The work

Facts first. Everything after depends on them.

Ocean industry has a ground problem. Every job on the water eventually needs the same thing — somewhere to stand, somewhere to stage equipment, somewhere to put people and machines. Floating platforms exist. That isn't the problem. The problem is getting the right amount of ground, in the right place, at the right time, in the configuration the job needs. When it isn't there, work waits. Or someone mobilizes an oversized asset from a hundred miles away. Or improvises. Or the work simply doesn't happen.

That last category is the interesting one, and nobody can measure it.

So we built the smallest useful piece of ground we could. Twenty feet by ten, welded steel, two hundred square feet. It is tied up in Southeast Alaska right now, in commercial service, working a derelict vessel removal.

It floats about four thousand pounds lighter than we predicted. We did not see that coming, and it is written down as a miss. The engineering record carries seven findings. One of them meant opening the hull back up before launch. We publish those.

There is one hull, one customer, and one person here full time.

The question we can't answer yet

How much work is waiting for ground?

We don't know.

We don't think a market report can tell us either, because existing barge revenue measures the work that already found a solution. We are interested in the work that waited, overpaid, improvised, or never happened. So we are measuring it from the bottom up, in one region, with real assets and real customers, until the answer is falsifiable.

That is the actual project. Everything below is in service of it.

Two watches are open

Steel

Turning drawings into hulls, and hulls into a process that repeats. Jigs, fixtures, sequence, yard discipline.

The first one was built by hand and it shows. The question isn't whether the second is better — it will be. It is what the tenth costs, and whether that answer came from a system or from heroics. We are not interested in heroics.

She owns the cost curve. She is judged on whether the tenth hull costs what the second one promised.

Systems

Turning a working instrument into a network. There is already a node aboard: it runs unattended on a few watts, keeps UTC-disciplined append-only records, and produces bundles that validate offline with no server in the loop.

It works on exactly one hull. Making it true for fifty — so we know where every Ode is, how it is being used, and what its history says, without the record ever becoming something you have to trust us about — is the work.

She owns the measurements the whole thesis rests on: occupancy, mobilization, maintenance, and whether customers come back.

What this is actually like

Southeast Alaska, on an island, in the rain. The fastest way here is a floatplane. One customer. Payroll is not yet a solved problem.

In her first month she will be handed a domain and asked to publish a weekly report on it that other people act on. The engineering record is kept with the mistakes still in it, and hers will go in it too, with her name beside them. When she is wrong she will be told quickly, by someone holding the evidence, and she is expected to do the same upward.

Nothing here is a rocket. It is steel, salt water, and arithmetic that has to be right.

The bar

We are hiring Wrights. A wright is one who makes a thing and answers for whether it holds — a shipwright, a millwright. A Wright at Thalor holds a watch.

So the question is not whether she is impressive. It is whether, six months in, her part of the company runs correctly when nobody is checking.

We hire slowly and we do not lower it. If she is the right person in the wrong year, we would rather stay in touch for three years than hire someone else.