste best will rule the future. Think of the
legendary Xerox PARC research lab, which in the 1970s produced not just some of
the best early members of the New Caste – like Alan Kay or John Seeley Brown– but
also a series of fundamental inventions that generated trillions of dollars of value 161 .
The mouse. Laser printing. A graphical user interface. PARC was arguably the most
economically significant small group in human history, an entire Renaissance
packed into a couple of dozen offices. Just as the nations that produced the finest
mariners once dominated global commerce and decided political questions, so
countries (or really any group – even terrorists or criminals or bankers) that can
breed, train and equip and deploy members of this New Caste will possess a really
unusual power. Calling them a “caste” is not accidental: There’s a way in which we
are all now tied to the virtual spaces they control, much as serfs were once bound to
the land by feudal lords. The Italian historian Giambattista Vico, in developing his
161 Think of the legendary Xerox PARC: Chunka Mui, “The Lesson that Market
Leaders are Failing to Learn from Xerox PARC”; Forbes Leadership Blog / Alan Kay
113
new, scientific view of history in the 18 th Century, once observed approvingly, “The
Egyptians reduced all preceding world time to three ages, namely, the age of the
gods, the age of the heroes and the age of men,” he wrote. “During these three ages,
three languages had been spoken….Namely the hieroglyphic language, the symbolic
language and the vulgar language of men.” 162 It’s hard not to feel the new age now
arrives with its own baffling and incomprehensible modes of communication. A
whole freshly demanded language.
In all of human history only a few languages ever evolve to become honestly global
in reach and influence – English now, French in the European centuries or Chinese
in the Asian imperial era. But why English? Why not more decades of French?
German? “Why a language becomes a global language has little to do with the
number of people who speak it,” the British linguist and historian David Crystal has
written. “It is much more to do with who those speakers are.” What made Latin a
global influential language wasn’t that millions of people spoke it, rather it was who
did: The elites at the very peak of 1,000 years of European power. What Latin had, in
a sense, was the ears and tongues of some of history’s most influential men. 163 The
private, technical language that connects the New Caste to their machines, to each
other and to us is one of the sources of their power. Their code marks, like a trail,
the path to the cores of a vast and modern power apparatus.
You could, if you wanted, compare the New Caste a bit to an earlier generation of
empire-deciding figures: Ocean explorers. Columbus, de Gama, Magellan. Backed by
a primitive version of venture capital, the “risk finance” of trading houses, these
discovery captains had a hunger to test their certain masteries – navigation, sailing,
trade – against the uncertanties of geography, weather and luck. There was as much
sheer nerve in these adventures as there was real knowledge. What lay five weeks’
sailing time away from Cadiz? If you were willing to endure the difficulties, to
believe in what might be out there – and your own ability to handle it – then fortune
awaited. “Early intercontinental travelers not infrequently had to pay for access to
distant shores by enduring bitter asceticism,” the German philosopher Peter
Sloterdijk has written. 164 Months at sea, risks of ocean turmoil, starvation, endless
boredom – all these sacrifices marked sailing adventure. But they knew the rewards
for a real mastery of the sea: Fame, riches, knowledge, adventure. Sloterdijk cites
Goethe, who reflecting on the power of nautical life in 1787, defined the edgy
advantage of that ocean-mapping caste in his age: They had perspective. “No one
who has never seen himself surrounded on all sides by nothing but sea,” he wrote,
“can have a true conception of the world and his own relation to it.”
162 “The Egyptians”: Giambattista Vico, The New Science, (Cornell Press, 1948) 69
163 What Latin had: David Crystal, English as a Global Language. (2nd ed.
Cambridge, UK: Cambridge University Press, 2003), 7 also Shahar Ronen, Bruno
Gonçalves, Kevin Z. Hu, Alessandro Vespignani, Steven Pinker, and César A. Hidalgo.
2014. “Links That Speak: The Global Language Network and Its Association with
Global Fame.” Proc Natl Acad Sci USA 111 (52) (December 15): E5616–E5622
164 “Early intercontinental travelers”: Peter Sloterdijk, In The World of Interior
Capital, Polity 2013, Ch. 13
114
We might say the same about our New Caste, of their sense of the massive datascape
on which we all sail and operate. They too head off to test their skills at coding,
encryption, system design. They are almost all auto-didact – self taught as
programmers or hackers or AI designers or investors. There is no established path
to power for them, no route that has been marked before. No wonder the greatest
figures of the era are college dropouts. This is a group that looks largely ahead:
Their work demands constant upgrade. Every device is shipped imperfect. It’s not to
say they are completely denuded of the Sixth Sense. That their feeling for history has
been stripped. But if to be honestly in love with version 1.0 of your product and then
to shred it for version 2,0 does demand a certain ahistoric distance. The New Caste
explorers have a hunger for fresh worlds and landscapes. To get there they endure
privations of long isolation – not so extreme as a long ship journey – but they
burrow deeper and deeper into systems, construct new links and new gates, in
closed rooms. “When building machine learning systems, making good decisions is a
strategic skill,” the computer scientist Andrew Ng has said. “Every day you wake up
and you are in some totally unique situation that no one in the planet has been in
before. It’s not a fact, there’s no procedure.” 165
This is our world now. No fact, yet. No procedure. A totally unique situation that
demands a fresh instinct. We can follow Goethe into our own age: No one who has
not seen and felt himself surrounded on all sides by networks can honestly
understand. If that motto of the last great epochal shift was Dare to Know, it was
certainly enacted by the sea captains. Discovery and risk were nothing but a nervy
kind of scientific experiment, where the laboratory was the Earth’s surface and the
tools of calculation were ships. And in our age? If we’re right that our historic
mandate is Dare to Connect, well the New Caste is certainly doing that, furiously.
They are working away at much that is hidden in our world, just like those old
explorers. The billions of phones or sensors or AI-rooted sensor engines spilling out
around our world are stripping away layer after layer of opacity. Pictures of far
away places, traffic lessons learned from movement patterns recorded by GPSenabled
phones, medical treatments refined by constant measurement – these are
all knotted, one after another, into the fishnet.
If the ocean voyages of that older explorer caste mapped our world, the voyages of
the New Caste are as much inward as out. They move inside the world of connected
systems. Remember: Network power is defined by dynamic tension – the pull
between center and periphery. The same dynamism is at work with the New Caste.
While they are, firmly, in the business of going out and opening our eyes to the
world with wider, faster connection, they are also draping huge, essential parts of
our lives with an impenetrable fustian: The essential algorithms of search engines,
the code of machine learning tools, the design of micro-targetted political influence
campaign. The sources of New Caste power are wrapped inside layers of complex
computer code, machine learning and data security protocols that few people can
165 “When building”: Andrew Ng, speech at GPU Technology Conference (March,
2015)
115
completely understand. The French philosopher Bruno Latour, the father of “Actor
Network Theory” has called this process “Black Boxing”. The better your phone
works, the less you notice it. The more precisely some machine feeds you your news,
the less you wonder what might I be missing? “Scientific and technical work,” he
says, “is made invisible by its own success.” 166
The operating system and network protocols of your tablet device are opaque to you
now in a way they never would have been two decades ago (when even the casual
user computer had to type to a C: prompt or wildly rage at a crashed “Blue Screen of
Death” from time to time). But, in fact, the system is incalculably more complex.
“Each of the parts inside a black box,” Latour remind us, “is a black box full of parts.”
And it is in the winding and linking of all these pieces that action in a connected
world is made possible. “It is by mistake or unfairness that our headlines read ‘Man
Flies’”, Latour says. “B-52s do not fly,” he writes. “The US Air Force flies.” Every
plane that ever makes it into the air does so because of the clicking coordination of
thousands of linked, black boxed systems. Your stock portfolio or your computer or
your bio-sensored heart is not a lone object; it’s a feature of a connected landscape.
We’re surrounded now, connected to, essential black boxes we’ve no way of
understanding and whose development and operation we’ve left to the New Caste.
Look around you, how many screens do you see? Each is a billboard: New Caste at
Work. It’s not only the hardware in our lives that I mean, but the bits of knitted
programming that decide how we search, when we communicate, and if we can
exchange information or money. The virtual and the real are in constant contact and
it’s the New Caste that does the stapling. In fact, one of the magic tricks of power in
the connected age is an ability to flop easily back and forth between network and
reality. It suggests other dangers too. As legendary machine systems designer Leslie
Lamport warns: Computer scientists collectively suffer from the confusion of language
with reality. 167 Anyone who’s ever written a computer program knows this sense:
You write some code. You compile and run the program to see what happens. You go
back and work on the code some more to refine what you’ve done. You run it again.
You touch the virtual; the real reacts. This seems in a way like the most trivial thing,
the writing of a computer program or an AI bot or a trading order, but in fact
166 The French philosopher: Bruno Latour, “On Technical Mediation – Philosophy,
Sociology, Geneaology” in Common Knowledge, Fall 1994 Vol 3 No 2. p. 23
167 It suggests other dangers: Leslie Lamport, “Computer Science and State
Machines”, Contribution to a Festschrift honoring Willem-Paul de Roever on his
retirement” (Redmond: Microsoft Research, 2008). In its entirety it runs: “Computer
scientists collectively suffer from what I call the Whorfian syndrome—the confusion
of language with reality. Since these devices are described in different languages,
they must all be different. In fact, they are all naturally described as state machines.”
The Whorfian problem is a linguistics observation about the way in which our
thinking is limited by whatever language we have to describe what we see or
contemplate.
116
something incredible is underway in this easy movement from machine to reality. 168
The New Caste takes these moves, this easy slip from their keyboards and programs
to our lives, for granted. They adjust code and networks and formulas; they watch
the effects on us. They do it again. The idea that such a move is natural, comfortable
even, reveals a new and important temprament. It draws a line betweent the people
writing the code and those who are snapped about in the world they are coding.
Do you know who decided what you see when you search? Do you understand what
the data on your phone reveals about you? Who will snip at and work on your DNA?
Your children? Are you trading stocks against some invisible high-velocity
connected master who will always be one profitable nanosecond ahead of you? In
this sense, network power involves something very much like the intentional
creation of concealment. Your Internet search results, for instance, contain a sharp
tension. Yes, data from all over the planet, from all of history sits rather amazingly in
front of you. But that bit of computer code deciding what you see is engaged in a
kind of digital book burning: It’s making whole sections of knowledge invisible even
as it is unearthing an ever more precise answer for whatever question you have.
What don’t you see? – is a question that hints not only at what is left out of your
search horizon, but generally at the way in which connected systems establish
necessary gates. Part of a Seventh Sense, then, is the ability not merely to look at the
virtual world and know how it becomes insidiously real, but also to feel that all the
connected points of the real world – markets, weapons, social movements – must be
pulled upon by code and links and networks. “Any technology depended upon,” as
The Critical Engineering Manifesto, says is “both a challenge and a threat.” 169 Human
experience is, we know, unboxable, uncontainable – our joy, hopes, sense of
freedom, these all defy boxing. Yet here, all around us, are containers that affect our
every choice. Who knows what happens inside all the difficult boxes?
The creeping, essential opacity of power now reveals a twisted puzzle, a really fresh
aspect of this New Caste and the revolution they are making: As much as they are in
the business of making knowledge widely and instantly available, they are also
madly black boxing our world. This breeds a sly, unintended (I think) tension with
Kant’s Enlightenment admonition to “Dare to Know.” Would you like to Dare to know
why your computer is secure? How your genetic information will be studied and
used? How encryption works? Mostly the answer to is: You can’t know. It’s too
complex – and, anyhow, if we told you it would make the whole system less secure.
There is nothing disingenuous here: You likely wouldn’t understand. It is too
complex. You’d be lost at the first turn into strange technical language, where simple
words like “object” or “edge” have specific, essential, different meanings. And telling
you would, in fact, expose you and everyone else to all sorts of risks. It’s as if we’ve
returned to that famous debate of millennia past, the one lingering between Athens
and Jerusalem: Could the world be known and atomized and understood as the
Greeks would have it? Or was mystery, inscrutability and opacity the nature of truth,
168 This seems in a way: See, for instance, Bret Victor in his speech, “Inventing on
Principle” at CUSEC 2012 Turing Complete Conference available online.
169 “ Any technology depended upon”:The Critical Engineering Manifesto, as above.
117
apprehensible, the Rabbis said, only by its movements. Are we back at the first
chapter of Genesis and its absolute prohibition against eating from the tree of
knowledge. Or, from the Talmud, “For him who reflects about four things – what is
above, what is below, what is before and what is behind – it would be better not to
have come into the world.” 170 We want to reflect about what goes on inside the
machines. Can we? Should we? How does Dare to know face off against these
impenetrable systems.
It is little surprise that places like Silicon Valley often leave a visitor with the feeling
of a town where work is done in rooms within rooms within rooms. To drive along
the dulled, anodyne asphalt stretch of road that runs in front of Sand Hill Road in
Menlo Park almost hurts your head: Inside the offices on revolutions are dreamed,
debated and funded. And it looks, for the most part, like a row of mildly prosperous
dental practices. The real import of the work is, on the outside at least, nearly totally
muted. The corporate structure of the most powerful tech companies are padded
with this sort of deadening fustian too. Founders control the majority of voting
stock; shareholders are more like lucky “users” than owners. Control, security and
speed in decision-making are secured from the inside, free of exploit risk or
interference. The companies are like computers. Of course the founders know where
real powers sits. But this shouldn’t distract us from the human energy breathing in
the code itself. The programs are “permeated by all the forms of contestation,
feeling, identification, intensity, contextualization and decontextualization,
signification, power relations, imaginings and embodiments that comprise any
cultural object,” the computer science historian Adrian Mackenzie has written. 171
Each of the parts of a black box is a black box. The famous billionaires of our
technology age operate for the most part as their systems do. Their tight, wellengineered
clusters of machines produce fortunes from connectivity, even as they
obscure some of the deeper nature of the connections that are essential to their
success. They are themselves at times obscured, human black boxes in a sense.
“Linux is just an enabler,” the genius programmer Linus Torvalds once observed
about the code language that undergirds much of the connected world. “It’s a solid
base, but like all good, solid bases, it really is something that should be almost
entirely hidden and out of people’s minds.” 172 It is a hard paradox for us. The work
of the black boxes, of connected systems or protocols such Linux is miraculous. It is
wonderful in so many ways. And the roots of it are, and seem like they have to be,
obscure. But this cuts very fast into the arteries of a healthy democracy.
“Democracy,” Arthur Schlesinger wrote in his famous post-war book The Vital
Center, “has no defense-in-depth against the neuroses of industrialism.” It’s easy to
see how the system might also have a weakened immunity to the subversive forces
170 Or from the Talmud: See Leo Strauss, Persecution and the Art of Writing,
(Glencoe, Ill.: Free Press, 1952), 21
171 The programs: Adrian Mackenzie, Cutting code: Software and sociality, (New
York: Peter Lang 2006) 5
172 “Linux”: Andy Meek, “Linux creator explains why a truly secure computing
platform will never exist” on bgr.com Sep. 25, 2015
118
of network power: Contagions of fear, manipulation of data, the subtle and invisible
influence of the boxes we depend upon but don’t understand. 173 Recall Francis
Bacon’s Enlightenment line? That human knowledge is human power? Well, what is
computer knowledge? It is human power? Or something else, in its entire, hidden
immensity? You have to wonder if this packing of insight and vision and control into
black boxes, or the hands of a small New Caste will bleed us of our liberty as a result.
“Our constitution is called a democracy because the power to make decisions is not
in the hands of a minority but of the whole people,” Pericles reminded Athens in his
Funeral Oration 2500 years ago. “We regard a man who takes no interest in politics
not as harmless, but as useless.” 174 Vital engagement is the food of democratic life.
To be baffled to the outside of the essential boxes of power then, seems an instant
sort of cancer on liberty. What do we make of a man who takes no interest in the
networks of networks that control the power to make decisions?
Perhaps you’ve heard of the famous manufacturing trilema: You can get something
made any two of good, fast and cheap. If you want that custom table made quickly
and well, it won’t be cheap. If you want it good and cheap, you had best be prepared
to wait. In networks a similar puzzle emerges in my mind. Systems can be any two of
fast, open or secure. A computer system that is really secure can be open, but it will
be very slow, inspecting each packet and instruction like a bank security guard
watching customers in a bad neighborhood. Think of the like an airport. Want it to
be fast? Secure too? Then it won’t be very fast. Mostly what we want today are fast,
secure arrangements for our markets, our nations, our data. So these will become, I
think, ever less open.
It used to be that history was made in public: Big visible wars and social shifts and
revolutions. Pericles in the Athenian square; the churning protests of Jefferson’s
Paris or the massing of armies. Now, however, subtle manipulations of technology,
invisible to most of us and maybe even accidental, maybe weird, will produce
historic-scale external effects. Changes to the network design will become political
and social exploits in a sense, living versions of that atomic-level “rowhammer” hack
that work on the connective energy of our world. Already social network analysis
can be used to manipulate voting patterns. Soon, it will be possibly to precisely
target any potential voter with a message engineered like a custom-made drug,
designed to bind right to the DNA of your habits and beliefs. It represents the
possibility for the complete technical perversion of politics.
As the New Caste operates on the systems that are at the core of connected politics
and economics, on how we vote or think or shop, they will vibrate the system in
invisible ways even when they don't mean to be nefarious. Improvements may be as
dangerously unpredictable as bugs. Small changes to algorithms or links or
protocols mean our whole system may be pwned before we’re quite aware. Such
173 “Democracy”: Arthur M. Schlesinger, Jr., The Vital Center: The Politics of Freedom
(Boston: Houghton Mifflin, 1949), 246
174 “Our constitution”: Thucydides, A History of the Peloponnesian War, (Oxford:
Oxford University Clarendon Press, 1881),119
119
adjustments will be started and managed and mined for fortunes, of course. The
Seventh Sense feels the way in which power has shifted, is shifting, from public to
private in this way, masked by coded language, hardware design, corporate
structure and the demands for speed and safety we all agree on. Huge shifts in
power will occur before we are even aware if we don’t have a sensibility to feel them
out before they occur. Decisions about code, search, machine intelligence, DNA
alteration rules in labs – all occurring in black box machines or corporations or
governments. Several years ago it occurred to me in an unsettling flash: “The most
important things that will happen in my life will happen in secret.”
I’m not sure I’ve quite recovered from this insight yet.
2.
This same unnerving worry has troubled many people who think about connected
machines, and for some time. It’s not only those among us who are pointlessly
nostalgic for a different era of devices, when the default setting of our instincts was
open, who now worry about this strange tension between function and opacity. And
remember – I’m running through all this here so we can all understand how to really
grab and use this new source of power in the service of what we desire, and to
protect the things we care about. But: What are we to make of systems that work
better when they are obscured from us? That we could never understand even if we
could see inside? Documents such as The Critical Engineering Manifesto, pulled
together by a collection of uneasy designers and engineers, reflect anxieties that stir
even inside the hearts of many in the New Caste. “The Critical Engineer considers
engineering to be the most transformative language of our time,” the Manifesto
begins. “Each work of engineering engineers its user, proportional to that user’s
dependency on it.” That is Conway’s or Wittfogel’s unsettling truth. The technical
layout of a social network, a medical diagnostic tool or a financial market affects
how we or our markets or our government (or our enemies) behave. We are, in a
sense, being engineered by the damn opaque systems we are using.
In 1965 the MIT computer scientist Joseph Weizenbaum found himself, somewhat
unexpectedly, considering this same problem. 175 Weizenbaum had written a
primitive computer program to perform what is now known as natural language
processing, essentially a bit of code designed to translate what a user tells a
computer into clear instructions. This is the work that converts your request to a
machine for weather updates into a query that can be turned into an answer.
Teaching a computer to answer a question such as “it is it raining?” demands layers
of logic and calculation.
In 1965, long before our era of simply talking to our phones or computers,
Weizenbaum’s program was a revolutionary breakthrough. So after developing his
algorithms, he wanted to find an intuitive demonstration. He decided to build a
175 In 1965: Joseph Weizenbaum, Computer Power and Human Reason, W.H.
Freeman and Company, 1976 p. 7
120
system that could, in a very limited sense, converse. The computer would take typed
input from a user and then, based on what it understood, ask a question back to
continue the conversation. The approach was borrowed from the psychological
theories of Carl Rogers, an American therapist who believed the best way to talk to
patients was with open ended, careful questions. Weizenbaum developed two
programs. The first was called DOCTOR, a script that determined what the computer
might say. The natural language processor responsible for carrying on the human
side of the conversation he called ELIZA, after the heroine from Pygmalion, a poor
cockney girl who is trained by a phonetics professor to pass as a duchess.
Here is a discussion between ELIZA and a user that Weizenbaum recorded in 1965:
Patient: “Men are all alike.”
ELIZA: “IN WHAT WAY?”
Patient: “They’re always bugging us about something or other.”
ELIZA: “CAN YOU THINK OF A SPECIFIC EXAMPLE?”
Patient: “Well, my boyfriend made me come here.”
ELIZA: “YOUR BOYFRIEND MADE YOU COME HERE?”
Patient: “He says I am depressed most of the time.”
ELIZA: “I AM SORRY TO HEAR YOU ARE DEPRESSED.”
Patient: “It’s true, I’m depressed most of the time.”
ELIZA: “DO YOU THINK COMING HERE WILL HELP YOU TO NOT BE UNHAPPY?”
Patient: “I need some help, that much seems certain.”
Weizenbaum’s program quickly became famous around MIT. (The joke was that it
was a better conversationalist than most of the awkward shoegazers studying at the
Cambridge campus.) ELIZA gave users a chance to have a taste of a machine with
something like a mind of its own. The program spread to other campuses, to
different research institutions, and with every step of its success, Weizenbaum
found himself more distressed. He summarized his worries in his 1972 masterpiece
Computer Science and Human Reason. What troubled him was not ELIZA, but her
users, the humans. Even when they were professional psychologists, many of them
quickly considered it reasonable to think that one day the work of diagnosis and
counseling might be turned over to machines. This felt like a natural next step in the
ceaseless progress they were used to in their lives. Better refrigerators, stronger
seat belts, faster jet planes, more plastic – why not a computer doing therapy? It
121
sounded kind of wonderful. “A number of practicing psychiatrists seriously believed
the DOCTOR computer program could grow into a nearly completely automatic form
of therapy,” Weizenbaum wrote. “I had thought it essential, as a prerequisite to the
very possibility that one person might help another cope with his emotional
problems, that the helper himself participate in the other’s experience.” To use a
machine for such a task? He was horrified. Weizenbaum knew the empathy ELIZA
was exuding was faked. It was just code. “Science,” he concluded, “has been gradually
converted into a slow-acting poison.”
“Would you mind leaving the room,” Weizenbaum’s secretary said to him once, lost
in a particularly personal discussion with ELIZA. “The reaction,” he wrote, “showed
me more clearly than anything I had seen hitherto the enormously exaggerated
attributions even a well-educated audience is capable of making, even strives to
make, to a technology it does not understand.” This was black boxing at its worst: “I
have no idea how this thing works. And it’s wonderful!”
What makes the New Caste so particularly powerful is that their essential work is to
build and operate the cores that control these systems. And the more people they
lure onto them, the more powerful the platforms – and the people who run them –
become. “The computer programmer,” Weizenbaum wrote, summing up his lessons
from ELIZA, “is a creator of universes for which he alone is the lawgiver.” Each of
these cores represents a fusion of power and politics and technology like nothing
the world has ever seen. They are assembled mostly from scratch, they represent
the concentration of billions of connections, and their direction is determined by
technological and market factors as much as by any democratic twitch.
The strategic power of societies that train the best of the New Caste is probably selfevident
by now. To educate and deploy masses of people capable of such
transcendent design genius will mark a difference, an electric gating line between
the nations that succeed and those that fail. But such training brings a real tension, if
this group is allowed to really rip away at their work. What won’t they attack?
Control over the protocols that answer questions, move money, protect data,
analyze your DNA – it’s hard to think of any single locus of power that will ever be
greater than the tight, gravitationally inevitable platforms emerging around us now.
These essential webs are filled, as we’ve seen, with complex bugs and errors and
loopholes. They depend on design decisions whose implications resonate for
decades – both inside the black boxes and the external world that vibrates to their
quiet demands. “If builders built buildings the same way programmers write
programs,” one famous coding lemma runs, “then the first woodpecker who came
along would destroy civilization.” Who would know if rot is spreading in these
systems? Who would stop it? Recall Paul Virilo’s line that trains produced train
accidents, planes produced airplane accidents. So: Black Boxes?
As much as the work of the New Caste looks tactical in nature – what protocols to
use, how to engineer networks or design machine boards – the reality is that most of
what they do would be blind without a strategic urge. Behind even the smallest
122
advance, whether it is fingerprint recognition on your phone or some new autotranslation
app, a set of careful, deterministic values and calculations linger. One
feature that defines the New Caste is unadulterated, unquestioned faith in the
continued network revolution, and that the values that underlie the best programs
may be the values that should gird the world that depends on that same code. The
danger here is clear enough. “Respect, understanding and love,” Weizenbaum wrote
as he considered ELIZA’s effects, “are not technical problems.”
The biggest of the platforms controlled by the New Caste herd together, remember,
billions of people, bind them with ever thickening cords. The revealing tics of every
movement in the virtual and every step or drive in the real world are marked down,
remembered and scored. To operate the strategic levers of such a force is, in all
reality, no less significant than leading a nation. The distinction between a CEO of a
major connected firm and a head of state lies less in the depth and efficacy of their
influence than in the questions of how they got such power, and how they might use
it. The New Caste has an admirable conviction near to faith that their products are
truly universal. They are absolute technological determinists. Watching their
services and influence expand often has that strange aura of the irresistible force
taking on an immovable object. They believe that their black boxes will bulldoze
concerns of politics or history. And soon.
Historical ambition of this scale, the sort that touches really countless lives, has
always blended a commercial and technical mastery -- the moves of the East India
Company turned as much on better ship design, maps and navigation as on imperial
objectives. But the aim of the New Caste is the same as it was for those three older
castes – the merchants, soldiers and sages: To put the tools they’ve mastered and
built in the service of still more dominance. The commercial calculations of the most
powerful figures of the New Caste carry a sense of seeing many moves ahead, a very
real kind of chess. Their billion dollar acquisitions, investment in moon-shot R&D
ideas, the hundred million dollar payouts for great engineers – all of these mark the
astonishing scale of what they have in mind: To have and control ever more
essential cores of power. Are they seduced by having a billion users? Sure, but not
because of the billion users, but rather because of the seductive allure of the black
box, of what it means to control such a central point of connection.
4.
I remember sitting with a member of the New Caste the week the first batch of
mimeographed and laser-scanned Snowden papers were released, as we both
discovered that everyone we knew was devouring the documents. Like a novel.
People were texting one another – “Have you seen this?!” – and you couldn’t get
through a dinner without a debate over the technical merits of what was on display.
The Snowden files were fascinating to the New Caste in a way few others might
understand, in the way a room of ballplayers might examine Ted Williams’s swing
mechanics. Let me try to explain it this way: When I was younger, people called our
generation – those of us born between 1965 and 1980 more or less – a generation of
slackers. Generation X. Generation Nothing. There was an argument to be had about
123
the Baby Boomers. Had they been the most destructive, selfish generation in
American history? A reaction against the selflessness of their parents? Retired to
leave the rest of us to pay their future medical bills and oggle their underfunded
pensions, to cope with the manipulated political system they’d sued into existence.
Or had they left a legacy of tolerance, an echo of 1968’s optimisitic energy, a firming
of American confidence. But, anyhow, Generation X? By comparison irrelevant: A
collection of sad, passive slackers.
But the great Internet companies were largely built by Generation X. The
foundational experience of 1989 – the fall of the Berlin Wall – bred optimism. It
created, in fact, the possibility for a new exploration. When we were told “Be
generous in what you accept,” this seemed reasonable and, eventually, lucrative. The
logic and power of networks became apparent by itself, the moment we began
connecting the world. So linkage in trade and finance and friendship was pressed
out into a new era of globalization, pushed as much by the smashing, enthusiastic
removal of so many historical limits as it was by the technology itself. Yes WiFi and
TCP/IP and other advances made wiring the world possible, but I wonder if they
would have developed so quickly if the context for using them hadn’t provided a
feedback loop of such quick profit and reward and, frankly, amazement. So, in this
fashion, we laid the groundwork for a world of billion-plus user platforms. For a
new concept of power.
But the Snowden papers were a shock. We knew him, in a sense. His mannerisms
and thinking and technical instincts harmonized with our own. It was as if the NSA
had enrolled most of the digitally visible world into a twisted panopticon of a social
network, one where your “membership” began the moment one of your data
packets was sniffed or chased along fiber optic lines. Here was a secret three billionuser
platform, in a sense, that had enmeshed, without their knowledge or
permission, a tremendous chunk of humanity. People of interest. The one or two or
three billion people swept into systems like the NSA’s Aurora fell victim to the
powerful leverage of scaled systems: The more people monitored by the NSA, the
more leads to follow, which meant still more people needed to be followed and
knitted unwittingly into the web.
No one in my generation who had been around tech for long was naïve. We knew
that like soldiers coming back from a war there were things done on the network –
the spreading of danger, of inequality, of pollution – that would have to be paid for
in balance by the benefits the system promised to bring. The immensity of what had
been built already in networks was clear enough; the even greater power yet to
come as trillions of more connections piled together was implicit, obvious even. The
perverse, sneaky side of this growth was known to anyone who’d spent time
working a connected machine. But that people of our generation could on the one
hand listen to “Karma Police” and on the other enact the sick OK Computer logic of
surveillance, even name a British GCHQ monitoring program after the Radiohead
song? The aim: Provide "either (a) a web browsing profile for every visible user on
124
the Internet, or (b) a user profile for every visible website on the Internet." 177 We’d
not honestly thought the end of this would be a need to so deeply question or defend
freedom or liberty.
The whole idea of connectivity had begun with liberty, after all. The Fall of the Wall.
We’d not known or expected to lose it through the very systems we’d constructed in
the open space a 20-year peace had presented. So: We’d been naïve after all. We’d
been united in a tacit claim that there was nothing bad about connectivity. We
believed it for the most part and convinced others. We didn’t mean it that way, I felt
the New Caste readers of the Snowden Papers were saying to each other, paging
through the sick, ineffably banal logic of those NSA Power Point presentations. They
were brochures for totalitarianism. If you have not read them, it is worth looking
them over. They have, you’ll see, all the anodyne banality of insurance company
pitches, even if the dull language was expressing something vivid, nothing less than
the potential murder of some very basic rights. We were reading and discussing and
debating the documents so avidly because, it seems to me, each of us had a kind of
horror of what we might have done.
Here was a massive technological and insidious web, a totalizing virtual machine of
collection and analysis and instant observation and reaction. It relied, for its safe
operation, on the humans in the loop of that sensing and seeing and machine
thinking. And what was it, exactly, that the humans had appeared to do in the face of
such urgent responsibility? Almost like that silly secretary chatting with the ELIZA
machine, they had suspended their sense of warm humanity in the face of the
magical charm of electrical promise. They had let the machine run. The whole
system had been built and operated by members of the New Caste who were, in
turn, beguiled and charmed out of a sense of certain essential limits as they stared at
the beautiful face of machine power.
Looking back on his formative years, before Europe was ground up in the First
World War, the powerfully brilliant and sensitive economist John Maynard Keynes
bitterly recalled the iron certainty of his set of friends and their confident, diffident
arrogance: “We were not aware that civilization was a thin and precarious crust
erected by the personality and the will of a very few, and only maintained by rules
and conventions skillfully put across and guilefuly preserved. We had no respect for
traditional wisdom or the restaints of custom.” 178 It was the war, then the
depression, then a war which taught them this expensive lesson. What is fracturing
around us now, with our own willful and diffident support, is that very thin and
precarious crust, cracked and assembled as Keynes said by “the will of a very few”.
What if those very few are the New Caste?
177 The aim: GHCQ “Pullthrough Steering Group Meeting #16” available from
theintercept.com
178 “We were not aware”: John Maynard Keynes, “My Early Beliefs” in Two
Memoirs: Dr. Melchior, a Defeated Enemy, and My Early Beliefs (New York: A.M.
Kelley, 1949) 99
125
The essence of the Seventh Sense will be not merely to be beguiled by our
technology, by the way it smashes old systems, not to ask “Could you please leave
the room while I use bend these electrons to even more omnipotent control,” but
rather – in the same instant as the networks snaps into its full power– to grasp the
full nature of the connected age. To see how it might be used to further, not erode,
the things we care most about. To ensure that if we are not among the very few, at
least we can guarantee that their will bends towards justice. We will see, in a
moment, just what that means in practice. How our best technology and our most
avid hopes for what a technological, connected future could look like might just be
yoked together. But before we can do that, there is one final question we need to
answer about the networks all around us now: What, in the end, are they really for?�
126
Chapter Eight:
“MapReduce”: The Compression of Space and Time
In which we learn what networks are really, rather wonderfully, meant for.
1.
Starting in the springtime of 1997, the American scientist and inventor Danny Hillis
began what has since become an every-few-months sort of ritual. He packed up
from his home in Encino, a short drive over the Hollywood Hills from Los Angeles,
and headed off for rural Texas for a few days that would largely defined by rock and
dynamite. Hillis, who was born in 1956, has spent most of his life working at the
electron level of the world, crafting some of the most significant computer
processing systems of our age. So the sort of paleolithinc earth moving he was
heading off to manage in Texas was a departure from his usual scale. His aim was to
work on blasting and then refining a space in an isolated mountainside for the
construction of a towering clock that he had designed, one intended to run for
10,000 years. That ten-millennia span was not accidentally chosen. Humans, when
Hillis began his work on the clock, had been around about that long already. We
were, as he pictured it, at a midpoint on that 20,000 year stretch of time. Hillis and
the group of tinkerers, thinkers, and engineers who had backed and designed the
clock – people such as Amazon’s Jeff Bezos, spreadsheet inventor Mitch Kapor or
investor Esther Dyson – were planning on a project that would stretch as close to
eternity as they felt reasonable. “The Clock of the Long Now” they called it. I
remember pulling into Danny’s driveway in Encino one afternoon as he prepared to
depart for Texas and being struck by the contrast between the lovely, innofensive
suburban blandness of Southern California and the tools he was taking with him to
make an assault not merely on a mountain, but on a whole conception of time.
I had met Hillis in an unusual fashion. I’d been asked to chair a committee that
would award a million dollars to a figure who had made an essential contribution in
the world of technology. The directors of the foundation behind the then-new prize
had been, from the start, slyly dropping big names – Bill Gates! Steve Jobs! They
hoped such a laureate would cast a bit of glamour on the first year of their award for
“Contributions to Man’s Present Condition.” But when our committee sat down to
talk it over, we knew that the boldfaced names didn’t want or need a prize. They
certainly didn’t need a million dollars. As we considered people we all knew who’d
made fundamental, essential contributions but had not been as boldfaced as they
might have been, Danny Hillis’s name came up immediately.
Hillis had developed a revolutionary “massively parallel” computer in the 1980s.
The machine had helped create an entire discipline of high-speed computing by
tying together tens of thousands of processors to tackle a problem at once.
Traditional computers worked problems the way you or I might, step by step. Hillis’
design was the equivalent of millions of minds, all moving at once. Coordinated,
connected and awesomely fast. In the years since, he’d played a key role in a dozen
127
other breakthroughs, from designing artificial intelligences to fine-tuning classified
military aircraft systems that depended on mathematics for their stability. When
you wander into a deep part of Google’s technical database systems, you’re touching
his work. When you talk to your phone, the interface bubbles with some of his
patents. How did Baran’s 1960s idea of a survivable, packet-based system at
ARPAnet become the Internet in the 1970s and 80s? Danny was part of a cluster of
dirty-fingernail engineers – along with engineers such as Vint Cerf and Jon Postel –
who’d done the work to make it possible. His centrality in that project was
memorialized in a famous speech he once delivered in which he described having
one of the very first email addresses in history – and then whipped out a sheaf of
bound pages that represented the entire Internet address list at the time. It ran
about 50 pages. To the extent there were membership cards in the New Caste,
Danny’s would have had a very low number. It was an easy decision for our prize
committee. No Bill Gates. No Steve Jobs. So, here’s how I met Danny Hillis: I called to
tell him he had won a million dollars. (I recommend this as a way to start a
friendship.)
Hillis had been a tinkerer since he was a child and never seemed to have lost the
pleasure of a wild intermingling of joy and practice. You couldn’t tell with him
where passion ended and work started. He was so technically adept that he could
inject even the coldest digital projects with a bit of hot emotion, like Bernini
breathing life into a block of carved marble with one, “just so” grace note of his
chisel. One of Hillis’s most famous projects, for instance, was a 15-foot high tic-tactoe
playing tinkertoy robot he’d built when he was 20 years old, in his second year
as an undergraduate at MIT in 1975. Made from 10,000 wooden spindles and poles,
it was an early attempt of his to show how machines, even simple ones, might
seduce us with both brains and looks. The effect of a giant tinkertoy pile sitting there
at The Massachusets Institute of Technology had to make you giggle, even as your
mind boggled at the fact that this heap of sticks, strings and dials was beating you
again and again at a child’s game. Hillis was an artist as much as an inventor – one
reason he’d not become Bill Gates or Steve Jobs. (And why Gates and Jobs
maintained a consistent, admiring respect for him.) He’d once spent a decade at
Disney designing rides or thinking up new dreams as a kind of real-world mayor of
Tomorrowland. He liked to joke that he knew he was at the right place when, on his
first day, he asked where he might find a parachute harness for an experiment and
heard, in response, “What size?”
Hillis was an avid reader, and he has the habit to think of his bleeding-edge work in
the context of long historical gulps. Conversations with him often tie back to
Paleocene era biology or some other deep root. That long-term view, married to his
unmatched hands-on feel for complicated systems, made him the ideal designer for
the clock, a machine intended to last millennia. The problems associated with such
an undertaking were, honestly, as unreal as you might expect. How to power the
clock? (Hand winding, the better to ensure it was not forgotten.) How to protect it?
(Put in the middle of nowhere.) Did you need to plan for global climate change? (Yes.
The design was adjusted to accommodate shifts in the earth’s spin when the planet’s
icecaps melt off.) Do you write a users’ manual for people 10,000 years from now?
128
(Yes.) Do you write it in English? (To be determined!) Working with composer Brian
Eno on the sound of the clock chime, and with a team of geologists and physicists,
Hillis had made the clock into a natural extension of his tinkertoy computer, a device
that both served a purpose and sent a message. If there was an emotion it conveyed,
a feeling that it tickled in the way Bernini’s Apollo and Daphne might inspire terror
or joy or faith, it was meant to be awe.
Stewart Brand, one of the supporters of the clock and an early member of the New
Caste too, would tell you that the idea for the clock had emerged from a desire to
emphasize, to physicalize in a way no one could forget, the importance of longerterm
thinking. We’d all arrived now, Brand and the other clock masters worried, at a
moment in history when no one had a view that extended much past their own lives
– or sometimes past the next election, or year, or the next financial quarter. Our
modern “on to the next thing” economics and politics were eroding every slow,
patient instinct. “Civilization is revving itself into a pathetically short attention
span,” one manifesto for the clock began. “What we propose is both a mechanism
and a myth.” 179 With its steady 10,000 year ticking, the Clock of the Long Now was
meant to make us think in longer jumps. The human winding mechanism, for
instance: Generations of clock-winders had to share in the work, and they would be
connected in a long thread over the 10,000 years. A sacred priesthood of time.
Moving slowly. As I spent time thinking, dreaming about the clock, I found myself
too craving the solidity and patient isolation it promised. Who among us these days
doesn’t want a break from the instant nowness of our age?
Yet, the more I understood the clock, the more I realized something else was at
work. Stop for a moment to consider who was backing and building the device. It
was a cluster of people who had, as a common link, the fact that they had their hands
honestly sunk into the guts of the Internet. Hillis, after all, had been waving more
than that slim book of email addresses when he talked about the early days of the
Internet. He was waving the credentials of a man who had been living in the virtual
cyber neighborhood of Web connections from its very first days. He was as close to a
native of the connected, fiber optic, light-speed world as you could find.
All the names supporting the clock smelled similarly of burning electrons: Jeff Bezos
had built Amazon into a high-speed marketplace whose backbone was the Web
itself. Another backer, Mitch Kapor, had cracked apart several centuries of slow
accounting habits when he created Lotus 1-2-3, the first successful computer
spreadsheet program in 1983, software that permitted you to see and change your
whole business one keystroke at a time. Kapor’s software helped move finance from
quarter-by-quarter calculations to a really instant-by-instant sort of business – more
or less the opposite of the “long time frame” the clock team was aiming to preserve.
Esther Dyson was one of the earliest, best investors in network companies. This was
a collection of men and women unified by a genius for connected change, sure, but
also by a desire for ever faster clock speeds, ever speedier delivery, ever faster
179 “Civilization is revving itself”: Stewart Brand, The Clock of the Long Now: Time
and Responsibility. (New York: Basic Books, 1999) 2
129
processing. They had lived this. Enabled it. Profited from it. If there was ever a
group you might hope to take aside, pull into a quiet room and ask gently What are
the really networks for, anyway? this would be it.
The act of keeping time, of marking it, is embedded in the nature of any age. Our
lives are, after all, dictated by timetables: School schedules, the seasons, rush hour,
the burning candle of birth-love-marriage-death. Time, in the days before industry,
was measured by nature’s schedule. How long it took a crop to mature. The
solstices. A beehive filling with honey. It was marked by moving tides and shifting
seasons, and it demanded a slowness, a personal presence on the shores, in the
oceans, atop the fields over generations. “Summer afternoon,” the novelist Henry
James remarked in a précis of a slower age he felt passing away in 1895. “To me
those have always been the two most beautiful words in the English language.” 180