Questions for you:
- What processes in your work do you treat as effectively random because their bias is too small to matter, and have you actually tested whether that is true at the scale you operate?
- Where does your organisation use apparently fair random processes — lottery selection, sampling, rotation — without checking whether those processes have systematic biases built in?
- Is there an equivalent of “call the side you can see” in your field — a tiny exploitable edge that most people ignore because it seems beneath serious attention?
Organisational applications:
The difference between approximately random and actually random: The coin toss story makes a specific and practically relevant point: a process can appear perfectly random, be widely accepted as perfectly random, and still have a measurable systematic bias. The Bartoš finding required over 300,000 tosses to detect a 51:49 split — at that scale, the bias is real and statistically significant, even though in any individual toss it is essentially invisible.
Many organisational processes that are described as random – sampling procedures, allocation methods, and selection sequences – have analogous small biases that only become significant at scale. The relevant question is not whether a process is approximately fair in individual instances but whether any systematic bias, however small, matters given the volume of decisions it is applied to.
Small edges at scale: the Brentford principle: The story’s observation about Brentford FC — that calling the side facing up gives a 1% edge, and that this might be worth attending to if you are seeking every available advantage — generalises to organisational contexts.
A pricing strategy that is fractionally more accurate than competitors’, a hiring process that slightly better predicts performance, an audit sampling method that catches marginally more fraud: each of these looks trivial in any individual instance but compounds into a meaningful advantage at sufficient scale. Organisations that are rigorous about identifying and protecting small systematic edges in high-volume, competitive processes tend to outperform those that focus attention only on large, obvious advantages.
Hidden structure in apparently random processes: The deeper point the story makes is that even the archetype of randomness — a fair coin toss — has detectable structure when examined carefully enough. This should prompt scepticism about any process described as random in organisational contexts.
Random sampling from a list depends on the order in which the list was sorted. Random workload allocation depends on how tasks were sequenced in the queue. Random selection of audit targets depends on the scope of the population from which they are drawn. None of these processes is perfectly random in the Bartoš sense; they all have structure. The useful discipline is not to assume randomness but to ask where the structure is and whether it matters.
Further reading
The original research:
Fair Coins Tend to Land on the Same Side They Started: Evidence from 350,757 Flips by František Bartoš et al. (2023). The preprint, subsequently published in the Journal of the American Statistical Association, reports the actual figure as 50.8% rather than the rounded 51% cited in the story.
Worth noting: the study also found considerable variation between individual tossers — some showed a same-side bias as high as 60%, others almost none. The bias appears to be related to how “wobbly” the individual’s tossing technique is, which means a determined practitioner could, in principle, increase their edge above the average.
On the physics and mathematics of coin tosses:
Struck by Lightning: The Curious World of Probabilities by Jeffrey Rosenthal. Rosenthal covers coin-toss physics and related probability puzzles with the same accessible style as in his walk-or-ride dilemma, providing the broader probabilistic context for the Bartoš finding.
Against the Gods: The Remarkable Story of Risk by Peter L. Bernstein. Bernstein’s history of probability covers the development of the fair coin toss as a theoretical ideal and the gap between the ideal and physical reality.
On small edges, marginal gains, and systematic advantage:
The Success Equation: Untangling Skill and Luck in Business, Sports, and Investing by Michael Mauboussin. Mauboussin’s framework for identifying exploitable edges in high-variance domains is directly relevant to the Brentford observation: the question is always whether a small systematic advantage is real and whether you are operating at sufficient scale for it to matter.
Moneyball: The Art of Winning an Unfair Game by Michael Lewis. The Brentford approach is a direct descendant of the Moneyball philosophy: systematic attention to small, measurable edges that are ignored by competitors who are not looking carefully enough.
On bias in apparently random processes:
How to Lie with Statistics by Darrell Huff. Huff’s account of how sampling and selection processes introduce systematic biases that are invisible in individual cases but detectable in aggregate is the classic treatment of the broader principle the coin toss story illustrates.
Noise: A Flaw in Human Judgement by Daniel Kahneman, Olivier Sibony and Cass R. Sunstein. The chapters on sampling and measurement cover how apparently neutral selection processes carry hidden biases, with practical guidance on identifying and correcting them.
About the image
My son Milo, about to flip a coin.
Photo montage and photo by Matt Ballantine, 2026
Random the Book 