The Point of Laziness (2011) | Hacker News Reader

The Point of Laziness (2011)

existentialtype.wordpress.com

45 pointsamelius6 comments

I can't understand anything written in this blog very well, but it was fun to try! (I don't mean this post, I mean the blog) Seems like good stuff if you know what the author is talking about.

My PL friends tell me he has a reputation for being a little bit of a troll (and the content of this post is likewise), but if you meet him in person, it's hard to dislike him. It's frustrating talking to him because there's a grain of truth to what he says.

For example, that laziness makes parallelism harder. I mean, cmon, the last thing I'm worrying about is whether the thread is going to evaluate its results to weak-head-normal-form.

So yea, mostly, he confuses people who aren't familiar with the material. ;)

From reading Harper's works and also listening to his lectures (but not meeting the guy, sadly) I'd back that up. He's more than happy to say something that's both insightful and... inciting! In a lot of communities this rallies the troops to descend upon what he's written and strip mine it for goodies while aggressively discarding the rhetoric—clearly a useful tactic.

But in other communities it feels really trollish and oversimplified. It's the equivalence of https://xkcd.com/793/.

Like physicists, though, he's got some phenomenally great points which most "practitioners" happily ignore all while eating the possibly enormous consequences. So if you can spend the time to figure out where he's coming from then you can really gain a lot.

Once I've gotten used to his style, I agree, the content is almost invariably very good and well thought out. I tend to now just mentally rewrite his insistence that certain things are "wrong" or "idiotic" or "useless", to a statement along the lines of, "X is widely believed, but I would like to point out some of its disadvantages, and advantages of a different view". Since to some extent he's in the position of an ML holdout, in a PLs field where the Haskell researchers are lopsidedly larger than the once-larger set of ML researchers, and Haskell is often seen as the current theoretically advanced language (while ML was the theoretically advanced language of the '80s/'90s), it's useful to have a strong advocate for the minority view that ML was on the better track.

edit: On the subject here, the response he links at the bottom is also interesting. In the first comment, Harper even acknowledges one thing he actually does like in Haskell more than ML: http://augustss.blogspot.com/2011/05/more-points-for-lazy-ev...

> it's useful to have a strong advocate for the minority view that ML was on the better track.

Total agreement here. I also really enjoy his rather principled, HoTT-inspired perspective on type theory. I vastly prefer PFPL to Pierce's TaPL due to PFPL just being less preoccupied with the diversity of reasoning possible and more interested in explaining the "one right way" efficiently.

<3 bob harper

Reply on news.ycombinator.com