[commandlinefan]

FTA: "even though such filtering saves you from having to see things you don’t like, it doesn’t stop the objectionable content from existing". He doesn't want upvoting/downvoting, he wants complete eradication (of whatever the majority happens to objects to right now).

[throwaway2245]

> of whatever the majority happens to objects to right now

I don't see that Martin Kleppmann is using 'democracy' to mean 'majoritarianism' here. He makes considered points about how to form and implement policies against harmful content, and appears to talk about agreement by consensus.

Democracy and majoritarianism are (in general) quite different things. This might be more apparent in European democracies.

[Viliam1234]

He plays a little trick by saying "ultimately it should be the users themselves who decide what is acceptable or not". This has two meanings, somewhat contradictory.

The straightforward meaning is that ultimately I decide what is acceptable or not for me, and you decide what is acceptable or not for you. We can, and likely will, have a different opinion on different things.

But the following talk of "governance" and "democratic control" suggest that the one who ultimately decides are not users as individuals, but rather some kind of process that would be called democratic in some sense. Ultimately, someone else will make the decision for you... but you can participate in the process, if you wish... but if your opinions are too unusual, you will probably lose anyway... and then the rest of us will smugly congratulate ourselves for giving you the chance.

> Democracy and majoritarianism are (in general) quite different things.

Sure, a minority can have rights as long as it is popular, rich, well organized, able to make coalitions with other minorities, or too unimportant to attract anyone's attention. But that still means living under the potential threat. I don't see a reason why online communities would have to be built like this, if instead you could create a separate little virtual universe for everyone who wished to be left alone... and then invent good tools for navigating these universes, to make it convenient, from user perspective, to create their places, to invite and be invited, and to exlude those who don't follow the local rules (who in turn can create their own places and compete for popularity).

[throwaway2245]

> The straightforward meaning is that ultimately I decide what is acceptable or not for me, and you decide what is acceptable or not for you.

I disagree that this is straightforward in meaning. Even if I do have a good idea of what is unacceptable to me, I need someone external to screen for that. If the point is to avoid personally facing the content that I find unacceptable, it's impossible for me to adequately perform this screening on my own behalf.

I can instruct or employ someone (or something) to do this, but then ultimately they will make the decision for me. It's only plausible to do this at scale, unless I'm wealthy enough to employ my own personal cup-bearer who accepts the harm. So, it makes sense to band together with other users with similar requirements.

Your claim seems to be that delegating these decisions is a bad thing that should be avoided, but it is an essential and inevitable part of this service - I have to delegate that decision to someone else, or I won't get that service.

This is not to mention legal restrictions on content in different jurisdictions, which define a minimum standard of moderation and responsibility, that may include additional risk wherever they are not fully defined.

[webmaven]

> I can instruct or employ someone (or something) to do this, but then ultimately they will make the decision for me. It's only plausible to do this at scale, unless I'm wealthy enough to employ my own personal food-taster, so it makes sense to band together with other users with similar requirements.

And here we run into the issue that economists and political scientists call "the Principal-Agent problem"[0].

Whether we're talking about the management of a firm acting in the interests of owners, elected officials acting in the interests of voters, or moderators of communication platforms acting in the interest of users, this isn't a solved problem.

And in fact, that last has extra wrinkles since there is not agreement on just whose interests the moderator is supposed to prioritize (there can be similar disagreement regarding company management, but at least the disagreement itself is far better defined).

This is deeply messy, and as hard as it is now, it is only going to get worse with every additional human that is able to access and participate in these systems.

[0] https://en.m.wikipedia.org/wiki/Principal%E2%80%93agent_prob...

[nine_k]

This is the crux of censorship. If anything, it hinges on hubris: the censor assumes to know which content deserves to exist at all.

The need for censoring content still exists because certain kinds of content are deemed illegal, and failure to remove that may end up in serving jail time.

On the other hand, moderation is named very aptly.

That said, I fully support the right of private companies to censor content on their premises as they see fit. If they do a poor job, I can just avoid using their services.

[InsomniacL]

-Devils Advocate

> I fully support the right of private companies to censor content on their premises as they see fit.

Those private companies don't have the right to censor content on their premises 'as they see fit' without giving up protections afforded to them in law as 'platforms'. The question is at what level of moderation and/or bias do they become a 'publisher', not a 'platform'.

> If they do a poor job, I can just avoid using their services.

Issues arise when the poor job spills over outside their service. As an example, The people who live around the US Capitol endangered by pipe bombs in part because of incitement organised on Twitter.

[chrisoverzero]

> Those private companies don't have the right to censor content on their premises 'as they see fit' without giving up protections afforded to them in law as 'platforms'.

Not only do they, but there’s no such thing as “protections afforded to them in law as ‘platforms’”: “No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.”

> The question is at what level of moderation and/or bias do they become a 'publisher', not a 'platform'.

This idea of “publisher vs. platform” has been entirely made up by people with no understanding of the state of the law. [1] “Bias” doesn’t play into it – they can do what they want, in good faith, on their website. Hacker News (via its moderators) has a bias against low-effort “shitposting” and posts which fan racial flames. It’s so frequent and well-known that it could become a tagline, “Hacker News: Please Don’t Do This Here”. At what level of curation of non-flamey posts does it become a publisher due to this bias?

[1]: https://www.eff.org/deeplinks/2020/12/publisher-or-platform-...

[nine_k]

They don't have the common carrier protections. That is, phone companies are not required to censor hate speech, and ISPs are not required to censor unlawful content that passes their pipes, because they are just, well, pipe, oblivious of the bytes they pass.

Platforms are in the business of making content available, so they are forced to control the availability, and censor unlawful content. They choose to censor any "objectionable content" along the way, without waiting for PR attacks or lawsuits. I can understand that.

(What is harder for me to understand is when these same platforms extoll the freedom of expression. I'd like them be more honest.)

[mytailorisrich]

Moderation is not upvoting/downvoting.

For example, when you moderate a debate you do not silence opinions you disagree with, you simply ensure that people express themselves within 'acceptable' boundaries, which usually means civility.

To me this means that 'decentralised content moderation' is largely an utopia: Whilst the rules may be defined by the community, letting everyone moderate will, in my view, always end up being similar to upvoting/downvoting which is a vote of agreement/disagreement.

[zozbot234]

On the other hand, decentralised filtering out of objectionable content might go hand-in-hand with replicating and thus preserving the most valuable content. Empirically, 90% of the content in most decentralized systems (e-mail, the Web etc.) is worthless spam that 99.99999% of users or more (a rather extreme majority if there ever was one) will never care about and could be eradicated with no issues whatsoever.