The strongest evidence on adult gaming and wellbeing is Vuorre, Johannes, Magnusson and Przybylski’s 2022 study in Royal Society Open Science, which tracked 38,935 active players across seven games (Animal Crossing: New Horizons, Apex Legends, Eve Online, Forza Horizon 4, Gran Turismo Sport, Outriders, and The Crew 2) over six weeks, using industry-cooperation telemetry to link verified play time to player-reported wellbeing. A one-hour daily increase in play produced a change of −0.04 [−0.15, 0.09] on an 11-point life-satisfaction scale and 0.03 [−0.09, 0.16] on a 13-point affect scale — effects the authors describe as “limited, if any.” This is the cleanest causal-inference design available for the question, and it points to a null average effect of gaming time on wellbeing. Halbrook, O’Donnell and Msetfi’s 2019 review in Perspectives on Psychological Science concludes that gaming’s effects are moderated by motivation, social context, presence of violence, and physical activity, with most recreational use sitting in the beneficial zone. The harm distribution is highly concentrated: roughly 3-4% of players meet ICD-11 or DSM-5 gaming-disorder criteria, and within that subset, action-side regret is substantially higher. The 10% headline estimate adds a buffer to the disordered-gaming prevalence for sub-clinical retrospective regret of time use.
The inaction side is harder to characterise. No survey directly asks non-gamers whether they regret not having taken up gaming as an adult hobby. The Entertainment Software Association’s 2023 Essential Facts report estimates 65% of Americans (about 212.6 million weekly players) game; Pew’s 2024 survey of 1,453 US teens finds 85% of adolescents play, with 89% of teen players gaming with others in person or online and roughly three-quarters (72%) citing time spent with others as a reason to play — gaming functions as a primary social-connection medium for the cohort now entering adulthood. Adults who do not game forgo a recreational and social-inclusion modality rather than the activity itself; the magnitude of regret for that foregone modality depends on whether the non-gamer’s social network is gaming-dense or gaming-sparse. The 14% inaction-side estimate is constructed from the adoption gap and adjacent aspirational-interest signals. Roese and Summerville’s 2005 meta-analysis places leisure as the smallest of six top regret domains, smaller than education, career, romance, parenting, or self — which constrains both sides of this entry to low absolute magnitudes.
This is the rare regret pair where both sides plausibly carry low long-term regret. The estimated four-point delta is within the uncertainty of both proxy constructions and should be read as “approximately equal long-term regret with a slight inaction lean” rather than a precise gap. Three caveats merit emphasis. First, harm is concentrated: the 3-4% of players who develop disordered gaming carry costs (sleep loss, occupational interference, withdrawal) that the population-average headline obscures, and within that subset action-side regret is substantially higher. Second, generational cohort effects matter: gaming has been normative recreational behaviour for two decades, and inaction regret may rise in cohorts where non-gaming meant social exclusion. Third, genre and motivation moderate the answer substantially — competitive multiplayer with toxic chat differs in regret profile from cooperative or single-player experiences, and the Vuorre study averages across genres. Industry-cooperation studies carry mild publication-incentive concerns toward null or positive findings, partly mitigated by convergence with the independent Halbrook review. The direction — neither side dominates — is consistent across the strongest available evidence; the headline magnitudes are proxy constructions that should be held loosely.