(function(){
      'use strict';
      var links = document.getElementsByClassName("titlelink");  
      for(var link of links){
          if(link.href.includes("twitter") ){
              var owner = link.closest(".athing");
              owner.nextSibling.remove();
              owner.remove();
          }
      }
  })();

I don't know any javascript, but this is my attempt at modification:

    let domains = "twitter\.com|\
                   cnn.com\
                  ";

    (function(){
        'use strict';
        var links = document.getElementsByClassName("titlelink");
        var owners = [];
        for (var link of links) {
            if (link.href.match(domains)) {
                owners.push(link.closest(".athing"));
            }
        }
        for (var owner of owners) {
            owner.nextSibling.remove();
            owner.remove();
        }
    })();

[0] https://chrome.google.com/webstore/detail/hackernew/lgoghlnd...

Optionally, you can publish the userscript for others.

    news.ycombinator.com##.itemlist > tbody > tr:has(.sitestr:has-text(/nytimes.com|economist.com/i))
    news.ycombinator.com##.itemlist > tbody > tr:has(.sitestr:has-text(/nytimes.com|economist.com/i)) + tr
    news.ycombinator.com##.itemlist > tbody > tr:has(.sitestr:has-text(/nytimes.com|economist.com/i)) + tr + tr

https://ontology2.com/essays/ClassifyingHackerNewsArticles/

based on these gripes

https://ontology2.com/essays/HackerNewsForHackers/

What I'd say based on that research project was that my intuition that I wanted to block things based on specific rules and keywords wasn't so hot. For instance I don't feel that Hacker News is crazy for Apple these days, when I was doing that project there was the combination of a high rate of Apple articles because Apple had just announced a round of products and also those products being derivatives of last year's products and not being exciting to me.

I still think paywall articles are worth either blocking or automatically routing to archive.is, and there certainly are a set of bad sites (medium, right-wing substacks, etc.) But actually the learning approach is highly effective against the Apple obsessions, dogpill theorists who can't stand that people are talking about anything other than Ivermectin and most of what is truly annoying.

A problem that still bugs me is what I was going to call "me tooism" when I was writing those articles but thankfully I didn't publish it before the #MeToo hashtag got popular. It isn't that particular phenomenon but the more general one that somebody sees an article about topic X got 500 votes on the front page of HN so then 10 people quickly write half-backed articles replying to the original article because they think if they write something fast enough it will also get 500 votes.

That's closely related to the problem of "news" where inevitably an article about something (say the BBC reports some people in Ukraine blew up a bridge to slow down Russian invaders) is followed by a number of other reports by other news agencies about the same thing. Maybe one of those articles is newsworthy but the rest of them aren't. Really they should be clustered as a single topic and filtered on that basis but what that means is not so simple: do you show only the first article on the topic or do you show the "best" article? Does the cluster end up including other bridges getting blown up in the Ukraine? Other bridges blown up everywhere, etc. It's easy to talk about but not so easy to implement.