Why Fixed Cost Doesn't Fix Itself
Follow the S-Curve
Why Fixed Cost Doesn't Fix Itself
Follow the S-Curve
Last week's piece made a claim: innovations arrive on emotion, but only fixed-cost reduction lets them survive. A fad is just an innovation that never clears that second hurdle.
That raises an obvious question. If fixed cost is the thing that decides whether something lives or dies, why doesn't every organization just go find it and fix it? Everyone agrees bottlenecks are expensive. Nobody's in favor of waste. So why does fixed cost sit there, visible, un-shrunk, year after year?
Because a bottleneck is almost never one problem. It's two, stacked on top of each other, and each one has a different reason for not getting fixed.
The 90/10 Split
In my experience, most organizational bottlenecks break down roughly like this:
90% is process conflict. Two systems, two teams, two procedures that were each built correctly for their own purpose, now butting heads at the seam where they meet. Nobody designed this on purpose — it accumulated. The fix is usually knowable. An engineer, a process owner, or a competent outside eye can look at it and see where the friction is. This is a structural problem, and structural problems are solvable in the ordinary sense: you can write the fix down.
10% is human problem-solving. Somewhere in the chain, a person has to make a call, own a decision, or admit that the way something has always been done needs to change. That's not a structural gap. That's a will gap.
Two different problems. Two different reasons the delay happens. And critically — two different kinds of cost.
Opportunity Cost vs. Emotional Cost
The process-conflict delay is an opportunity cost. The fix is known or knowable, the org just hasn't spent the resources — time, budget, attention — to implement it. Every day it sits unfixed, you're not paying a penalty so much as forfeiting a gain you could have already banked. It's the cost of the thing you didn't do yet.
The human-issue delay is an emotional cost. This is the part that doesn't respond to a whiteboard. Avoidance, ego, fear of being visibly wrong, sunk-cost loyalty to "how we've always done it" — these delay a fix that everyone privately knows is coming anyway. This isn't a resourcing problem. Throwing more budget or more headcount at it doesn't touch it, because the constraint isn't capacity. It's a person, or several people, who have to move first.
Here's the part that matters for the S-curve: these two delays don't just sit side by side. They compound.
The process conflict makes the org visibly slower, which raises the emotional stakes of admitting whose process has to change — which makes the human delay longer — which lets the process conflict compound further, because nobody's touched it. The 10% makes the 90% harder to fix, and the 90% makes the 10% harder to face.
Why "Naming the Bottleneck" Isn't the Fix
There's a popular idea in AI-and-org-design conversations right now: identify where the organization loses the most momentum — usually coordination, decision-making, fragmented information — and that's where technology becomes transformative.
That's directionally right, and it's also incomplete. Naming the bottleneck is diagnosis. It's not treatment. You can have a perfectly accurate map of the 90% process conflict and still not move, because the 10% human issue is what's actually holding the org in place. A tool that automates the process conflict away does nothing for the person who doesn't want to be the one who admits their team's procedure was the wrong one.
This is why so many technology rollouts — AI included — look like they should collapse fixed cost on paper and don't, in practice, for months or years longer than the business case predicted. The math was right. The 10% wasn't accounted for.
What This Means for the AI S-Curve
If AI is going to climb its own S-curve the way the last piece argued — surviving on fixed-cost reduction rather than emotional appeal — it has to clear both parts of this split, not just one.
The 90% is the easy sell: AI genuinely can resolve a lot of process conflict, faster and more visibly than most prior technologies. That's real, and it's why the emotional excitement isn't baseless.
The 10% is where most of these efforts will actually stall. Someone has to admit a role changed. Someone has to stop being the bottleneck they've quietly been for years. Someone has to make the call that the old process — the one they built, defended, or inherited — isn't the one going forward. No model ships that decision. It still has to be made by a person, on a Tuesday, in a room, at some emotional cost to them personally.
Fixed cost only goes down when both halves move. Track the 90%, and you'll see where the opportunity is. Track the 10%, and you'll see why it's still sitting there.
Herbert Roberts, P.E. is a licensed professional engineer with 30+ years in aviation research and development across two companies, and has spent eight years analyzing accidents for attorneys under his P.E. license