I guess another way of phrasing the original article's claim is that although communities do form, they have no straightforward way to exclude strangers, interlopers, and even griefers: and except for incessant use of hashtags (which could eat up valuable space), there's no simple way for a person who is part of multiple communities to direct comments to only one community.
That's a recipe for both serendipity and uncomfortable moments (or worse) if someone has strong opinions on, say, computer science, religion, and animal rights. People who chose to follow them over one thing will constantly see their opinions on other things. That might be great under some circumstances because it will promote more interesting and broader discussions or lead people to learn about ideas that they wouldn't have naturally come across in their own filter bubble. But it might produce some serious disruption in the conversation too, especially if that person's views are offensive or upsetting to some readers.
I know a Twitter user writes a lot about computer science and a lot about sex and sexuality. I find both sets of posts frequently insightful, but the latter would be off-topic in a forum devoted only to computer science, and they do sometimes produce offense.
It seems like the best case for avoiding really bad forms of conflict is when a group of people tweet almost exclusively on a single topic that outsiders don't find upsetting or offensive (or simply don't know about). But a lot of people do want to have at least some discussions that others will inevitably be offended by, and the broadcast medium can be a challenge for that if you didn't want to get into it with the strangers (or for that matter have some of them insult you, threaten you, or even dox you).