Caffeine and alcohol are timing problems

You can fall asleep after an evening coffee, so you concluded caffeine doesn’t affect you. You sleep better after a drink, so you concluded it helps.

Both conclusions are drawn from the same evidence — how you felt at the moment of falling asleep — and that is exactly the evidence neither of these things reveals itself in. What they do happens later, and mostly to a part of the night you were not conscious for.

Caffeine outlasts feeling alert

Caffeine works by blocking a signal. Across a long day, a substance called adenosine accumulates in your brain, and the accumulation is part of what makes you feel progressively more like sleeping. Caffeine occupies the places that signal is read, so the pressure is still building and you stop noticing it.

Two things follow, and the second is the one people miss.

The sleepiness was never removed, only masked. When the caffeine clears, the accumulated pressure is still there in full. Nothing was paid off.

It clears slowly. The commonly cited figure is a half-life of a few hours — very roughly four to six for most adults, with genuinely wide variation between people. What that means practically is not “it lasts five hours”. It means that after five hours or so, half of it is still working. After ten, a quarter. A large afternoon coffee is still partially present at bedtime, whether or not you can feel it.

And you cannot feel it, which is the trap. The subjective alertness fades long before the substance does. Falling asleep fine after an evening coffee is not evidence that it did nothing; it is evidence that sleep pressure won a race, which it often does. What the caffeine may still have altered is the shape of the night.

The individual variation here is unusually large — how fast people clear caffeine differs several-fold, and it is affected by other things going on in the body. Some people genuinely can drink coffee at nine and be unaffected. Most who believe that are judging by onset.

Which makes the useful intervention a cutoff time rather than a quantity. Pick a time in the early afternoon, and let the amount before that be whatever it is. The instruction “cut down on coffee” is harder to follow and addresses the wrong variable.

Alcohol does the opposite of what it feels like

This one is worth being precise about, because the felt experience is genuinely misleading rather than merely unreliable.

Alcohol is sedating. It shortens the time it takes to fall asleep, and it does that noticeably, which is why the belief that it helps is so durable — the part of the night you consciously evaluate is the part it improves.

Then it is metabolised, over the first several hours, and the second half of the night runs without it. What is well established about that second half is that it becomes fragmented: more wakings, lighter sleep, and a characteristic pattern of coming fully awake in the small hours and struggling to get back. The 3am waking after an evening of drinking is not a coincidence and it is not dehydration alone; it is the shape of the night after a sedative has worn off.

Alcohol also suppresses REM sleep early in the night, which then tends to rebound later — part of why the back half is restless and why the dreams are vivid and odd.

So the honest summary: alcohol trades the first half of the night for the second, and you are awake for the half it improves.

The lever, again, is timing. The gap between the last drink and going to bed is what determines how much of the metabolising happens before you are asleep for it. Several hours changes the picture considerably; an hour changes nothing.

The wind-down

THE WIND-DOWN — caffeine and alcohol

  · A caffeine cutoff in the early afternoon
                    → the whole intervention. Half is still
                      working hours after you stop feeling
                      it.

  · Judging caffeine by whether you fell asleep
                    → useless as evidence. Onset is the one
                      thing it doesn't reliably show up in.

  · Hours between the last drink and bed
                    → decides how much of the metabolising
                      happens before you are asleep for it.

  · Alcohol as a sleep aid
                    → does the opposite of how it feels.
                      Faster onset, fragmented second half,
                      the 3am waking.

  · Cutting down on the amount rather than
    moving the time
                    → MARGINAL by comparison. Both of these
                      are *when* problems.

  · How fast you clear caffeine
                    → VARIES several-fold between people. Some
                      genuinely are unaffected in the
                      evening; most who think so are judging
                      by onset.

  · Sleeping badly in a dark, cool room with
    nothing in your system
                    → WON'T FIX. Removing these removes two
                      obstacles. It does not produce sleep,
                      and it explains nothing about a night
                      that was already clean.

What the cutoff is doing, and why afternoon

Working backwards from the half-life is the whole reasoning. If a meaningful fraction is still active five hours later and a smaller but real fraction at ten, then a cutoff sometime in the early afternoon puts bedtime far enough down the curve for what remains to be small for most people.

That is not a rule with a correct hour attached. It is an arithmetic you can do yourself: count back from when you want to be asleep, and be sceptical of your own sense of whether the earlier coffee is “still in there”, because that sense is precisely what caffeine interferes with.

Two practical notes. Decaf is not caffeine-free, though it is close enough that it works as the replacement for the ritual — and the ritual is often what the late coffee was actually for. And caffeine arrives in more places than coffee: tea, cola, energy drinks, chocolate. Nobody needs to audit this forensically, but a large evening tea is not the neutral choice it is treated as.

The awkward case: the job needs the coffee, the evening involves a drink

Advice that assumes you can rearrange your consumption around your sleep has the causality backwards for most people. The coffee exists because the morning is hard. The drink exists because of who you were with.

So the version of this worth actually using is small and unglamorous. Move the last coffee earlier by an hour rather than eliminating it — the arithmetic rewards that more than it looks like it should, because you are moving down an exponential curve. Put the gap in on the alcohol side rather than the quantity: finishing earlier in the evening is more achievable in company than drinking less in it, and it is the variable that matters.

And notice the loop, because it is the actual mechanism keeping most people in this pattern. A bad night makes the next day harder, which makes the caffeine later and larger, which makes that night worse. The cutoff is not a virtue exercise; it is the point where the loop is easiest to interrupt.

What this changes and what it doesn’t

Moving both of these earlier removes two of the most common and most invisible causes of a bad night — invisible because neither shows up where people look for it. That is worth doing and it is often the single highest-yield change available to someone who has already got the room right.

But it is subtraction, not addition. A clean system does not produce sleep; it stops actively interfering. If the room is dark and cool, the wake time is steady, nothing was drunk after lunchtime, and the nights are still bad for months — none of these levers is the explanation, and the next step is a doctor rather than a stricter cutoff.