It's Not Normal to Go to These Lengths

Altitude: the founder journey

I wrote a while ago about not knowing what I wanted, for decades, and about the contract that finally made the question urgent rather than theoretical. I left that story at the turn: I stopped asking organisations what I was worth and started building something myself.

What I didn't say is what "building something myself" actually turned into, because at the time I couldn't have told you. It looked like making a task app. It wasn't.

Four questions, and only one of them was really the question

Can I build it. Can I market it. Can I sell it. And, the one underneath the other three: can I do it properly. Professionally. To the standard I'd have demanded from a supplier.

The first three are ordinary founder questions and everyone asks them. The fourth is the one I'd been carrying without ever saying it out loud, because it came from the other side of the desk. I'd spent my career watching organisations choose the appearance of rigour over rigour, and then paying for that choice for years afterwards, usually in a room I was called into to explain the overrun. I wanted to know whether the version I'd been arguing for the whole time actually works when nobody makes you do it.

The lengths

So here is what got built alongside a task app.

A test harness that doubles as a video pipeline. A written list of known issues, each one with a decision attached to it. Stop conditions written before the work started, so shipping was a bar I'd set rather than a mood I'd arrived at. Proper tickets, with acceptance criteria and judgement calls flagged as product decisions rather than guessed. A review process where the reviewer doesn't get to see what the author intended. Registers, an honest planning document, and a discipline of writing down what I decided and why.

For a solo project. With no team, no stakeholder, and nobody to hand the artefacts to.

A reasonable person looks at that and says most of it wasn't needed. And that person is right, on the terms they mean it. I could have shipped sooner. Plenty of people ship perfectly good products without any of this.

The requirement was never "ship a task app"

The requirement was to know, not to hope.

Those two words describe completely different projects. Hoping it's good enough is cheap and available and gets you to a launch date faster. Knowing costs you months, and the only thing it buys is that when someone asks whether you can do this properly, you're not guessing at the answer.

Which reframes what was actually being produced. The proofs were the deliverable, and the app is what they were made of.

That sounds like a line, and it's the plainest description I have of the last seven months. Every register, every review, every test run was evidence gathered on a question I'd never been able to answer.

The part I didn't expect

Here's the bit I'd tell anyone doing this alone, because I didn't know it going in and it turned out to matter more than any of it.

Most solo attempts don't die of bad products. They die of demoralisation.

There's a specific evening, and it recurs, where nothing works and there's no one to check the reading against and the honest interpretation feels like this isn't going to happen. On a team, that evening is survivable because someone else has the other half of the picture. Alone, it's just you and a feeling.

What the apparatus does, accidentally, is answer that feeling with data. The known-issues list says which of these things are already understood and decided, and which are actually new. A stop condition says whether you're near the bar or nowhere close. A harness says the thing that worked yesterday still works. None of that is encouragement. It's better than encouragement, because it's checkable: it tells you what to mould, instead of leaving you to conclude it's over.

A career spent applying programme governance to other people's work, and the first time I pointed it at my own it turned out to be doing a job I'd never noticed it doing for anyone else.

The proof I can't award myself

So: can I do it properly? On the evidence of the last seven months, yes, and I can show my working.

But I've noticed the limit of that answer. Every proof I've gathered so far is one I marked myself. The registers are mine, the standard is mine, the bar was set by me and cleared by me, and a person who sets their own exam and passes it has proved something real and also something quite narrow.

The next set doesn't work that way. Can I market it, can I sell it, will anyone but me want the thing — those need witnesses. Other people, deciding, without me in the room.

That's the part I can't do to a standard. It's the one proof I can't award myself, and it's why the interesting chapter is the one that starts now rather than the one that just finished.