The 20-minute rule: how I stopped losing whole evenings to one problem

Last year I spent three hours on one problem. One integral. I did not solve it. I looked up and it was 11:40 pm and I had done nothing else that evening. I told myself that was perseverance.

It was not. It was ego.

I am in Kota for my second attempt, and I will be honest about what the first attempt taught me. Struggling with a problem is useful. Struggling for three hours is not. The first ten minutes teach you a lot. The next two hours teach you almost nothing, because you keep running the same failed idea in a circle and calling it effort.

So I made a rule. I call it the 20-minute rule. It is not clever.

The rule

1. Start a timer when you meet a problem you do not know how to begin. Not a problem you are calculating. A problem you are stuck on.
2. For those 20 minutes, only try. Draw the diagram. Write what is given and what is asked. Try one approach, then a second. Write everything on paper, not in your head.
3. When the timer rings, stop. Even if you feel close. You are almost never as close as you feel.
4. Write one line: where did I get stuck, and what did I try. This is the most important step. Do not skip it.
5. Now you may look, but not at the whole solution. Look only at the first line or the hint. Cover the rest with your palm. Try again for 5 more minutes.
6. Still stuck? Read the full solution. Then close the book and redo the problem from nothing. If you cannot redo it, it goes on the revisit list.
7. Next day, redo it again without looking.

Why 20? There is no science behind it. It is roughly how long I take to run out of ideas. For a heavy maths problem I allow 25 or 30. For a physics MCQ it is closer to 3 minutes. Pick your own number but pick it before you start, because during the problem your brain will always say "just five more minutes".

The integral that ate my evening

Here is the problem:

$I = \int_0^{\pi/2} \frac{\sin x}{\sin x + \cos x}\,dx$

Last year I tried $t = \tan(x/2)$. Yes. The full Weierstrass substitution. It turns everything into a rational function of $t$, and the algebra gets ugly fast. I was 40 minutes in before I had even finished one clean version of the integrand.

The actual solution is four lines. Use the property

$\int_0^a f(x)\,dx = \int_0^a f(a-x)\,dx$

With $a = \pi/2$ we have $\sin(\pi/2 - x) = \cos x$ and $\cos(\pi/2 - x) = \sin x$, so the same integral is also

$I = \int_0^{\pi/2} \frac{\cos x}{\cos x + \sin x}\,dx$

Add the two forms of $I$:

$2I = \int_0^{\pi/2} \frac{\sin x + \cos x}{\sin x + \cos x}\,dx = \int_0^{\pi/2} 1\,dx = \frac{\pi}{2}$

So $I = \dfrac{\pi}{4}$.

What I was missing was not intelligence. I just did not know the trick. And no amount of sitting would have given it to me. Staring harder at a problem only works if the missing piece is already somewhere in your head. If it is not, staring is just sitting.

After this I wrote on the first page of my notebook: limits 0 to a, trig functions swap under a minus x, so try the a − x property. That one line has since saved me at least four times.

The stuck log

This part changed my results more than the timer did. After every stuck problem I add one row to a table at the back of my notebook:

Problem	Minutes stuck	What was missing	Type
$\int_0^{\pi/2}\frac{\sin x}{\sin x+\cos x}dx$	20	the $a-x$ property	trick I didn't know
Block on a moving wedge	20	drew forces in the wrong frame	concept
Equilibrium constant question	15	confused "initial" with "at equilibrium"	reading
Probability with three boxes	20	did not see it was a Bayes situation	concept

Four types: trick I didn't know, concept, reading, silly mistake. After one month I counted the rows. This is only my own notebook, not a study, but the pattern was loud. Most of my rows were "trick I didn't know" and "reading". That told me what to fix. I did not need more hard problems. I needed to read the question slower and I needed a small notebook of tricks.

What the rule is not

It is not "quit at 20 minutes". If you are making progress, keep going. Progress means a new equation, a cleaner diagram, a simplification you did not have before. If at minute 20 you are still staring at line one, that is not progress.

It is also not "look at the solution when it feels hard". You need the struggle. I have watched people read the solution after two minutes and feel productive. That is the opposite mistake. Give the 20 minutes honestly or the rule does nothing.

The Kota part

My hostel room has a window with a view of a wall. For a long time I believed that if I sat in front of that wall long enough, answers would come. My roommate used to get up in the middle of a stuck problem to fill his bottle. He would come back and solve it in two minutes. I thought it was luck. It is not. The brain keeps working in the background when you walk away. So walking to the mess for tea is part of the rule. It is not a break from the work. It is part of it.

I am not a talkative person. I am writing this only because a few juniors asked how I plan problem sets. This is the answer. Time-box it, log it, take a hint, redo it tomorrow. Nothing else.