What FSRS is, in plain terms
FSRS stands for Free Spaced Repetition Scheduler. It is an open-source scheduling algorithm that treats every card as having a memory state: how difficult the card is, how stable your memory of it has become, and how likely you are to recall it at this very moment. Each time you review, you report how it went, and the model updates its estimate for that card.
The older approach is a table. A card is due after one day, then six, then some multiple of that, adjusted by a simple ease number. It works, and it was a real step forward for its time, but it knows very little about you or the card. If you want the side-by-side, FSRS vs SM-2: what actually differs walks through both algorithms step by step.
Why Cirello uses FSRS-6
FSRS-6 is the current version of the algorithm. It has 21 parameters that shape the forgetting curve and how stability grows after each review. The default parameters are a sensible starting point, but they are an average over many people. Your memory is not an average.
So Cirello retrains the parameters every night against the review history you have actually built up. Early on, with little data, the schedule leans on the defaults. As your history grows, the schedule becomes a better description of how you specifically forget. You do not configure anything for this to happen. The mechanics are covered in Inside Cirello's FSRS-6 pipeline.
What you notice day to day
You will not see a dashboard of weights. What you see is the effect:
- Cards you know well drift toward longer gaps, so they stop taking your time.
- Cards that keep slipping come back sooner, until they hold.
- Each day's queue is whatever is actually due, not your whole library.
- If you blank on a card, it is offered again in the same session instead of waiting for tomorrow.
The goal is a short daily review, a few minutes, that stays proportionate to how well you know the material.
The one setting worth understanding
Every FSRS scheduler is aiming at a target: how likely you should be to remember a card when it comes up. Raise the target and you remember more but review more often; lower it and you review less but forget more. Desired retention in FSRS, explained covers that trade-off and how to think about it.
Beyond the algorithm
A good scheduler is only useful if cards get into it. Cirello is built so that capture is cheap: add front and back text, paste a whole list at once with duplicate detection, attach images to either side, or point a card at a book or website with a short instruction like "page 12" so you review from the real source. The interface is deliberately calm. There is a streak, shown as a plant that grows as the run lengthens and fades, rather than resetting, if you skip a day. Nothing shames you for missing one.
If you are new to the idea, start with what spaced repetition is and why it works, or see the general overview on the flashcard app page.
What Cirello does not do
It is a web app today, with iOS and Android in progress. There are no cloze deletion cards yet, and your library is organized by you into folders and decks rather than sorted automatically. What Cirello can and can't do today keeps that list current. The beta is free, no card required.