You read seven posts
about breaking the monolith
into services.
You watched three talks.
Every one of them
went the same way.
It was hard.
It took longer than planned.
And then deploys got faster,
teams got freedom,
and everyone was happier.
So you wrote the proposal.
Eighteen months.
Twelve services.
You had the references
to back every claim.
In the review,
a quiet engineer
who had been at the company
for nine years
asked one question.
How many teams tried this
and stopped.
You did not know.
Nobody in the room did.
There is a reason for that.
During the Second World War,
a group of statisticians in New York
was shown where the bombers
that came home
were full of holes.
Wings.
Fuselage.
Tail.
The obvious move
was to armour those places.
Abraham Wald said
armour the places
with no holes.
Engines.
Cockpit.
Because the planes hit there
were the ones
that never came back
to be counted.
Your reading list
is a hangar of returning planes.
A team that rewrites
and succeeds
gives a talk.
A team that rewrites
and quietly rolls back
does not.
The people move on.
The post never gets written.
The failure
is not hidden.
It is simply missing,
and missing evidence
feels exactly like
no evidence of a problem.
So you see a success rate
of one hundred percent,
built entirely
from the people who survived.
So ask for the denominator.
Before you copy a success,
ask how many tried the same thing,
and where they are now.
Go looking
for the abandoned branches.
Ask the people who were there
for the attempts
that were never announced.
Read the post mortems,
not just the launch posts.
Count what did not come back.
A rewrite that failed
is still evidence.
It just never got a stage.
The evidence you can find
was chosen by who survived.
The answer you need
is usually in the part
that did not.
– Serguey Asael Shinder
Top comments (0)