Data Cascades

Compounding events causing negative, downstream effects from data issues, resulting in technical debt over time.

Data cascades share three properties: they are opaque (no clear indicators or metrics), triggered by conventional AI practices that undervalue data quality, and produce multiplied negative impacts that manifest long after the initial data decisions were made.

The term captures how small upstream data decisions compound into large downstream failures, often taking months or years to surface, and typically discovered only through system metrics, user feedback, or outright malfunction.

Related: [None yet]