There’s a version of adjusting things that feels like progress and isn’t. You move the app to a different folder. You rename the categories. You try a new capture tool for a fortnight, then another one. Nothing is wrong, exactly, and you’re certainly doing something — but a year passes and you’d struggle to name a single thing you decided. The distinction between experimentation vs tinkering is the one that explains why.
An Experiment Has an End in Mind
An experiment asks a question you intend to answer. It has a shape: here’s what I’m trying, here’s roughly how long I’m trying it for, and here’s the decision I’ll make at the end. Keep going, or stop. That last part is the whole thing. Without a decision waiting at the end, you haven’t run an experiment — you’ve just changed something.
Tinkering is what happens when the endpoint never arrives. It looks almost identical from the outside and feels considerably better from the inside, because adjustment is pleasant and decisions are not. Bob Ross adds a little more to the waterfall, then a little more, and the only reason he ever stops is that the show ends. With a painting, that might be fine. With the systems you’re running your working life on, the absence of an endpoint means you never find out whether any of it worked.
Parameters Are What Make an Experiment Possible
The instinct is to treat constraints as the thing getting in the way of the experiment. They’re usually what makes it answerable.
Take video podcasting. The question “should I make this a video show?” is unanswerable in the abstract. The question becomes tractable the moment you name the parameters: I can’t control the other person’s camera, their lighting, or their background, which means I can’t produce a consistent result at a standard I’d be happy with. That’s not an obstacle to the experiment. That is the experiment’s finding — and it arrived from running a single one-off in a studio and noticing what conditions made it work. One session produced a real decision, because there was a question attached to it.
That’s also why “I’ll try this for a while and see” tends to produce nothing. There’s no fence line. Nothing can be concluded, so nothing is.
Where Experimentation vs Tinkering Gets Uncomfortable
Here’s the part people skip: stopping is an experiment too.
Ending something that’s working — a project at its peak, a revenue stream you’ve grown tired of, a commitment that still pays but no longer interests you — is a genuine experiment, because you don’t know what happens next. What will I do with the time? How will I fill the gap in the income? Those are open questions with real stakes, and running the experiment is the only way to answer them. Patrick Rhone killed Minimal Mac at the height of its popularity, while it was still bringing in money every month, because he’d said everything he had to say about minimalism. That’s not quitting. That’s testing a hypothesis about what the time is worth.
The reason this matters for anyone thinking about habits: habits are the thing you keep, and experiments are how you find out what’s worth keeping. Atomic Habits tells you how to build one. Tiny Experiments is closer to the question underneath it — how do you know which habit, at what cadence, for what reason? Both books belong in the same conversation, and only one of them tends to get invited.
You Don’t Get to Choose What Lands
One more thing, offered as consolation rather than technique. After thirty years of writing online, the most widely linked post of Patrick’s career came out of a conversation with his wife in the car. It took him twenty seconds to type. It wasn’t even his idea.
Which means the results of an experiment are not the point, and can’t be — you have no reliable access to them in advance. The ride is the point. You run the thing, you find out, and then you make the choice the experiment was there to inform. That’s productiveness in its most practical form: not knowing the outcome, and going anyway, with an end date in your pocket.
