Every twenty minutes, I get off the chair.
Twenty push-ups. Twenty sit-ups. Then I go back to work.
Sometimes I add a few weights now as well, because getting older apparently means voluntarily buying objects whose principal function is to make gravity more inconvenient.
I did not invent this as a productivity system. I first started doing something like it in my mid-twenties while studying for a law degree and spending long stretches at a desk revising.
There was no study behind the numbers. No app. No carefully optimised protocol. Twenty was simply easy to remember.
What I discovered was that a short burst of physical effort broke the day into pieces and, more importantly, interrupted the point where thinking stopped being useful and started becoming circular.
Years later, while spending equally unreasonable amounts of time at a computer building software, I found myself doing it again.
This time I had a name for the state it interrupted.
The perpetual cycle of erroney.
Yes, erroney. Derived unnecessarily from erroneous. I am keeping it.
There is a point where a problem turns into a hole
If you work on something complicated for long enough, you eventually encounter a problem that stops behaving like a problem and starts behaving like a hole.
You fix one thing.
That breaks something else.
You fix the second thing. The first thing changes shape. A test fails. You undo a change. Another error appears. You ask an agent. It gives you a plausible answer. You try that. Now there are three errors, one of which has somehow acquired a database migration.
At some point you are no longer methodically solving the problem.
You are reacting to it.
That is the perpetual cycle of erroney.
It is different from a genuinely difficult problem. Difficult problems can be enjoyable. You are learning something, narrowing possibilities, following evidence.
The bad version is when your thinking has narrowed enough that every attempted fix comes from the same assumptions that produced the previous failed fix.
There is always one more thing to try.
That is precisely why you should probably stop trying things for a minute.
Twenty is not magic
Before somebody turns this into a fitness graphic, there is nothing scientifically sacred about twenty push-ups, twenty sit-ups or twenty minutes.
It is just my rule.
I suspect it has lasted because it requires almost no thought. Twenty. Twenty. Twenty.
No habit tracker. No twelve-week programme. No notification congratulating me on maintaining a streak of possessing abdominal muscles.
The interruption is deliberately simpler than the work.
That matters because when I am already stuck inside a complicated problem, the last thing I need is another system to administer.
I need to stand up and do something with an obvious beginning and an obvious end.
Twenty push-ups finish.
Software, regrettably, often does not.
A physical problem has the decency to end
I think the definite ending is part of what works for me.
Desk work can be completely open-ended. There is rarely a natural moment where a difficult technical problem announces that enough thinking has happened for one sitting.
There is always another log to inspect, another test to run, another function to change, another document to read or another model to ask.
Physical effort is much less ambiguous.
You either did the twentieth push-up or you did not.
There is also very little scope for a stakeholder to return two days later and explain that, having reflected on matters, push-up seventeen should perhaps have been more dynamic.
For a few minutes the entire problem is: move your body, count, finish.
Then sit down again.
That temporary certainty is surprisingly useful after spending an hour inside something that refuses to provide any.
AI can make the hole deeper faster
Modern software work has made this old habit relevant again for a new reason.
AI can generate, inspect and change things extremely quickly. That is a huge advantage when the direction is right.
When the direction is wrong, congratulations: you have mechanised digging.
There is almost no friction between one attempted fix and the next. You can make six changes in the time it once took to consider one properly. Each result creates the next suggestion. The machine is happy to keep going because you keep asking it to.
A physical interruption introduces a small amount of useful friction.
I stand up. I move. I stop looking at the problem.
When I come back, the bug is not guaranteed to have developed a conscience and fixed itself. Quite often it is sitting exactly where I left it, looking smug.
But I am approaching it again rather than continuing the previous reaction.
That distinction matters.
The science supports moving more than it supports my peculiar numbers
The broad idea of breaking up long periods of sitting is not especially controversial. Current UK guidance advises adults to minimise prolonged sedentary time and break it up with activity where possible.
The more specific question — whether short activity breaks reliably make you cognitively better in the next few minutes — is much less tidy. Reviews have found some small benefits in areas such as executive function and memory, but the evidence is mixed and often low-certainty.
That is fine by me because I am not claiming Twenty Every Twenty is a scientifically optimised cognitive-performance system.
My claim is much smaller.
For me, movement is a useful interruption.
It changes the state of the problem because it changes what I am doing.
That is enough.
Sometimes thinking harder is not more thinking
There is a particular kind of vanity in assuming that if thinking has not solved the problem, the solution must be more thinking.
Sometimes it is. Some problems genuinely require longer concentration.
But sometimes you are not thinking more. You are repeating yourself with increasing intensity.
This is especially easy to miss when you work alone because nobody else is there to point out that you have been staring at the same screen for an hour and are now audibly arguing with PostgreSQL.
A physical break is an unsophisticated way of forcing that interruption.
It does not have to be push-ups. Walk downstairs. Go outside. Stretch. Make tea. Do a few weights. Walk around the block.
The useful part is that it is physically different from the thing you have become stuck inside.
The whole point is not to optimise it
There is an obvious temptation, particularly if you spend your life around software, to improve the system.
Should the interval be eighteen minutes?
Would squats be more balanced?
Should heart rate determine the break?
Could an app detect the onset of the perpetual cycle of erroney and deploy a resistance band automatically?
No.
Or at least not for me.
The interruption works because it is less complicated than the thing it interrupts.
Once it has configuration, metrics, notifications and a settings page, we have recreated the problem in activewear.
Twenty Every Twenty survives because I do not have to think about it.
Different work, same chair
The odd thing about returning to a habit from my twenties is realising that the original problem has barely changed.
Back then I was sitting at a desk trying to force legal material into my head for hours.
Now I can be sitting at a desk trying to understand why an application, an agent, a database and several years of accumulated intent have collectively decided that today is the day they will become philosophical.
Different work. Same chair.
The useful moment is still the interruption.
Stand up.
Do something definite.
Finish it.
Come back.
If the problem genuinely needs your attention, it will still be there.
You just do not have to remain inside it continuously.