On finding the exceptional in the unranked majority.
There is a reality that no one wants to look at directly. In the world of programming, decisive results are not produced by average "excellence." They are produced by a tiny minority of people who sit at the far right tail of the distribution. Those who have watched the field long enough know this. Yet saying it openly ruffles feathers, so it is usually wrapped in vague language.
From my own personal experience, roughly the top 10% of the most elite institutions can, with fairly high probability, be considered "confirmed" as people who can at least step onto the ring of high-level intellectual competition. Of course, not all of them will become exceptional programmers. But in terms of the foundational capacity for reasoning and comprehension, they rarely fall far short.
The problem lies in the next tier down. People with credentials that the world calls "excellent" — those who narrowly missed the most elite institutions, yet are still considered more than capable by society's standards. This group is genuinely talented. But the cruel truth is that, as the result of long rounds of selection, they have largely converged into a tier that is "highly capable, but probably not 3σ or 4σ material." That is how the odds have settled.
This is not a denial of their ability — quite the opposite. They are critically important people for most organizations and form the backbone of practical operations. However, the probability that someone at the extreme right tail of the distribution will emerge from this tier is, statistically speaking, already quite low.
Turn, then, to the vast majority of society — the undecorated, ordinary public, so to speak. Among them lies an entirely different structure. Because they exist outside the formal systems of selection, their abilities have not yet been confirmed as either high or low. The average is by no means exceptional, but within that enormous population, there are certainly individuals of 3σ or even 4σ capability who have been quietly overlooked — never picked up by the system.
In other words, the layer where genuine potential still remains is precisely this unconfirmed group. People with lightning speed of thought, an intuition for grasping structure, and an uncanny power of concentration exist there, in silence.
The challenge is how to find them. Most organizations take comfort in selecting people with polished résumés. As a result, they gather those who are competently excellent — but rarely encounter someone with the decisive, breakthrough kind of force.
The real challenge for a technical organization is not to collect people whose value has already been assessed. It is to identify the buried 3σ and 4σ individuals within a vast population that has not yet been evaluated at all.
As we enter the age of AI, this structure becomes even more pronounced. AI is not a device that averages out ability. It is a tool that applies ferocious leverage to the power of thought and the ability to frame problems. Those who were already exceptionally capable will therefore reach even further, and the gap with the average will widen — not shrink.
What will matter going forward is not how many people you have gathered. It is who you put on the ring — and whether you can find the talent whose name is not yet known to anyone. The intelligence of a technical organization, in the end, reveals itself in exactly that one point.