Every sport eventually optimises the fun out of itself, because the people playing are trying to win and nobody is trying to make it enjoyable to watch.

Voting for the least bad option is not cynicism, it is the entire mechanism working as designed, and I have got tired of being told otherwise.

Why is every app now trying to be a social network? I want the thing to do the thing.

The best manager I ever had appeared to do almost nothing. It took me three years to work out that was the job.

A backdoor for one government is a backdoor for everyone, and this is arithmetic rather than politics. A key that exists can be stolen.

Football has too much football now.

The four-day week only really works if your job is thinking. Every pilot anyone quotes was an office, and nobody ever ran one in a warehouse.

Football fans do not want their club to be successful, they want it to be theirs. That is why every ownership row goes the way it does.

Every organisation says it wants honest feedback and has exactly one person who gives it, and that person is always halfway out the door.

Most people would be happier with fewer options. Ask anyone who has spent forty minutes scrolling and then rewatched something they had already seen twice.

Nuclear is the safest energy we have per unit generated, and thirty years of being squeamish about it is a large part of why we are still burning gas.

I would rather pay for a thing once than be in a relationship with it forever, and almost nothing is sold that way any more.

Everything now assumes you have a phone on you, and we made that assumption into infrastructure without ever voting on it.

Cash is going and I notice I spend more without it, which is embarrassing but seems to be true of nearly everybody I ask.

A thing about going to the doctor that took me far too long to work out. For years I went in and described my symptoms in the order they had bothered me, which is the order they occurred to me, which is almost exactly the wrong order. The thing I was most worried about came last, usually in the doorway, usually prefixed with "oh, and probably nothing, but". What changed it was a GP who interrupted me about forty seconds in and asked what I was actually frightened of. I said it out loud, she dealt with it in two minutes, and the remaining eight were useful for the first time in my adult life. I now write down one sentence before I go in, which is the thing I am afraid of, and I say it first. I am aware this sounds like a productivity tip for being ill. It has changed every appointment I have had since.

The algorithm has quietly killed the thing where you find something you were not looking for. Every recommendation is a slightly worse version of what I already like.

A university degree is now mostly a signal about who your parents were, and only incidentally about what you know. The tell is that employers screen on institution rather than transcript, and institution is set by admissions, which is set by schooling, which is set by money.

A thing I cannot resolve about giving money to people begging. The argument against is well rehearsed and mostly correct: it may fund an addiction, it does not address the cause, the money would do more good given to an organisation set up to help, and the impulse is partly about how it makes me feel rather than what it does. The argument for is much shorter and I keep finding it harder to dismiss. Someone in front of me has asked for something I can trivially provide, and every framework I use to say no is a framework I constructed after the fact, sitting comfortably somewhere else. What I actually do is inconsistent — sometimes yes, sometimes no, and the variable is mostly my mood and how much of a hurry I am in, which is the worst possible basis. I would rather hold either position honestly than keep operating a policy that is really just weather.

Everyone in this country has a story about a builder and I have started to think they are all the same story. Someone comes, quotes low, starts, disappears for eleven days, returns, discovers something behind a wall, revises the price upward by forty percent, and finishes to a standard you would describe as fine. You pay it. You tell the story at dinner parties as a comedy. Nothing happens to the builder. The reason it keeps happening is not that builders are dishonest. It is that almost nobody commissions building work often enough to get good at it. You will do this twice in your life. He does it every week. The information is entirely on one side, and no amount of getting three quotes fixes that, because you cannot evaluate the three quotes either. Which is why every piece of advice you get is useless. Get it in writing, agree a schedule of works, hold back a retention — all sound, all assume you know what should be in the writing. The one thing that would actually help is having done it before, and by the time you have, you are finished.

There is no book everyone ought to have read, and the list is always suspiciously full of whatever the person making it was assigned at nineteen.

Are smart wearables a fad?

Everyone in this country agrees the trains are bad and nobody agrees about why, so here is an attempt at the actual causal chain rather than the usual list of villains. Start with the fact that almost every decision is made on a different timescale from the thing it affects. Rolling stock lasts thirty to forty years. Franchises ran seven to ten. Ministers last about eighteen months in post. Nobody who commissions a train is in the job when it arrives, and nobody who signs a timetable is there when the consequences of it show up. That mismatch produces a specific failure. It is always rational to defer maintenance, because the cost of deferring lands after you have gone. It is always rational to promise a route rather than fix an existing one, because the promise is legible now and the fix is invisible. Do this for thirty years and you get exactly what we have: a network that is simultaneously over-promised and under-maintained, staffed by people who mostly know what is wrong and have no mechanism to say so upwards. I do not think this is about nationalisation. Both models have produced it here, and both models have avoided it elsewhere. It is about whether the person making a decision is still around to receive its consequences.

Everyone I know is one broken boiler away from a genuinely bad month and we have all agreed not to bring it up.

Right, the thing nobody says about "learn to code" advice: It was decent advice in about 2012 and terrible advice by about 2020, and the people giving it never updated. It got handed to redundant workers as though the shortage was permanent and the barrier was purely knowledge. Both of those turned out to be wrong within a decade. What actually happened is that the entry level filled up while the senior end stayed short, which is the worst possible shape for someone retraining at forty. You can learn the language in a year. You cannot manufacture eight years of having seen things go wrong, and that is what the remaining jobs are actually asking for. I do not think the advice was cynical. I think it was given by people who got in early and genuinely could not see which parts of their own path were the ladder and which were the timing.

A thing I have noticed after fifteen years of interviewing people: almost every interview process is measuring how comfortable someone is being evaluated, and then reporting that as a measure of how good they are at the job. The tell is that the same candidate performs completely differently depending on the room. Give someone a whiteboard and three strangers and you learn how they handle being watched. Give them a real problem and an afternoon and you learn something about how they think. These correlate weakly, and we have built an entire industry on assuming they correlate strongly. What makes it hard to fix is that the noisy signal is cheap and the good signal is expensive. An hour of whiteboarding costs an hour. A realistic work sample costs a day of the candidate's time and half a day of yours, and multiplied across a funnel that is a real budget line that somebody has to defend. So we keep the cheap test and then add stages to compensate for its noise, which is how a process ends up with six rounds that are each individually indefensible. I do not have a clean answer. I have stopped pretending the current thing measures what we say it measures.