What co-founders actually share
Ownership is a solved documentation problem. The thing two people are actually holding between them has no unit and no cut-off.
Contents
Writing down who owns what is a solved problem, and I spent a decade inside the solution. There is a way to record a share, a way to record a change in a share, a way to record what happens to it under a dozen circumstances, and none of it is difficult. It has all been argued out by people who are better at it than I am, and the answers are sitting in standard documents anybody can adapt.
Which is why it is strange that the split is what everybody asks about when they hear a venture has two founders. That is the easy part. It is legible, it can be made as precise as you like, and it is the one component of the arrangement that was fully specified before anything happened.
The part with no unit
What is actually shared is an expectation, held by each person about the other, that they will keep working on this when nothing is arriving from outside to tell them to.
That expectation is the load-bearing element and no document contains it. Documents cannot contain it. Effort under uncertainty has no unit, so there is nothing to state. It has no cut-off, so there is no moment at which it is assessed. And there is no reconciliation available, because there is no second record to compare against, and I know what a reconciliation requires because I did them for years.
Everything the paperwork covers is downstream of the thing the paperwork cannot reach. The document allocates the proceeds of an effort it does not describe.
Which explains the disputes
The received wisdom is that founder disputes happen because people did not document things properly, and I no longer think that is usually what happened.
More often the documents work precisely as written. What broke was the expectation, weeks or months earlier, quietly and without any statement being made. Then the argument gets conducted on the paperwork, because the paperwork is the only surface in the relationship that has words on it. The clauses become the vocabulary of a disagreement that is not about the clauses, and everybody involved sincerely believes it is a documents problem, because that is where all the sentences are.
There is a measurement difficulty underneath this that I find genuinely hard. Before anything is external, the only evidence you have about the other person’s continued intent is their behaviour in weeks where nothing was at stake. That is a noisy signal. A quiet fortnight can be commitment operating normally or the first part of a withdrawal, and the two look identical for a while, by definition, since a withdrawal that announced itself would not need observing.
The thing I tried, which does not work the way I expected
My instinct was the one my training hands me. If the important thing is unwritten, write it down. Define the roles. Agree what each person is responsible for, in what volume, by when.
That fails, and not for the reason people usually give. It is not that the specification is too rigid for early stage work, though it is. It is that specifying effort changes what is being produced. The moment effort has a number, both people begin satisfying the number, and satisfying a number is a different activity from the one you wanted. What you were trying to secure was the behaviour when nobody is looking, and any instrument that looks at it destroys the property it was built to observe.
So the honest position is that the central thing in a shared venture is unspecifiable, and every attempt to specify it converts it into something else that is easier to measure and worth less.
Where that argument gets luxurious
It is worth noticing who gets to say the documents do not matter.
Somebody with a working partnership. The paperwork is invisible exactly while things are fine, and it becomes the only thing that exists the moment they are not. Every argument of the kind I have just made is available only to people for whom it has not yet been tested, and I am one of those people, months into something that has not launched and has therefore not produced a single occasion on which the documents would be consulted.
So I would not read any of this as an argument against being careful with the documents. Be extremely careful with them. They are cheap now and they are the only part of the arrangement that will still be there if the other part stops.
The evidence is a long run of ordinary weeks
The evidence for the shared thing is a long run of ordinary weeks, and there is no way to get it early.
You cannot know whether you have it until you have had it, and by that point the question has answered itself for reasons neither person can take much credit for. Nobody chooses to keep showing up for two years. They either did or they did not, and the explanation arrives afterwards, assembled out of whatever happened.