Let's see:
Network effect hits Mastodon specially hard as it competes not just with Twitter, but also Threads and Bluesky. In those situations, a smaller userbase means that people will outright ignore you as an option.
The way that federation was implemented; as linearchaos mentioned in another thread, if you settle in a smaller instance (the "right" thing to do), you won't get "good collections of off node traffic". So it creates a situation where, if you know how federation works you'll avoid big instances, and worsen your own experience; and if you don't, well, Mastodon's big selling point goes down the drain.
Federation itself introduces a complexity cost. That's unavoidable and the benefits of federation outweigh the cost by far; however, the cost is concrete while the bigger benefit is far more abstract.
Branding issues. Other users already mentioned it, but you don't sell a novel tech named after an extinct animal.
And this is just conjecture from my part, but I think that microblogging is becoming less popular than it used to be; people who like short content would rather go watch a TikTok video, and people who want well-thought content already would rather read a "proper" blog instead.
On a lighter side: the very fact that we're using the ActivityPub now helps Mastodon, even if we're in different platforms (like Lemmy, MBin, PieFed, SubLinks). Due to how federation works, you're bound to see someone in Mastodon sharing content with those forums and vice versa; it could be a bit less clunky but it's still more content for both sides.
On the text: I think that the author reached the right conclusion through the wrong reasoning. The activity peaks don't matter that much, when there's a huge influx of users you're bound to see some leaving five minutes later. The reason why Mastodon is struggling is this:
See those slopes down? They show that the stable userbase is shrinking. Even users engaged enough with the platform are slowly leaving, but newbies who could fill their place aren't popping up.