It's good for HN to defy expectations, because predictability and repetition are the worst things for curiosity. But it's good to defy expectations in an interesting way. The distribution of defied expectations is bimodal; when it's bad it's very bad.
On the third hand: it's also good if—more rarely—there is an occasional bit, just a wafer-thin bit, of "why is this on HN" agitation of the bad kind too. It gets the juices flowing. It means readers care. I think it bonds users more closely with the site in the end. But it should be rare. Since there's so much variance in what people like to see, and HN is a non-siloed site (i.e. we're all in the same theater), every user sees a lot of things they dislike, so we mostly have the opposite problem.
One other point. People sometimes react to these unpredictable submissions by saying "oh, so stories about little kids driving [or whatever $story was] are on topic now?" But of course that is exactly the wrong conclusion. Since this story made HN's front page, a follow-up about, say, a 7-year-old driving a tractor, would have power-law-dropoff levels of uninterestingness. And a 3-year-old driving a motorboat would be right out. Rather, there's some other unexpectedly interesting off-beat story waiting to play this role for next time, and the point is that you can't predict it from any sequence. Actually such submissions are surprisingly uncommon—I feel like they show up maybe 2 or 3 times a year, if that. That's why I'd be curious to see links.
Edit: just to be clear—I'm not criticizing your comment! I'm saying you're an ideal HN user.