On the surface, you are correct. Think a little more insidiously and you'll start to see where the value comes in.
Let's say a person with ties to the Coca Cola corporation buys a popular instance. They are in control of that instance including where instance wide rules get enforced or not. It would be unwise to openly spout pro-Coca Cola messages and ban dissenters, so they'll be sneakier about it.
They'll create bot instances that create, upvote and boost posts and downvote dissenters, not enough to stick out, but enough to manipulate the feed algorithms early in the posts lifetime. And occasionally upvote and downvote some random posts to add noise to the user history. Otherwise, they let the instance run as it always has.
There will be accusations, but because it won't be provable or actionable outside of defederation or the banning of individual accounts. And other instances will hesitate to do the former because these accusations are not proven and the instance is still putting out content that their users are interacting with.
If the compromised instance admin needs to put out a fig leaf or two, they can ban the bot accounts and silently create more later.