Cheating in Online Chess (Part II): The Analysis of Engine Use – Chess.com

Hctor Laiz Ibez and Ral Snchez Garca are currently conducting a qualitative study on cheating in online chess, whose preliminary findings they describe in a two-part blog entry. This blog post is the second of the two parts.

Missed the first part? Read it here!

Chessable provided support to their research. University students and faculty research sponsors starting or continuing chess-themed research may apply before May 15 at https://www.chessable.com/research_awards for Chessable Research Awards.

To study the topic of cheating in online chess we conducted a qualitative study involving 24 Spanish male chess players, each with an Elo rating between 2100 and 2500. The players were divided into three different experimental groups: (A) human; (B) human playing autonomously with the help of a chess engine during the whole game; (C) human playing with the restricted help of a chess engine. See Part I for a more detailed methodological account of the experiment.

As part of the study, we sought to dissect the implications of engine assistance on the integrity of competitive play, focusing on the behaviours of two distinct groups under experimental conditions. This post will focus on the strategies and impressions arising from participants who had access (unlimited or restricted) to a chess engine: Group B and Group C.

Group Bs engagement with the chess engine was defined by an unrestricted access policy, allowing players to utilize technological assistance throughout their games. Participants were informed that their objective should not necessarily be to disguise the use of engines, but rather to secure the win. Nevertheless, the interaction with the engine varied among players, revealing a diverse range of strategies from heavy reliance to more selective use.

The engines influence was described as addictive by participants, who noted that it significantly boosted their confidence. However, this reliance also posed its own set of challenges, particularly affecting players ability to engage in independent strategic thinking and making it difficult to delve into any precise or sharp line calculations. Regarding the technical setup for engine assistance, Group B players typically played their moves on a phone and checked the engine on a PC/laptop, although a minority swapped programs on the same PC/laptop.

Concerning expectations of opponent play, most players initially assumed that their adversary could, as they themselves, likely cheat or have access to a chess engine. However, gaining an advantage or observing an opponents mistake led them to conclude that (a) the opponent was playing independently; or (b) if they had access to an engine, it was in a more limited fashion compared to theirs. All Group B players won and dominated their games, except for one draw.

For Group C, experimental conditions restricted engine assistance to three specific consultations per game, provided the player had more than 2 minutes on the clock. This limited access to Stockfish 15 capabilities introduced a strategic element to engine use, requiring players to judiciously decide when to seek help based on the games critical moments. This assistance included the best engine move (only one move, not the entire line) and the position evaluation.

Group C participants focused on leveraging these limited opportunities to gain a competitive edge, primarily using consultations in sharp positions where the correct move could significantly alter the games course. Trust in the engines suggestions was absolute. In one case, a player, despite mishearing a move (misinterpreting f for e) and verbally expressing concern about its suitability, played it anyway. Other players followed moves that, in post-game interviews, they admitted were contrary to their playstyle and that they would never choose in a real game.

Most Group C players believed their opponents might also employ the 3-wild cards or other types of computer assistance, given the lack of information about the conditions under which the other player was playing. This occasionally led to players hoarding wildcards, relying as much as possible on their own analysis, and seeking help when they felt their opponent was getting a favourable position.

Engine evaluations were generally considered more useful than the specific moves, which were described as confusing without the follow-up moves to justify them as the best choice. This sometimes caused nervousness among players. The overall impression was that this type of assistance, although helpful, did not create a significant imbalance or provide a sizable advantage under the experiments conditions.

Nevertheless, all participants agreed that, given this edge and the opportunity to use it in a larger number of games (200-300), they would greatly improve their use of wildcards, even adapting their game to take full advantage of this type of assistance, to maximize its impact.

These preliminary findings offer a glimpse into the dynamics of engine use in online chess, providing valuable perspectives on how players navigate the challenges and opportunities presented by technology. By examining the strategies and motivations behind engine assistance, the research contributes to a broader understanding of cheating in chess, highlighting the need for ongoing dialogue and action to ensure fair play and maintain the games integrity.

Hctor is an honorary fellow of the Department of Business Management and Economics at the University of Len. His research focuses on the digital economy and emerging technologies. He also works full-time at the Spanish National Cybersecurity Institute (INCIBE), dealing mainly with matters related to international relations and EU initiatives. He is a FIDE Master and plays for Club de Xadrez Fontecarmoa. Email: hlaii@unileon.es

Ral is a lecturer on motor learning and the theory of play at the Sports Science school of the Polytechnic University of Madrid. He is also closely connected to the Embodied Design Research Laboratory (EDRL) of the University of California, Berkeley. His research blends social and cognitive sciences to study skill acquisition from an embodied perspective. His interest in chess deals with the question of distributed cognition and distributed agency between humans and computers. Email: raul.sanchezg@upm.es

Originally posted here:

Cheating in Online Chess (Part II): The Analysis of Engine Use - Chess.com