The Scheduler: A Study Plan Run by an Algorithm, Not an AI Guess

The hardest question in studying isn't how do I learn this. It's what should I actually be doing today. Get that one wrong and you either spend three hours cramming something that didn't need it or you drift and quietly fall behind without noticing.

Most tools answer it in one of two unhelpful ways. Either they dump everything overdue on you at once and let you sort it out, or they pass the question to an AI and let it guess. Norudit's scheduler does neither, because it runs a deterministic algorithm, and that word is basically the entire point of this post.

Deterministic, meaning you can check it

Deterministic means that given the same inputs, the same knowledge state, the same things due, the same deadlines in front of you, it produces the same plan every time. That makes it reproducible, and more usefully it makes it explainable, because the plan follows from your data rather than from what a model felt like emphasising that morning.

Now compare that to asking a chatbot what you should revise. You're trusting a box you can't see into, one that might give you a different answer tomorrow for no reason you'll ever discover, or invent a deadline you haven't got, or drift toward whatever it happened to weight heavily on the day. You can't check the working. And study time is the one resource nobody gives you back.

So the schedule gets built out of two research-grounded algorithms and your actual obligations, and nothing else:

• FSRS-6, the modern spaced-repetition algorithm that also runs under current Anki, which models your memory and predicts the point you're about to forget something.
• Bayesian Knowledge Tracing, out of decades of intelligent-tutoring research, which estimates how well you genuinely know each concept from the pattern of how you've been answering.
• Your fixed obligations, meaning the exams on your timetable, your coursework deadlines, and the classes you actually have this week.

There's no "and then trust the AI" step hiding anywhere in the middle of that. The plan is computed, which is what lets it be not merely plausible but close to optimal for the way memory behaves.

How it decides the order

Four things get weighed every day and turned into one ranked list.

What's due, from FSRS-6. Every concept you've studied has a predicted moment where it's about to slide out of memory, and FSRS-6 aims the next review to land just before that point. Too early and you've spent a repetition for nothing. Too late and you've forgotten it and you're relearning from scratch. So things surface at the most efficient moment available rather than on a fixed interval.

What's weak, from BKT. It estimates how firmly you hold each concept from your answers and updates that every time you're tested, so weak concepts get pulled forward and the ones you've clearly got step back. Your effort ends up going at the gap rather than spreading itself evenly over material you already know, which is what happens by default when you're choosing for yourself.

What's fixed, from your obligations. Exams and deadlines are hard anchors and the scheduler works backwards from them, so an exam a fortnight out changes what lands on today instead of what lands on the night before it.

What fits the day you've actually got. The plan is shaped around your real timetable, so it won't hand you eight hours of work on a day you've got one free hour.

Those four collapse into a single ordered list of the highest-leverage things available right now. Not a wall of cards, not a vibe, and every item on it traceable back to a reason if you wanted to go and check.

A backlog that doesn't turn into a cliff

Everybody misses days, and most spaced-repetition tools punish you for it. You skip a week, you come back to a few hundred overdue cards, and the demoralisation finishes the job. The maths was never what went wrong there. Nobody designed for the fallout.

So the backlog gets managed rather than dumped. Miss sessions and the plan re-solves, spreads the catch-up out sensibly, and puts the most important and most at-risk things first. Falling behind becomes something you climb out of over a few days rather than a cliff you drop off, and because the backlog is recomputed rather than guessed at, what's waiting for you when you come back is still the right plan.

Exam prep pointed at the actual exam

Because the scheduler knows your exam dates, it doesn't just tick over doing maintenance review forever. As the date gets closer the plan bends toward it, and it can assemble a timed mock aimed at your weakest topics and tie that mock to the specific exam on your timetable, so your final practice sharpens toward the real thing at roughly the right moment instead of landing at random.

Reminders, because a plan you have to go and check is a plan you'll drop

Norudit emails yours to you. A short daily note of exactly what to do today, so the decision has already been made before you've opened your laptop and the "hang on, what was I meant to be doing" friction never gets a chance to start.

It re-solves the second anything moves

Add a deadline, miss a day, finish something early, have an exam shifted, and the whole thing recalculates at once, so you're never working off a stale schedule. Because it's an algorithm rather than a fresh guess, that recalculation is instant and consistent every time. Move one thing and everything downstream quietly rearranges itself around it.

The unglamorous engine

Scheduling is the least interesting part of studying and the first thing nearly everybody gets wrong. It's also the engine keeping the rest of the method turning: the right review at the right moment, the right practice before the right exam, and a sane way back in when life gets in the way for a fortnight. And it works for the same reason you can trust it, which is that it refuses to guess.

Read more

• How Norudit's flashcards rebuild themselves so you can't just memorise the card: The Deck That Rewrites Itself
• How Norudit fits its schedule to your own forgetting curve: How Norudit Actually Personalises Your Learning