Design Rationale

01 Why This Column Exists

Most consulting sites tell you what they sell. Fewer tell you why the thing is built the way it is, and fewer still name the alternatives they rejected.

This column is the argument. It sets out the reasoning behind the practice, including the parts a skeptical reader would push on. If you are deciding whether to fund this kind of work, these pages are the case you would be funding.

They are also, deliberately, arguable. Each one takes a position that a competent person could disagree with.

02 The Case for Cause and Effect

Two pages carry the core claim.

Decision Intelligence makes the practical case. Prediction tells you what is likely to happen. Decisions require something harder: what happens if you act. Those are different questions, and most analytics answers only the first while being presented as if it answered the second.

Pearl’s Ladder is the framework underneath everything else on this site. Three rungs: what tends to happen, what happens if we act, and what would have happened had we chosen differently. Each rung requires more than the one below it, and no amount of data moves you up a rung on its own.

If you read only one page in this column, read Pearl’s Ladder. Everything else assumes it.

03 What We Argue Against

Two pages name the alternatives and say plainly why they fall short. Both are aimed at methods that are widely used and often defensible, the objection is to using them for decisions they cannot support.

Against Regression takes on the workhorse of corporate analysis. Regression is excellent at describing patterns. It cannot tell you what would happen if you intervened, and the standard output gives no warning when you have crossed that line.

Against Black Boxes argues that a model whose connections are named and visible is more defensible than one whose accuracy is merely higher. In a regulated setting, being right is not sufficient. You have to be able to show why.

04 Where the Ideas Come From

Two pages place the work in context rather than arguing for it.

Foundations traces the intellectual roots, where these ideas come from in statistics, computer science, and the study of causation itself. None of this was invented here, and the page says so.

The Competition is the honest survey: who else sells causal modelling, what they do well, and where this practice sits among them. Written on the assumption that you will look them up anyway.

05 What It Looks Like in Practice

One page closes the gap between argument and output.

Inference takes a single model and asks it three different kinds of question, a forward prediction, a what-if, and a look back at what would have happened otherwise. Same model, three answers, three different uses.

This is the page that makes the ladder concrete. The other pages tell you the three rungs exist; this one shows you the same model climbing them.

06 A Reading Order

The column is numbered, but that is not the order most readers want.