Kruger and his collaborators started by studying sarcasm. In their first
experiment, they gave a group of participants a list of topics. For each topic,
they were asked to write two sentences: one normal and one sarcastic. They
then emailed their sentences to participants in another group, who were
tasked with identifying which sentences were meant to be sarcastic. “As
expected, the participants were overconfident,” the paper explains. The
sentence writers predicted that the readers would essentially get every
choice right. In reality, they failed nearly 20 percent of the time.
In a follow-up experiment, half the sentence writers got to record
themselves reading their sentences on a tape recorder, while the other half
still emailed their creations. Perhaps not surprisingly, hearing sentences on
a recording made it easier to determine whether or not they were sarcastic.
What was surprising was that the sentence writers predicted there would be
no difference: they believed the recipients would have an equally easy time
detecting sarcasm in written and recorded sentences.
To test the claim that egocentrism was the source of the participants’
overconfidence, the researchers turned their attention to humor. They now
provided each sender with a short humorous passage. In particular, they
drew from humorist Jack Handey’s Deep Thoughts: absurdist mini
monologues delivered as scrolling text, read by a deadpan narrator and set
against a relaxing backdrop. These appeared as a regular feature on
Saturday Night Live during the 1990s and early 2000s. To make this
experiment more concrete (and to provide me an excuse to replicate the
funniest passage I’ve ever read in a peer-reviewed article), here’s an
example Deep Thought that was actually used by the researchers:
I guess of all my uncles, I liked Uncle Caveman the best. We
called him Uncle Caveman because he lived in a cave, and
because sometimes he’d eat one of us. Later on we found out he
was a bear.
To test egocentrism, the researchers randomly divided the senders into
two groups. Each participant in the first group was simply provided a Deep
Thought to send via email. Those in the second group were shown a clip of
the thought being delivered on Saturday Night Live, with the oh-so-perfect
calming music, the deadpan narrator, and the shocked laughter from the
audience. After watching the clip, this group also emailed just the text. In
both cases, the senders were asked how funny they thought the passage was
and to estimate how funny the recipients would find it.
“Participants in the videotape condition thought that the [Deep
Thoughts] were funnier than did participants in the control condition,” the
paper reveals, “and the same was true of participants’ predictions of the
recipients’ evaluation of the jokes.” Seeing the video clip provided the
minds of the participants with a richer accompaniment to play alongside the
text they were typing into the email. Like Elizabeth Newton’s song tappers
hearing the music in their head, the videotape group couldn’t shake the
funny visuals and laughing crowd when trying to judge how well their
email would be understood. The richer the sender’s subjective experience of
what she’s trying to communicate, the bigger the gap grows between her
understanding and that of her correspondent—evidence that egocentrism is
at the core of the measured overconfidence.
The conclusion of this work is that emails are commonly
misunderstood because of the “inherent difficulty of moving beyond one’s
subjective experience of a stimulus and imagining how the stimulus might
be evaluated by someone who does not share one’s privileged perspective.”
To make matters even worse, the researchers found that the recipients of
these ambiguous messages were as overconfident as the senders. They were
confident that they were correctly detecting sarcasm or identifying humor,
even when they weren’t doing well at all. This last observation applies a
particularly devious twist to our understanding of email’s many confusions.
It’s not just that we’re less clear than we think, but we’re often completely
misunderstood. You were sure that you were sending a nice note, while
your receiver is equally sure you were delivering a pointed critique. When
you build an entire workflow on exactly this type of ambiguous and
misunderstood communication—a workflow that bypasses all the rich, nonlinguistic social tools that researchers like Alex Pentland documented as
being fundamental to successful human interaction—you shouldn’t be
surprised that work messaging is making us miserable.
We don’t need research studies, however, to emphasize something that
many of us already experience on a daily basis. In her book Reclaiming
Conversation, MIT social scientist Sherry Turkle catalogs stories of the
issues caused when workplaces shift more of their interaction to written
text. One such case study focuses on the trials of a technology director
named Victor, who manages a team at a large financial services firm.
“Typically, things get into trouble when too much has been done by email,”
Victor tells Turkle. He keeps having to convince his team that when
problems arise with a client, they need to talk to them in person. “This is
not something they would come to themselves,” he explains. “I’m usually
facing someone who wants to send twenty-nine emails to fix a problem.”
His solution is simpler: “Go talk to them.” As Victor elaborates, his
younger colleagues see electronic communication as a “universal language”
that provides a more efficient way to interact. Increasingly, Victor sees his
role as convincing them that this couldn’t be further from the truth: email is
not a universal form of interaction, he keeps trying to explain; it is instead
an impoverished simulacrum of the types of complex and nuanced
behaviors that through most of human history defined our communication.
We all increasingly feel the effects of this mismatch.32
Email Creates More Work
In 2012, a research team led by Gloria Mark published one of my favorite
studies on the impact of email.33
Their experiment was brilliant in its
simplicity: they selected thirteen employees at a large scientific research
firm and had them stop using email for five workdays. The researchers
didn’t make elaborate contingency plans or develop alternative workflows
in advance of the experiment: they simply shut down the subjects’ email
addresses and sat back to watch what happened.
Though the study includes many interesting results, I want to focus on
an observation that was not reported in the published paper but was instead
brought to my attention more recently in a conversation with Gloria Mark.
As she explained to me, one of the subjects was a research scientist who
needed to spend around two hours each day setting up a laboratory for an
experiment. He reported that he was frequently frustrated because his boss
had the habit of sending him emails during this preparation period, asking
him questions or delegating work. This required the scientist to stop what
he was doing to attend to his boss’s wishes—significantly slowing down the
lab setup. The reason Mark remembers this scientist’s plight was because
during the five days when he was without email, his boss stopped bothering
him during his lab setup. What makes this observation remarkable is that
the boss’s office was only two doors down the hall. The small amount of
extra difficulty required to walk a few steps and poke his head through the
door was enough to prevent the boss from handing off extra work to the
scientist. “He was thrilled,” Mark remembers.
This vignette of the frustrated scientist and his distracting boss
underscores an important cost of email that we often miss. Tools like email
almost completely eliminate the effort required—in terms of both time and
social capital—to ask a question or delegate a task. Viewed objectively, this
seems like a good thing: less effort equals more efficiency. As I’ll show,
however, the side effect of this transformation is that knowledge workers
began to ask more questions and delegate more tasks than ever before,
leading to a state of perpetual overload that’s driving us toward despair.
— One way to examine our changing workloads is to look at the systems we
use to keep track of them. As productivity guru David Allen argues in his
canonical 2001 bestseller, Getting Things Done, the period in which email
spread was defined by significant changes in time management approaches.
As late as the 1980s, the “essence of being organized” involved keeping a
pocket-sized Day-Timer (a paper calendar) and making a daily to-do list to
help figure out how to spend the time in between appointments. Especially
organized workers would use prioritization schemes, such as Alan Lakein’s
ABC Method or Stephen Covey’s Four Quadrants, to help determine the
order in which to complete the handful of tasks they identified as important
for the day.
“The traditional approaches to time management and personal
organization were useful in their time,” Allen notes. But as the 1980s gave
way to the 1990s, the idea that your day could be captured by a short list of
coded tasks became quaint. “More and more people’s jobs are made up of
dozens or even hundreds of e-mails a day, with no latitude left to ignore a
single request, complaint or order,” Allen writes. “There are few people
who can . . . maintain some predetermined list of to-dos that the first . . .
interruption from their boss won’t totally undo.”
Allen rose to fame within time management circles at the same time
that the hyperactive hive mind rose toward ubiquity. He sold over 1.5
million copies of his book in large part because he was one of the first
business thinkers to take seriously how much this new workflow was
increasing the amount of work dumped on our plates. He told his newly
overwhelmed readers that they needed to capture every last one of these
obligations into a “trusted system,” where they could be clarified and
organized—providing the foundation for a frenetic work style in which you
attempt to execute existing items faster than new ones arrive.
Getting-things-done rookies are often shocked by the length of their
task lists. As Allen recalls, in his consulting work, he soon found he needed
two full uninterrupted days to help executives go through and clarify
everything they were supposed to be doing. The process of simply listing
tasks for which they were responsible often took “six hours or more.”35
Gone are the days of the “productive” executive consulting his Day-Timer,
then carefully listing out the six things he hoped to accomplish. In the
modern world, knowledge workers now feel under siege by obligations.
The relevant research literature also helps clarify this sense of
overload. In their original 2004 study on attention fragmentation, Victor M.
González and Gloria Mark partitioned the efforts of the employees they
observed into distinct working spheres, each representing a different project
or objective. They found that on average their subjects worked on ten
different spheres per day, spending less than twelve minutes on one before
switching to another.
A follow-up study in 2005 found the observed
employees touching on eleven to twelve different working spheres per day
on average.37
The large number of different spheres these subjects tackled in
a given day, combined with the reality that each sphere demands the
accomplishment of many smaller tasks and presumably dozens of emails,
provides a harried portrayal of modern knowledge work. “At night, I often
wake in a panic about all the things I need to do or didn’t get done,” writes
journalist Brigid Schulte in Overwhelmed, her 2014 book on this busyness
epidemic. “I worry that I’ll face my death and realize that my life got lost in
this frantic flotsam of daily stuff.”38
This brings us back to my original contention that we can blame email
—or more accurately, the hyperactive hive mind workflow it enabled—for
much of this shift toward overload. One piece of evidence for this claim is
timing. This rise in busyness seems to have occurred somewhere between
the late 1980s and the early 2000s: the same period when email spread
throughout the working world. Another piece of evidence comes from the
experts themselves. David Allen and Gloria Mark, among other relevant
commentators, specifically connect email with our current state of frenetic
We can also identify a plausible mechanism that helps explain how
email might have increased our workload. I opened this section with the
story of the frustrated scientist fending off requests from his boss. When the
scientist’s email was temporarily removed, the boss stopped handing off the
extra requests, even though his office was only two doors down from the
scientist’s laboratory. Simply adding a small amount of friction significantly
reduced the requests coming the scientist’s way. For many knowledge
workers, this story probably makes sense—how many of the quick asks for
someone else’s time and attention that you dash off over email during a
normal day would you still make if you had to instead walk down the
hallway and interrupt someone’s work?
This effect implies there’s something irrational lurking in this system
we use to allocate cognitive resources in the workplace. If slightly
increasing friction drastically reduces the requests made on your time and
attention, then most of these requests are not vital to your organization’s
operation in the first place; they are instead a side effect of the artificially
low resistance created by digital communication tools. The idea that
eliminating friction can cause problems might sound unusual, as we’re used
to thinking about more efficiency producing more effectiveness, but among
engineers like me, this concept is commonly understood. Too little friction
can lead to feedback loops that spiral out of control, as happens when a
microphone gets too close to a speaker and the self-amplification
recursively explodes into a deafening screech.
Something like the workload equivalent of the microphone screech is
happening in modern knowledge work. When the friction involved in
asking someone to do something was removed, the number of these
requests spiraled out of control. I frantically try to grab other people’s time
and attention to make up for the time and attention they’ve already grabbed
from me. Soon everyone is like Brigid Schulte, up late at night, drowning in
the “frantic flotsam of daily stuff.”
What would happen to this “stuff” if some friction was reintroduced to
the system (as in Gloria Mark’s email freedom experiment)? My guess is
that a lot of these urgent tasks would simply disappear: the vital question I
dashed off in a quick Slack message suddenly becomes less vital when
asking it requires me to go interrupt what you’re doing and confront that
look of annoyance on your face. I might drop it or just handle it myself.
Many other tasks would probably get consolidated into more reasonable
chunks. What used to unfold over a few dozen ad hoc messages might
become a larger discussion at a regular status meeting. This is slightly more
annoying in the moment, as you now have to keep track of things you need
help with until the next meeting, but everyone ends up much less distracted.
Friction also motivates the development of more intelligent processes.
Imagine that I frequently need you to sign a certain type of requisition form.
With low-friction communication tools like email, I might simply shoot you
copies of the forms to sign whenever I need them, as this gets the
responsibility off my plate with minimal effort. Without email, however, the
pain of having to come physically find you for every signature will
motivate me to develop a better system, such as one where I put these forms
in your mailbox on Friday morning and you promise to sign them and have
them back to me by Monday morning. This system is much better for you,
as it frees you from yet another source of unscheduled requests for your
time and attention, but it was unlikely to emerge in a setting where just
firing off the forms electronically generated essentially zero cost.
To summarize, we often overestimate the rational nature of our
workloads. If a task is on our plate, we believe, it’s because it’s important—
part of the job. But as I’ve just argued, both the type and quantity of the
efforts that make up our day can be strongly influenced by less rational
factors, such as the relative cost of asks for someone else’s time and
attention. When we made communication free, we accidentally triggered a
massive increase in our relative workloads. There’s nothing fundamental
about these newly increased workloads; they’re instead an unintended side
effect—a source of stress and anxiety that we can diminish if we’re willing
to step away from the frenetic back-and-forth that defines the hyperactive
hive mind workflow.
Clarifying the Misery Mechanisms
Most knowledge workers intuitively feel a sense of unhappiness emanating
from their overflowing inboxes. The reason this reality hasn’t catalyzed a
revolt, however, is that it’s often portrayed as unavoidable—the sine qua
non of work in a hyper-connected, high-tech age. As a 2018 article from the
MIT Sloan Management Review explains: “The ‘keep everybody busy’
theory remains alive and well . . . in knowledge work.”39
(The article
elaborates that the manufacturing sector, by contrast, figured out in the
1980s that relentless busyness was not an optimal way to run things.)
In this chapter, I attempted to push back against this generalized
fatalism by detailing three specific ways in which the hyperactive hive mind
workflow makes us unhappy: the psychological anxiety of an inbox that
fills up faster than we can empty it, the frustrating ineffectiveness of textonly communication, and the out-of-control overload that results when
friction is eliminated from office interactions. When we isolate these
sources of unease, they no longer seem unavoidable; rather, they are
unfortunate and largely unexpected clashes between the specific ways we
work and the natural operation of our brains. The solution is not to shrug
our shoulders, but instead to pursue the obvious fix: replace the hyperactive
hive mind with alternative workflows that still get things done, but sidestep
the worst of these misery-inducing side effects. A world without email, as
we’ll explore in part 2, is largely a happier world. We have one last stop,
however, before we begin this discussion of what works better. The
challenge we take up in the next and final chapter of part 1 is trying to
understand how we ended up wedded to such an unproductive and miseryinducing approach to work in the first place.
Chapter 3
Email Has a Mind of Its Own
The Rise of Email
Why did email become so popular? One clue can be found in an unlikely
place: hidden behind the walls of the Central Intelligence Agency’s original
headquarters building in Langley, Virginia. Here you’ll find more than
thirty miles of four-inch steel tubing, installed in the early 1960s, as part of
an elaborate, vacuum-powered intra-office mail system. Messages, sealed in
fiberglass containers, rocketed at thirty feet a second among approximately
150 stations spread over eight floors. Senders specified each capsule’s
destination by manipulating brass rings at its base; electromechanical
widgets in the tubes read those settings and routed the capsule. At its peak,
the system delivered 7,500 messages each day.
According to oral histories maintained by the CIA, employees were
saddened when, in the late 1980s, during an expansion of the headquarters,
this steampunk messaging system was shut down. Some of them reminisced
about the comforting thunk, thunk of the capsules arriving at a station;
others worried that internal office communication would become
unacceptably slow, or that runners would wear themselves out delivering
messages on foot. The agency’s archives contain a photograph of a pin that
reads “Save the Tubes.”
Why would the CIA invest the significant amount of resources required
to build and maintain such an unwieldy system? By the mid-twentieth
century, much more common and inexpensive methods for office
communication had already become standard. When this headquarters was
built, for example, internal telephone exchanges had been around for
decades. Isn’t it unnecessary to send you a note through a pneumatic tube
network when I could just as easily call you directly using the telephone on
my desk?
But the telephone was no panacea. It represents an example of what
communication specialists call synchronous messaging, which requires all
parties in the interaction to participate at the same time. If you’re not at
your desk when I dial your extension, or if your line is busy, then the
attempted interaction is a bust. In a small organization, tracking people
down on the phone might be manageable, but as the nineteenth century
gave way to the twentieth, single-room countinghouses and small
managerial suites tucked in the backs of factories gave way to huge
edifices, like the CIA headquarters, that could house thousands of whitecollar employees under the same roof. At this scale, the overhead of
arranging synchronous communication becomes onerous, leading to drawnout games of secretarial phone tag and piles of missed-call message slips.
An alternative form of interaction that avoids the overhead problem is
asynchronous messaging, which doesn’t require a receiver to be present
when a message is sent. The intra-office mail cart is a classic example of
this communication type. If I want to send you a note, I can drop it in my
outgoing mail tray when it’s convenient for me, and once it’s delivered to
your incoming mail tray, you can pick it up and read it when convenient for
you—all with no coordination between us required. The problem with the
mail cart, of course, is that it’s slow. It might take the better part of a day for
my note to actually make it from my outbox to a sorting station, then on to a
cart on your floor, where eventually it will be pushed past your desk and
manually delivered. This might be fine for conveying static information, but
it’s clearly an impractical means to efficiently coordinate or share timesensitive news.
What the rise of the large office really needed—a productivity silver
bullet of sorts—was some way to combine the speed of synchronous
communication with the low overhead of asynchronous communication.
Which brings us back to the CIA. This is exactly what they were trying to
achieve with their pneumatic tube system. Their electromechanically
routed, vacuum-driven capsules were the equivalent of a turbocharged mail
cart: I can now asynchronously deliver you a message within minutes
instead of hours. It’s not surprising, therefore, that the CIA employees were
saddened to see the tube system shut down when the headquarters was
expanded in the 1980s. But this sadness didn’t last long, as this same period
marked the arrival of a newer, cheaper, even faster method for practical
asynchronous messaging: electronic mail.2
— Most organizations lacked the resources to build a system similar to the
CIA’s tubes, so for them, the arrival of email was the first time they could
enjoy high-speed asynchrony. We’re so familiar with this capability today
that we take it for granted, but during the 1980s and 1990s, when it began to
spread widely, its impact was profound.
We can find nice snapshots of email’s rapid ascendency in the archives
of The New York Times from this period. One of the paper’s earliest
mentions of this technology in a business context is in a 1987 article that
places the word e-mail in quotation marks throughout.3
“Although ‘e-mail,’
as it is called, has not spread as rapidly as its proponents predicted,” it
explains, “it has established itself as a niche market, and it has a small but
increasing following in the corporate world.” As the article clarifies,
professional email at this point still required a special application that
would dial into a server to establish a connection, allowing you to send and
receive messages before disconnecting. If you needed to reference the
information from a message later, a laborious process was required to save
it to a disk. Given the complexity of this technology at this early stage, the
article’s caution about its importance is understandable. But this soon
Appearing just a few years later is another instructive article—this time
without quotation marks around e-mail.4
The article describes the embrace
of this technology within the entertainment industry. In 1989, we learn,
Mike Simpson, the cohead of the powerful motion picture department at the
William Morris Agency, connected three hundred computers in their
Beverly Hills and New York offices with an early computer network
technology offered by Steve Jobs’s post-Apple start-up, NeXT, Inc. “A
cornerstone of our business is the quicker you get information, the quicker
you can use it,” Simpson says. “E-mail has already given us an edge.”
The article contains other examples of early admiration for email’s
potential. “It’s fast information, replaces telephone calls, is environmentally
correct and allows more people to know things at the same time,” explains
one agent. Another talks about his experience shifting to the rival Creative
Artists Agency, where, to his “horror,” he discovered that they were still
delivering paper notes with runners. He insisted his new colleagues adopt
email. We also learn that at Disney, Jeffrey Katzenberg set up a private
email network connecting twenty high-level executives. “We had to love email because Jeffrey loves it,” explains the vice president of feature
publicity at Disney, before helpfully clarifying: “You communicate by
computer instead of by phone.”
Email was still new enough in 1992 that not everyone understood its
potential. “E-mail is fun, but it’s a toy,” says a story analyst at Columbia
Pictures, providing a quote he probably now wishes he could take back. He
then adds: “E-mail encourages people to chatter and say things that don’t
need to be said.” The article also notes that at this point, most motion
picture studios still depended on a primitive communication device called
the Amtel, a combination of screen and keyboard that was used to send
short text messages. (A common use of the Amtel in Hollywood was to
allow assistants to inform executives, without interrupting their closed-door
meetings, about who was holding on various phone lines.)
In a 1989 article, the venerable technology writer John Markoff
provides more insight into the dynamics that helped accelerate email’s
“Electronic mail, which has taken a secondary position to the
facsimile machine through the personal computer boom of the 1980’s,” he
writes, “is finally coming into its own.” As Markoff’s piece clarifies, in the
late 1980s, email was largely used to connect employees within the same
company. In 1989, under pressure from the Aerospace Industries
Association (a group of fifty aerospace companies with over six hundred
thousand total employees), the main email network providers “grudgingly”
agreed to interconnect their networks using an early email protocol called
X.400, allowing users from one network, for the first time, to communicate
with users from another.
Markoff presciently argues that once email becomes global, it will
largely eliminate the need for fax machines and therefore spread rapidly. He
wasn’t the only one to see this potential. In the article, Markoff quotes
Steve Jobs—identified as “Steven P. Jobs”—providing what turned out to
be an accurate prediction: “In the 1990’s, personal computing will
transform personal communication roughly by the same magnitude that, in
the 1980’s, spreadsheets transformed business analysis and desktop
The case studies in Markoff’s long piece paint a picture of a technology
on the rise. “We found that electronic mail dramatically improved the way
in which we communicated,” explains a hospital executive. “It took off and
permeated our organization.” Markoff later elaborates: “In large and small
offices throughout the country, [email] is being seized on as a means of
communication more efficient than the telephone.”
By 1992, the Times reported that email had become a $130-million-ayear business, projected to be a $500 million business by mid-decade as
many big software companies, including IBM and Microsoft, began
preparing to enter the market.6
A couple of years later, email’s dominance
was unquestioned. “Ever since the Lotus 1-2-3 spreadsheet was anointed a
decade ago as the first killer app . . . people have been asking, ‘What’s the
next killer app?’” writes Peter Lewis in a 1994 article. “In my mind, there is
no doubt: electronic mail is the killer app for the 1990s.”7
As portrayed by these articles, the speed with which email spread
through the business sector is astounding. In 1987, it’s a clunky tool useful
to only a “niche market.” By 1994, it’s the “killer app” of the decade and
the foundation of a half-billion-dollar software industry. That’s about as
close to an overnight transformation as you’re likely to find in the history of
commercial technology adoption.
We shouldn’t be surprised that this tool spread so fast. As I established,
it solved a real problem—the need for high-speed asynchronous
communication—and did so in a manner that was relatively inexpensive
and easy to master.