Garbage Can Model Of Decision Making

Why Your Decisions Keep Failing and What to Do About It

I spent three years trying to run organized decision-making processes in a company where nobody actually agreed on what problem they were solving. We had weekly meetings where four different problems got dumped onto the same agenda, three solutions were already floating around from other teams, and two people with decision-making authority weren't even in the room. It was exhausting and it produced garbage results every single time. That's when I actually started paying attention to the Garbage Can Model Of Decision Making instead of pretending our organization worked like the textbooks said. The model came out of March and Olsen's 1972 paper, but reading the academic paper won't help you much unless you understand what it's actually describing in practice. Most organizations don't make decisions through rational steps. They have streams flowing independently. Problems, solutions, participants, and choice opportunities all move at their own pace. Decision making happens when these streams accidentally collide, not when someone follows a process.

Breaking Down the Garbage Can Model Of Decision Making

There are four independent streams. Problems float around looking for solutions. People drift in and out of decision situations based on scheduling, mood, or whoever happens to be available. Solutions search for problems to attach to. Choice opportunities are the moments when the organization pretends something needs deciding. In a rational model, you identify the problem first, find the right people, develop the solution, then execute. The garbage can model says that order rarely exists. What actually happens is messy and temporal. The key insight beginners miss is that problems don't always get solved. Sometimes a solution finds a problem that wasn't looking for it. Sometimes a decision gets made without any real problem attached. Sometimes the same problem circulates for months while the wrong solution gets attached to it because the right people weren't in the room. This isn't dysfunction. This is how these organizations work by default.

How to Actually Use This When You're Dealing With It

First, map the four streams in your specific situation. Not abstractly. Write down which problems are currently floating, which solutions exist that nobody's using, who the actual participants are including the ones who never show up, and when the next choice opportunity is on the calendar. This takes about ten minutes and changes everything about how you approach a decision. I ran into a specific problem last year where we needed to decide whether to rebuild our internal reporting platform. The rational approach would've been to define the problem clearly first. Instead, I mapped the streams and found something unexpected. The real problem wasn't the reporting tool at all. It was that our data ownership was ambiguous and everyone was blaming the dashboard for symptoms of a governance issue. The solution stream had three different platform candidates floating around, each championed by a different team. The participant stream was a mess because our VP of Engineering rotated out every six months and the new person brought a completely different agenda. The choice opportunity was coming up in a quarterly review where leadership expected a decision. My workaround was to decouple the problem from the choice opportunity. I scheduled a separate session two weeks before the quarterly review specifically to define what problem we were actually solving, with the right people who had context about the data flows. We didn't touch the platform question. We spent forty five minutes agreeing on the actual problem statement. Then at the quarterly review, when the choice opportunity hit, we already had alignment on what we were choosing between. It went from a three month argument down to about two weeks of actual decision time.

Another practical step is to track decision opportunities as events rather than assuming they'll arrive when you're ready. In most organizations I've worked with, choice opportunities cluster around budget cycles, leadership changes, and compliance deadlines. If you're waiting for the perfect moment to make a decision, you'll miss the window. Get your solution and problem stream aligned ahead of time so when the opportunity arrives you're actually ready.

What People Get Wrong About This Model

The biggest mistake is treating it as an excuse to do nothing. It's not. It's a description of reality that lets you work with the chaos instead of fighting it. When you understand that decisions are outcomes of stream coupling, you start manipulating the timing deliberately. You bring problems and solutions together before the choice opportunity arrives. You figure out which participants actually matter in any given moment rather than assuming the org chart tells you who decides things. A second counter-intuitive point is that organized anarchies have higher decision throughput than you'd expect once you stop fighting the model. Companies that accept the garbage can dynamics tend to make more decisions, even if those decisions are messier. The rational model looks cleaner on paper but produces paralysis because every decision requires perfect problem definition, perfect participant alignment, and perfect solution matching before anything happens. In practice, the garbage can approach gets things moving in eighteen to twenty four months of equivalent organizational time.

Where This Approach Completely Fails

The garbage can model doesn't work in safety critical environments. If you're running a nuclear plant, flying an airplane, or managing a patient in surgery, you cannot rely on streams colliding by chance. Those domains require structured decision processes with explicit problem definitions and verified solution paths. Trying to apply garbage can logic there will get people killed. It also breaks down in small startups where everyone knows everyone. When participant flow is low and problems are visible to all, the model collapses into something closer to rational choice because there's no organizational darkness hiding the streams. The model is most useful in mid to large organizations with high participant turnover, ambiguous problem definitions, and multiple competing agendas. If your organization has clear hierarchies and stable membership, stick with traditional decision frameworks. The garbage can model adds noise in those environments rather than clarity. You're better off investing that energy in improving communication structures than trying to map streams that already move predictably.

A Quick Reference for Getting Started

When you're about to face a decision, spend fifteen minutes writing out the four streams before you schedule any meetings. List the problems actively circulating. List solutions already available. List who will actually be in the room and who won't. Note when the choice opportunity is locked in. Then look for mismatches. Is the problem undefined? Pull it into focus before the decision meeting. Is the wrong person showing up? Find out who actually has influence and get them there. Is a solution being pushed for the wrong problem? Separate them explicitly. This doesn't turn decision making into a clean process. It just makes the mess visible so you can navigate it instead of being crushed by it. I've used this approach in about forty decision situations over the past decade across three different companies. The ones where I mapped the streams beforehand consistently took less time and produced outcomes that stuck. The ones where I ignored it and went straight into the meeting usually ended with someone blaming the process afterward, which was exactly what the process actually was. If you want to read the original source, March and Olsen's "A Garbage Can Model of Organizational Choice" from 1972 is still the primary reference. It's dense but short. There are also decent follow-up papers by Cohen, March, and Olsen that expand on the participant stream dynamics. Nothing beats the original for understanding the core mechanism though.