You’re right that it won’t be completely undesirable for bots, ever. However, you can make it less desirable, to the point that the botters say “meh, who cares? That other site is better to bot”.
I’ll give you an example. Suppose the following two social platforms:
Orange Alien: large userbase, overexcited about consumption, people get banned for mocking brands, the typical user is as tech-illiterate enough to confuse your bot with a human.
White Rat: Small userbase, full of communists, even the non-communists tend to outright mock consumption, the typical user is extremely tech-savvy so they spot and report your bot all the time.
If you’re a botter advertising some junk, you’ll probably want to bot in both platforms, but that is not always viable - coding the framework for the bots takes time, you don’t have infinite bandwidth and processing power, etc. So you’re likely going to prioritise Orange Alien, you’ll only bot White Rat if you can spare it some effort+resources.
The main issue with point #1 is that there’s only so much room to make the environment unattractive to bots before doing it for humans too. Like, you don’t want to shrink your userbase on purpose, right? You can still do things like promoting people to hold a more critical view, teaching them how to detect bots, asking them to report them (that also helps with #4), but it only goes so far.