Math Is How We Try to Make Sense of Systems
Systems are complex.
They involve interaction. Variation. Change over time.
Too much to hold all at once.
So we simplify.
We measure.
We model.
We use math.
Three Ways We Use Math
Math doesn’t just describe systems.
It shows up in three distinct ways:
- Descriptive — what happened
- Predictive — what might happen
- Prescriptive — what should happen
At first, this feels clean.
- The numbers first tell a story.
- That story becomes a model of what might happen next.
- Those models can then shape what comes next.
Where It Starts to Break
The further you move along that path,
→ from describing,
→ to predicting,
→ to prescribing.
The further you move from direct observation.
Each step introduces a gap between what is known…
…and what must be inferred.
That gap is bridged through models, assumptions, probabilities, and judgment.
Which is why systems like weather are modeled by many forecasts — not one.
Weather is a System Math Can’t Fully Contain
Weather may be one of the best examples of math operating and interpreting within a system.
Meteorologists don’t rely on one model.
They compare many:
- ECMWF (European)
- GFS (American)
- NAM
- HRRR
Each model uses different assumptions, inputs, and weighting.
Each produces slightly different predictions.
Why?
Because the system is too complex for any single model to fully capture.
So forecasters compare.
→ They look for overlap.
→ They study divergence.
→ They interpret the gaps.
Not because math failed — but because even strong math is still modeling a system larger than itself.
And even then, different forecasters can study,
- the same models,
- the same overlaps,
- the same gaps…
And reach different conclusions.
Because math informs the forecast.
But interpretation shapes it.
Even Probability has Limits
Take something simple. The forecast might say:
”12% chance of rain today.”
Many people interpret that differently.
Does that mean:
- 12% of the day will be rainy?
- 12% of the area will get rain?
- A light chance of rain all day?
Not exactly.
Weather probability is often more complex than that.
This is where systems, math, and interpretation start to separate.
The forecast is not describing certainty. It’s expressing modeled probability under specific conditions.
That doesn’t mean rain for 2.88 hours. It doesn’t mean exactly 12% of your day.
And it doesn’t always mean the same geographic assumption people imagine.
It means the system is being modeled through probability, what meteorologists commonly call PoP: Probability of Precipitation
At its simplest: PoP = Confidence × Area
Which resolves to:
How likely is rain somewhere… combined with how much of the forecast area it may affect.
So a single number — 12% — may represent very different underlying conditions.
Not because the math is wrong.
- But because the system is complex,
- the model is compressed,
- and interpretation varies.
This is the Real Point
A number can feel precise…
While still representing uncertainty, assumptions, and multiple moving parts.
Math can simplify systems into something usable.
But usable is not the same as fully understood.
The limit is rarely math alone.
It’s the complexity of the system math is trying to model.
The Shift
This matters more than it seems.
Because once math moves from describing…
to prescribing…
It starts to influence behavior.
- A target is set.
- A metric is defined.
- An outcome is expected.
And the system begins to respond to the model.
Not just the underlying reality.
Consider Something Simpler: The 18-second Call
Imagine a customer service system where a customer call must last at least 18 seconds to count.
Why 18?
At some point, there was likely math behind it.
- Patterns were analyzed.
- Outcomes were measured.
- Optimization decisions were made.
→ Maybe shorter calls correlated with poor outcomes.
→ Maybe average handling time mattered.
→ Maybe customer friction was modeled differently under older conditions.
→ Maybe it was a known rule copied from a leading competitor.
The exact reason may be unclear.
But the number itself suggests something important:
- A threshold was chosen
- based on assumptions, data, and priorities at a specific point in time.
Then the Deeper Question
What happens when the system changes?
- If voicemail replaces live inbound calls…
- If customer profiles are surfaced automatically…
- If language patterns shift…
- If friction changes…
Does 18 seconds still mean what it once did?
Or is the system still operating on inherited math?
Because many metrics aren’t just measuring the present.
They are carrying forward past assumptions.
Sometimes that’s useful.
Sometimes conditions change… but the number stays.
The real question isn’t just: Is the metric data-informed?
It’s also: Is the model still current?
When Math Moves From Observer to Participant
Math once primarily analyzed systems after the fact,
like a scientist outside the petri dish…
Observe.
Analyze.
Respond later.
Then computers shortened the delay. Math moved closer to the action.
- Transactions triggered formulas.
- Thresholds triggered responses.
- Systems began reacting in real time.
These highly complex digital systems often rely on predefined assumptions.
The formulas may be layered.
But they are still built on logic and outcomes chosen and planned in advance.
AI Pushes this Further
AI introduces something different.
- Not just measuring.
- Not just reacting.
- But dynamically adjusting.
🔄Evolving interpretation.
Instead of only following fixed rules…
AI can identify patterns, adjust weighting, and respond in ways that feel less hard-coded.
The logic is no longer limited to observation.
It operates inside it.
- Responding in real time.
- Shaping conditions as behavior unfolds.
Which raises a bigger question
If older systems used math to measure…
And digital systems use math to react…
What happens when systems evolve to interpreting and adjusting on their own?
The scientist may no longer be outside the dish.
The system now contains the observer.
The Limitation
Math can describe part of the system.
It can approximate where it’s going.
It can suggest what to do.
But it cannot fully contain the system itself.
What this Changes
When you see a number, a model, a target — It’s worth asking:
- Is this describing the system?
- Predicting it?
- Or trying to direct it?
And just as important: What is it leaving out?
Because math didn’t just evolve.
Its relationship to the system changed too.
The Connection
We now clearly see that:
- Systems produce outcomes.
- Systems behave differently.
- Boundaries shape what you see.
- Modern systems can reshape boundaries without you noticing.
Math is how we try to make sense of all of it.
But sense-making is not the same as reality.
An unsettling reality to hang with… when math stops being neutral
Math once primarily measured systems after the fact.
Observe. Analyze. Respond later.
Digital systems collapse that delay.
The more immediate the observation… the harder it becomes to separate measurement from influence.
Because when systems calculate in real time…
- Scoring behavior
- Filtering options
- Adjusting incentives
- Triggering responses
The model is no longer just looking.
At some point, the model stops watching behavior…
… and starts participating in it.
Challenge
A team is measured on:
● number of calls completed
● average call time
At first, the numbers describe performance.→ Then they’re used to forecast output.
→ Eventually, they become targets.
What changes when measurement becomes instruction?
Curious how your answer compares?
The companion discussion on LinkedIn includes the MNKY Math take and is where the conversation continues.
Continue the publishing sequence
We've explored how systems can be understood. The next step is recognizing that your own experience inside a system is information too.