How do you keep working effectively when you are under heavy time pressure?

My first move under pressure is to stop working and write everything down, because what actually hurts me is not the volume, it is holding it all in my head at once. Then I split the list into what breaks a user and what merely annoys me. I put that split somewhere visible — a comment on the tracking issue, or a message in the project channel — and I say which parts I am not doing. That helps twice: it stops me polishing something nobody is waiting on, and it gives anyone the chance to correct my ranking before I have spent the time. Second habit is timeboxing investigation. If I cannot reproduce a report in about forty-five minutes, I ask, rather than treating asking as an admission of something. A concrete case: during a hardening week on the open-source project I contribute to, my verification queue reached forty-one unconfirmed reports. I posted my ranking, worked through the eleven that touched the default configuration path first, and asked another contributor to take a second pass on the rest. That batch closed with fourteen percent of the fixes coming back open, against the twenty-two percent we had been running. And I stop at a fixed hour. Verifying while tired is how bugs come back, so the extra hour usually costs more than it buys.

I treat pressure as a coordination problem rather than an endurance problem, so most of what matters happens in the first hour. I write a cut line and publish it: what ships, what ships with a known-issue note, what moves. Anyone who disagrees gets to disagree early and cheaply, and if the maintainers push back I would much rather learn that on the first day than the night before a freeze. Then I make status boring and frequent — a short written update at the same time daily, with blocked items named explicitly. When I have been under real load, the thing that has repeatedly saved me is someone reading that update and mentioning they had already solved the problem I was about to spend an afternoon on. I also shape the load for people around me. In the last hardening push I was mentoring a newer contributor, so I gave them one narrow lane — regression verification against the candidate build — with a firm rule that twenty minutes stuck means ask, not grind. They completed seventeen verifications that week and were still enjoying it afterwards. On the measurable side, that push ended with seven percent of verified fixes reopening, down from the mid-teens, largely because we stopped accepting fixes that arrived without a reproduction. The one thing I refuse to do is go quiet. Silence under pressure is what turns a slip into a surprise.