Being Wrong in Public
Newton, Popper, the AWS Correction of Errors process, and how mistakes made in the open become the shoulders others stand on
Being Wrong in Public
Newton, Popper, the AWS Correction of Errors process, and how mistakes made in the open become the shoulders others stand on
The Milken Center for Advancing the American Dream, Washington DC
In a 1675 letter to Robert Hooke, Isaac Newton wrote his most famous line: if he had seen further, it was by standing on the shoulders of giants. Generations have read it as a model of humility, a great man crediting those who came before.
The reading is probably wrong. Newton and Hooke were in a long and bitter feud over who first understood the inverse square law, and the letter was a move in that fight. Hooke was slight and stooped, afflicted since youth with a severe curvature of the spine. When Newton praised his rival for standing on the shoulders of giants, several historians read a buried insult: you, specifically, are no giant. If they are right, the best-known tribute to intellectual inheritance began as an insult.
The phrase was not Newton’s either. Bernard of Chartres taught it in the twelfth century, describing scholars as dwarfs perched on the shoulders of giants. The line runs through John of Salisbury in 1159, Robert Burton in 1621, and George Herbert in 1651 before Newton used it, five centuries after Bernard.
Strip away the feud, and the sentence still states something true. Newton could see further because the work of his predecessors was available to him: their books, their results, and above all their mistakes, already made and corrected so he could skip them. The shoulders are built from errors someone was willing to commit in the open.
The method is institutionalized error
Science gets described as a system for producing truth, and it works more like a system for producing wrongness efficiently. A hypothesis is a guess exposed on purpose, framed so that reality can knock it down, and most are. A guess that survives a serious attempt to kill it is worth more than a hundred that were never tested.
Karl Popper built a philosophy of science on exactly this asymmetry. No number of observations can prove a universal claim true, since the next observation might contradict it, but a single observation can prove it false. So the mark of a scientific claim, Popper argued, is not that it can be confirmed but that it can be refuted: it has to stick its neck out, predict something that could fail, and risk being wrong. A theory compatible with every possible observation tells you nothing, because it forbids nothing. Astrology and a rigorous physics differ less in confidence than in exposure: physics risks refutation on every prediction, while astrology cannot be caught.
Popper was careful in a way his popularizers often are not. He did not think a theory should be thrown out at the first contrary result, since a good theory defended stubbornly reveals possibilities a timid one never would, and he allowed a limited, deliberate dogmatism in its defense. The willingness to be wrong asks something more specific than a reflex to abandon ideas: expose them to the test, and accept the verdict once the test is fair and repeated.
Wrong once, so the next person is right sooner
Put the two ideas together, and the shoulders stop being a metaphor about reverence and become a description of how knowledge compounds. Each generation inherits not just the successes of the last but its failures, cataloged and corrected. The failures are the more valuable inheritance, because a documented dead end is a place no one has to walk into again.
This only works if the wrongness is public. A mistake made in the open, written down and checked by others, becomes part of the shoulders. The same mistake made in private, hidden because it was embarrassing, stays a private pit that the next person falls into unwarned. A giant’s shoulder and a hidden pit both start as errors; they diverge on whether the person who made the error was willing to be seen making it.
The blunders of famous scientists illustrate the point precisely because they were public. Kelvin calculated the age of the Earth at tens of millions of years, far too low, because radioactivity had not yet been discovered; the gap between his figure and the geological evidence was itself a signpost that something was missing, and it pointed the way to the discovery that corrected him. Being wrong in the open, by a specific and measurable amount, let the next person find the shoulder.
Inside an organization
An organization decides under uncertainty all day, which means it is wrong at scale, every day. The only open question is where the wrongness surfaces: early and in public, where it becomes a shoulder, or late and in private, where it becomes a pit.
Most organizations are built, without meaning to be, to hide their errors. A status report with no problems in it is rewarded over an honest one that names three. A project stays green until the week it turns red. A decision that isn’t working keeps its funding because the person who made it has more to lose from admitting the mistake than from letting it run. Each of these quietly turns a correctable early error into an expensive late one, and each is driven by the same fact: being seen to be wrong carries a penalty, so people arrange not to be seen.
The costly failures documented in engineering and business almost never come from a single bad decision. They come from a bad decision that no one was willing to call wrong while it was still cheap to reverse. The information existed. Someone knew. The organization had made it more dangerous to report the error than to let it compound, and so the error compounded, in private, until it was too large to hide.
Slogans about embracing failure usually celebrate it after the fact while punishing it in the moment. The structural correction is the scientific method pointed inward. Make hypotheses explicit, so a decision states what it expects to happen and can therefore be shown wrong. Shorten the loop between a decision and the evidence on it, so wrongness surfaces while it is still small. Reward the person who surfaces an error over the person who hid a bigger one. And separate the admission of a mistake from punishment, because an organization that punishes honest wrongness is training its people to produce hidden pits.
A worked example: Correction of Errors
Amazon Web Services runs the scientific method as a formal operations process, and it publishes the mechanics. You can read about it here, right in the AWS Well-Architected documentation. After any significant incident, even one no customer noticed, AWS teams write a Correction of Errors document, which its own documentation describes as a closed-loop mechanism for post-incident analysis. The operating premise is stated plainly in the AWS material: everything fails, all the time. The COE assumes the error rather than treating it as an aberration, and its focus is on preventing that specific event from recurring in the workload where it happened.
Two features make it a working instance of the argument rather than a corporate ritual. The first is that the document is a deliberate act of being wrong in writing. It records what went wrong, why existing tests did not catch it, and what will change, and AWS’s own guidance names the failure modes to avoid, chief among them assigning blame rather than finding the cause, which it warns creates a culture of fear and shuts down the open communication the process depends on. The point of writing the error down is defeated if doing so is punished.
The second is that the learning does not stay local. AWS built a second process, the Operational Readiness Review, precisely to carry lessons from one team’s COE into a checklist every other team runs before launch, so a failure discovered once becomes a question every other team has to answer before they can repeat it. One team’s documented mistake becomes a step the next team does not have to fall into, on the explicit reasoning, in AWS’s words, that failures are a given and the work is learning from them.
What it costs and what it buys
Being wrong in public has a real price, and pretending otherwise is why the advice usually fails. It means exposing a belief before it is safe, accepting that a prediction on record may fail on record, and giving up the comfort of a claim so hedged it can never be caught. Newton’s giants paid it, and so did Kelvin. Every published result that a later result overturned was a bet its author could lose in front of the field, and many did.
What the price buys is the only kind of progress that accumulates. A field, a team, or a person that makes its errors in the open builds shoulders others can stand on, including its own future self. One that hides them builds nothing, and repeats the same falls in the dark. Newton saw further because five centuries of corrected mistakes sat underneath him. The only way to add to that stack is to be wrong where the next person can see it.
By Joshua McDonald on August 11, 2026.
Exported from Medium on August 26, 2026.
Reader discussion