For about six years, my default working rhythm was the canonical Pomodoro — 25 minutes of work, 5 minutes of break, four rounds, long break, repeat. I learned it back when I was juggling a part-time software job in Buenos Aires with side projects nobody had asked me to build, and the structure genuinely saved me. I went from “open laptop, stare, panic, close laptop” to actually shipping things. For years I assumed 25/5 was just how focused work worked.
Then one specific night broke it for me. I was debugging the audio pipeline for what would eventually become the lofi stream — a bug where the ffmpeg concat was inserting a single frame of black between every track at the loop point. Almost imperceptible, but viewers in chat had started catching it. I sat down at maybe 11pm, opened the logs, and the 25-minute timer felt like a fly buzzing in my ear. Every time it went off, I’d just gotten the trace into my head — the buffer offsets, the timestamps, the way the encoder was rounding — and the bell was telling me to stop and stretch. I ignored it three times in a row, felt guilty about ignoring it, then finally turned the timer off altogether and worked straight through until about 3am, when I finally caught the off-by-one in the keyframe alignment.
That was the moment the orthodoxy cracked for me. Not in a “Pomodoro is broken” way, but in a “the 25-minute interval is not sacred” way. The next morning I started experimenting, and over the following months I cycled through pretty much every variant in the literature. Some I still use. Some I tried and dropped immediately. The point of this post is to walk through them the way I’d describe them to a friend over coffee, rather than as a feature comparison. If you’ve already tried the classic Pomodoro and felt vaguely wrong about it, you probably need a different block, not a different method. For the basic Pomodoro plus lofi pairing, the main Pomodoro guide is where to start; this is the post about variations.
Let me start with the 25/5 itself, because I want to be fair to it. I still recommend it as the entry point for anyone who has never seriously practiced segmented work. The reason is psychological more than physiological. When you’ve never sustained deliberate focus, even 25 minutes feels long, and the short break is reassuring — you know relief is around the corner, so you don’t keep checking the clock in dread. It’s particularly good for tasks with high distraction potential (think email triage, light reading, problem sets that mix easy and hard sections), and it’s the variant I most often suggest to viewers who write into the chat saying they have ADHD or just genuinely cannot get started. The short ramp-up is the whole point. The 25-minute number itself is mostly arbitrary — Francesco Cirillo named it after his tomato-shaped kitchen timer, not after some peer-reviewed cognitive cycle — so I treat it as scaffolding rather than scripture. Where it fails me, personally, is exactly the situation I described above: deep flow state work, dense academic reading, anything where the cost of context switching is high. By the time I’ve loaded the problem into my head, the bell is ringing.
The next variant I tried, almost by default, was 50/10. This is the academic standard, the one Cornell’s study materials have recommended for decades, related to the Cornell note-taking system more broadly. Fifty minutes of work, ten minutes of break. I’ll be honest — this is my workhorse interval, the one I use for most of what I’d call “structured but not heroic” work. Writing the descriptions for new wallpaper themes on the site, drafting blog posts like this one, working through a stack of viewer DMs, reviewing code from contributors, doing my Anki reviews. It’s long enough that I actually settle into the task and stop fidgeting, short enough that I haven’t burned my whole cognitive battery before the break arrives. The ten-minute break is also genuinely useful in a way that the 25/5’s five-minute break isn’t — you can walk to the kitchen, make a mate, look out the window for a minute, and come back without the feeling that you just sat down. If you’re past the beginner stage and you have a reasonable attention span, 50/10 is, in my biased opinion, where you should be living most of the time.
Where 50/10 doesn’t reach is the kind of work that actually does need ninety minutes to load. That’s where 90/20 comes in, and that’s the variant I use when I’m doing the work that the lofi stream operations are actually built around — debugging the live encoder, writing new pipeline scripts, building out the recommendation system for the wallpaper gallery. Ninety minutes maps reasonably well to the brain’s ultradian rhythm, which Anders Ericsson identified years ago in his research on elite performers across fields — musicians, athletes, chess players — who all seemed to converge on roughly 90-minute peak work blocks before fatigue forced a break. Cal Newport’s “deep work” framework leans on similar timing. What I notice in myself is that the first twenty or thirty minutes of a 90 are spent loading context — recalling where I left off, paging variables back into working memory, reconstructing the mental model — and only after that do I actually get the productive part. A 25-minute block on this kind of work is almost pure setup cost. The trade-off is that 90/20 is brutal on tired days. If I’m already drained from a bad night’s sleep, a 90-minute block becomes a 90-minute slow grind, and I get less out of it than I would from two 50/10s. So I treat 90/20 as the formal-occasion variant, not the daily uniform.
Then there’s flowtime, which I find conceptually beautiful and operationally tricky. The idea is that you don’t set an interval at all — you start working, you note the time, you work until you genuinely lose focus, you note that time too, you take a break appropriate to the block you just did, and you repeat. The data accumulates over weeks, and you eventually learn your actual personal rhythm rather than guessing. I tried this for about a month and it taught me something useful: my real natural block, in the late evening when I do most of my work because I’m a night owl living on Buenos Aires time, is between 70 and 80 minutes. Not 25, not 50, not 90. Knowing that was genuinely valuable. The catch is that flowtime requires honest self-assessment, which is hard when you’re tired or anxious or working on something boring. Beginners read “feel restless” and “lose focus” as the same signal, and they’re not — restlessness usually passes if you sit with it for two minutes, while real loss of focus does not. So I’d recommend flowtime as a diagnostic tool you run for a few weeks, rather than a permanent practice.
A variant I’ll mention briefly because I actually use it more than I expected to is the reverse Pomodoro. You take the five-minute break first — some small low-stakes activity, like loading the dishwasher or watering plants or just walking around the apartment — and then sit down and start the 25 minutes of actual work. It sounds counter-intuitive, but the activation barrier on hard tasks is enormous, and warming up first makes the real work feel less like jumping off a cliff. I lean on this most on days when I’ve been procrastinating something for a week. The trick is to make the warmup something that involves your body, not your screen. If your “5-minute warmup” is checking Twitter, you’ve just lost ninety minutes. Pair this with the 2-minute commitment from our procrastination post and it becomes pretty hard to not start.
The 52/17 variant deserves a mention if only because viewers in chat ask about it every couple of weeks. It came out of a 2014 study by DeskTime that analyzed knowledge worker activity logs and found that the highest-performing 10% averaged 52 minutes of work followed by 17 minutes of break. I tried it for about ten days and honestly couldn’t tell it apart from 50/10 in any subjective way. The exact numbers are an artifact of where the data happened to cluster, not a universal optimum. If 50/10 isn’t quite right for you, by all means try 52/17, but don’t treat the specific numerals as load-bearing — the underlying principle is just “moderate work blocks with substantial breaks,” and you’re welcome to land at 48/12 or 55/15 or whatever fits your actual day.
For viewers who write in saying they have ADHD or chronic attention difficulties, I usually point them at a structure I think of as 3-2 — three short work blocks of fifteen to twenty minutes each, with two short breaks of three to five minutes between them, then one longer break of fifteen to twenty minutes before the cycle repeats. The shorter blocks reduce activation energy, and the more frequent breaks prevent the “I lost focus ten minutes ago but kept staring at the page” pattern that wastes so much time and makes you feel worse than if you’d just stopped. I cover this in more depth in our ADHD-friendly study post. The principle generalizes beyond ADHD — anyone working on a low-energy day or recovering from sleep deprivation can benefit from shrinking their blocks rather than abandoning the practice altogether.
I also want to talk briefly about layering daily planning on top of block timing, because the two operate on different scales and people sometimes conflate them. The 1-3-5 rule, which I borrowed years ago from someone whose name I’ve embarrassingly forgotten, says that each day you plan to complete one big thing, three medium things, and five small things. That’s it. It replaces the infinite to-do list with something that fits in a single mental frame, and it pairs well with any block-timing variant. The block timing is for how you work in any given hour; the 1-3-5 is for what you actually work on that day. Independently, Cal Newport’s time-block planning takes this further — you schedule every block of your day in advance, including meals and breaks and rest, before the day starts. The discipline isn’t sticking exactly to the schedule, which never survives contact with reality. The discipline is making the schedule, which forces you to confront how much is actually possible in a finite day. I don’t time-block every day, but on weeks when I’m under a real deadline, it’s the only thing that keeps me from drift.
If you’re trying to figure out your own pattern from scratch, here’s roughly how I’d structure the experiment. Spend the first week on 50/10 for everything. Spend the second week on 90/20 for everything. At the end of each week, look back honestly: which week did you finish more actual work, not in number of blocks but in number of tasks completed; which week did you feel less wrung out at the end; which fit better with the realities of your class schedule, job, or family. If neither of those felt right, run a third and fourth week on 25/5 and flowtime. By month two, you’ll have a default. And don’t be surprised if your “default” ends up being multi-modal — most experienced practitioners I know, myself included, use different blocks for different work. I use 50/10 for writing and reading, 90/20 for coding and infrastructure work, 25/5 for routine tasks and email triage, and pure flowtime when I’m in the middle of something genuinely engaging and don’t want any timer interrupting me at all.
Time of day matters too. I’m a night owl, which is partly geographic — when you run a lofi stream that serves a mostly North American and European audience from Buenos Aires, your evening is everyone else’s prime studying time, and you start adapting your hours to the chat — so my heaviest 90/20 blocks happen between roughly 8pm and 1am. Morning people should invert this entirely; the 90/20 belongs in their first three or four hours after waking, when willpower and alertness are highest. Afternoons for nearly everyone are a slump zone where shorter blocks (25/5 or 50/10) generally beat heroic 90-minute attempts. I unpack the morning-versus-evening question more in our morning vs night studying post, but the short version is: do your hardest work at your sharpest hour, whenever that happens to be, and shrink your blocks during your dim hours rather than fighting them.
A note on tools, because this is the section where most posts turn into an affiliate-link festival and I’d rather just tell you what I actually use. For 90% of my work I use a kitchen timer — a literal plastic dial timer next to the keyboard. No notifications, no phone unlocking, no temptation. When I’m away from the desk I use my phone’s built-in timer with multiple alarms set. If you want something a little fancier with task tracking attached, TickTick has a competent built-in Pomodoro, Toggl Track does decent time tracking with Pomodoro mode, and Forest is the gamified one viewers seem to love (I cover it more in our apps post). For day-level time-blocking, plain Google Calendar is honestly fine; you don’t need Notion or Sunsama unless you genuinely enjoy the act of configuring tools.
Finally, the mistakes. Across years of doing this myself and watching thousands of viewers describe their attempts in chat, the failures cluster around the same few patterns. People obsessively check the clock during work blocks, defeating the entire purpose of the timer, which is to manage time on your behalf. People skip breaks when they feel they’re in flow — sometimes legitimately, but more often it’s an avoidance of the discomfort of stopping, and they pay for it with a wall in hour three. People “take breaks” by scrolling Instagram, which is not a break at all but a different and arguably worse kind of cognitive load. People count Pomodoros instead of finished work, ending the day proud of eight completed blocks that produced nothing shippable. And people force themselves through low-energy days with sheer willpower, generating six grimly mediocre Pomodoros where the honest move would have been to sleep more, do less, and let tomorrow handle it. Whenever I catch myself in any of these, I try to name it out loud — sometimes literally, in the empty apartment — and reset.
The lofi stream is calibrated for any of these variants, deliberately. Tracks rotate every couple of minutes, so even a 25-minute block has variety; a 90-minute block has even more variety without ever crossing into the “this song again?” fatigue that a tight playlist eventually produces. The audio is meant to sit underneath the timing structure, not compete with it. For the deeper science on why ambient music supports different block lengths in different ways, see our science of ambient music post.
If I had to compress this whole post into something useful, it would be this. The 25/5 is famous because it’s a great entry point, but most experienced focused-work practitioners eventually move to longer blocks or to flowtime. The real principle underneath all of these variants is that structured time with deliberate breaks beats unstructured continuous work. The numbers — 25, 50, 90 — are just tuning. The practice of segmenting and resting is what makes the difference. Run the two-week experiment, find your default, keep the other variants in the toolkit for specific tasks and specific days. And then pair the whole thing with the rest of the system: 10 study techniques, a sensible daily routine, a cozy desk setup, and the 24/7 lofi stream running in the background. That’s the environment I’ve built for myself over years, and the one I keep recommending to viewers, because in the end it’s the only setup I’ve found that produces real progress without burning the operator out.
