15 minutes a day beats a 5-hour crunch.

I've had this conversation more times than I can count. A student comes back after two weeks of silence, tells me they spent five hours on Saturday going through problems, and asks why they don't feel like they've improved. The five hours happened. The effort was real. But the two weeks of nothing before it swallowed most of the gain, and they can't see it because the Saturday felt so productive.

Competitive programming skill doesn't accumulate in chunks. It accumulates in layers, and layers need time to set. The five-hour Saturday is not five times more valuable than one hour a day for five days. In my experience, it's often worth less — not because effort is bad, but because the mode of effort matters, and cramming is the wrong mode for this kind of work.

What fifteen minutes actually looks like

When I tell students the floor is fifteen minutes a day, they sometimes assume I mean fifteen minutes of warm-up before the real session. I don't. I mean: on the days when life is full and you have nothing left, fifteen minutes of focused work on one problem is enough to keep the thread alive. That's the floor. Not the goal, not the ceiling — the floor.

In practice, fifteen minutes usually means: open the assigned problem, read it fully once, think about the structure. Maybe you write nothing. Maybe you sketch a few lines. Maybe you realize you've seen this shape before in something I assigned you three weeks ago — and that recognition is the whole point. The graph-traversal intuition you built in October is sitting there; you just need to reach it. You can reach it in fifteen minutes if you show up. You cannot reach it if you haven't shown up in two weeks, because the path back is overgrown.

Most days it doesn't stop at fifteen minutes. Once you're in the problem, you tend to stay. But the commitment is to the floor, not the ceiling, because the floor is what shows up on the streak calendar as an active day. The ceiling is variable and unpredictable. The floor is a choice.

Why cramming doesn't compound

Here's the honest framing: the brain processes what it learned during downtime, not during the session itself. This isn't neuroscience cosplay — it's just a pattern I've watched play out across years of coaching. A student who solves a hard problem on Monday and then does light review on Tuesday and Wednesday tends to own that technique by Thursday. A student who solves the same problem and then does nothing for a week tends to feel vaguely like they've seen this before when I assign a variant.

Cramming produces a different problem: you can hold a lot in working memory for a few hours, which makes you feel sharp. But working memory isn't retention. The student who cranked for five hours on Saturday often can't reconstruct the key insight by Tuesday, because the five hours was all load and no consolidation.

The goal isn't to maximize hours per week. The goal is to maximize the number of things that actually stick, which means spacing out the encounters and giving each one time to land.

This is why I sequence problems the way I do. When I assign something, I'm usually thinking about what I assigned two or three weeks ago and whether you've seen enough variants of the idea to stop having to reconstruct it from scratch each time. If you crammed five problems in one afternoon and then vanished, that sequencing breaks. The variant I assign next week lands with no foundation under it.

The streak calendar is the unit

The activity calendar on your profile isn't there to stress you out. It's there because days are the right unit for tracking this kind of work. Not hours, not problems — days. A day you worked is a day the thread stayed alive. A day you didn't is a gap, and gaps are honest: they're visible, they're countable, and they compound the same way consistency does, just in the wrong direction.

I mark class days on the calendar too — the sessions we do together. Those are fixed points. The question the calendar is really asking is what happens between them. A student who has a class Tuesday and Friday and does fifteen minutes on three other days is doing meaningful work six days a week. That student improves at a rate that surprises them. A student who has the same class schedule but treats the days between as rest days is essentially training twice a week, and then wondering why progress feels slow.

I'm not asking for a lot. The platform shows your current streak and your longest streak, and I'd be lying if I said I don't look at those when I'm deciding what to assign next. A student with a 30-day floor streak has been thinking about problems every day for a month. I can give that student something harder and trust that they'll stay with it. A student who crammed last weekend needs a different kind of problem — something that rebuilds the habit before it tests the ceiling.

The honest part about hard weeks

Finals weeks exist. Family things happen. Some weeks the floor is genuinely not reachable and that's fine — I'm a coach, not a streak counter. If you message me and say you have your math final and two essays due and you're dropping to zero this week, I'll tell you to sleep and come back after. That's not a failure of the system.

What I'm pushing back against is the pattern of treating every week like a hard week. The student who crams when they feel motivated and disappears when they don't is accidentally training the motivation to be the gating factor. And motivation is unreliable. It spikes on weekends and evaporates at 6pm on a Tuesday. If your training only happens when you feel like it, you're training maybe 30% of the available days and wondering why the compound interest isn't showing up.

Fifteen minutes. One problem. Sometimes it turns into two hours; sometimes it stays at fifteen. Both are fine. The point is that you showed up, the calendar has another active day, and when I assign you the variant next week, you'll have something to reach for.

That's the whole thesis. It's not complicated. It's just harder to do than it sounds, which is why I track it and you see it on your profile, and why I'm writing this instead of just assuming everyone already knows it.