Yesterday I attended the Systems Thinking & Design mini-conference at Zone01 Kisumu. I was surrounded by like-minded people who shared ideas I plan to carry into every project I build from here on. This post is my attempt to organise what I took away, add the reading I did afterwards, and show how it connects to the code I write.

So what is systems thinking?

The first thing I had to do was understand the term. Systems thinking and design means looking at software not as isolated lines of code, but as an interconnected ecosystem where every piece, the user, the server and the process, affects the others.

That sounds abstract until you learn the vocabulary. The work of Donella Meadows, probably the best-known writer on the subject, breaks any system into a few parts:

  • Stocks are things that accumulate: users waiting in a queue, bugs in a backlog, patients in a hospital.
  • Flows are what fill or drain those stocks: sign-ups per hour, bugs fixed per week, admissions and discharges.
  • Feedback loops are how a system reacts to itself. A reinforcing loop amplifies change (more users bring more content, which brings more users). A balancing loop pushes back and keeps things near a goal (a thermostat, a rate limiter).
  • Delays are the gaps between an action and its visible result. They are why systems overshoot and why people give up on interventions that were actually working.
    Once I had those words, I started seeing them in software I already knew:

  • A client that retries instantly when a server is struggling adds load, which makes the server struggle more. That is a reinforcing loop that can turn a small outage into a big one.

  • A rate limiter, a circuit breaker or a queue with a maximum size is a balancing loop. It exists to stop the first kind of loop from running away.

  • A message queue is a stock. If messages arrive faster than they are processed, the stock grows, and no amount of staring at a single function will explain why.
    None of these problems live in one file. That is the point of the whole discipline.

The workshop: every group got a constraint

Beyond understanding the concept, we got to try it. The conference included a hands-on workshop where every group was handed a problem with its own constraints to think around.

My group was lucky to end up with the Covid-19 and Ebola outbreaks. We joined forces and tackled the topic from its depth rather than from the surface.

Why outbreaks are such a good systems problem

Looking back, I think we were handed a perfect topic, because an outbreak is never only a medical problem. It is people, culture, logistics, information and trust, all pushing on each other. Here is what I found when I read more afterwards.

Ebola, 2014 to 2016

The scale alone changed how people thought about the disease. Before 2014 the largest recorded Ebola outbreak was in Uganda in 2000 and 2001, with 425 cases and 225 deaths. The West African epidemic ended with roughly 28,600 reported cases and over 11,300 deaths, and it was the first time the virus spread across several countries at large scale.

The system view explains why:

  • A cultural practice became a transmission loop. Traditional funerals involve washing the body and close physical contact, which is a deeply human way to honour the dead and also an extremely high-risk activity during Ebola. WHO reporting linked about 80 percent of cases in Sierra Leone and 60 percent in Guinea to funeral practices, and health authorities traced 365 deaths to a single funeral in Sierra Leone.
  • Trust was a stock that ran low. Rumours, fear and distrust of both local and foreign responders led people to avoid clinics. When people avoid care, cases go uncounted and untreated, and the outbreak grows in the dark.
  • Capacity was missing at the start. In early summer 2014, no hospital in Liberia had an isolation ward. WHO also noted that over 70 percent of Ebola patients treated in West Africa died, compared with 26 percent of foreign medical staff who were evacuated to specialised care, a gap it called unacceptable.
  • Good interventions worked on the system, not just the virus. Sierra Leone used an anti-malaria campaign to rebuild public trust in health workers. Responders later allowed small, safe ceremonies at the graveside, which respected the ritual while cutting the risk. Both are fixes aimed at the human parts of the system. ### Covid-19

Covid-19 stressed a different set of connections:

  • Stealth. Because people can pass the virus on before they feel ill, the old approach of waiting for symptoms, then isolating and tracing, struggles to keep up. By the time WHO declared a pandemic on 11 March 2020, spread was already underway in many places.
  • Capacity. "Flatten the curve" is a stocks-and-flows idea. The aim was not only to reduce the total number of cases but to spread them out so hospital demand stayed below capacity. One operations researcher makes a sharp point: the fatality rate is not a fixed number. It depends on whether hospitals have enough beds, staff and equipment to treat people.
  • Supply chains. Personal protective equipment shortages showed how a delay or a shortage in one part of a chain leaves frontline workers without what they need at that moment.
  • Side effects. Hospitals that cleared space for a surge saw patients stay away out of fear and elective services pause, which created financial strain. Fixing one part of the system created pressure in another. ### Side by side
System element Ebola (West Africa) Covid-19
Main transmission loop Close contact, including at funerals and in home care Spread, often before symptoms appear
What ran short Isolation beds, trained staff, trust Hospital capacity, testing, protective equipment
Role of information Rumours and fear kept people away from care Mixed messages and delayed signals
Effective lever Community engagement, safe and dignified burials Slowing spread so demand stays under capacity

Finding the leverage

One idea from Meadows stayed with me more than any other: leverage points. These are the places in a system where a small push can shift the whole thing. She ranked them, and her ranking surprised me. Tweaking numbers (a budget, a threshold, a target) is the weakest move. Stronger moves include changing who has access to information, changing the rules and incentives, and letting the system organise and adapt itself. The strongest of all is changing the mindset the system grew out of.

Looking at the outbreaks through that lens:

  • Information flows: getting accurate information to people who had been uninformed, and getting case data back to decision makers.
  • Rules: how burials were done, who could attend, and who was responsible.
  • Self-organisation: community leaders and local volunteers who adapted the response to their own context.
  • Mindset: treating a community as a partner to work with rather than a problem to manage. ## What this changes about how I build software

The conference was about systems, but I left thinking about my own projects. These are the habits I want to build:

  1. Map before I code. Who are the users, what data moves where, and what happens when one part fails? A rough sketch on paper is enough.
  2. Look for the loops. Where does a small fault get amplified, and where is the brake? If I cannot find the balancing loop, I should probably add one.
  3. Respect delays. A change in a database, a cache or a deployment may not show its effect right away. Judging it too early leads to bad decisions.
  4. Ask who is affected beyond the user. Support staff, other services, people who cannot access the product at all. An outbreak response that ignores a community's customs fails, and a product that ignores real users does too.
  5. Fix the structure, not just the symptom. If the same bug keeps coming back, the problem is probably in how the pieces connect.
  6. Design for trust. Ebola showed how fast a system degrades when people stop believing it. Clear errors, honest messages and predictable behaviour are part of the design. ## Closing thoughts

I came to the conference expecting a talk about diagrams. I left with a different way of looking at the world: very little exists in isolation, and the most useful question is often not "what is broken?" but "what is connected to this?"

Thank you to the Zone01 Kisumu team and everyone who spoke, facilitated and shared ideas. If you attended, I would love to hear what stayed with you. If you did not, find a mini-conference near you and go.