A forecast is a structured expression of uncertainty. Almost every failure in this area comes from stripping the uncertainty out — converting a probability into a promise, a range into a date, and a set of assumptions into a commitment nobody can trace back.
These articles cover the arithmetic and the psychology together: why ten sensible assumptions multiply into an unlikely plan, why more data cannot make an adaptive system predictable, and why "data-driven" is not the same as evidence-based when the data still requires interpretation.
Articles in this guide
-
How to Explain Statistical Distributions Without Starting With the Maths
What Monty Python, Douglas Adams and Groundhog Day can teach us about uncertainty, probability and Monte Carlo analysis
Interactive simulation: Hitchhiker's Guide to Distributions →
-
Probabilistic Delivery Forecasting for Product Managers
Why a fixed deadline hides uncertainty, and how probabilistic forecasts support better product decisions.
-
Compound Probability in Business Plans: Peep Show and Football Accumulators
Mark Corrigan explains why a ten-leg football accumulator and a ten-assumption business plan fail for exactly the same mathematical reason.
Interactive simulation: Compound Probability — Why Programme Plans Fail Like Accumulators →
-
AI Football Forecasting and Open Systems: Chris Kamara Explains
Why more data cannot make an open system predictable, and what goes wrong when a 57% probability gets reported as a promise.
Interactive simulation: Open Systems and the Limits of Prediction — An Interactive Model →
-
Data-Driven Product Management: E.H. Carr and the Interpretation Problem
E.H. Carr and the Myth of Data-Driven Product Management.
-
Stakeholder Objectives and Misaligned Incentives: Lessons from the Likely Lads
Two failure modes that stakeholder management usually treats as one: people who do not share your objective, and people who share it but are quietly optimising for something else.
Interactive simulation: Stakeholder Objectives — Misalignment, Side Quests and Invalid Goals →
Where this gets applied
Work Simulator is a Monte Carlo platform built precisely to produce confidence ranges instead of single dates.