Why lowering the synchronous share makes an organization faster
Execution got cheap, but delivery dates did not move. Amdahl's law explains where the bottleneck went — meetings, waiting for replies, approvals and confirmations.
AI made the doing of work dramatically cheaper. Code, documents, research — all of it lands in a fraction of the time it used to.
And yet the date the project finishes has barely moved. If that sounds familiar, this post is about why.
The part you cannot parallelize governs the whole
Amdahl’s law comes from parallel computing:
The upper bound on speeding up a program is set by the fraction that cannot be parallelized.
If 40% of the work is serial, running the other 60% across ten thousand cores still caps you at 2.5×. Only once the serial fraction drops to 1% does 100× come into view.
The formula is simple. With a serial fraction of s:
maximum speedup = 1 / sThe smaller the denominator gets, the more sharply it pays off. That is the counterintuitive part.
What “cannot be parallelized” means in an organization
Think of your organization as one parallel machine. What parallelizes is execution: writing the code, making the deck, doing the research. That is exactly what AI made cheap.
So what does not parallelize? Work that cannot move until you synchronize with someone else.
- Nothing is decided until the meeting happens
- Nobody moves until the reply arrives
- It sits still until the approval lands
- Someone is waiting on “is this right?”
Adding people does not parallelize any of this. If anything, more people means more synchronization points.
In a world where execution is effectively free, this synchronous region is the only thing setting your speed.
The goal: from 40% to 1%
At a 40% synchronous share — roughly “two days a week disappear into waiting and meetings” — your ceiling is 2.5×.
- 40% → 2.5× ceiling
- 10% → 10× ceiling
- 1% → 100× ceiling
Cutting from 10% to 1% feels far less dramatic than cutting from 40% to 10%. But the payoff goes from 10× to 100×. The law pays out most at the end, which is its most practical implication.
Not “work faster” — “synchronize less”
The important distinction: lowering the synchronous share is not the same as hurrying. Shortening a meeting does not remove the synchronization point as long as that meeting is still the only place decisions happen.
What you reduce is not time. It is dependency.
- Turn decisions that needed a meeting into ones that can be made asynchronously
- Make “whose ball is this” visible without having to ask
- Give deadlines structure instead of leaving them verbal
- Delegate approvals to rules wherever a person is not actually required
None of that is about working faster. It is about structure.
And you cannot change a structure you cannot see
The catch is that in most organizations nobody knows what their synchronous share actually is.
Which task, whose decision, how long it has been sitting. When none of that is visible, you are left with “things feel slow lately.”
What UTSUROI does first is turn that into data: extract the state of work from everyday communication and task movement, and point at where it is stalling. Not for oversight — to get it flowing.
UTSUROI — 移ろい — means to keep changing without ever stagnating. We are building the structure that lets an organization be that.
Ready to try it? Start with onboarding.