Hacker News Reader

> suppressing noise and increasing the signal-to-noise ratio.

So apparently here lies the gory details. There was a short interview article with one of fired Twitter engineer who claims to had been "basically responsible for messing up your timeline", and as a followup to the article, someone on Twitter who is believed to be that person defended that the end state that GP considers a worse state(so do everyone!) showed better retention through A/B testing.

They must have been doing what they think is good, backed by scientific methods and hard data, SNR ever improving in pandas. Yet everyone is in good agreement that Twitter had betrayed and sabotaged the user community, engaged in artificial manipulations on honest good people, accelerated divisions, radicalization, even promoting terrorism, worldwide.

I think we have elephants in the room, definition of SNR being one thing, or one of its legs. Something is off.

1: https://www.businessinsider.com/pro-elon-musk-twitter-employ...

> hate speech is the thing that some people try to defend under the banner of free speech.

I think it's more of an acknowledgement that there's a price to freedom/liberty and there are fundamental trade-offs involved. I.e. can't have our cake and eat it too.

The difficulty is in finding where exactly the draw the line. For better or worse, currently there are many technically legal yet unsavory views allowed under "freedom of speech".

You've spelled out pretty well the dangers of free speech but if we err to much on the side of caution, that itself causes animosity and discord from (perceived or actual) loss of liberties. And of course, if left unchecked, it can create a culture which embraces more centralized concentration of power, inviting corruption and further concentration.

I'm still wrestling with my own thoughts on where to draw the line and how to balance freedom with safety.

Exactly. I'm not going to claim there's an easy solution to any of this. The line between "valid political opinion" and "hurting someone" can be very thin indeed. Sometimes they overlap, and then what?

All I'm saying is: we should be honest about these trade-offs, and not pretend there's such a thing as "absolute free speech". And I also think we'd better err on the side of allowing too much speech than on the side of too little. But at the same time, I also think events of the past couple of years warrant a reevaluation of those tradeoffs and there we draw those lines. And I don't think Twitter, Musk, or any other social media should be making those decisions for us, but they probably should be a voice in that discussion.

Completely agreed. Although I'm not sure the balance of trade-offs needs to change due to recent events, I certainly believe it's worth discussing.

There’s free speech from a legal perspective and free speech from PR perspective. I’d argue that free speech from a legal perspective is actually pretty free in most western countries.

From a libertarian perspective private platforms should freely decide what they allow and what they won’t. The public can also freely criticise them and move away. Advertisers can freely decide if they want to advertise there.

Every libertarian should visit a true neo-nazi site like Stormfront to see how much we are allowed to say.

Started off strong there but your points and logic fall off.

Porn is not "completely legal". There are many limits to porn within the law.

Inciting violence isn't protected speech. Hate speech isn't protected speech.

The first ammendment states explicitly that the government cannot create laws to restrict the human right to free speech and association. That does not mean every expression is legal. Comparing legal things and illegal things is muddy-ing the water

> All meaningful speech has consequences. Calls to violence definitely so. Is this really something that needs to be defended when spam and porn don't?

To me it's all pretty simple. The platform has its loyalty to the listener. It is up to the listener to decide whether he wants to engage with different kinds of speech, and most networks already can determine that, either by separation into forums and communities or just by general engagement metrics.

If you prevent exposure because the listener doesn't want to be exposed, that's OK.

If you prevent exposure because you don't want the listener to be exposed, that's censorship.

Censorship is suppressing information because it might be listened to, while spam filters and porn filters is suppressing information because it won't be listened to. The ethical and moral boundary is clear, and it is strictly the platform's loyalty to its consumers.

Anything else is inherently positioning the platform in an assumed moral superiority to its users, which given the recent Twitter revelations, is an incredibly bad assumption.

Moderation can be both good and bad. Good moderators are those that remove content nobody will want to see. Bad moderators remove content because people might see it.

There are plenty of calls to violence from the Ukraine side against Russia. None of those got censored. Should they?

I'm not defending the essence of the speech. It should be irrelevant. I'm defending what is a moral intent and what is an immoral intent of the platform. There's meaningful discussion and then there are constructing strawman and fabricated threats and vague ambiguous terms like hate speech.

If your method for deciding whether to censor foofoo is whether there exists a story where foofoo leads to bad outcome, and someone wants foofoo censored, they will create that story (potentially including real life actions). Therefore deciding whether to censor foofoo should be independent of the existence of those stories. They also tend to overgeneralize and use broad categories when the stories themselves are specific anecdotes.

To me the boundaries are pretty clear, and everything else is just people telling pretty meaningless stories, conflating terms, and having inconsistent standards.

Understanding where things stand morally is something you do after discussing them, not before.