Daily Chess Puzzle: A Habit, Not a Retention System

Open any "daily chess puzzle" page and you get the same shape: one puzzle, refreshed at midnight, the same position shown to every visitor that day. Chess.com puts it plainly in its own materials: "every day, more than a million members come to Chess.com to solve the exact same puzzle." Lichess, Listudy, Chessigma, ChessMood and Shredder Chess all ship the identical format underneath their own branding. It is a genuinely good habit. A short, low friction puzzle every day is easy to keep up, and doing it once a day rather than in one long binge gives you one sleep consolidation cycle instead of none.

What none of those six products do, checked directly today, is decide for you when a puzzle you got wrong should come back. That is a narrower claim than "no memory at all," and it is worth being precise about it, because Chess.com specifically does let you go dig up your misses by hand. The honest version of the claim survives the correction: a daily puzzle is a habit. Turning that habit into something you actually remember needs scheduled re-exposure, not just repetition of the ritual.

What six daily puzzle products actually give you free

Chess.com is the most built out version of the format. Its Daily Puzzle refreshes at midnight in your own local time zone, and each attempt gives you five hearts, losing one per wrong move or hint used; running out of hearts does not end the attempt, it only costs you the streak eligible "perfect" state for the day. A streak survives a missed day within limits: Chess.com's own support pages say each Daily Puzzle "must be completed within 48 hours of its release," and a calendar interface lets you page backward and open any specific past date. That calendar is a real archive, not scheduling. It shows the same fixed puzzle for a given date to everyone who opens it, and it never decides which past puzzle to put back in front of you or when.

Chess.com does go one step further than the rest of this list: a Custom puzzle mode carries a filter called "Puzzles I Have Missed," per a Chess.com forum reply walking a member through where to find it. Nothing in that thread describes an interval, a priority order or a forgetting curve model deciding when a missed puzzle resurfaces. You have to remember the filter exists and open it yourself, on no particular timetable, which is functionally the same gap this article is about, just with an extra manual step available. As of September 2026, free "Basic" accounts, and guests, are capped at three puzzles a day; unlimited puzzles, unlimited Puzzle Rush and unlimited Puzzle Battle all unlock together at the Gold tier, per Chess.com's own membership comparison, not the higher Diamond tier some secondary sources claim. Chess.com's own forums also document a real, dated glitch in this same random delivery mechanism: starting 26 February 2026, a cluster of roughly twenty complaints reported the identical puzzle repeating, one user counting "Same puzzle 7 times in a row," a paying subscriber writing "fix it! we pay to play puzzles!!!!" That is a bug, not a design choice, and it may already be fixed by the time you read this. But it is a real illustration of what "no per-user memory" can do when the random selection layer misfires: even a paying account can get stuck reviewing nothing new at all.

Lichess is the cleanest counterpoint on price, and a real, structural advantage worth crediting rather than downplaying. It is, in its own words, "100% free" and ad-free, funded by patron donations rather than by selling ads or user data, with no premium tier of any kind anywhere on the site. Its Daily Puzzle is the same one-a-day format as everyone else here, and its Puzzle Dashboard organizes your results by tactical theme, which tells you that you are weak on pins in general. It does not track that you specifically missed today's puzzle and decide when a version of it should come back.

Listudy, an open source, volunteer run site, is free with no account required; at the time of this check its daily page was showing "Tactic #3436." Beyond the daily puzzle it also runs a separate, unlimited Random Tactic mode plus drill variants like Blind Tactics and Pieceless Tactics. Nothing on the page tracks a past attempt, a performance history, or any scheduling logic at all.

Chessigma is the sharpest competitor on this list, and the strongest single piece of evidence for the argument this page is making. Its puzzles page advertises "Unlimited puzzles daily," directly contrasts itself against "Chess.com's measly 5-puzzle limit" (a figure that, per Chess.com's own support page, is actually three, not five, so even the competitor most invested in the comparison has the number wrong), and picks each next puzzle's difficulty from a real, continuously adjusting Elo rating. But as of September 2026 its own page states plainly that Woodpecker-style focused drilling is "coming soon," and that weakness-specific tracking is a planned coaching feature, not a shipped one. Its actual, current retention advice is a habit-frequency claim rather than a scheduling system: "15 puzzles daily beats 100 puzzles once a week." The Woodpecker Method itself, the one it is promising, is worth two sentences here and no more: solve a set of roughly 1,100 exercises once, then re-solve the same set repeatedly, aiming to roughly halve the time each cycle across up to about seven cycles. We cover the book, the grandmaster result and the honest criticisms of that method in full in our piece on why puzzle rating doesn't match game rating. Nobody in this survey, including Chessigma, ships real scheduling as a live feature yet.

ChessMood requires a login just to solve its one puzzle, rewards a correct first-try solve with "500 MoodCoins," and shows a "Recent puzzles" list of past dates you can revisit, again a static browse list rather than a system that decides when your specific miss should come back. Its own framing credits "Grandmaster coaches" for hand-picking the puzzle, not an algorithm.

Shredder Chess is the bare version: three difficulty levels, updated daily, free, with no login mentioned and no tracking, archive or review system of any kind described on the page.

The habit case, taken seriously

Chessigma's own blog post about its daily format makes a real argument, not just marketing copy, and it deserves a straight answer rather than a dismissal: "No backlog guilt, there's no pile of missed puzzles waiting for you. Just today's puzzle, done or not done." The post argues that thirty separate daily sessions across a month give you thirty overnight sleep-consolidation cycles, while doing the same hundred puzzles in one sitting gives you exactly one. That is a legitimate case for the single-daily-puzzle format as a habit device, and nothing in this article disproves it. A short daily session you actually keep up beats an ambitious plan you abandon in week two.

The problem is that a habit and a retention system solve different problems, and the format only solves one of them. Doing today's puzzle every day tells you almost nothing about whether you will still recognize the specific pattern from three Tuesdays ago when it shows up in an actual game. That is not a knock on the habit. It is a gap the habit was never built to close.

What the forgetting curve actually says about one exposure

The underlying research is old, well replicated, and specific. Hermann Ebbinghaus's 1885 forgetting-curve experiments were independently replicated by Jaap Murre and Joeri Dros in a 2015 PLOS ONE paper, relearning the same nonsense-syllable lists after gaps ranging from 20 minutes to 31 days. At short intervals the replication tracked the original almost exactly: a savings score of 0.558 at 20 minutes and 0.422 at one hour, in both the 2015 data and Ebbinghaus's own 1885 numbers. At the longest interval tested, the two diverged sharply, 0.041 savings at 31 days in the replication against 0.165 in the original. The authors' own conclusion is that the curve "is not completely smooth but most probably shows a jump upwards starting at the 24 hour data point," consistent with a sleep-related consolidation boost. Translated plainly: a single exposure, with no re-exposure, loses the great majority of its value within about a day. That is exactly the window a once-a-day puzzle format never revisits for a specific miss, because tomorrow's puzzle is a different position entirely.

A 2006 meta-analysis by Cepeda, Pashler, Vul, Wixted and Rohrer, published in Psychological Bulletin, pooled 839 assessments across 317 experiments in 184 separate studies and found the ideal gap between two study sessions is not one fixed number. It scales with how long you actually need to remember the material: the gap that maximizes final retention grows as the retention target grows. A 2008 follow-up by Cepeda and colleagues in Psychological Science, run on more than 1,350 participants and retested up to a year later, put a concrete figure on that scaling: the optimal gap before a single review session is roughly 20 to 40% of a one-week retention target, shrinking to roughly 5 to 10% of a one-year target. A fixed, generic "every day" cadence, which is the only cadence any product in the head-to-head above actually offers, is not what that research says optimizes long-term retention. It is one specific interval, applied uniformly to everything, regardless of how well or badly you already know it.

What a real scheduler computes on a miss

FSRS, the Free Spaced Repetition Scheduler, is the open algorithm this section is built on. It tracks three numbers for every individual item: Difficulty, Stability (how many days the item can go before your recall probability drops to a target level), and Retrievability (its current estimated recall probability right now). Every single review recalculates all three, and the recalculation depends on whether that review was a pass or a fail. A failed review specifically drops Stability sharply, which is what pulls a missed item's next appearance much closer instead of leaving it wherever it happened to land. A static once-a-day format has no equivalent mechanism at all. The open-source community behind FSRS, building on research by Jarrett Ye at MaiMemo and extending Piotr Wozniak's older DSR model, reports it needs roughly 20 to 30% fewer reviews than the older SM-2 algorithm to hold the same retention level.

A worked example, with the caveat stated plainly

Here is what that mechanism actually computes, not as a marketing claim but as real output from Chessdrive's live scheduler, run directly against ts-fsrs 5.4.1, the exact library and version the shipping app depends on. One thing needs to be said plainly before the numbers: this dataset comes from the app's opening-repertoire spaced-repetition drills, not from a puzzle-specific run. Chessdrive's puzzle feature today uses Woodpecker-style repeating cycles, covered above, not this FSRS scheduler. The numbers below illustrate the general mechanism a real scheduler computes, not a puzzle-specific statistic, and none of it should be read as one.

At a fixed 0.80 retention target, a fresh card that gets a correct ("good") review every time it comes up climbs fast: 8 days after the first review, 75 days after the second, 521 days after the third. Miss that fourth review instead, grading it "again," and the interval does not shrink a little, it collapses, down to 18 days. Two more correct reviews rebuild it, to 73 days and then 251 days, still well short of where it would have landed without the lapse.

That is the concrete difference between scheduled and not scheduled. A once-a-day puzzle format has no way to notice that review 4 was a miss and shorten anything in response; the next puzzle you see is whatever the format was going to show you regardless. Nothing about this number is puzzle-specific, and it should not be mistaken for one. It is one worked example of what the FSRS algorithm computes on any card, in any Chessdrive feature that actually uses it.

More reps isn't the fix either

It's tempting to conclude the answer is just doing more, faster, inside the existing daily-puzzle format: Puzzle Rush, Puzzle Storm, whichever timed mode is on offer. The one real head-to-head test of that idea argues against it. The Say Chess newsletter ran a 14 day test of Puzzle Rush against Puzzle Storm, 89 participants starting and 50 completing, a 56% completion rate, each doing at least one daily session of 30 to 60 minutes, Puzzle Rush on Chess.com and Puzzle Storm on Lichess. The two formats diverged in an interesting way: the heaviest-volume Puzzle Rush solvers showed lower average improvement than the rest of the group, and the heaviest-volume Puzzle Storm solvers actually got worse, their scores declining over the fourteen days. The study's own conclusion: "mere volume might not be as important as the quality of practice." Speed-solving under a clock can train guessing over calculation, and grinding a bigger number inside an already unscheduled format doesn't substitute for deciding, on purpose, when a specific miss should come back.

What actually closes the gap

None of this is an argument against keeping a daily puzzle habit. Keep it. It's low friction, it's consistent, and Chessigma's sleep-consolidation case for doing it daily rather than in one binge holds up on its own terms. What it needs is something sitting on top of it that decides, for you, when the puzzle you actually missed should come back, instead of leaving that decision to a manual filter you have to remember exists, or to nothing at all. That's the difference between a habit and a system: a habit gets you to show up, a system decides what you see once you do.

Chessdrive's own puzzle feature is built around exactly that gap, with the Woodpecker Method rather than FSRS specifically: a curated set at one of four difficulty tiers, with the cycle number and your place in it tracked for you, so a puzzle you got wrong on cycle one comes back on cycle two rather than whenever you happen to remember to go looking for it. That's a narrower, more specific promise than "the strongest engine" or "the most puzzles," and it's deliberate. For the fuller picture of what a paid coaching app actually includes, and what it costs, see our buyer's guide to chess coaching apps.

The honest summary: a daily chess puzzle is worth doing. Every product surveyed here proves that much simply by existing. On its own, though, it is not a memory system, it is a ritual. What turns solving into remembering is deciding, on purpose, when a specific miss comes back, and that is what scheduled re-exposure means, and what none of the six products checked here has actually shipped yet, Chessigma included, by its own admission.