The answer is twofold 1) Network effect combined with 2) coordination problem. One or few figures dropping off is just net loss for them, does not affect the X at all. Only way to do the change is to coordinate and do it with critical mass.
Individual action can't solve problems effectively when the cost of changing behavior is too high when others don't follow immediately. Many public figures started their mastodon/bluesky/threads accounts but came back to Twitter when they didn't get following there.
And actually it isn’t that cheap because if you post to say 4 (X, Bluesky, mastodon, Facebook), you now have to choose where you’re going to spend time reading and responding to others. On all of them or just on X, where the gravity is.
The problem I rarely see discussed is that if one user I follow posts to Twitter and Bluesky simultaneously and another totally left Twitter for Bluesky I as a Bluesky reader will have a mix of posts I have already seen on Twitter and original Bluesky posts on my timeline. There needs to be a way to somehow mark and filter crossposted content.
Yeah, that's a fair question and I usually don't do that if user posts more than 1-2 posts per day. But it's hard to know how the account will be used in advance when user says "I made an account on Bluesky, follow me there".
To me, you numbers demonstrate that it works very well: compare the whole userbases of the two platforms and you will see that the engagement on Mastodon is larger and growing fast.
Of course it can. Just publish posts on two platforms simultaneously. Or even delay the Twitter post.