Innovation is driven disproportionately by a small slice of people. The top thinkers and builders, the 1% of any field, solve the problems the rest of us benefit from.
I believe talent, how they’re found, created, evaluated, and placed is arguably the most interesting hard problem there is. It’s the machine behind the machine. And right now, it’s badly built. The current process is to widen the funnel, collect more applicants, and build better filtering tech. This leads to a longer time-to-hire without a proportional improvement in hire fit.
Hyper-growth, tech companies have figured out the solution. These companies behave like VCs: roughly 80% of output comes from the top 20% of employees. They accept high churn because a single outlier can dominate the results, and they reward outliers disproportionately.
Current State
Inbound is filled with fraud. Gartner predicts that by 2028, one in four candidate profiles worldwide will be fake. Fraud runs in both directions, fake recruiters also target candidates. To be fair, fraud-detection tools are getting better, but that’s a false reality. As resistance improves, fraud attempts improve. Parts of the hiring stack get better, but historically there’s always more being added, and time-to-hire keeps increasing.
This slows overall economic productivity, and it deters top-tier talent who value speed. That last part matters most.
Top-Tier Talent
Not only is 80% of company output coming from 20% of staff, but the gap between low tier and top tier talent is widening. Productivity per person is increasing disproportionately with top tier talent being able to do more (btw by tier I don’t mean seniority). As a result, demand for top-tier talent is rising and claiming a growing share of the labor market. The natural response is a shift toward outbound. The really good people are generally busy doing something else, so you have to go find them rather than expect them to find you.
Right now there isn’t enough top-tier talent to go around, so competition is fierce. Competition isn’t sustainable long term which means new ways of finding untapped high potential talent is crucial. And the current methods suck.
Resumes, GitHub profiles, and LinkedIn are weak signals. They tell you almost nothing about ability, initiative, or culture fit. AI tools aren’t a complete fix, using past hiring decisions puts biased human judgment back into the system. You’ll notice that wherever you found your last untapped talent its never where you find the next, those places get patched once discovered. The traits that matter most are qualitative, hard to fake in person, and largely invisible to any online system.
There’s also fewer agencies and more in-house recruiters now. The best companies don’t want to pay high fees they want to build their own self-selection systems, because that’s how they find untapped, high-potential talent.
Self-Selection Systems
Self-selection systems work because they are open-ended, provocative, novel, and mission-driven, all things that attract high-potential talent.
A few examples:
Netflix’s Culture Deck was a viral 2009 manifesto that replaced HR rules with radical autonomy, expecting employees to act like a “pro sports team” where high performance is mandatory and underperformers are quickly replaced.
HRT’s 2015 intern class quit traditional recruiting entirely, sourcing strictly from the global Math and Computer Science Olympiad circuits and attracting only elite competitive programmers through an extremely high technical bar. The 10-person cohort built low-latency systems, worked on hands-on projects, and formed a tight-knit peer group. They went on to become a tech “mafia,” launching multi-billion-dollar companies like Scale AI, Cognition AI, and Hyperliquid.
SpaceX’s early recruitment was essentially a public warning label: insanely long hours, harsh factory conditions, low pay relative to the rest of tech, and real risk of total failure, all justified by the mission of making humanity multi-planetary.
Counterintuitively, these systems are hard to build. They require lots of upfront design time and don’t produce volume quickly. But if built properly, time-to-hire becomes instant, the moment the right person notices theres no friction to join.
Open-ended: There’s no application roadmap the system is defined by how you interpret it. If you interpret it as “I’m going to show up and prove I’m the right fit,” it worked. If you interpret it as “this is too hard for me,” it also worked. These systems let character do the heavy lifting.
Provocative: They spread a message or call to action that breaks from the norm, sometimes stating outright that the company fires quickly and doesn’t treat people like family. Balancing attention with intention is key in this case.
Novel: These systems work best when they’re brand new. A first cohort, a new role, a new layer of the company. New usually reads as, underdeveloped, risky and low-ROI, but people who see the upside, and aren’t chasing the superficial prestige of an established reputation, are the ones it attracts.
Mission-driven: You have to actually believe in the mission.
The Future
Automation will manage high-volume tasks like parsing and fraud detection, but human judgment and taste remains essential for spotting outliers. The new focus of candidates is past work quality, how someone thinks, and team fit. All this information lives in human memory and references, not in public datasets .. yet. As AI speeds up company processes, talent becomes the ultimate bottleneck making it an exciting time to build in the space.