4 ms·
"We simplify the causal discovery problem by assuming no confounding, selection bias and feedback." That's a pretty big caveat. Not saying the paper is not im
by jacobn 12y ago
"We simplify the causal discovery problem by assuming no confounding, selection bias and feedback."
That's a pretty big caveat.
Not saying the paper is not important, just that it's not (yet?) a full solution to the overall conundrum.
- UhUhUhUh 12y agoParticularly since the confounding issue is really enormous in science. And sits at the core of the example the article gives in introduction... That would be an achievement in itself to build an experiment without confounder(s).