My father was an artisan. His name, Omenka, means a gifted craftsman in Igbo, and it was not a courtesy — he earned it at a bench, with his hands, over a working lifetime.
At seventy-five he lost his sight.
He kept working.
Not a compensating gift
I want to be careful here, because there is a sentimental version of this story that is much less interesting than the true one.
He did not develop some remarkable heightened sense. He did not memorise the workshop after going blind. What happened was far more ordinary, and far more useful to anyone who designs systems for a living:
His workshop had been kept in such rigorous order, for so long, that it no longer required eyes to operate.
Every tool had one place. Every place held one tool. Nothing was ever set down "just for now." He had spent a working life paying a small daily tax in discipline — the two seconds it costs to return something to where it belongs — and when the catastrophic failure came, the system he had built simply carried on running in a degraded mode.
He could still find every tool. He could still perform basic operations. The loss of a primary sense reduced his capability. It did not stop him.
I asked him how, the way I had asked him about the circle when I was a child. He said:
When you train your hands and your eyes, train your mind more. When those fail, your mind stays with you till the end of life.
Which is a safety case
I now spend my working life on systems that must behave when something has already gone wrong. Advanced reactor components. Handling equipment in a bay nobody can enter. Wellhead devices whose entire purpose is the case where everything else has failed.
The vocabulary is formal — defence in depth, single-failure tolerance, graceful degradation, common-cause analysis — and the documentation is extensive and expensive. But the idea underneath is the one my father applied to a workshop by instinct, for free, and it is this:
Build the system so that losing one sense does not lose the system.
Nuclear codes make you demonstrate it. He just did it, because he could not afford to lose a day's work to a misplaced tool.
What it changes about how I design
Three things, concretely.
1. The discipline goes in early, where it is cheap.
The two seconds to put a tool back is trivial. Searching a workshop while blind is not possible at all. The cost curve is enormously steep, and it runs the same way in engineering: an interface definition written at concept stage costs an afternoon; the same definition discovered during fabrication costs a change order, a schedule slip and somebody's confidence.
Configuration management, interface control, design basis written as you go — these are the same daily tax. They feel like overhead right up to the moment they are the only reason the project is still viable.
2. Ask what happens when a sense fails.
Not "what if this breaks" — that gets you a fault tree, which is useful but narrow. Ask what happens when a channel of information is lost. The instrument reads wrong. The operator cannot see into the cell. The drawing revision in the shop is not the current one. The vendor's data sheet was for a different variant.
Systems that only work when everything works are not systems. They are demonstrations.
3. Degrade, do not stop.
The interesting property of my father's workshop was not that it was fault-tolerant in the binary sense. It was that capability fell off gradually. He could not do everything he had done before. He could do a great deal.
That is the property worth designing for, and it is often available cheaply if you ask for it early. A cooling plant that can carry 70 % of load on a failed unit is worth far more than one that carries 100 % or nothing. A handling tool that can still be recovered manually when its actuator fails is worth more than a faster one that cannot.
The uncomfortable part
Order is not tidiness. I want to be clear about that, because it is easy to hear this story as a parable about being neat.
Tidiness is appearance. Order is a structure that holds information — where things are, what version is current, which surface everything else references, what this component is for. Tidiness makes a workshop look good. Order is what you have left when a sense fails.
You cannot retrofit it. That is the whole point. He did not organise the workshop after he went blind; he could not have. The order was already there, built up over decades of days when it did not appear to be earning its keep.
Most of the engineering disciplines I care about have that same shape. They look like overhead right up until the day they are the only thing still working.
*He also drew a perfect circle bare-handed, which I have never managed. There is an instrument on this site that measures precisely how badly I fail at it.*
