THINKING IN SYSTEMS

A Primer

DONELLA H. MEADOWS

The Big “So What”

KEY QUOTES

We live in a complex world of inter-locking systems. Systems thinking “We know a tremendous
helps us to understand how things work, so we can better identify amount about how the
root problem causes, see new opportunities, make better decisions world works, but not nearly
enough.”
and adapt to changing circumstances. In this summary, we will:

• Give an overview of systems thinking concepts and principles; and

• Explain how to apply systems thinking in real-life.

Systems: Concepts and Principles

What are Systems?

Definition: A system is a set of interlinked elements that are

organized to achieve a goal.

• Every animal, plant, organization and society is a complex

system. A system’s structure defines its behavioral pattern.
While systems may be influenced by external forces, the way
they respond to these forces tend to come from their inherent
characteristics.

• This means that the economy will move in cycles regardless of “Systems happen all at once.
They are interconnected not
what political leaders do. Companies that lose market share
just in one direction, but in
would’ve done so regardless of how competitors respond. many directions
Unless we can see how systems operate and create their own simultaneously.”
problems, we can’t identify and address the root causes. We’ll
be like the proverbial blind men, each touching a different part
of an elephant and drawing different conclusions about what
an elephant is like.

Copyright © 2019 Skool of Happiness Pte Ltd. • All Rights Reserved.

A system is more than the sum of its parts. Its elements are
interrelated, e.g. the organs in your digestive system jointly break
down the food you consume. Systems can be a part of larger
systems, and may be able to achieve goals, evolve or even self-
repair.

• System elements can be physical (e.g. machines, people,

products) or intangible (e.g. culture, values, skillsets).
KEY QUOTES
• Due to the sheer number of elements and sub-elements,
it’s more useful to examine the interconnections between “The behavior of a system
cannot be known just by
elements—these may be physical (e.g. how products move knowing the elements of
through a factory) or information-based (e.g. how a customer which the system is made.”
decides to buy something).

• A system’s purpose is the hardest to decipher—the only way “Purposes are deduced from
to do so is to observe exactly how the system behaves. Since behavior, not from rhetoric
or stated goals.”
each sub-system has its own goals, a system can have a
complex network of sub-systems and goals. For example, the
students and staff in a university will have different goals from
the university itself.

• Elements are the most observable but least crucial. The greatest “A change in purpose
impact comes from changes to the system’s purpose, followed changes a system
profoundly, even if every
by its interconnections and elements. element and interconnection
remains the same.”
Stocks, Flows and Feedback Loops

Stocks and flows help us to understand how a system behaves.

• A stock is an observable or quantifiable aspect of a system, e.g.

population size or inventory levels. It can be accumulated or
depleted, and thus captures the history of inflows and outflows
in the system. Stock levels rise when inflow is greater than
outflow and vice versa. Stock levels stay constant and achieve
dynamic equilibrium when inflow = outflow.

• The best way to understand a system is to (i) map out a system

diagram and (ii) plot the variables/stocks on a time graph. For
example, the simple diagram below shows the key factors
that affect the water level in a reservoir. When you examine it
alongside time-graphs, you can get a better grasp of how each
variable affects the water levels.

Copyright © 2019 Skool of Happiness Pte Ltd. • All Rights Reserved.

 Inflow Outflow

Rainfall Water in Evaporation

reservoir
River Inflow Discharge

Water
level (ft)

900 Reservoir water level over time

KEY QUOTES
850 “A stock is the memory of the
history of changing flows
within the system.”
800

Jan Apr Jul Oct Months

• Stock levels tend to change gradually, thus providing stability or “A stock takes time to
momentum in a system. We use stock levels to make decisions change, because flows take
(e.g. offer discounts when there’s excess inventory, spend less time to flow.”
when the bank balance is low). Remember: flows take time
and things can only happen as fast as they happen. To achieve
your desired stock levels, you need to take such delays into
account.

A feedback loop is formed when a change in stock affects the

inflow/outflow for that stock, i.e. it triggers a process to increase,
decrease or maintain the stock level. When you see a consistent
pattern of behavior, there’s always a feedback loop at work.

• A balancing loop is goal-seeking. For example, a thermostat

regulates temperature—it warms up a room if it gets too cold,
and cools it down if it gets too hot. Such loops bring stability
but also resist change.

• A reinforcing loop amplifies an existing effect to trigger a

virtuous or vicious cycle. For example, the more people catch
a virus the faster it spreads, or the more you save the faster
your bank balance grows from the compound interest. Such
loops bring exponential growth but can also cause things to
spin out of control and collapse.

Systems Behaviors
“Systems with similar
feedback structures produce
Just like how animals in the zoo give us a snapshot (but not the similar dynamic behaviors.”
full picture) of wildlife, systems diagrams represent a simplified
3

Copyright © 2019 Skool of Happiness Pte Ltd. • All Rights Reserved.

model of the world. Two systems that appear very different can
exhibit similar patterns so long as they have similar feedback
loops.

A one-stock system can have many feedback loops.

• During winter, the room temperature is affected by the

thermostat setting (which controls heat released from the
heater) and the outside temperature (which affects heat loss). KEY QUOTES
These are 2 conflicting balancing loops that affect the room
temperature. A feedback loop can break down (e.g. if all the
windows are opened to cause excessive heat loss).

Heat from Room Lose heat

heater Temperature to outside

Gap between Gap between

Thermostat actual & inside &
Outside
setting desired outside
temperature
temperature temperature

• A country’s population is determined by the nett effect of a “Dynamic systems studies

reinforcing loop (# births) and a balancing loop (# deaths). These usually are not designed to
predict what will happen.
are in turn driven by fertility and mortality rates. It’s always Rather, they’re designed to
easier to predict behavioral patterns than actual numbers, explore what would happen
e.g. the exact population size could fluctuate as (i) fertility/ if a number of driving factors
mortality rates change over time, (ii) a feedback loop becomes unfold in a range of different
ways.”
more dominant, and/or (iii) additional factors emerge (e.g.
migration).

Population
# Births Size # Deaths

Fertility Mortality
Rate Rate

• There’s always a time lag between something happening and “We can’t begin to
the feedback loop registering it and taking corrective action. In understand the dynamic
behavior of systems unless
balancing loops, a delay in 1 part of the system is likely to lead we know where and how
to oscillation (e.g. delay in product delivery → shop runs out of long the delays are.”
stock → perceived shortfall → place larger order → overstock).
This effect, when combined with imperfect information and
multiple feedback loops operating concurrently, can create
massive ripples throughout a system.

A two-stock system can be much more complex, especially when “Any real physical entity is
always surrounded by and
1 stock limits another stock, e.g. a mining company’s money being exchanging things with its
limited by the availability of precious metals in the mines. environment.”

Copyright © 2019 Skool of Happiness Pte Ltd. • All Rights Reserved.

 • No physical entity/system can grow indefinitely—at some point,
balancing loops will kick in to limit growth, e.g. a company can
only grow insofar as it has manpower, customers, materials and
resources. The nature of the constraints depend on whether the
critical resources for growth are renewable or non-renewable.

• Non-renewable stocks are fixed in supply, e.g. once an oil field

is depleted, there’ll be no more oil. Your rate of extraction is
only limited by your capital or resources. However, there’s KEY QUOTES
an inherent balancing loop in place: the faster you grow, the
faster you’ll drain the resources, making it progressively harder “A quantity growing
to grow until the stock is totally depleted. exponentially toward a
constraint or limit reaches
that limit in a surprisingly
• Renewable stocks can be living (e.g. fish or grass) or non- short time.”
living (e.g. sunlight or wind), and can be regenerated through
a reinforcing loop or a constant supply. They’re limited by
flow, i.e. they’re inexhaustible so long as they’re harvested at
a rate slower than their rate of regeneration. However, once
they’re driven below a critical threshold, they can become
non-renewable. For example, in overpopulated waters, fishes’
rate of reproduction may improve initially if a slight decrease “Nonrenewable resources
in fishes leaves more food for the remaining fishes. However, are stock-limited…
once the population drops beyond a critical point, reproduction Renewable resources are
plummets and the fish can become a non-renewable resource. flow-limited.”

Applying Systems Thinking

3 Characteristics of Effective Systems

Systems thrive with 3 key characteristics: resilience, self-organization

and hierarchy.

Resilience in this book refers to elasticity, or the ability to bounce

back to the original state after being strained or manipulated.

• Resilient systems are dynamic. A system is more resilient when “Systems need to be
(i) it has multiple feedback loops (so when 1 loop fails, another managed not only for
productivity or stability, they
kicks in) and (ii) there’s at least 1 loop that can repair other loops. also need to be managed
The most resilient systems are those that can self-organize. for resilience—the ability to
recover from perturbation,
• Unfortunately, resilience is not obvious unless you have a full the ability to restore or
repair themselves.”
system perspective. We tend to sacrifice resilience for stability
or productivity. For example, growth hormones can increase a
5

Copyright © 2019 Skool of Happiness Pte Ltd. • All Rights Reserved.

 cow’s milk-production but reduces its long-term immunity.
Pesticides kill pests, but they also kill off other species to
disturb the eco-system’s bio-chemistry, reduce its resilience
and trigger long-term ecological disasters.

• Look beyond productivity or observable properties to

consider how you can strengthen a system’s resilience and self-
regeneration capabilities. For example, instead of simply giving
food and money to the needy, develop programs that increase KEY QUOTES
people’s ability to generate their own food and money. “Productivity and stability
are the usual excuses for
Self-organization is the ability to learn, evolve and create new turning creative human
complex systems/structures to survive, rebuild and thrive. beings into mechanical
adjuncts to production
processes.”
• For example, the human body is developed from a single cell
and constantly repairs itself. A company converts raw materials
into a much more sophisticated form (e.g. a car or computer).

• Self-organization is spontaneous and unpredictable; it requires “The ability to self-organize

freedom and experimentation. However, complex creations is the strongest form of
are often built upon a few simple rules, like how all lifeforms system resilience. A system
that can evolve can survive
are captured in a series of DNA, RNA and protein molecules. almost any change, by
changing itself.”
Complex systems require hierarchies to organize the huge
amounts of information/interactions and operate stably.
“Hierarchical systems evolve
• For example, each cell in our body is part of an organ, which is a
from the bottom up. The
part of a system (e.g. digestive or reproductive system), which purpose of the upper layers
is a part of our human body. Each of us in turn is a part of a of the hierarchy is to serve
family, community, nation etc. Ideally, each sub-system should the purposes of the lower
be able to manage itself while serving the needs of a larger levels.”
system. The larger system in turn should integrate/augment
the sub-systems synergistically.

• A system is sub-optimal when hierarchies don’t work in sync. In “When a subsystem’s goals
such cases, imposing centralized control won’t work and may dominate at the expense of
the total system’s goals, the
even backfire. The only sustainable approach is to understand
resulting behavior is called
and address the needs of the sub-systems so the entire system suboptimization.”
can flourish dynamically. When hierarchies malfunction, the
entire system can break down, e.g. when certain cells in the
body multiply out of control and attack other cells.

Copyright © 2019 Skool of Happiness Pte Ltd. • All Rights Reserved.

 Thinking in Systems

We tend to get surprised by systems because we over-estimate

how much we know and under-estimate how complex systems
are.

• Real-world systems are complex networks of stocks, flows,

and feedback loops. Systems thinking helps us to consider KEY QUOTES
the relationship between system structures, behaviors, and
“Structure is the key to
events. By asking what-ifs, we can consider possible scenarios
understanding not just what
so we’re less likely to be surprised. is happening, but why.”

• Yet, in most cases, it’s impossible to accurately predict the future.

For example, there’s no way to account for all the variables,
relationships and events that affect economic growth or stock
market indices. That’s why complex modeling may still work
for short-term predictions, but not for long-term forecasts.

Watch out for these thinking gaps:

• Linear vs nonlinear relationships. We tend to think in linear “The world is nonlinear.

relationships (i.e. assume things run in a straight line) when Trying to make it linear for
our mathematical or
most relationships are actually non-linear (i.e. there’s no
administrative convenience
proportionate relationship between variables). For example, is not usually a good idea.”
once vehicle density reaches a certain level, traffic jams get
much worse.

“Remember that boundaries

• Imaginary boundaries. For simplicity, system diagrams use clouds
are of our own making, and
to denote the stocks at the start/end. In reality, there’s no that they can and should be
finite start or end to a system. For instance, the earlier diagram reconsidered for each new
on population size doesn’t address the drivers for mortality/ discussion, problem, or
death rates, nor other factors that affect population size e.g. purpose.”
migration. It’s unrealistic to try and capture everything—the key
is to be aware that all system boundaries are imaginary.

• Multiple constraints. Every outcome is influenced by multiple “Ultimately, the choice is not
factors, but we tend to think in terms of singular cause-effect to grow forever but to decide
relationships. Over time, a system’s limiting factors (what’s most what limits to live within.”
vital for growth) will also change, e.g. a startup faces different
constraints as it grows to a mid-sized firm. At some point, you
must choose your limits (e.g. the maximum population size to
maintain acceptable living standards), or the system will decide
it for you (e.g. resource shortages, declining air quality).

Copyright © 2019 Skool of Happiness Pte Ltd. • All Rights Reserved.

 • Pervasive delays. Things take time to happen, and there’re
delays in most feedback loops—delays in shipping, perception,
processing etc. Mail takes time to reach the recipient and viruses
take time to develop into diseases with observable symptoms.
If there’re long delays in the feedback loop, it’ll be too late
to take action when the symptoms surface, e.g. if you only
start to build power plants when there’re electricity shortages,
you’ll end up with cycles of under-capacity and over-capacity.
KEY QUOTES
• Bounded rationality. We make the best possible decisions
based on what we know to satisfy a current need. However, “To act only when a problem
becomes obvious is to miss
we (i) won’t have perfect information, (ii) can’t see all options, an important opportunity to
viewpoints or implications, and (iii) will always interpret info solve the problem.”
subjectively. Thus, a sound decision that meets a personal
short-term goal can create long-term problems for ourselves
and others. The best way to get a bigger picture is to experience
“Change comes first from
the system from a different position, e.g. an activist going to stepping outside the limited
work for the government body he’s been campaigning against. information that can be seen
Each shift gives you a wider system perspective to make better from any single place in the
decisions. system and getting an
overview.”

Identifying System Traps and Opportunities

Learn to recognize the “behavior archetypes” or system structures

behind common problems, so you can deliberately restructure them
to fix the problems at cause.

Resistance to change. Balancing feedback loops can sustain

behavioral patterns to make them change-resistant.

• In 1967, the Romanian government tried to increase birth rates “The most effective way
by making abortions illegal for women under 45 years old. of dealing with policy
resistance is to find a way
There was a temporary spike in birth rate, then people turned
of aligning the various goals
to illegal abortions or abandoned the children they couldn’t of the subsystems, usually
afford to raise. So long as the sub-systems have different goals by providing an overarching
with players pulling in different directions, there’ll be policy goal that allows all actors to
resistance. The harder 1 player pushes an outcome, the harder break out of their bounded
the other players in the balancing loop will resist. rationality.”

• A better approach is to step back, identify all the players’ goals

and how to incentivize them to move in the desired direction. For
example, to increase birth rates, you need to know why people
aren’t having more children. In the 1930s, Sweden’s birth rate
was declining. Instead of trying to forcibly increase family sizes,

Copyright © 2019 Skool of Happiness Pte Ltd. • All Rights Reserved.

 the Swedish government rallied the country behind the
common goal of improving the quality of child care so every
child is wanted and nurtured.

The “tragedy of the commons” comes about when a shared stock

is eroded or destroyed because people exploit more and more of
the stock instead of conserving it.

• Such problems arise in any common resource (e.g. public parks, KEY QUOTES
common toilets) where users benefit directly from the stock,
but share the cost of stock depletion. For example, a farmer “The tragedy of the
commons arises from
can directly use every tree he cuts down, whereas the costs missing (or too long delayed)
of deforestation is shared with everyone else. The more users feedback from the resource
there are, the scarcer the resource and the more people rush to the growth of the users of
to consume it before it’s depleted. that resource.”

• Solutions include: (i) educating people on the implications

of uncontrolled usage and appeal to their good sense, (ii)
privatizing/splitting the resource so people are accountable for
their own usage and (iii) using regulations (e.g. taxes, quotas,
permits) to limit consumption. Paid parking lots, traffic lights,
broadcasting licenses etc. are all examples of ways to manage
the commons.

Systems may experience sliding standards or a “drift to low

performance” if (i) the players perceive things to be worse than
they are, and (ii) such perceptions affect their future targets.

• For example, you may accept lower-than-desired air quality “Standards aren’t absolute.
because it doesn’t seem any worse than before. The new (lower) When perceived
standard becomes the norm and the negative loop continues. performance slips, the goal is
allowed to slip.”

• Solutions include: (i) sticking to absolute goals (vs being

influenced by past/existing performance) or (ii) benchmarking
only against the best historical performance.

Escalation arises when competing players try to outdo one

another in a reinforcing loop, e.g. price war, shouting match,
weapons race.

• Even escalation for positive things (e.g. efficiency, morality) can

become negative when taken to the extreme. If left unchecked,
it usually ends up with at least 1 of the players collapsing.

Copyright © 2019 Skool of Happiness Pte Ltd. • All Rights Reserved.

• Solutions include: (i) one party taking the first steps to
unilaterally reduce the stocks (and stop the reinforcing loop)
or (ii) negotiating balancing loops that can keep escalations in
check.

Widening winner-loser gaps occur when winners get rewarded

with resources that help them to win even bigger in future. This
creates a reinforcing loop that could eliminate losers from the
system, e.g. the rich using their wealth and access to gain even KEY QUOTES
more wealth and access.

• Solutions include: (i) Diversification (e.g. a company moving to

a new niche to exploit new resources), (ii) adding balancing
loops to prevent domination (e.g. antitrust laws) or equalize
the playing field (e.g. handicaps for weaker players), and (iii)
finding ways to break the reinforcing loop (e.g. progressive
taxes, public welfare).

Addictions. Just like how people can get addicted to alcohol or

drugs, players in a system can get addicted to a solution that
eases problem symptoms without addressing the root causes.

• Such interventions are usually introduced to boost a stock. “Addiction is finding a quick
However, if the root cause isn’t addressed, problems will and dirty solution to the
resurface and demand an even bigger fix. Worse still, the symptom of the problem,
intervention could cause a degeneration of the original system which prevents or distracts
one from the harder and
capabilities, and shift the burden increasingly to the intervener.
longerterm task of solving
For example, kids’ reliance on calculators can reduce their the real problem.”
capacity for mental arithmetic, and the use of pesticides can
destroy the pests’ natural predators so farmers need to use
even more pesticides.

• Solutions include: (i) withdrawing from the addiction (gradually “The secret is to begin not
or going cold-turkey), or (ii) avoiding it from the onset by with a heroic takeover, but
strengthening the system’s abilities instead of taking on its with a series of questions.”
burdens.
“Wherever there are rules,
In any system, there’ll be people trying to break the rules, i.e. there is likely to be rule
they seem to be following the rules but are actually working beating.”
around them, e.g. drivers keeping to the speed limit only when
there’re speed cameras or police cars. Use rule-beating as a form “Systems…have a terrible
of feedback to refine the rules, or design rules that specifically tendency to produce exactly
and only what you ask them
incentivize people to self-organize in the desired way. to produce. Be careful what
you ask them to produce.”

10

Copyright © 2019 Skool of Happiness Pte Ltd. • All Rights Reserved.

Systems will deliver whatever they’re designed to deliver—if you
set the wrong goals, you’ll get disastrous outcomes. For example,
an education system will deliver vastly different outcomes
depending on whether you measure “good education” using
expenditure or test results. Don’t confuse efforts with results.

Creating Effective System Changes

KEY QUOTES
The best way to change a system is to (i) identify the leverage
points where a tiny shift could trigger a large change in behavior, “There are no cheap tickets
then (ii) figure out the right direction to push. This is a difficult to mastery. You have to work
process that requires hard work and an open mind (to explore hard at it.”
new ideas and options).

Use these leverage points in order of impact:

“If no paradigm is right, you
• Every system or society comes with its own paradigm, i.e. a can choose whatever one
set of unspoken assumptions and beliefs (about what’s good, will help to achieve your
right, fair, etc.). Paradigm shifts can be achieved in any system purpose. If you have no idea
with the right reinforcements and change agents. However, where to get a purpose, you
can listen to the universe.”
the ultimate leverage comes from transcending paradigms, i.e.
realizing that there’s no single “right” or “true” paradigm.

• A system’s goals define its behaviors. Most systems (including

cancer cells) seek to survive and grow. The key is to deliberately
choose how far to grow and to what end, so we prevent 1
system’s unchecked growth from destroying others.

• Increase a system’s self-organization abilities so it can change

dynamically. The best way to do so is to promote learning and
evolution through diversity and experimentation.

• Use rules—including rewards, penalties and constraints—to

directly shape behavior.

• Typically, system players will organize themselves with the “Missing information flows is
one of the most common
right info. Design information flows to incentivize players to causes of system
act in the desired way, e.g. show the increasing cost per unit malfunction.”
to discourage people from consuming too much of a limited
resource.

• Build in checks and balances to curb the momentum of

reinforcing feedback loops to prevent long-term system collapse.

11

Copyright © 2019 Skool of Happiness Pte Ltd. • All Rights Reserved.

 • A complex system comes with many balancing feedback loops,
each designed to maintain a sub-goal and jointly define the
overall system strength. Ensure that each balancing loop is
strong enough relative to the impact it’s designed to achieve,
e.g. governments can only manage big industry players if they
have enough power. Be careful when removing balancing loops
e.g. an emergency cooling system may seem unimportant since
it’s usually inactive, yet its removal will weaken the system.
KEY QUOTES
• Adjust the length of delays in feedback loops (where possible)
to influence system behaviors and the level of oscillations.

• A system’s behaviors are determined by the structure of stocks, “The only way to fix a system
flows and their interconnection. Your best leverage is to design that is laid out poorly is to
a great system from the onset, since it’s hard to change a rebuild it, if you can.”
system once it’s built. You can only improve existing systems
by addressing the bottlenecks and limiting factors.

• Build stocks that are big relative to the flows, so they can act
as a buffer to stabilize the system, e.g. a lake is much less likely
to overflow than a bathtub.

• Numbers seldom result in real change (e.g. tweaking minimum

wages or tax rates won’t restructure the economy), unless the
change is big enough to trigger other leverage points above.

Conclusion
“The future can’t be
Ultimately, it’s impossible to fully understand dynamic, self-organizing, predicted, but it can be
nonlinear systems, much less predict or control them. The real value envisioned and brought
of systems thinking is in helping us to understand the complexity of lovingly into being.”
the world we live in, so we can manage our expectations, learn from
surprises, and actively design/refine systems and find creative ways to
bring our visions to life.

The book is packed with examples and diagrams (ranging from business “The thing to do, when you
don’t know, is not to bluff
to sociology and economics) to help us understand systems at work. and not to freeze, but to
Meadows also presents several rules of thumb to help us navigate learn.”
more effectively in a world of systems. For more details, please visit
academyforchange.org.

12

Copyright © 2019 Skool of Happiness Pte Ltd. • All Rights Reserved.

 About the Author

Donella H. Meadows (1941-2001) was

an American environmental scientist,
teacher, and author. She received her B.A.
in chemistry from Carleton College and her
Ph.D. in biophysics from Harvard. She was a
research fellow at MIT, taught at Dartmouth KEY QUOTES
College for 29 years, and received several
“Transition support in a team
awards for her work on conservation, sport.”
environment and sustainability. She passed
away in 2001.
In 1996, Meadows founded the Sustainability Institute, which was
renamed the Donella Meadows Institute in 2011, then the Academy
for Systems Change in 2016.

About ReadinGraphics

ReadinGraphics captures pearls of wisdom from

the best books around the world, chunks down
concepts into simple, actionable steps, and
presents them in graphics that you can absorb at
a glance.

We cut through the clutter to bring ideas to life, making it easier to

see, organize and apply insights to create breakthroughs.

Find out more about us at https://readingraphics.com

www.readingraphics.com ReadinGraphics ReadinGraphics

13

Copyright © 2019 Skool of Happiness Pte Ltd. • All Rights Reserved.