I have minimal insight into Polygon's specific economics, but in my experience (former consumer tech journalist/reviewer and later EIC of GDMag), in-depth feature writing is almost never what pays a web publication's bills, and you're lucky if it breaks even.
With print magazines, features are what sell each issue; they give you a high-impact story to put on the cover that persuades people to buy the book for $6 or whatever, and hopefully they like the feature and all the stuff that came before and after it enough to subscribe. If your ad sales team is on point, you can even sell your advertisers on higher rates to put their ads closer to the print feature in the magazine, or pitch specific clients on advertising in issues that are more relevant to their products. Features basically carry the rest of the magazine.
In web publishing, features don't really have the same impact. Subscribing to a magazine is basically paying for the entire book (even if you don't actually read it all, which most people don't); pretty much no one is going to click on every single page linked from that initial feature, and even if they do, there's no guarantee that they'll come back to your site day in and day out. At best, they'll probably be more inclined to click on your site's content when it shows up in their social channels/news aggregators, or go to your home page if they think you might have covered something important in games, but when you look at how much time and money goes into a really great feature, it's highly unlikely that it'll ever pay for itself. And mind that some of these features can tie up multiple people -- writers, editors, web devs, designers -- for weeks of work.
Meanwhile, writing opinion pieces based on current events is really easy to write and get people to click on, costs next to nothing, and the great thing is that people will click through to read them if they agree with the headline, or click through to hate-read them if they disagree.
Conversely, the best features are usually way more nuanced and require more time and mental energy from the reader than a silly op/ed, which means it's actually harder for people to click through. In my experience, the smarter an article is, the fewer people are going to read it, and it takes genius writers and editors to write something that is smart AND can convince people to read it.
For a fun exercise, track all the content you click on (articles, videos, memes, whatever) over the course of the day and read to completion, then go through that list and see whether the mix of that content (its topic, format, sophistication) reflects the mix of content you would like to see in any given publication you follow. It probably doesn't!
From what I can tell, the Vox ad team is putting in more work in monetizing specific high-profile features than any other online pubs I ever saw put in (probably in part because they have a pretty snazzy in-house ad agency, IIRC), but it's still a pretty tricky game to play.