That doesn't seem like too much to ask.
And while threats are fairly clearly defined, "harassment" is very nebulous. One person views a non-directed statement as harassing, the other doesn't, the community around that exchange might take one side, society-wide may vote differently, countries have different norms & attitudes, etc. How can everybody share a censorship system where everybody's view of what constitutes harassment is slightly different? It'll never go far enough to cover many people's definition (which is your complaint against the current system), and go too far for others. Sorry, but I think that having the world's systems catered to one person's comprehensive subjective view of "bad" (whoever it is) really is too much to ask. We should stick to those things which are more cut and dry, which won't ever be comprehensive.
Not everything has to be a social network, but many things (including software engineering) are better if they are. Not every social network needs moderation, but many of them are better if they have it.
AFAIK YouTube ranking also prioritizes controversial comments, which increases interactions (metrics!) but also highlights bad stuff.
Google may get software and systems (and even hardware now), but Google has failed spectacularly when it comes to social.
Do we have any reason to suspect they even tried? Maybe it's because they just slapped comment functionality onto their pages to increase user engagement?