Efficiency and flexibility are opposing qualities in a team or business. Slack represents how far
from perfectly efficient you are willing to go, to achieve some level of flexibility.
You can keep a whole work force busy (fully utilized) if you allow work to buffer at each point in
the network (on each person's desk, or on each team's backlog). This has the downside of reducing
the velocity of tasks through the system.
Most employees equate control in their career as their main opportunity for growth
"You have to give her some leeway, some opportunity to choose her own directions and make her
own mistakes. Mistakes are important here. If she has control over her choices only to the
extent that she makes the same ones that you would have made for her, she has no control at
all."
This can be called "control slack" — enough slack in the system to absorb suboptimal choices,
which provides the benefit of employee control.
"Slack is the way you invest in change. Slack represents operational capacity sacrificed in the
interests of long-term health."
On too much process:
"As defense against failure, standard process is a kind of armor. The more worried you are about
failure, the heavier the armor you put on. But armor always has a side effect of reduced
mobility. The overarmored organization has lost the ability to move and move quickly."
The first requirement of vision is to know who we are. Someone who can say "product X might be
nice, but it just isn't us."
Prelude
The efficiency-flexibility quandary: the more efficient you get, the harder it is to change.
Resource availability implies some inefficiency.
You increase efficiency by eliminating slack in the system. But slack is required for change.
Madmen in the halls (chap 1)
The social convention for the effectiveness of someone is busyness — "utilization". Think of
Hollywood's standard depiction of a successful company: every employee is running around and "on
it."
Flattening the hierarchy and removing middle management for efficiency wins are rewarded in the
short-term by the stock market.
In cutting middle management, we cut the capacity to change.
"The main activity of those managers is reinvention. It is the middle of the organization where
reinvention takes place. This is where the dynamic of today's organizational functioning is
examined, taken apart, analyzed, resynthesized."
Busyness (chap 2)
"Very successful companies have never struck me as particularly busy; in fact, they are, as a
group, rather laid-back. Energy is evident in the workplace, but it's not the energy tinged with
fear that comes from being slightly behind on everything."
The value of a secretary that's not entirely utilized
"She's available to do stuff that you or your people find you need to have done. That's part of
what's so great about Sylvia: when something comes up, she can usually get cracking on it right
away."
You can keep a whole work force busy (fully utilized) if you allow work to buffer at each point in
the network (on each person's desk, or on each team's backlog). This has the downside of reducing
the velocity of tasks through the system.
The myth of fungible resources (chap 3)
A common strategy in restructuring is to allocate people into a pool so their partial time can be
finely allocated to stakeholders.
"A highly partitioned worker can't be obsessively involved in any of his/her many fragmented
tasks, and so tends not to bind into the team. Whatever the productivity boost due to team action
may be, the partitioned worker does not benefit from it."
Task switching penalty =
mechanics of moving to a new task +
rework due to inopportune abort of previous task +
immersion time for think-intensive tasks +
frustration (emotional immersion in previous task) +
loss of team binding effect
From a live study of software developers in 1984, "we modeled performance against task-switching
frequency and came to the conclusion that the best way to understand task switching was to assume
that each switch imposes a direct penalty of 20m of concentration."
So, restructuring which increases fragmentation of work for each worker carries a large
productivity penalty.
Author estimates ("in his experience") that no less than 15% loss in productivity (6hrs/week)
occurs from having a person do two jobs versus one (i.e. cover two bases, or work with two teams).
The penalty increases with many jobs.
Managing Eve (chap 5)
For curious, hungry employees, work must be structured to provide growth. It's as essential as the
paycheck.
On trying to control the work of volunteers in a non-profit, which robs them of their motivation:
"If you looked over their shoulders enough or imposed standards that were different from their
own, they would shrug and walk away from the work, leaving you to do it yourself. Control, as
they see it, is their payment for working."
Most employees equate control in their career as their main opportunity for growth
"You have to give her some leeway, some opportunity to choose her own directions and make her
own mistakes. Mistakes are important here. If she has control over her choices only to the
extent that she makes the same ones that you would have made for her, she has no control at
all."
This can be called "control slack" — enough slack in the system to absorb suboptimal choices,
which provides the benefit of employee control.
"You have to create a real sense that control is not completely centralized in your hands, but
spread generously over the whole of your organization. Like a gifted helmsman, who knows that all
use of the rudder increases drag and thus holds the vessel back, you have to steer with the
lightest possible touch."
It is not healthy for the management hierarchy lines to be the only communication conduit between
partner orgs. These hierarchies are far too narrowband.
Business instead of busyness (chap 6)
The benefits of having slack designed into the organization:
Responsiveness
Flexibility
Better people retention
A capacity to invest
Ability to change is valuable, but it costs money. How much should you invest in having an ability
to change?
Ability to change cannot be siloed to dedicated "change specialists." It should be implemented in
a diffuse way, as some portion of everyone's mindshare.
"Everybody needs to have some capacity to devote to change. This is time that people dedicate to
rethinking how their piece of the whole works, and how it ought to work. Once the change is
under way, more time is required to practice new ways and master new skills. That's the cost.
The benefit is vitality and a firm grip on the future."
"Slack is the way you invest in change. Slack represents operational capacity sacrificed in the
interests of long-term health."
"The right way to think about domain knowledge is as a corporate capital asset, as dollars of
investment in the head of each knowledge worker, put there by organizational investment in that
employee. When that person leaves, the asset is gone. If you did a rigorous accounting of this
human capital, you would be obliged to declare an extraordinary loss each time one of your people
quit."
The cost of a departing employee is: their salary * (rampup time / 2) + opportunity cost +
training cost.
Multiply this by churn per month, and this is the cost the company is paying for human capital
drain.
High utilization of a worker makes them feel "used", and leads to churn.
"I am much more concerned when smaller companies invest outside of their own product areas. I see
this as bankruptcy of inventiveness. It is particularly evident when companies find themselves
with extra leverage due to run-up of their stock price. Their willingness to spend this found
capital outside their own backyard is a signal that they have no real vision, no idea of how to
grow in the arena that they know best."
This bankruptcy can happen if there's not enough slack for resources to be set aside for
invention.
"Companies keep their people too damn busy to invent."
The cost of pressure (chap 7)
Management adding pressure doesn't produce much gain; maybe 10-15%, according to the author's
experience. In a dysfunctional org where there's lots of wasted time, it might deliver a larger
improvement.
Supplying additional pressure doesn't free up as much time as you'd expect within an already
functioning org.
"In a healthy knowledge-worker organization, people don't waste a lot of time anyway, since
wasted time is an affront to them as much as it is to their management. They are more likely to
be frustrated by wasted time than enjoy it."
Overtime (chap 9)
Sprinting at the end of a marathon makes sense, but not the full duration of the marathon.
Many types of valuable knowledge work are error-prone, and can't be done well in the later
portions of a long work day. So a longer day doesn't increase the output for this type of work.
"When we measure human capital carefully and use the measurements to give a cash quantification to
personnel turnover, it often becomes the second or third largest cost category."
Overtime increases turnover; people feel used because they their personal lives took a hit.
Eliminating overtime puts pressure on the work culture, to be more efficient, so that deadlines
can be hit. E.g. fewer, smaller meetings, and higher friction to interrupt people.
Power sweeper (chap 11)
It's bad optics to hire clerks and secretaries, but we really should. High-paid managers shouldn't
be doing their own tech support or tweaking fonts in a deck.
"We have become so obsessed with getting rid of people who are burdened with the
characterization of overhead that we have ended up with organizations where many high-priced
knowledge workers and managers are spending as much as a quarter of their time being their own
overhead."
Low-level support roles should add less interaction overhead to a team than a full IC, because the
interaction paths are "thinner" between them and each full IC, because interacting with a support
role should require less maintenance and brain power.
His argument: "by dividing up the work into four pieces instead of five, the total time spent
serving the interactions is reduced. The capacity of the four-person team plus clerk is thus
greater than that of the five-person team. Four developers plus a clerk also cost less than a
five-developer team."
The second law of bad management ("put yourself in as your own utility infielder") (chap 12)
The mindset that leads someone senior to spending so much time on lower-level responsibilities:
"All the rest of your people are busy as hell; you don't want to further burden them with
another task, particularly not one that upper management found to be of so little importance
that they 'trimmed' the person who was doing it. Yet that 'whatever' still has to be done. Oh
well, you add it to your own burden and do it yourself."
"Assigning yourself to an unfilled position in your domain means that you unassign yourself (at
least partially) from the task of managing that domain."
Organizations discount the value of management, because managers are not making the product, and
their interventions interrupt that process in the short term. But the reality is that the changes
made by good managers greatly increase output in the long run.
"Getting rid of management to save cost is like losing weight by giving blood."
Managers do lower-level jobs partly because they under-value the management work they're doing.
Blending management with your old IC work
"Now I was not only a manager, but able to spend my days doing sublimely black-and-white work.
It seemed like the best of all worlds. But it wasn't. I was walking away from the challenge of
management to return to work I knew cold. The relief it gave me was the relief of retreat."
Why management is hard
Unlike engineering, the problems are not well-defined, and what makes a good solution is not
black and white.
The skills are inherently difficult to master.
Culture of fear (chap 13)
The paradox of over-staffing
"The nature of project work is that whatever it is you're about to build, the early conceptual
phases are crucial. But this kind of conceptual work can't be done with a crowd of people. A
staff of no more than six might make perfect sense while the first-cut design decisions are
made. Burdening the project with an extra fifty people at this stage will only make the work go
slower."
In response to a larger-than-necessary team: "You're forced to partition the whole — this kind
of partitioning is the essence of design — along lines that are dictated by personnel-loading
considerations rather than design considerations. The result is sure to be a mediocre or poor
design, something that will encumber the project from this point on."
Litigation (chap 14)
In an environment where the bottom X% get cut each year, "any failure by managers beside you on
the org chart thus has the effect of reducing pressure on you."
Process obsession (chap 15)
"Process obsession — developing a standard process for each problem — is an epidemic. It's
ill-suited for knowledge work."
Product standards — the interfaces of things — are very valuable. How-to standards are not.
On letting the team own its process, instead of standardizing a process across all of the teams,
such that each team loses some ownership:
"In both of these examples, ownership of the process is pushed downward. Instead of being a
corporate asset, it is a team asset. Somewhat lost is the flexibility to move people from one
team to another, since over time the teams may begin to deviate substantially from each other.
Offsetting this loss is a much more interesting workday for the workers, enhanced identification
wit the product (and its customer), lower turnover, and strongly felt loyalties to the team and
to the corporation."
Empowerment means "putting process ownership largely into the hands of the people doing the work.
That doesn't mean there should be no standard, only that whatever standard evolves should happen
at the level of the work itself. Ownership of the standard should be in the hands of those who do
the work."
"Process standardization from on high is disempowerment. It is a direct result of fearful
management, allergic to failure. It tries to avoid all chance of failure by having key decisions
made by a guru class (those who set the standards) and carried out mechanically by the regular
folk. As defense against failure, standard process is a kind of armor. The more worried you are
about failure, the heavier the armor you put on. But armor always has a side effect of reduced
mobility. The overarmored organization has lost the ability to move and move quickly."
Quality (chap 16)
He argues that the corporate headline of "quality programs" really means headlining "defect
prevention." While important, this tends to crowd out the team's and company's focus on much more
impactful things of product value, like uniqueness, and usefulness to the customer.
"Defect prevention and removal efforts may add sufficient overhead to the overall process so that
it is too slow and unresponsive to market needs."
"Any risky new endeavor is likely to result in an increasing pattern of defects; the Quality
Program thus may pit itself squarely against risk-taking efforts."
QA is tacked on at the end of a project and doesn't have the influence to extend the delivery time
to ensure QA is done well.
Efficient and/or effective (chap 17)
"Directing an entire organization is hard. Seeming to direct it, on the other hand, is easy. All
you have to do is note which way the drift is moving and instruct the organization to go that
way."
There is an "unfortunate side effect to optimization, first noted by geneticist R. A. Fisher, and
now referred to as Fisher's fundamental theorem: 'The more highly adapted an organism becomes, the
less adaptable it is to any new change.'" (e.g. the giraffe).
"The more optimized an organism (organization) is, the more likely that the slack necessary to
help it become more effective has been eliminated."
Management by objectives (chap 18)
"A fad from the 1950's" (ha)
It's a tool for incremental changes along the current direction.
"MBO is always based on stasis, the organization's present steady state. MBO sends the message 'Do
everything the same as last year, only this year do more of X.'"
I.e. it assumes the current direction is correct, and that we should just travel faster.
"Stasis plus just a bit of tinkering with the particulars is hardly a recipe for success in the
new economy."
Rather than a company being a vehicle of "production", which implies an ongoing steady-state, "the
new central organizing principal is the project. A company in this kind of flux can be viewed as a
portfolio of projects. Each project seeks to effect some change, and there is no longer a
long-running status quo.
Vision (chap 19)
The first requirement of vision is to know who we are. Someone who can say "product X might be
nice, but it just isn't us."
Drucker argues that culture is the one thing that cannot and should not change. It's the bedrock.
"When a vision statement walks perfectly between what is and what could be, and the could-be part
is wonderful but not impossible, acceptance by those listening is almost assured."
"People want to sign on. They want to be enrolled."
Leadership and "leadership" (chap 20)
"Meaningful acts of leadership usually cause people to accept some short-term pain (extra cost or
effort, delayed gratification) in order to increase the long-term benefit. We need leadership for
this, because we all tend to be short-term thinkers."
"Blueprints" are a powerful and necessary part of leadership, because they make actionable the
challenging proclamation.
Fear and safety (chap 22)
Starting with a new way of doing things: "people can make this kind of change, but they can only
make it if they feel safe. In an unsafe environment, people are not likely to let themselves be
thrust into a position of inexperience."
Change has an intrinsic cost. While time may have made the old way no longer the best way, the old
way has the advantage that it's familiar and people have mastered it.
Trust and trustworthiness (chap 23)
Effective leaders quickly acquire trust by giving it
"The giving of trust is an enormously powerful gesture. The recipient gives back loyalty as an
almost autonomous response."
Good leaders "give responsibility well before it's been completely earned. They know when to
turn their backs and take their chances."
This is how parents progressively empower their kids — with responsibility and trust on the
edge of their capabilities.
Timing of change (chap 24)
In periods of growth, people feel like they're winning, and have less anxiety about change. Change
is perceived differently, as "growth-related change", e.g. necessary scaling and refactoring now
that our team or business is larger. It's good. This is the easiest time to introduce chance.
Introducing change during decline carries the added baggage that people are anxious about
corporate health, and maybe their jobs.
What middle management is there for (chap 25)
Middle managers are the change agents — the implementers of reinvention. If they get trimmed for
the sake of efficiency, the org will find it much harder to execute change.
Companies "have hurt themselves by encouraging their middle managers to stay extremely busy. In
order to enable change, companies have to learn that keeping managers busy is a blunder."
They need slack for non-routine or unplanned work; reinvention.
Danger in the white space (chap 27)
Competition between middle managers, creating silos
"Over the years, I have come to believe that this kind of competition is almost never explicitly
designed into an organization; rather, it happens without anyone really wanting it."
"In knowledge work, all internal competition is destructive." It inhibits cooperation, which is
essential in knowledge work.
"Slackless organizations tend to be authoritarian. When efficiency is the principal goal, decision
making can't be distributed. It has to be in the hands of one person (or a few), with everyone
else taking direction without question and acting quickly to carry out orders. This is a fine
formula for getting a lot done, but a dismal way to encourage reinvention and learning."
A key element of training is the slow-down characteristic: the person learning is given the space
to perform a new skill at a much slower speed than an expert. Argues that efficiency-obsessed orgs
do not allow for long, slow training/ramp-up periods.
Uncommon sense (chap 29)
Don't expect your organization to overcome all adversity. It's unrealistic, and ends up
discouraging risk-taking.
"An organization that has suffered no important setbacks has in fact taken no real risks."
"Risk management" means planning for failure, and deploying enough slack in the system to
accommodate failure.
Stochastic control: control over a variable (e.g. employee churn), with influential, but imperfect
levers (e.g. employee comp). This form of control doesn't feel sufficient when managing a single
team or project, but at a company level with a portfolio of teams and projects, it can be
sufficiently effective.
"Risk management is the explicit quantitative declaration of uncertainty."
Delivery date of a project should be shown as a probability distribution function.
Risk management: the minimal prescription (chap 30)
"Risk diagrams": for each component, a list of outcomes, and a probability for each.
Risk containment: when risk materializes and costs you unplanned money or delay, this gets paid
out of a reserve (i.e. a buffer). Intentional risk planning sets aside an explicit reserve.
"The major effort is to keep the risk reserve from being eliminated by someone who wants
desperately to hear lower numbers."
In corporate life, there is constant downward pressure on explicit risk reserves.
Many risk containment strategies require pre-materialization work. Like installing fire
extinguishers in a school before the crisis of a fire occurs. This work is often not prioritized,
because we're busy.
Learning to live with risk (chap 32)
Management by default favors zero-risk projects. The author argues those resources are better
spent on big, transformational wins, given the modern business environment.
"The only new initiative you can afford to take on today is one that is full of risk. It's got
to be something that thrusts you into a new market or exploits a brand-new technology, one that
transforms your company."