Fleet Street is not actually where any of this happened — the newspapers had already left Fleet Street by the nineties, which is one of those facts Londoners know and visitors find confusing. Reuters was on South Colonnade in Canary Wharf by then, Bloomberg a short walk away, and the whole financial district had a specific texture in those years: glass and granite and the permanent smell of coffee and dry-cleaning, people in suits moving between buildings at a pace that implied urgency even when nothing particularly urgent was occurring.

I had been at Reuters for a while. Technical support, the kind that involves answering questions from customers about why their terminal is displaying something unexpected, and from salespeople about why they can’t promise the customer a feature that doesn’t exist yet. The customers were usually fine. The salespeople were a variable quantity.

The salesperson in question — I’ll call him Gary Prestwich, which has the right energy — had called in to ask me to confirm, to a customer he had on another line, that a particular thing the product did not do was in fact something it was about to do very soon. This was not the first time Gary had made this request, or a version of it. It was, perhaps, the fourth or fifth time. I had previously responded with some variant of “I can’t confirm that because it isn’t true,” which Gary had taken as a failure of team spirit rather than basic professional honesty.

This time I was less measured. I was, by the account given later in the room where these things are discussed, quite direct. I used one word that should not appear in a professional telephone exchange and I used it with some emphasis. Gary Prestwich, or whatever his name was, escalated. The escalation was, from his perspective, entirely reasonable. From the perspective of the person doing the escalating, there is a strong case that what I said was disproportionate even if it was accurate.

The conversation with my manager — a decent man named something like Phil Beresford who had been dealing with the Gary Prestwich problem for longer than I had and was visibly tired of it — was brief. He said that what I’d said had created a situation. I agreed that it had. He said that the situation was one he couldn’t manage around. I said I understood. There was a pause. “You’re quite good at this, you know,” he said, which was unexpectedly kind. “Just not here.”

I picked up my things. It was a Friday afternoon.

What happened next sounds like I planned it, but I didn’t, not exactly. I knew Bloomberg was nearby — everyone in that world knew Bloomberg was nearby, the two companies existed in a state of permanent parallel awareness, like neighbours who are polite to each other and have been for years and both know perfectly well they want the other’s parking space. I had contacts there; not close contacts, but the kind you accumulate when you work in a sector long enough. I rang one of them from a payphone — this was the nineties, I did not have a mobile, nobody had a mobile in the way people now have mobiles — and said something like, “I’ve just left Reuters, is there anything going?”

There was a role. There had been a role for about three weeks that they hadn’t filled. Could I come in now?

I walked there. It was, I think, twelve minutes. I had my CV in my bag because I had learned, somewhere along the way, to carry one. The interview was informal in the way that late-Friday interviews are informal, which is to say the person doing it had already mentally clocked off but was going through the motions efficiently. I told them what I could do. They said they’d been looking for exactly that. I started the following Monday.

I have told this story a number of times over the years and it always sounds more calculated than it was. But here’s the thing that actually happened: I assessed risk under incomplete information and made the move anyway.

I knew Bloomberg recruited heavily from Reuters — the entire sector did. I knew they had an open role. I knew the person I was calling owed me no favours but also owed no antagonism. What I didn’t know was whether they’d actually hire me, whether the role would still be available on Monday, whether I’d like working there, whether they’d find out the specifics of why I’d left. None of that was knowable. What was knowable was that staying was worse. That Phil had said I wasn’t a fit. That the alternative was definitely wrong.

When you face a choice between a known-bad and an unknown, the move-cost can be worth it. Not always — sometimes you should sit with bad and wait for clarity. But when the bad is actively corroding you and the alternative is only unknown, not downside-clear, the confidence to move is a feature, not a bug.

The few-weeks-to-think-about-it option is sometimes correct. In that specific moment, in the City, in the middle of the nineties, with a Friday afternoon and twelve minutes of walking between me and a job that was slightly better than the one I’d just lost — it would have been the wrong call entirely.

Phil Beresford was right, incidentally. I was quite good at it. Just not there.