The Science

The ChUnKs platform is an attempt to create a training tool based upon scientific research on the nature of chess knowledge, how such knowledge is most efficiently created, and how it is retained. Research has demonstrated that the most efficient means to increase chess playing strength is by focusing on increasing chess knowledge (Gobet & Charness, 2006).

Studies have shown that the time spent on increasing chess knowledge and the size of ones chess library predicted chess playing strength better than playing tournament games, the age at which the subject became serious about chess, and presence of coaching (Charness, Krampe, & Mayr, 1996) (Charness, Tuffiash, Krampe, Reingold, & Vasyukova, 2005).

Further, studies suggest that it is more efficient to build chess knowledge than it is to try to increase the number of moves you can look ahead (Gobet, 2016) because the ability to foresee moves as well as “chess intuition” (the feeling that you know a move is good without being able to explain why) are side effects from a well-developed knowledge-base (Gobet & Jansen, 2005).

The three components of the ChUnKs platform under development, which we refer to as ChUnKs_e, ChUnKs_c, and ChUnKs_r—“e”valuation, “c”omprehension, and “r”etention—are, therefore, intended to enable chess players to efficiently increase chess knowledge.

What is Chess Knowledge?

Chase and Simon (1973a) proposed an influential theory of chess knowledge, referred to as “chunking” theory, later refined by Gobet and Simon (1996a), among others, as “template” theory. These researchers proposed that the fundamental unit of chess knowledge can be described as a “chunk,” a small arrangement or pattern of pieces stored in long-term memory that is perceptually treated as a whole (Gobet, 2019).

For beginners, chunks consist of individual pieces on a given square. As players progress, the chunks increase in number and size, include larger configurations of pieces, and become associated with particular actions (given this pattern of pieces, knights should be exchanged), referred to as procedures, and other forms of higher-order knowledge, such as tactical motifs (if a queen is on the same diagonal as a knight, consider using a bishop to pin the knight to the queen), strategic principles (two bishops are usually better than two knights) and endgame principles (passed pawns should be pushed) (Gobet, 2019).

With further progression, chunks develop into templates, which are themselves chunks and consist of stable information with slots for information that is variable (Gobet, 2019). For example, a template may be comprised of a particular pawn structure with a slot consisting of whether a bishop or a knight occupies a certain square relevant to the structure. Just like chunks, templates provide useful information for decision-making, such as plausible moves and standard plans.

Gobet and Simon explain that chunks and templates can be recalled instantly because they are stored in long-term memory (Gobet & Simon, 1996a). This feature explains how chess players are able to come up with plans and find strong moves very quickly without consciously considering all possible moves available (Gobet, 2019).

It also explains chess intuition. Specifically, Gobet and Jansen state that intuition is comprised of chunks and templates stored in long-term memory that are associated with emotions (the gut feeling of whether a move is good), but not specific conscious thought (Gobet & Jansen, 2005).

Some have estimated that chess masters have learned between 10,000 and 100,000 chunks (by comparison, approximately 50,000 words comprise an average college student’s vocabulary), and that grandmasters may have as many as 300,000 stored in long-term memory (Gobet, 2016).

Chunks and templates do not exist as isolated knowledge units. As chess players become more skillful, chunks, templates, and procedures become better organized or indexed and cross-referenced with one another in the course of a learning process that occurs at both the conscious and subconscious level. This knowledge-base enables chess masters to use strategies adaptively and flexibly. With simple problems or high time pressure, masters may rely more on intuition. With complex problems and enough time to think, they will use a combination of intuition and deliberation (Gobet, 2016).

A chess master’s knowledge-base, therefore, can be characterized as a large set of chunks, templates, and procedures that have been richly indexed and cross-referenced, providing the master with an apparent seamless ability to play better than less skilled players.

How is playing strength increased?

According to the research, increasing playing strength requires chess players to build a highly indexed and cross-referenced base of chess knowledge (chunks, templates, and procedures). To build such a knowledge-base, the information to be learned should be clear and appropriate to the skill level of the learner, is best learned through a specific type of practice, referred to as deliberate practice, and is optimally transferred into long-term memory through a specific type of review, referred to as spaced-repetition learning. There is no short-cut to chess mastery. It has been estimated that it may take 10 years or between 3,000 and 23,000 hours of deliberate practice to become a chess master (Gobet, 2019).

First, regarding how information should be learned, Gobet and Jansen outlined three principles for guiding information processing (Gobet & Jansen, 2005).

The first principle is that learning is optimized when it proceeds “from the simple to the complex.” The basic building blocks of knowledge must first be acquired before more complex knowledge, such as detailed templates, can be created. Second, learning is optimized when the “elements to be learnt are clearly identified.” This assists with indexing and cross referencing the concepts learned as well as enabling the player to generalize concepts to novel positions.

Finally, learning is optimized when following an “improving spiral.” This means that instruction should begin with the basic material, and thereafter regularly return to this material, gradually adding more complex information (Gobet, 2016).

ChUnKs_c, the comprehension component of the training platform, will be designed to facilitate learning in this manner by enabling live coaching sessions or sessions with a fellow player, making a chess engine available, providing a palette (chess symbols and other notation type functions, such as highlighting certain squares and drawing arrows) for annotating positions and games, and, most importantly, by allowing chess players to create virtual flash-cards reflecting important concepts distilled from analysis.

Second, chess knowledge is increased (and, as a result, playing strength) through what researchers describe as deliberate practice (Gobet, 2019). Deliberate practice means engaging in goal-directed activities at an appropriate level of difficulty over long periods of time. These activities are highly structured and are designed to improve performance by eliminating weaknesses through optimizing opportunities for error correction. The activities should be monitored so that regular and detailed feedback can be provided, a role typically filled by a coach (Gobet, 2016).

ChUnKs_e, the evaluation component of the platform, and ChUnKs_c will enable players to engage in such deliberate practice. ChUnKs_e will provide specific feedback on a players’ strengths and weaknesses. This information will guide training by enabling players to focus training time on weak areas. At regular intervals, the player will take additional tests to gauge improvement and obtain guidance on new areas that need attention (weaknesses will change as the player’s playing strength increases, as an area that was a strength when one was a novice, may be weak when one is more advanced). Similarly, ChUnKs_c will support the provision of feedback, as well as enabling the regular monitoring of progress, through live coaching sessions or sessions with a more experienced fellow player.

Finally, research indicates that information to be learned is best transferred into long-term memory (retained or internalized) through a method referred to as spaced repetition learning. This method is based on the theory that information retention is maximized when it is reviewed at certain intervals (days, weeks, and months) following instruction (Smolen, Zhang, & Byrne, 2016). In other words, people begin forgetting information learned very shortly after instruction unless the information is consciously reviewed time and again (Shrestha, 2017). First identified by Hermann Ebbinghaus in 1885, the amount forgotten is exponential in nature, meaning that it falls along a “forgetting curve” (Chun & Heo, 2018).

This means that memory retention is 100% at the time of learning, but may drop to 40% within the first few days. If, for example, the information is reviewed again on the first, third, and sixth day from instruction, research has demonstrated that 80% to 90% of the information may be retained.

As summarized above, a critical aspect of chess players’ playing strength is the ability to recall from memory patterns of pieces and higher order knowledge (chunks, templates, and productions). ChUnKs_r, the retention component of the platform, facilitates the transfer of chess knowledge into long-term memory by enabling chess players to create virtual flashcards containing the information that the player distilled in the comprehension component of the platform.

The platform would then automatically prompt the chess player to review the virtual flashcards created at intervals designed to maximize retention (days, weeks, and months in accordance with an optimal spaced repetition algorithm). Through this cycle—evaluation, comprehension, and retention—chess knowledge, and, therefore, playing strength will be efficiently and maximally increased.

Charness, N., Krampe, R. & Mayr, U. (1996). The Role of Practice and
Coaching in Entrepreneurial Skill Domains: An international Comparison of Life-Span Chess Skill Acquisition. In K. A. Ericsson (Ed.) The Road to Excellence: The Acquisition of Expert Performance in the Arts and Sciences, Sports and Games (pp. 51-80). Mahwah, NJ: Erlbaum.

Charness, N., Tuffiash, M., Krampe, R., Reingold, E. M., & Vasyukova, E.
(2005). The Role of Deliberate Practice in Chess Expertise. Applied Psychology, 19, 151-165.

Chase, W. G., & Simon, H. A. (1973a). Perception in Chess.Cognitive Psychology, 4, 55-81.

Chun, B. A., & Heo, H. J. (2018). The Effect of Flipped Learning onAcademic Performance as an Innovative Method for Overcoming Ebbinghaus’ Forgetting Curve. ICIET 2018, January 6--8, 2018, Osaka, Japan.

Gobet, F. (2016). Understanding Expertise. A Multi-DisciplinaryApproach. Macmillan Education & Palgrave.

Gobet, F. (2019). The Psychology of Chess. Routledge. at 13-15.

Gobet, F., & Charness, N. (2006). Chess and Games. Cambridge

Handbook on Expertise and Expert Performance. Cambridge University Press.

Gobet, F., & Jansen P. (2005). Training in Chess: A Scientific
Approach. In Redman T. Education and Chess.

Gobet, F., & Simon, H. A. (1996a). Templates in Chess Memory:
A Mechanism for Recalling Several Boards. Cognitive Psychology, 31, 1-40.

Shrestha, P. (2017). Ebbinghaus Forgetting Curve, in Psychestudy,
November 17, 2017, Link.

Smolen, P., Zhang, Y, & H. Byrne, H. J. (2016). The Right Time to
Learn: Mechanisms and Optimization of Spaced Learning. Nat Rev Neurosci. 2016 February, 17(2): 77–88.