New method claims better accuracy at detecting cause and effect from data

Researchers developed two techniques to automatically determine causal direction in paired numerical datasets, achieving 77.9% accuracy on real-world examples compared to prior methods at 63%. This matters because most data analysis tools can't reliably tell whether A causes B or B causes A — knowing the direction is essential for making predictions and policy decisions based on data.

If this method becomes reliable enough, it would reduce researchers' need to run expensive controlled experiments to establish causal relationships; they could infer direction from observational data alone, accelerating research in fields from medicine to economics where experimentation is costly or impossible.