DAN HEISMAN

THE IMPROVING
CHESS THINKER

Second Edition
Revised and Expanded
 Table of Contents

Preface to the First Edition ...................................................................... 7

Preface to the Second Edition ................................................................. 8
Introduction ............................................................................................ 9

1. The Exercise .....................................................................................16

2. Class F ..............................................................................................27
3. Class E..............................................................................................40
4. Class D .............................................................................................60
5. Class C .............................................................................................78
6. Class B............................................................................................100
7. Class A ...........................................................................................119
8. Expert and Above ............................................................................140
9. Thought Process Basics ...................................................................171
10. The Thinking Cap ...........................................................................185
11. The Most Common Thought Process Mistakes................................224
12. The Basics of Time Management.....................................................253
13. Beyond the Exercise ........................................................................263
14. Additional Exercise and Lesson Tales ..............................................282

Appendix A: Dr. Max Euwe Protocol of ‘de Groot A’ ............................286

Appendix B: Computer Analysis of the de Groot Positions ....................290
Glossary ...............................................................................................297
Bibliography .........................................................................................308
About the Author ..................................................................................310
 Preface to the Second Edition
This edition is a major upgrade. Here’s an overview of the most significant
changes. We have:

• Added protocols to all the class (rating under 2000) chapters but signifi-
cantly boosted the “Expert and Above” chapter with numerous addition-
al protocols by players in the 2000-2500 range, for a total of almost one
hundred protocols;
• Added the chapter “The Most Common Thought Process Mistakes,”
which provides discussion and examples of these errors;
• Added a short chapter providing additional humorous and interesting
stories about giving lessons and using the de Groot exercise;
• Made corrections and numerous additions to the material from the first
edition;
• Expanded the Glossary to include more terms used within the book;
• Expanded Chapter 13, changed the title, and included a section about
finding positions suitable for giving de Groot exercises, with an addition-
al half-dozen sample positions; and
• Moved Chapter 9, “Thought Process Basics,” to after the protocols. It
was formerly Chapter 2.

The net result is a more extensive examination of the typical thinking pro-
cesses of players below grandmaster level, and the lessons that can be learned
from them. I believe that this work represents the most extensive compilation
of examples illustrating how players think at all levels from beginner up to
2500, and hope that the reader will find it entertaining, instructional, and
valuable from a historical viewpoint.

8
 Introduction

I enjoy performing chess research in the following two areas:

1) How do players learn and improve?

2) How do players think during a game?

These two questions may seem to be only remotely related, but they have
many aspects in common.

When someone is taught to play baseball, they don’t primarily learn about
innings, outs, and bases. Instead, they are taught how to bat, throw, catch,
and run the bases. But when players learn chess, the only things they are usu-
ally taught – even by competent beginner’s books – are the basic rules such
as checkmate and draws, and how to move the pieces. Then they are taught
more and more about what the pieces can do. A few principles like “Get all
your pieces into the game,” “At the start of the game try to control the cen-
ter,” “For your first move, push a pawn two squares in the center,” “Castle
your king early,” or “Knight on the rim your future is dim” are thrown in for
good measure.

But chess is a thinking game, and few beginners are taught how to think
to find their move. No wonder everyone learns their “chess thought pro-
cess” in a non-systematic way that quickly leads to bad thinking habits. A
thought process represents all the generic “steps” or sequences of logic that
go through the player’s mind during both his move and his opponent’s move.
A thought process does not include the content of what is being thought.

For example, systematically searching for threats after an opponent’s

move is part of a process, but deciding where to put a bishop in a particular
position is content.

Beginners are first taught basics such as how to differentiate the queen
and the king, how each piece moves, and that the players alternate making
moves. It is not effective to immediately thereafter attempt to teach a thought

9
 The Improving Thinker

process via suggestions such as, “The first thing you do is examine your op-
ponent’s move to see its effect on the position…” Beginners need time to
assimilate basic concepts before attempting to implement higher-level ideas.

However, it is also true that never to teach someone the basics of the think-
ing process can lead to bad habits and, consequently, to unnecessarily chaotic
mental habits. Some aspects of the thought process are common among good
chessplayers and can be taught relatively early in a player’s development.

I have explored this phenomenon in many places, notably online via my

columns Novice Nook at Chess Café and The Thinking Cap at IM Jeremy
Silman’s website; in blogs and articles for Chess.com; and with books such as
Everyone’s Second Chess Book, A Guide to Chess Improvement, and Looking
for Trouble.

The Improving Chess Thinker explores the results of “think out loud” ex-
ercises performed by subjects of differing levels of ability. These exercises
are based on those performed by the first serious researcher into the chess
thought process, Dr. Adriaan de Groot.

In the late 1930’s, Dr. de Groot, a Dutch professional psychologist and

chess master, recorded the thought processes of dozens of players at all levels
(mostly stronger players). His purpose was to investigate scientifically how
chessplayers really think. Players were given a position and asked to find a
move while at the same time verbalizing everything they were thinking. Each
subject’s verbalization was recorded and transcribed; these transcriptions
were called protocols.

Dr. de Groot analyzed dozens of protocols and published his results. His
publication was a doctoral-type thesis that was eventually translated into En-
glish as the book Thought and Choice in Chess (1965; reprinted in 2008 by the
Amsterdam Academic Archive). Thought and Choice in Chess is not so much
a “chess book” as it is a psychology book about chess.

With the 1938 AVRO tournament nearby, Dr. de Groot garnered some of
the best players of the day to participate in his experiment: Alexander Alekh-
ine, Reuben Fine, Dr. Max Euwe, Paul Keres, and right on down the line to
a few class players. Most of the protocols were by players of at least master
strength. Dr. de Groot wanted to find out how players arrived at a move in a

10
 The Improving Thinker

Appendix A contains Dr. Max Euwe’s protocol (recorded transcript) of

the “de Groot A” position, plus some observations. Appendix B contains
computer-aided analysis of each exercise position.

How to Use This Book

This book can be used in multiple ways:

• As a text supporting psychologists, demonstrating how chessplayers think

at various levels;
• As a “chess book” showing players how the thought process works in
practice. This includes aspects such as which approaches are effective
and which are not, how much time players use in determining a move,
how much effort is spent on various tasks such as identifying candidate
moves, and so on;
• As a manual for improving one’s chess thinking process. This can be ac-
complished by doing any or all of the following:

o The protocols in the book are for the six positions found in Chap-
ter 1. Before reading the rest of the book, take each of the six posi-
tions and carefully choose which move you would play, taking no
longer than you would in an important slow event. That way you can
compare the moves you chose and the thought process you used to
choose them with the players’ protocols throughout the book;
o Read the chapter (2 through 8) representing your rating class. Then
read the chapters one and two classes above your level to see what
knowledge, process, and logic is applied by superior players to arrive
at their move;
o Focus on the sections of the book which address the thought process
and its cousin, time management, found primarily in Chapters 9-13;
o Skim the protocols and focus on the comments after each and on
the summaries for each section. Much of the instructional value in
the book is contained in these comments;
o Learn from the general principles involving the thought process.
These principles are placed in italics outside the protocols.

14
 Chapter 1

The Exercise

The de Groot experiment allows a researcher to determine how chessplayers

find their moves during competitive play. This is clearly different from how a
player solves a puzzle. In a puzzle, the solution is guaranteed. Thus a player
can adopt the attitude that, “If this attempt does not solve the puzzle, then I
will try something else. The solution has to be there.”

In the de Groot exercise, the players (or subjects of the experiment) are
asked to find moves just as they would in a tournament game. During a game
there is no guarantee that there is anything good to find, such as a mate or
the win of material. The position may contain no clear ideas or candidate
moves that lead to winning or even drawn positions. In many practical po-
sitions there is no “best” move – there may be several almost equally good
alternatives.

Thus a researcher performing the de Groot exercise is interested in how

players find their moves when the goals are open-ended and a clearly best
move may or may not exist. For the chessplayer, performing the exercise and
learning from its results can have enormous practical benefits.

Since chess is a mental sport – a thinking game – the process used to

find a move is of importance not only in examining the source of the player’s
current strength, but also in determining his future possibilities for improve-
ment. If a player has a poor process for move selection, then his ability to
increase his playing strength is impaired, even if his other chess skills and
knowledge improve. An exercise that can diagnose a player’s process and ex-
pose him to a superior player’s process is a beneficial tool for both instruction
and psychological study.

In Thought and Choice in Chess, de Groot relied on one position for many
of his conclusions. He also included two other “primary” positions and a few

16
 The Exercise

supplemental ones as well. In giving this exercise to hundreds of my students,

friends, and acquaintances, I also utilized mostly de Groot’s first position,
which he labeled “de Groot A.” But I also occasionally used the other two
primary de Groot exercises (“de Groot B” and “de Groot C”) as well as a
few of my own.

The following six diagrams are the ones discussed in this book. The first
three are from de Groot’s book; “de Groot Shafritz” is from one of my
games; and the final two were from my students’ games.

If you have not looked ahead and attempted to play the positions, I sug-
gest you now set aside some time and decide which move you would play in
each of the following six positions, assuming a 40/2 (40 moves in 2 hours)
time control and sufficient time on your clock. If you want to make note of
how you came to your conclusions and the salient points behind the move
you selected, that can be additionally helpful.

de Groot A: White to play

w________w
[wdrdw4kd]
[0pdwgpdp]
[w1bdphpd]
[dwdnHwGw]
[wdw)wdwd]
[)wHQdwdw]
[B)wdw)P)]
[dw$wdRIw]
w--------w

17
 Chapter 1

de Groot B: Black to play

w________w
[rdwdwdkd]
[0wdwdpdp]
[wdbdwdpd]
[dwdwdwdw]
[wdwdwgPd]
[dwdPdwdw]
[P)Pdw)Pd]
[$wdw$wdK]
w--------w
de Groot C: Black to play
w________w
[rdbdw4kd]
[dp0wdw0p]
[wdq0phwd]
[0Ndwdwdw]
[PhP)wdwd]
[dPdwdN)B]
[wdwdQ)w)]
[dwIRdwdR]
w--------w
de Groot Shafritz: White to play
w________w
[rdbdn4kd]
[0pdndqgp]
[wdw0wdwd]
[)w0P0pdw]
[wdPdwdwd]
[dQHwdw)w]
[w)wGN)B)]
[dwdw$RIw]
w--------w
18
 The Exercise

de Groot Zyme: White to play

w________w
[rdbiwdw4]
[0wdwdpdw]
[wdwGw1w0]
[dpdwdw0w]
[wdpdQdnd]
[dw)wdwdw]
[wdwdP)P)]
[$wdwIBdR]
w--------w
de Groot Ernie: White to play
w________w
[rdbdw4wi]
[0p1pdw0w]
[wdwdwdw0]
[dwdQdBdw]
[w)Pdwhwd]
[dw)wdNdw]
[wdwdw)P)]
[$wdw$wIw]
w--------w
All of the above positions are at least somewhat analytical. Among the
six, the de Groot Shafritz is the most “quiet” but it still depends partly on
analysis. The most frequently used position, de Groot A, requires analysis
primarily. Thus the bias in this book is toward the analytical thought process.
While it is possible to do a de Groot exercise with a purely non-analytical
position, it is not as instructive. So I always choose a position requiring at
least some analysis.

The best ways for a student to improve his thinking process in non-ana-
lytical positions are the “traditional” best ways to improve chess judgment:
extensive experience in slow games, reviewing one’s games with stronger
players, reading annotated master games, learning principles and how to ap-

19
 Chapter 2

Class F and Below

This chapter includes players rated below 1000 USCF, or roughly below 1150
Internet Chess Club (ICC) standard.

This beginning level is composed primarily of youngsters or adults who

have just started playing. Their thought processes (especially youngsters’) are
usually way too fast, quite superficial, and rarely systematic, and they contain
little regard for safety.

For each protocol, I list the de Groot position (for example, de Groot A,
de Groot B, de Groot C, etc.; see Chapter 1 for those positions); the age of
the subject; the subject’s rating; and the time the subject spent to choose his
move.

Comments in parentheses indicate outside actions or comments that re-

flect the action of the player, such as (silent) if he paused for a long time.
Comments in brackets are my thoughts. My frequent use of [sic] means that
the subject is making a clear mistake in analysis or visualization. In contrast,
[!] indicates that the player has made a comment that is either very insightful
for their level of ability or a surprising error. A frequent note is [no eval],
meaning that the subject did not try to evaluate which side stood better, by
how much, and why.

Protocol F-1 (de Groot A; adult; 750; 13 minutes):

1.Eh6 [no eval]. 1.Cd7 no good. 1.Cxf7. 1.Gfe1 doesn’t help. 1.Eh6 to at-
tack rook. 1.Cxd5 he can just recapture and get his rook open. 1.Exd5: no, that
doesn’t work. 1.Cd7 attacks the rook but then 1…Cxd7 2.Exe7 Cxe7. The
knight on e5 is safe. 1.Ia5 does not make much sense [?]. Hmm. 1.Cd7 then
1…Cxd7 or 1…Exd7. 1.Cxg6 fxg6  2.Ih3 he takes my bishop. The knight
on f6 guards h7, so I have to remove that knight with 3.Exf6 Exf6. Doesn’t
work. 1.Ih3 with the idea of 2.Ih6. 1.Exf6 Exf6 no good. 1.Exf6 Exf6
2.Exd5 exd5 3.Cxd5. Push clock.

27
 Chapter 2

Unlike many beginners, Subject F-1 did take his time, which is good.
However, the density of his analysis per unit time is extremely low, indicating
he was spending a lot of time just trying to figure out what might be possible.
In the end, without verifying his analysis, he indicated a “principal varia-
tion” (his guess for best moves for both sides; PV) which involved trying to
remove the guard. In his PV, he forgot about the bishop on c6:

de Groot A
Black to play after PV 1.Exf6 Exf6 2.Exd5 exd5 3.Cxd5?
w________w
[wdrdw4kd]
[0pdwdpdp]
[w1bdwgpd]
[dwdNHwdw]
[wdw)wdwd]
[)wdQdwdw]
[w)wdw)P)]
[dw$wdRIw]
w--------w
White’s last move, the capture 3.Cxd5, is a visualization error since 3…
Exd5 would win the knight.

Anyone can make this mistake in analysis. However, if you feel that you
are winning material or giving checkmate, then this is a “red flag” indicating
you should be extra careful. Whenever there is a “red flag” in your analysis,
you should verify that analysis before committing yourself to the first move of a
critical sequence. If the sequence doesn’t work, it can either cost you the game or
just the opportunity to have played a much better move instead.

In addition, if the subject were trying to get in Cd7, then he is making the
common mistake of sacrificing a piece to win the exchange, and that is not
good, either. We will see this “give up a piece to win the exchange” mistake
quite frequently among the de Groot A protocols of weaker players.

Protocol F-2 (de Groot A; adult; 800; 4.5 minutes)

1.Eh6 [no eval] 1…Cf4 attacks the queen, so better is 1.Ih3. Would like

28
 Class F and Below

if you feel that you are winning material or giving checkmate, then this is a “red
flag” indicating you should be extra careful. Whenever there is a “red flag” in
your analysis, you should verify that analysis before committing yourself to the
first move of a critical sequence. If the sequence doesn’t work,
it can either cost you the game or just the opportunity to have played a much
better move instead.

the queen on h6. That leaves the pawn open on d4 for queen to capture so
1.Ih3 Ixd4 not so great. How can I protect the d-pawn and make an offen-
sive move at the same time? He could not do that [why?]. 1.Eh6 Cf4 threatens
queen then 2.Exf4. So 1.Eh6 is decent. Still leaves pawn on d4 open – no it
does not; Id3 stays. He has the bishop pair [sic]. So I could play 1.Cxc6 to
win the bishop pair immediately and he could capture with the queen or rook.
Let’s see – game is tied [finally counts material]. OK. Pick off bishop pair with
1.Cxc6. Push clock.

This protocol contains some typical “beginner” thought process errors.

The subject did not notice that Black was threatening 1...Ixb2 and did not
consider all capturing sequences. At least the reason for the move – to win
the bishop pair – is a lot better than those of many intermediate players!
Notice that the subject’s one big idea is to checkmate the black king on the
dark squares. Therefore, he focuses on getting the queen and the bishop
over there. Many weaker players go right for a checkmate pattern, no mat-
ter how hopeless or uncalled for by Steinitz’s Rules, which require a player
to have (or be able to generate) some advantage near the opponent’s king
before he can expect to successfully attack there. However, weaker play-
ers often get into bad habits of making otherwise counterproductive check-
mate threats because their weak opponents sometimes allow such check-
mates! At least in this case the subject properly saw that the checkmate at-
tempt was hopeless. When the idea failed, his fallback – the bishop pair –
was reasonable.

Protocol F-3 (de Groot A; age 11; 950; 2.5 minutes )

1.Eh6 [no eval] attacks the rook and then… or 1.Cxd5 exd5 2.Exd5 Exd5
no good. 1.Exf6 Exf6 something. 1.Gfe1 or something. 1.Ca4 to attack the
queen. What other piece to move? 1.Ce4 Cxe4 2.Ixe4 Exg5 not good. I think
that’s it [what’s “it”?]. I think that’s it. 1.Eh6. Push clock.

29
 Chapter 2

It’s good that this subject considered the capture on d5. However, he re-
jected the capture simply because it did not win material. In other words, he
treated the position as if it were a problem. This is a common error among
players rated below ~1600.

The subject did not consider all of his opponent’s recaptures after 1.Cxd5.
But he is correct that, if you find one reply which does not suit you (as he did,
although for the wrong reason), then you don’t have to consider the others.
That would be a waste of time.

As it turns out, the subject’s chosen move – 1.Eh6 – forces the rook to a
better position (d8). Therefore, unless there is a strong follow-up, this kind of
threat should usually be eliminated quickly as a candidate move.

Protocol F-4 (de Groot A; adult; 900; 13 minutes)

Piece Safety – protected vs. adequately protected. Checks, captures. Knight
takes pawn [?]. 1.Exf6 Exf6. 1.Ixg6+ takes the pawn. 1.Cxd5. 1.Ce4 to-
ward a knight fork. I had better make a move soon [7.5 minutes]. I can take the
bishop on c6 [does not mention it wins the bishop pair]. I could play 1.Eh6. If
I take the knight I am subject to a pawn fork [?]. Running out of things to think.
I haven’t come up with a good move. I am contemplating 1.Ce4. 1.Ce4 Cxe4
2.Ixe4 Cxc3 [sic] with discovered attack. I can’t find the helpful move. I will
play 1.Ce4.

The subject first looks at safety – good. Of course, counting the material
would help, too! Then he considers checks and captures, again good. Unfor-
tunately, he does not consider all captures and see how good each recapture
is. This subject only had a few random thoughts in the first seven minutes and
was, by far, the least verbal subject (per unit of time) in the book!

Protocol F-5 (de Groot A; adult; 850; several minutes)

1.Ixg6+ no good [no eval]. Captures: 1.Exd5, 1.Exf6 (lots of silence).
1.Gfe1 strengthens the center or 1.Gce1 then 2.Cg6 Gfe8. 1.Ce4 pressures
f6 – protected. If 1…Cxe4 2.Ixe4 Cxc3 [sic!].

30