Did average income become larger after inflation?
Hello, wanderer.
A little work.
A lot of curiosity.
Things I make, stories I collect,
and ideas I'm still figuring out.
Make yourself at home.
On the workbench
All workThese days
Questions becoming evidence
The workbench.
Interactive case studies, public-data analysis, build logs, and working questions. Open one to inspect the reasoning, not just the outcome.
Nothing here by that name.
Try a different word, or return to everything.
Decision Lab 01 · simulated case study
Capacity without
tunnel vision.
How should a small fulfillment team respond to changing demand without quietly trading away quality or a sustainable pace?
Decision reading
A workable plan—with room to recover.
Delivery and quality clear their targets without pushing operating pressure past its guardrail. Preserve the balance; room to recover is useful capacity too.
Good orders shipped1,00644 below demand
On-time delivery95.8%Target ≥ 95%
Weekly contribution$15,514Illustrative · fixed cost excluded
A balanced view
Target markerDelivery95.8%
Higher is better · target 95%
First-pass yield96.4%
Higher is better · target 96%
Operating pressure53 / 100
Lower is better · guardrail 55
The decision
Not “How do we maximize output?”
The better question is how to meet demand while keeping the operation capable of doing it again next week. That changes which measures belong together.
Implementation
React where state earns it.
The Lab is a self-contained React interface, loaded only when this route opens. It connects each management lever to service, quality, financial contribution, and pressure while the rest of the portfolio stays lightweight.
Inspect the model assumptions
- Base scheduled capacity is 1,050 orders per week.
- Each overtime hour adds 2.2 scheduled orders, with small reliability and quality penalties.
- Each process-care hour uses 2.5 orders of near-term capacity while improving reliability and first-pass yield.
- Contribution uses $18 per shipped order, $22 per overtime hour, $45 per process-care hour, and $8 per defective order.
- The pressure index is an illustrative proxy for overtime and backlog strain; it is not a validated safety measure.
Decision Lab 02 · economic thought experiment
Growth for
whom?
A country becoming richer on average does not tell us who benefited—or what that income can actually buy. This short experiment separates those questions.
Did lower-income people grow faster or fall further behind?
How much can the same income buy at local prices?
Loading the economic model…
Optional readingMethod, limits, and sources
Why three lenses?
A richer average is not automatically a shared gain.
Real income growth asks whether resources per person increased. Bottom-40 growth asks whether that improvement reached people lower in the distribution. PPP changes the price-comparison unit so similar baskets can be compared across economies.
What not to conclude
This model does not estimate inequality itself.
A faster or slower bottom-40 path indicates convergence or divergence from the average under these assumptions. A Gini coefficient requires the full income distribution, while welfare also depends on services, security, health, environment, and more.
Model assumptions
- Both paths begin as hypothetical annual income per person; bottom-40 income starts at 45% of the overall average.
- Growth rates are real, annual, and compounded over the selected horizon.
- The PPP view divides market-converted income by the selected relative price level. The price level is held constant, so it changes the displayed level—not the modeled real growth rate.
- Bottom-40 growth follows the logic of the World Bank's shared-prosperity measure, but the model contains no actual country observations.
- Relative-price comparisons are broad statistical tools, not a personal cost-of-living calculator.
Conceptual sources
Definitions before conclusions.
Decision Lab 03 · family statistics
Count what?
Compared with whom?
A marriage, a divorce case, and a birth are events. A useful rate must say which population and time period those events are being compared with.
Loading the family-statistics model…
Optional readingDefinitions, coverage, and sources
A rate is not a cause
Later marriage can affect fertility—but it is not the whole explanation.
Education, health, contraception, economic conditions, preferences, age structure, migration, and policy can also change family patterns. This lab does not estimate causal effects.
Administrative coverage matters
Recorded events are not automatically every event.
BPS notes that marriage and divorce series in Statistics Indonesia can cover Muslim marriages and divorces from specific administrative systems. Always inspect the current table footnotes before calling a count national or comparing years.
Primary sources & data
Follow the definition, not just the number.
- World Bank · Crude birth rate ↗
- World Bank · Total fertility rate ↗
- World Bank · Population aged 0–14 ↗
- BPS · Union before age 18 ↗
- BPS · Recorded marriage and divorce events ↗
- Kemenko PMK · 2019–2024 event series ↗
- BPS · First-marriage-age metadata ↗
- Download demographic trends (.csv) ↓
- Download recorded events (.csv) ↓
Evidence study 01 · Indonesia · 2010–2025
One number.
Two stories.
Indonesia's manufacturing share of GDP fell. Manufacturing output still grew in fourteen of the next fifteen years. Both statements are true.
My reading
A denominator can change the story.
A share is relational: manufacturing can grow while other parts of the economy grow faster. I paired the share with real annual growth and a broader industry-employment measure before drawing a conclusion.
2025: manufacturing was 19.1% of GDP, real manufacturing output grew 5.3%, and industry accounted for 22.5% of employment.
What changed
−3.0 percentage pointsManufacturing's GDP share moved from 22.0% in 2010 to 19.1% in 2025.
What did not follow
14 of 15 years grewReal manufacturing value added increased year over year in every subsequent year except the 2020 shock.
What needs care
Not the same denominatorThe employment series covers industry—including construction—not manufacturing alone. It adds context, not proof of manufacturing jobs.
Decision implication
Never diagnose a sector from its share alone.
Before recommending action, separate level, rate of change, and composition. A metric can be accurate and still answer a different question from the one a decision-maker thinks they asked.
See the “map vs territory” mental model →Source & reproducibility
World Bank World Development Indicators. Snapshot retrieved 20 September 2026; source data updated 13 July 2026.
Inspect all 16 observations
| Year | Manufacturing share | Real growth | Industry employment |
|---|
A few things worth keeping
On my shelf.
My longer-running film, reading, and writing trails live elsewhere. Personal recommendations will settle here when I have a reason—not just a rating—to share them.
These covers are examples, not personal recommendations.
The shelf is taking shape.
For now, the complete diaries below are the honest view.
A small thinking toolkit
Try a mental model.
These are lenses, not conclusions. Pick one to look at a question from a different angle, then follow a nearby model when the first view is not enough.
Open full notebook (opens in a new tab)Local preview · start Quartz first
Nearby lenses complement or challenge each other · Choose one
Current lens
A person still becoming
More than
what I do.
Hi, I'm Your Name.
Say helloHow I try to work
Three rules I keep returning to.
- 01Start with the decision.
A dashboard, model, or meeting should clarify what someone needs to choose—not simply display what can be measured.
- 02Keep assumptions visible.
Confidence is more useful when another person can inspect where it came from and disagree productively.
- 03Leave room for reality.
Models simplify. Operations surprise. A good answer should survive contact with information that arrived late.