Even better would be if there were user-configurable "lists", whereby you could decide upfront what you want / don't want to see (like many sites do right now with NSFW content) - the default filter would be very "protective" (no porn, no violence, no gore, no hate speech) but users could turn off any or all of these "filters". The next step is the addition of user-curated "lists" / "filters" (e.g. "no democrats", "no republicans", "no vegans", "no dog lovers", ...).
It's a tough call. In some sense, for any global website that doesn't want to impose its own moral code upon the world, it makes the most sense to be hands-off and let users judge for themselves what to see and what not to. On the other hand, doing so would amplify the echochamber effect that's already strongly present on Twitter.