Skulls as nodes

I wanted to pick up on my earlier piece about Fidelity of Comprehension and address the scale problem any SME should hope to have one day: how do you grow without destroying the very quality and productivity that got you there?

The default response is usually: more systems, more processes, more documentation. All of those are essential tools. But you don't lead with them. You lead with key people and clear roles and responsibilities.

To scale, every role does of course need to be distilled and parameterised to a degree:

• This is the outcome expected.
• This is how we measure it.
• This is where your responsibility ends and how escalation works beyond that.

That last point is crucial. Without a clear path of escalation, you end up in the classic failure mode: "Not my problem. That's outside my process."

I've seen process-driven scaling work at serious size first hand — north of 25,000 employees at a major US tech giant. The processes worked. But here's what many leaders miss: that company also had an average retention rate of seven years. That's not a trivia fact. It's the whole point.

To scale successfully — rather than boom, bust, and burn people out — you have to take your people with you. The tech-bro fantasy of "hire fast, fire fast" is fatal over any meaningful time horizon.

Tacit knowledge is the business

Process, boundaries, escalation, documentation — that's hygiene. That's blatantly obvious. That's like saying, "Our people have laptops and chairs." Correct, well done — have a cookie. But it's completely missing the root issue: retention.

Your company is a bunch of skulls on legs, running on coffee and fructose corn syrup. For every document those people produce, what they truly know and need to know to be effective is far broader than anything that ever gets written down. The gap between written process and lived reality is enormous, and that's before you account for the fact that, historically, >99% of information is lost to time.

People cannot write down every little thing they know and do without stopping themselves from actually doing the thing they're writing about.

Can AI help here? Maybe — help being the operative word. Some knowledge retention is possible through embeddings, RAG, and LLM-assisted search. But there are limits. And if you're not careful, you simply create a massive black box of dependency on non-intelligent "AI" systems that you neither fully control nor fully understand.

Don't delete the node

When you think about large systems, projects, products, or companies, you must think about the key people who got you to the point where scale is even an option. To believe you can continue at the same momentum with a completely different set of people, purely because you've "documented everything," is naïve and fatal.

Identify your key personalities. Invest in them. Understand that your people are your business.

If we're going to say "your systems are your business," then think one layer deeper: who built those systems? Could they really be reproduced from documentation alone? The answer, obviously, is no.

So when you're staring at a balance sheet, deciding which poor life gets to be uprooted in the name of "reducing overheads," remember: you're not just losing the specific function that person performs. You're losing the tacit knowledge they carry — the understanding they need to execute their tasks and, in turn, enable everyone else connected to them. You're not just cutting a headcount line. You're deleting a node in the network — skull and all.