> Unfortunately, literally none of the tweets we saw even considered the possibility that a problematic graph specific to software tasks might not generalize to

> Unfortunately, literally none of the tweets we saw even considered the possibility that a problematic graph specific to software tasks might not generalize to literally all other aspects of cognition. How am I not surprised?