In the old Internet (forums and Usenet), sure, there was crowd psychology too. There always is.
However, these social media designs were far less prone to hiding unpopular opinions. As a general rule, unless you had an overtly strict moderator or something, you could read posts with unpopular opinions, even if the "crowd" disagreed.
Big social media sites (Facebook/Twitter/Reddit/etc.) in contrast use algorithmic reordering, shuffling, and tweaking of your feeds / timelines / etc. Likes / retweets / etc. are one component to what gets displayed on your feed. The net result here is that unpopular opinions are often effectively "silenced". IMHO this leads to a negative and polarizing effect.
I can't call it 100% censorship though. It's different. On most social networks, you are free to, say, start your own little space (a subreddit or your own Twitter space and whatnot) where, say, the less fortunate sides of diversity initiatives are discussed. Unless the subject strays into TOS violation territory, you can probably say what you want. However, there's a good chance in said subreddit (in this polarized environment and considering crowd psychology) that it is the positive side of diversity initiatives that are downvoted to oblivion. And the reverse is true as well. Basically, crowd psychology and the binary like/retweet system ensures that no one hears what the "other side" has to say. Maybe it's a "dark pattern" of sort? Tweak the emotions of the user in the right way, not the wrong way, and they might spend a bit longer looking at that advertisement...
Hacker News doesn't feel as "bad" to me as the bigger social networks for some reason -- I'm not a big fan of upvote/downvote systems, obviously, but at least HN includes tools like the showdead option that's pretty easy to enable. There was a large discussion about diversity initiatives during relevant stories (the James Damore memo for instance), but for others, I know that some of the "crowd" here is wary of overtly political articles period, favoring only pure tech articles. So you are going to get some front page selection bias against articles of a "political" nature.