A world Without Email: Reimagining Work in an Age of Communication Overload

The Protocol Principle
Designing rules that optimize when and how coordination occurs in the workplace
is a pain in the short term but can result in significantly more productive operation
in the long term.
The remainder of this chapter explores case studies of the
protocol principle in action. You’ll learn about the usefulness of
corporate office hours and how restricting clients’ access to you can
make them happier. You’ll also learn what happened when an
academic research group began running structured daily status
meetings like a software development team, and you’ll hear an
argument for why you should never again try to schedule a meeting

---

over email. All these protocols are more complex than just rocking
and rolling with your email inbox or Slack channel, and some make it
more likely that the occasional bad thing will happen. But guided by
Shannon’s fundamental insight, they embrace the central idea that
sometimes a little extra complexity can unlock a lot more
performance.
Meeting Scheduling Protocols
In 2016, I spoke on a panel at a business event. One of my fellow
panelists was a New York–based technology entrepreneur named
Dennis Mortensen. As I learned when we later chatted, he was the
CEO of a start-up that was in the process of leaving stealth mode and
taking on beta testers. It was called x.ai, and its product deployed
cutting-edge artificial intelligence technology to tackle a mundane
task: scheduling meetings.
In its original iteration, x.ai implemented a digital agent named
Amy. When you needed to schedule a meeting with someone over
email, you would cc a special email address connected to Amy and
then, in natural text, ask the agent to help set up the gathering. For
example, you might write: “Amy, can you set up a meeting for me
and Bob next Wednesday?” At this point—and this is where the
magic happens—Amy would interact with Bob over email to find a
time on Wednesday that worked for both his and your schedules,
then add the event to your calendar. This might sound like a minor
improvement to office life, but it attracted major investment. By
2016, when I met Mortensen, x.ai had already spent more than $26
million of investment capital on the Amy natural language interface.
By 2018, they had received $40 million in total investments.
6
There’s a reason why automated meeting-scheduling companies
like x.ai are receiving so much attention from investors: even the
most die-hard hyperactive hive mind booster can’t ignore the raw
time-wasting inefficiency of the way most knowledge workers
currently tackle this increasingly common task. The standard
protocol for setting up meetings is what I call energy-minimizing
email ping-pong. At some point during an email conversation it
becomes clear that a meeting is needed. Because this task is
annoying and non-urgent, all participants involved initiate a game

---

whose unspoken rule is to see how quickly you can bounce the
responsibility for the scheduling to someone else, even if just
temporarily:
“We should meet. Let me know when works for you.”
“Should we shoot for next week?”
“Sounds good to me. Generally speaking, Tuesday and Thursday are probably
best.”
“I’m sort of swamped those days. Friday?”
“Sure, when?”
“Morning?”
“Maybe I could do 11:00 if not too late?”
“I leave for an off-site meeting around then. How does the following week look?”
And so on . . .
The cognitive cost of this protocol is large, as each one of these
back-and-forth messages requires time spent in your inbox. To make
matters worse, once a scheduling conversation is in progress, you
have to check your inbox frequently while waiting for the next
message to arrive, as it would be impolite to disappear for many
hours in the middle of one of these quasi-synchronous back-and-
forth interactions.
It would be bad enough if you had just one such meeting to
schedule at any given time, but in reality, most knowledge workers
find themselves juggling many different scheduling conversations
simultaneously. As reported in a 2017 Harvard Business Review
article, dramatically titled “Stop the Meeting Madness,” the average
executive now spends twenty-three hours a week in meetings.
7
The
sheer volume of the scheduling required to set up those meetings
becomes a major driver of hyperactive inbox checking, and therefore
induces a major cognitive cost. When you have to continually return
to your inbox to nudge along one of many different meeting-
scheduling conversations, your ability to perform valuable cognitive
work significantly diminishes. This is why investors are willing to
spend $40 million to see whether artificial intelligence might

---

drastically reduce this cognitive cost—this price is small compared
with the massive amount of productivity that would be unlocked if
the knowledge sector could abandon energy-minimizing email ping-
pong altogether.
—
When seeking better meeting-scheduling protocols, there are several
solutions that are significantly less costly on average than ad hoc
emailing. The first, and most extreme, is to hire an actual flesh-and-
blood assistant who has access to your calendar and can schedule
meetings on your behalf. There was a time when this option was
prohibitively expensive for all but the highest-level executives, as it
involved paying a full-time salary to a dedicated employee. This is no
longer the case. Online freelance services have made it simple to hire
assistants to work remotely for a limited number of hours on specific
tasks. When I hired my first part-time virtual assistant, using a
service called Upwork, I was surprised to discover that she could
easily handle my meeting scheduling in no more than two to three
billable hours a week. The real cost of meeting scheduling comes
from the numerous interruptions required to check your inbox and
keep the conversations moving, but all these costly interruptions
don’t actually add up to a large total amount of billable time when
handed off to an assistant.
8
Though hourly rates differ depending on the experience of the
assistant, given the reality of how much time is actually involved in
scheduling, it shouldn’t be difficult to off-load the bulk of your
meeting scheduling for around forty dollars a week on average. An
extra $160 a month, of course, is a nontrivial amount of money. In
my experience, the type of knowledge workers most likely to make
this investment are entrepreneurs who are already used to investing
money in themselves and their businesses to keep things growing.
For those who work as employees for large organizations, on the
other hand, the idea of trading your own money for increased
productivity is more foreign, and in this context bringing in an
outside assistant to interact with your colleagues might be viewed
with suspicion, if not outright hostility. That’s why in my
professional life, I use my assistant to manage the overwhelming
number of meeting and interview requests I receive in my writing

---

business, not, for the most part, to deal with the demands of my
other job as a university professor.
Successfully working with a part-time assistant to schedule
meetings requires two things: access to your availability and a way to
add new events to your calendar. There are many tools that can
satisfy these requirements. I’ve been using an online scheduling
service called Acuity. At the beginning of each semester in which I’ve
hired an assistant, I’ll manually enter into the system all the times
I’m available for meetings in the months ahead. When my assistant
needs to schedule a meeting, she uses Acuity to select a block within
these available times. What makes this service useful is that it
synchronizes with my Google Calendar. When my assistant books an
appointment in Acuity, it shows up automatically on my calendar.
Equally important, if I directly book something on my calendar,
Acuity automatically removes that block of time from my availability.
The obvious question, of course, is why I don’t just directly use
Acuity to accelerate my meeting booking: if someone wants to meet
with me, instead of passing them off to my assistant, I could just
send them straight to Acuity to book a meeting time that works for
both of us. The reason I don’t default to this simpler and cheaper
option is that I work with a diverse set of possible appointments, and
they’re not all created equal from a scheduling perspective. When
booking a meeting that will be held in my Georgetown office, for
example, I want to consider only times when I’m on campus. When
booking a podcast interview, by contrast, I want to do the opposite,
offering only times when I’m working from home and can make use
of my in-home studio. Some meetings are urgent, and I want to find
the nearest available time slot, while others are not, and I want to
defer them to a less crowded period in the future. It wouldn’t work
for me to respond to each meeting request with a list of all times that
I’m available; I can instead let my assistant navigate these different
demands.
For most knowledge work jobs, however, these types of
distinctions are less relevant. You have a standard workweek during
which you block off some times for uninterrupted work, leaving the
rest open for meetings and appointments. In this case, there really
isn’t a need for an actual human to help you with your scheduling.
Tools such as Acuity, ScheduleOnce, Calendly, and, of course, x.ai (to
name a few examples among many) make it easy for other people to

---

set up meetings with you during times when you’re available. When
someone requests a meeting, you simply send them a link to your
scheduling service and tell them to pick whatever time works best for
them. Days of energy-minimizing email ping-pong have now been
reduced to a single message and some clicking on a scheduling
website.
If the meeting involves multiple people, then avoiding email
ping-pong becomes even more urgent, as the number of messages
required for scheduling often increases exponentially with the
number of attendees. In these cases, it’s worth using a group polling
service like Doodle. For those who are unfamiliar, these services
require you to set up an online poll by entering in multiple date and
time options that work well for your calendar. You then send the poll
to the other meeting participants, who each check off which of these
times work for them, allowing you to easily identify a time that works
for everyone.
I would go so far as to say that anyone whose job requires more
than one or two scheduled events in a typical week absolutely should
be using a scheduling service or, if the work demands it, a part-time
assistant. There’s really no reason why anyone should still have to
waste cognitive cycles in dragged-out scheduling conversations. You
might think that the gains here are small—how hard is it to send