I think he reversed his opinion on scrolling a few years back, but his methodology is (or was) flawed; it seems to have a strong bias towards what his testers are already used to. More importantly, we can’t see how the test was run exactly (questions, etc) or how significant the results were, so we can’t critique it properly.

Interesting anecdote: http://www.dourish.com/nielsen.html

Come now, don’t be hating on the Jakob.

The thing about him is that he makes recommendations based on his studies, and he even revises them once in a while. Surely he can’t still be dissing the scrolling?