On Facebook and most other social networks, at least the subject of harassment can at least block the perpetrator, and refuse to friend an alt account. Since the primary use case for Twitter is to have public tweets and open replies/mentions, blocking a user making death threats won't solve the problem; they will just open a new account and flame you from there. On Youtube, there is less reason to look at the comments; a celebrity can focus on the number of views as a metric for success.
The reach of abusers on Twitter is far greater than on other networks. Add to that a lack of defined communities, as in Reddit, Facebook Groups, and other message boards: there are no community moderators on Twitter to react to harassment and the posting of personal information.
I see the handling of abuse on Twitter as a hard problem that will take a mix of pattern-matching algorithms for brigade detection, machine learning on tweet sentiment and content (doxxing, death threats, other TOS violations), and least scaleably, a large pool of tier 1 support staff to act on recommendations to ban users.
That's (as you note) putting it very lightly. It's basically unusable as a public platform if you're moderately famous. Go watch Jimmie Kimmel's bit with celebrities reading mean tweets on youtube and imagine being on the receiving end of it. The ability to lock down your account misses the point, though it does say that Twitter thinks abuse might possibly be a problem (as does comments by its CEO).
Screening out abusive users is a hard problem that hasn't been properly solved yet, anywhere in the world. An eternal September, to revive the phrase. G+/Youtube and Facebook's real name policy hasn't managed to solve it, and Facebook is only able to screen out the worst - gore and pornography, with devolving political discussions leading to many unfriendings this election season. Maybe not quite abuse, but still unpleasant.
If fighting abuse is your life's mission, then join Twitter. See what they're doing about abuse, talk to people who have tried to fight abuse for years, find out what's succeeded, what's failed. There may be some Twitter org-specific implementation barriers, but that's still a learning experience for the future. If you find some Twitter org-specific barriers that are insurmountable, get out.
My bias, here, isn't that I work for Twitter (I don't) but I doubt Twitter's inability to quash abuse has anything to do with corporate governance (other than an inability to spend infinite amounts of money on an infinite amount of moderators).
Humans are still currently the best at understanding humans, and we still do a terrible job sometimes. Computers aren't able to parse sarcasm, idioms, or subtlety, and being fed in plain text instead of speech makes it even harder as there's no vocal tone to analyze either. Twitter's 140 character limit hurts quite a bit here.
I agree that making a difference is key, but the problem itself also has to be considered - the early teething pains at Twitter in reaching "webscale" have been fixed by now (when was the last time you saw the fail whale?), but "people can be assholes to each other" is probably the oldest problem in the world.