There’s a sentence that tells you everything about how a company works: “let me check with the team.” Sometimes it’s literal and fine. Usually it’s the sound of a question beginning a long journey through people who each own a sliver of the answer, none of whom can say yes alone, all of whom can say no. By the time the answer returns, it has been collaborated on. Nobody made it.
We built fryga to never say that sentence. Four people, no project managers in the middle, no bench, and before a sprint starts the client knows the names of the two people in their codebase. That’s a position on what actually produces good work, and it has a century of evidence behind it.
The rope and your standup
Joan Westenberg wrote a deliberately rude essay called “Collaboration Is Bullshit” (worth the read, and the comment war) that I keep returning to. Her core claim, in her words: serious work gets done by people “with clear authority and sharp accountability,” and the teamwork story gets written afterward.
The evidence she assembles is old and keeps replicating. Ringelmann’s rope experiments, published back in 1913, showed that the more people share a rope, the less each one pulls, while everyone involved feels perfectly reasonable about their effort. Brooks documented the software version in 1975, watching coordination costs compound until adding people to a late project made it later. What both measured downstream is effort; what dissolves upstream is responsibility. A team of four can feel a failure with their own names on it. A team of forty has a process that failed.
Her sharpest point is about what the collaboration industry built on top of that dissolution: a culture where being visible in the thread feels the same as being accountable for the outcome. Tools multiplied, meetings spawned meetings, and the warm sense of collective progress replaced the uncomfortable clarity of one person being on the hook.
What we do instead of collaborating
Westenberg is careful to separate communication, which work needs, from collaboration as an organization’s main occupation, which is where output goes to die. I’d push it one step further: the cure is structure that makes ownership impossible to dodge. Ours looks like this on every engagement.
Everything has an owner with a name. Mateusz has owned FastTravel’s payments from day zero, for years, and I’ve argued elsewhere that this continuity is itself an architectural decision. Katarzyna owned Domestika’s project editor from user research to production rollout, one person carrying the design decisions, the product decisions, and the consequences. When something is everyone’s, it’s the thing that doesn’t ship.
No handoffs, because every handoff is where responsibility goes to dilute. The person who talks to the client is the person in the codebase. The person who made the architecture call is the person who lives with it at 3pm on a Friday when it breaks. Layers between deciding and doing mostly add places for intent to die and for blame to get lost.
Best argument wins, not the loudest voice or the longest title. Ownership without that rule curdles into fiefdoms. With it, disagreement is cheap, decisions are fast, and being wrong is survivable because somebody owns fixing it too. We tell each other the truth, even when it’s uncomfortable, and a four-person room is small enough that the truth arrives undiluted.
The honest ceiling
Small teams lose some games, and pretending otherwise would undercut the whole argument. Some jobs simply need more hands than we have, and a small team that says yes anyway is selling what it can’t ship. Some problems carry round-the-clock operational weight that four humans cannot honestly promise. And the ownership model concentrates risk in named people: our answer is that the four of us have worked together across companies for years and the money paths have a continuous owner, but “what if the owner is hit by a bus” is a fair question, and small teams answer it with overlap and writing, not with redundant headcount.
What changed recently is where the ceiling sits. Implementation capacity used to be the reason you scaled headcount, and headcount was the reason you needed the coordination apparatus, and the apparatus was where ownership went to dissolve. Agents now carry implementation in a way that lets a small senior team punch far above its size without hiring its way into the Ringelmann zone. (They don’t fix the 24/7 concession; some ceilings are about humans needing sleep.) The leverage that used to require forty people requires four people and a harness, and four people can still feel their names on the outcome.
That’s the actual case for small teams: arithmetic about responsibility, not coziness or culture-deck virtue. Responsibility is the input that doesn’t scale, so you build so it never has to.