Only when I don't fully understand the arguments then I read the article
I do try to read the article / thing if I am going to comment - but that's not many.
Many times the comments let me know that other needed thing is not released yet - or that no pricing info is published - love those saved clicks.
I also do that. Most non-news articles aren't better than the average two paragraphs HN comment, but they are 10 times as long. Time is precious and at some point who has the time for yet another "[strong opinion about some technology/pattern/that's how we do at our company]". Almost 100% of the time the comments contain better information on the topic tackled by the article and with higher information density.
If it's a news article there's always a few comments quoting important parts of the article which is usually enough to understand what's being reported.
The articles that I tend to prefer and to read are the ones that highlight a feature that I wasn't aware of about a language that I'm interested in.
The articles have a filter that someone has gone to the "effort" of writing up an article, and it takes a lot of upvotes to get to the front page. The commentary has no such filter.
It's very easy to shoot down ideas on HN in the comments and get rewarded for it, especially if it's to dismiss a product with alternatives. Not just the now notorious dropbox comment, but I'm often seeing comments here suggesting alternatives that if you're in the know you'll realise just don't solve the problems in a pain free way.
when people mention mac only (big thanks to those who do this!) - I tend to just believe and move on.
with some projects I open tabs for the article and all the alternatives people mention bookmark to dig into later.
So my methods are not always this or that - but as a general 'way of HN' for me, these methods are what makes HN a great time saver in general, but not always.
Perhaps it is time to retire one of the HN guidelines namely - Please don't comment on whether someone read an article. "Did you even read the article? It mentions that" can be shortened to "The article mentions that." since it seems a lot of people haven't actually RTFA when posting comments.
Have you ever seen a "Did you RTFA?" just as a single comment with nothing else?
Honest question because I've never seen that but may have missed some.
e.g. "Why I like small projects"
Conversely, if I haven't read the article, I try not to comment as if I have.
For science postings I read the artcle, half of wikipedia and then humbly read the comments.
And often the first few comments are enough for me to decide if the article is worth reading.
The exceptions arise when I have already read up on the topic, so I might read the comments to see how the article effects others, then go read it to see what slant it has.
So I'm mostly using HN as an ersatz message board.
I do, however, read the article before I write a comment of my own.
Occasionally, I find an interesting article and actually read it, but most of the time such articles come from personal blogs of developers who just wanna write about an interesting thing they learned. Coincidentally, such writers tend to share my love for simplicity and terseness - they contain no ads, no distractions and in most cases no necessary JavaScript code - they just write about the thing they're writing about and don't distract you. The j3s.sh website is a nice example that I could find amongst my bookmarks.