Skip to content

The science

The science of getting better at chess

Learning researchers have spent more than a century testing which ways of studying change what people remember and can do. A handful of findings are strong enough to build a product on. Here are eight of them, what each one means for chess, and what ChessOS Player does with it, with the sources so you can check us.

In Player today means it is built and you can use it.Coming soon means it is not built yet. We never put a feature in the first group until it works.

The research

Eight findings, and what we do with them

Each one has the same four parts: what the research found, what it means at the board, what Player does today, and what is still coming.

  1. Spread your practice out

    What the research found

    In 1885 Hermann Ebbinghaus measured how fast we forget: quickly at first, then more slowly. Reviewing something just before it fades makes the memory last longer, and each successful review lets the next gap be longer. Reviews of study techniques (Dunlosky and colleagues, 2013) rank practice spread over time among the most useful things a learner can do. The details are still argued about: one well-known experiment found equally spaced reviews did better for long-term memory than steadily growing gaps (Karpicke and Roediger, 2007), so we do not claim any one schedule is perfect.

    At the chessboard

    An opening line you studied once on Sunday is mostly gone by the next tournament. A line you met again after a day, a week and a month is still there.

    Sources: Ebbinghaus (1885); Dunlosky, Rawson, Marsh, Nathan and Willingham (2013); Karpicke and Roediger (2007).

    In Player today

    • Training mixes puzzles from your own games with puzzles near your rating.
    • A puzzle you miss comes back in about 3 days (about 7 for a library puzzle). A puzzle from your own games that you solve comes back after about 12 days, then roughly 42, then roughly 147.
    • A library puzzle you solve cleanly the first time is not repeated; there is nothing to forget.

    Coming soon

    • The same spacing for every lesson question and opening line you have studied, with one daily Review that picks what is due.

    See the idea

    0%25%50%75%100%Day 0102030405060reviewreview
    Reviewed on days 1, 4, 11 and 27: about 98 percent still there on day 30, and 77 on day 60, with four short reviews. An illustration of the idea, not a measurement of any learner.
  2. Answer before you look

    What the research found

    Trying to pull an answer out of memory strengthens it more than reading it again, even when the attempt fails. This is called the testing effect or retrieval practice. Re-reading feels productive because the material looks familiar; familiarity is not the same as being able to produce it (Roediger and Karpicke, 2006).

    At the chessboard

    Seeing the winning move in a lesson is not the same as finding it at the board. The finding is where the learning happens.

    Sources: Roediger and Karpicke (2006).

    In Player today

    • Lessons ask you to find the move or answer the question before the explanation appears.
    • Puzzles are answered by moving the piece on the board.
    • Test-out exams draw questions at random for each try, so passing means you can answer, not that you remember the last attempt.

    Coming soon

    • Review sessions made entirely of questions you have already met, answered by moving the pieces.
  3. Feedback that says what to try next

    What the research found

    Across hundreds of studies feedback is one of the strongest influences on learning, with a catch: it helps when it is about the task, close in time to the attempt, and points to a next step. Feedback that judges the person, or that arrives with no direction, often does nothing or does harm (Kluger and DeNisi, 1996; Hattie and Timperley, 2007).

    At the chessboard

    "Blunder" tells you something went wrong. "This move allowed Qxg6#; next time look at your opponent's checks first" tells you what to do.

    Sources: Kluger and DeNisi (1996); Hattie and Timperley (2007).

    In Player today

    • When a puzzle move fails, the board shows the reply that punishes it, with a short tip.
    • A two-step hint ladder gives you the theme to look for, then the square, only if you ask.
    • Game review tags every move with one of eleven labels and ties each explanation to the position it describes, on the board.
  4. Play it yourself

    What the research found

    People remember actions they perform better than actions they only read about or watch, which researchers call the enactment effect. Imitation matters too: learners copy what experts do and often absorb patterns before they can explain them. Leaving learners to discover things with little guidance has a poor record; guided practice does better (Kirschner, Sweller and Clark, 2006).

    At the chessboard

    Playing the moves of a line, on a board, from a position, builds the pattern in a way reading the same moves in a list does not.

    Sources: Feyereisen (2009); Kirschner, Sweller and Clark (2006).

    In Player today

    • Lesson steps make you play the move on the board.
    • Line lessons play a line out one move at a time and pause on the move that matters, with Back, Next and Replay so you can go at your own speed.

    Coming soon

    • A mode where you play a whole line from memory, one move at a time, with the next move hidden.
  5. Practice at the right level

    What the research found

    Practice helps most when it is hard enough to make you think and not so hard that you are lost, the zone educators call the region of proximal development. Studies of how people spend study time show they often labor in vain: hours on material far above their level (Nelson and Leonesio, 1988). Deliberate practice, meaning a clear goal, repetition, and informative feedback, separates the best performers (Ericsson, Krampe and Tesch-Römer, 1993). It does not explain everything: chess studies found large differences between players who put in the same hours (Campitelli and Gobet, 2008).

    At the chessboard

    A beginner working through a master's opening course is not studying; they are copying moves they cannot yet use. A strong player redoing the basics is not either.

    Sources: Nelson and Leonesio (1988); Ericsson, Krampe and Tesch-Römer (1993); Campitelli and Gobet (2008).

    In Player today

    • Library puzzles are chosen near your puzzle rating, and your own mistakes come back as puzzles.
    • Lessons unlock in order, and a test-out exam lets you skip what you already know.
    • Course pages and cards tell you, quietly and only when it matters, that a course looks like a stretch for your rating (more than 150 points above you) or like review (more than 100 points below you).

    Coming soon

    • Placement puzzles at sign-up that find your level before you pick a course.
  6. Make learning visible

    What the research found

    John Hattie's large synthesis of education research found that the biggest gains come when learners can see what they are trying to learn, how far they have come, and what to do next, and when they become their own teachers: monitoring, evaluating and adjusting their own work (Hattie, 2009). Seeing progress also keeps people going.

    At the chessboard

    "Study more" is not a plan. Knowing that you lose points in the middlegame after move 20, or that you convert winning endgames only half the time, is.

    Sources: Hattie (2009).

    In Player today

    • A progress bar on every course, and a record of what you have finished.
    • Insights from your own games: accuracy by phase of the game, where blunders cluster, which tactics you miss, how you handle your clock, and which winning positions you throw away.
    • A puzzle dashboard with a performance rating that says how many puzzles it rests on.
    • Settings that put you in charge: hints, goals, timers and feedback can each be turned on or off.

    Coming soon

    • A view of how well you still remember each lesson and opening line.
  7. Reward the work, carefully

    What the research found

    Games feel good, and feeling good about studying helps people keep studying. Researchers link reward to the brain's dopamine signals, and dopamine to the willingness to put in mental effort (Westbrook and colleagues, 2020). But the step from "dopamine exists" to "points and badges make you learn faster" has not been shown, and we do not claim it. What studies of rewards do show is a warning: rewards that feel controlling or that are handed out for simply taking part can reduce interest in the task itself (Deci, Koestner and Ryan, 1999). People who care about improving, more than about proving themselves, keep going longer.

    At the chessboard

    The best reward in chess is the moment a position you once lost makes sense. Badges and sounds are a nudge toward that, not a replacement for it.

    Sources: Westbrook and colleagues (2020); Deci, Koestner and Ryan (1999).

    In Player today

    • Badges with tiers, for lessons finished, puzzles solved, games reviewed and days practiced.
    • A streak counter that only you can see, and badge totals that only go up: a rest day takes nothing away from them.
    • Board sounds for moves, captures and checks, and a small celebration when you play a brilliant move.
    • No public leaderboards and nothing for sale: every reward is earned by learning.

    Coming soon

    • A daily goal, levels, and unlocks such as new board looks that you earn by doing the work.
  8. Mistakes are information

    What the research found

    Carol Dweck's mindset research proposes that believing ability can grow changes how people react to mistakes: they try again instead of giving up. Large trials of school mindset programs have found small and uneven effects, so we use the idea for the tone of the app, not as a promise. One result is worth keeping in mind: how people respond right after an error, not only whether they make one, differs from person to person (Moser and colleagues, 2011).

    At the chessboard

    Everyone loses games. What separates players is what they do with the loss.

    Sources: Dweck (2006); Moser and colleagues (2011).

    In Player today

    • A written standard for how Player talks about mistakes: it describes what happened and what to try, never labels you, never compares you with other players. Our own text is tested against it.
    • An exam you have not passed yet says so, then lists the lessons that will help.
    • Your insights compare you with your own earlier games, not with other people.

Putting it together

How the ideas fit in one study session

No single idea does the work. A good session uses several at once.

  1. Step 1

    Start where you are

    Puzzles near your rating and a course that fits (idea 5), with your progress in view (idea 6).

  2. Step 2

    Find, do not read

    Every question asks you to find the move on the board before you see the answer (ideas 2 and 4).

  3. Step 3

    Learn from the miss

    A miss shows why and what to try next, and the puzzle returns later (ideas 1 and 3).

  4. Step 4

    Leave on a win

    Badges and a private streak recognize the work, and a mistake is described as information (ideas 7 and 8).

Honest limits

What we do not claim

  • We do not promise a rating gain, or that ChessOS is faster than any other way of studying. How fast you improve depends on far more than an app: your starting point, your age, how much time you give it, and how consistently.
  • We do not say points, badges or sounds make your brain release more of any chemical, or that this makes you learn faster. The research on rewards is real; that conclusion has not been shown.
  • We do not say there is one best review schedule. Researchers disagree about the details, and the gaps ChessOS uses today are a sensible start, not a proven optimum.
  • We do not use research as decoration. A feature that is not built says "coming soon," and a finding whose evidence is mixed says so on this page.
  • Chess is a hard game and talent is not the only thing that matters, but nobody can honestly tell you your ceiling. We do not try.

How we check ourselves

We will measure whether it works

A method that is right in a study can still be wrong for chess puzzles, or for you. Once the lesson-wide Review is built, ChessOS will record how likely it thought you were to remember each item and whether you did, compare the two, and publish what we find, including where we were wrong, the way the brilliant-move page publishes its checks. Coming soon

Read the sources

Sources

Most journal articles can be found by title on a search engine or Google Scholar. We are not able to host them ourselves.

  • Campitelli, G. and Gobet, F. (2008). The role of practice in chess: a longitudinal study. Learning and Individual Differences, 18(4).
  • Deci, E., Koestner, R. and Ryan, R. (1999). A meta-analytic review of experiments examining the effects of extrinsic rewards on intrinsic motivation. Psychological Bulletin, 125(6).
  • Dunlosky, J., Rawson, K., Marsh, E., Nathan, M. and Willingham, D. (2013). Improving students' learning with effective learning techniques. Psychological Science in the Public Interest, 14(1).
  • Dweck, C. (2006). Mindset: The New Psychology of Success.
  • Ebbinghaus, H. (1885). Memory: A Contribution to Experimental Psychology.
  • Ericsson, K., Krampe, R. and Tesch-Römer, C. (1993). The role of deliberate practice in the acquisition of expert performance. Psychological Review, 100(3).
  • Feyereisen, P. (2009). Enactment effects and integration processes in younger and older adults' memory for actions. Memory, 17(4).
  • Hattie, J. (2009). Visible Learning: A Synthesis of Over 800 Meta-Analyses Relating to Achievement.
  • Hattie, J. and Timperley, H. (2007). The power of feedback. Review of Educational Research, 77(1).
  • Karpicke, J. and Roediger, H. (2007). Expanding retrieval practice promotes short-term retention, but equally spaced retrieval enhances long-term retention. Journal of Experimental Psychology: Learning, Memory, and Cognition, 33(4).
  • Kirschner, P., Sweller, J. and Clark, R. (2006). Why minimal guidance during instruction does not work. Educational Psychologist, 41(2).
  • Kluger, A. and DeNisi, A. (1996). The effects of feedback interventions on performance. Psychological Bulletin, 119(2).
  • Moser, J., Schroder, H., Heeter, C., Moran, T. and Lee, Y. (2011). Mind your errors: evidence for a neural mechanism linking growth mind-set to adaptive posterror adjustments. Psychological Science, 22(12).
  • Nelson, T. and Leonesio, R. (1988). Allocation of self-paced study time and the "labor-in-vain effect." Journal of Experimental Psychology: Learning, Memory, and Cognition, 14(4).
  • Roediger, H. and Karpicke, J. (2006). Test-enhanced learning. Psychological Science, 17(3).
  • Westbrook, A. and colleagues (2020). Dopamine promotes cognitive effort by biasing the benefits versus costs of cognitive work. Science, 367(6484).

See it working on real games

The demo account has a player's games, puzzles, lessons and insights already filled in. No sign-up, nothing to download.