Online Chess and Working from Home

by Stefan Löffler
12/4/2020 – Can the performance of world class players at online tournaments be used as an indicator for the quality of work done from home? Three economists, among them German IM Dr. Christian Seel, do think so. A report by Stefan Löffler highlights their sobering conclusions. | Image: Christian Seel (private)

ChessBase 16 - Mega package Edition 2021 ChessBase 16 - Mega package Edition 2021

Your key to fresh ideas, precise analyses and targeted training!
Everyone uses ChessBase, from the World Champion to the amateur next door. It is the program of choice for anyone who loves the game and wants to know more about it. Start your personal success story with ChessBase and enjoy the game even more.


Bad marks for work from home

When the pandemic struck in spring, Christian Seel began playing online more frequently. He soon got the impression that he was making more mistakes while playing on a screen than he did on a real board, and wondered whether other players felt the same. When the Magnus Carlsen Chess Tour began in April, Seel followed the games of the world's best players – and knew he was onto something.

Among chess players Seel is best known from the Bundesliga, where he plays on board one for the SK Aachen. However, Seel is also a professor of microeconomics at Maastricht University. He is not the only chess enthusiast in his faculty. There is also employment market researcher Stefan Künn, who just recently proved that a higher concentration of particulate matter in the air increases the probability of errors during games of chess.

In call centers or banks, it is expensive or even impossible to reliably assess the quality of services. Chess games, on the other hand, can be analysed very efficiently with engines.

Dainis Zegners of the Rotterdam School of Management, a colleague and fellow chess enthusiast of Seel and Künn, was involved in their research. At the time, Zegners was  working on  a different study, which also used chess data. Said study relates to the notion that work is becoming increasingly more challenging from a cognitive standpoint. How does our cognitive performance develop over the course of our lives, and how do different generations compare in this regard? Error quotas in chess games are easy to evaluate, which is why Zegner used them to answer this question.

The three researchers did the same for the games played at the Magnus Carlsen Invitational. As a benchmark, they used games played by the same players at the Rapid World Championships 2015 and 2019, as all these tournaments were played with the same time limit: 15 minutes for the whole and with an additional 10 seconds for each move.

Another important factor from an economic point of view were the incentives offered to the players in the form of prize money, which were at least on a comparable level. This meant that there were a grand total of 27,000 moves to be analysed. This was done with Stockfish, at a search depth of 25 plies.

Seel's personal observation was confirmed. Online, the top players blundered more often than in live games. This held true for every single player for whom data had been available: Magnus Carlsen, Ding Liren, Anish Giri, Alireza Firouzja, Hikaru Nakamura, Ian Nepomniachtchi and Maxime Vachier-Lagrave.

At least Carlsen seems to be sharing this sentiment. Although he has won amost every single online tournament up to this point, he has also repeatedly criticised the quality of his own performance. For one particular tournament, his second Peter Heine Nielsen even rented a holiday lodge to get Carlsen out of "home office mode" and into "tournament-mode".

According to Seel, it is of course plausible to assume that players first need to get used to the new situation. The three authors are considering a follow-up study on whether performance is going to improve over time, and if so, to what extent. However, Künn, Seel and Zegners first wrote a paper on cognitive performance during work from home. Considering the increased popularity of this approach during the pandemic, their work understandably managed to strike a chord. Their findings were picked up numerous times by the Dutch press.

By the way: Next Sunday, December 6, Christian Seel and Dainis Zegners will be discussing their chess research at the online conference ChessTech 2020. Their session will be preceded by a joint introductory lecture by Fernand Gobet and Andrea Brancaccio, titled "Using chess databases to answer psychological questions: A survey".


Translation from German: Hugo B. Janz

Stefan Löffler, a journalist and International Master based in Vienna and Lisbon, is member of FIDE's Education Commission, a consultant at ChessPlus Ltd. and Programme Director of the London Chess Conference:
Discussion and Feedback Join the public discussion or submit your feedback to the editors


Rules for reader comments


Not registered yet? Register

Tabamsey Tabamsey 12/6/2020 07:37
Thank you for the comments. A few clarifications below:
@sligunner: You are correct. There are at least three differences: 1) home environment vs. tournament hall 2) no physical player on the other side of the board and 3) 3D vs 2D task. It is not possible to distinguish between them empirically within the study, but this would be interesting to scrutinize further.
@fgk..: The analysis indeed controls for Elo and Elo opponent, complexity of the position, number of games on the given day.... To furthen weaken the first concern, every player in the tournament was below 35, so if anything they should improve. Risk-taking due to tournament situation is not controlled for. Any take on how you would control for this (discard certain games near the end)? Psychological stress due to COVID is discussed as a potential confound and argued against, but cannot be fully ruled out.
@cyronix: Clearly depends on the task. The study just talks about purely cognitive tasks. For call-centers another study shows that working from home does not yield lower productivity, but there the task is not entirely cognitive and also noise levels are arguably lower at home. The esports reference is interesting.
@Schurick: Indeed, the study just talks about cognitive tasks and many jobs do not consist mainly of cognitive tasks. For a lot of jobs which involve cognitive tasks, it is hard to get a good assessment method (How good is a contract that someone set up? How good is a document on restructuring?) so this is a first step in that direction.
Schurick Schurick 12/5/2020 07:20
Due to the nature of chess - it is not comparable with a typical office work. If an office worker crunches numbers and sends emails at work and at home - she/he might not notice such a big difference (except for lacking communication to colleaguea and bosses). A chess player works "over-the-board". An assembly line worker suddenly working from home on a 2-D visualization of a 3-D process might be closer to a chess player, than an office worker. But again, we can't compare chess to work. No verbal commincation during real chess games. High importance of non-verbal behaviour (reading and sending). At office - both verbal and non-verbal are important.
cyronix cyronix 12/5/2020 01:19
I dunno, I think you can pretty well concentrate in a "tournament" setting in a fullinformation game with a motivating aim (i.e. beat other player) ... a lot of esports show this ... home office work is different, its not a single day, its all the time, and there is no motivator like beat other player, its much much more complex, there is no direct motivator, and also relies on communication ... in home office you only see people a bit on microsoft teams and often only audio only ... sitting all day in the same room unobserved is not the same as going some official place ... the study is comparing apples to oranges
fgkdjlkag fgkdjlkag 12/4/2020 08:39
The methodology is flawed. Because chess favors younger players, you cannot look at the SAME players in 2 tournaments that are 4 years apart, as overall you should expect the performance to be lower, depending on the specific ages of the players (eg, Firouja might be expected to have improved). Also one has the players' ratings, so at least that could have been used to adjust the analysis. If Nakamura, eg, is rated lower now than he was 4 years ago, then of course his performance will be lower now.

A second major factor is the tournament format, which includes the number of games. Since one needed to qualify for the final of the Magnus Carlsen Invitational, there are circumstances in which players will take more risks, and therefore have a higher number of blunders. Such considerations need to be taken for the World Rapid as well. Number of games is very important due to fatigue - if the Magnus Carlsen invitational had more games per day than the World Rapid, it will negatively impact performance.

A final factor is the psychological stress induced by covid-19.

If I were doing such a study, I would have selected games for analysis based on rating, not player, and then controlled for number of games per day and tournament factors.
Johannes Fischer Johannes Fischer 12/4/2020 03:33
@Somewhat Experienced
Thanks for the hint. The title was changed.
Somewhat Experienced Somewhat Experienced 12/4/2020 02:36
Dear Stefan,
"Home Office" is Ginglish - in England "Home Office" is the term for the Innenminsterium, and people either "work from home" or use telework...
sligunner sligunner 12/4/2020 12:24
So, did they blunder more often because,
a) Playing in a 'home' environment was not conducive to high levels of concentration, or
b) It is harder to visualize the moves on a two-dimensional computer screen than on a three-dimensional board?
The article doesn't answer this, but I would say from my own experience (FIDE rated 1800 club player) that it is harder the analyse a 2D computer screen than a board.
What do the top players say?
This article could have done with more investigation. Maybe Chessbase needs a journalist to write for them (just contact me and I'll let you know my availability!).