Research paper · 2026 · 19 pages

The Science Analytics Never Shipped

An AI agent can generate findings continuously, at a cost approaching zero. The person receiving them has the same working memory they had in 1984 — the last year a finding from cognitive science made it into an analytics product. This paper is about the research that piled up in between.

It is written by someone who spent twenty-five years building the analytics defaults it argues against.

What is inside

Section by section

The gap that is not ignorance — why one 1984 result became every BI default and the rest stayed in the literature
The capacity that does not scale — working memory, cognitive load, and the arithmetic of forty findings
The obvious fix, and what the data did to it — what explanations actually buy
What works, and what it costs — cognitive forcing and the satisfaction trade-off
The variable underneath — verification cost, and how to make it a number
Where attention goes — surprise, bad news, the information gap
The single-format assumption — the hospital dashboard we would no longer defend
The cost of telling a story — narrative transportation and the reader’s guard
The management implication — four questions that rarely appear on a procurement checklist
A note on method — where AI was used, and what three verification passes found

“Some of what we have been calling a culture problem is a design problem wearing a culture problem’s clothes.”

— from Why I wrote this

Ali Yıldırım — Founder and CEO, D-CAT Group. Twenty-five years in enterprise business intelligence. Computer Engineering, Boğaziçi University; MBA, Istanbul Bilgi University.

Ten sections, twenty sources, one argument: the defaults were set before most of the research arrived. Each source is listed with what it actually found — including the ones that cut against the argument here.