Hacker News Reader

Comment by 6gvONxR4sf7o | Hacker News Reader

> ... have been removed from Facebook with no explanation given

At the very least, this should be unacceptable. You shouldn't get banned for something if you aren't told what you did. If it was for use of the word revolutionary, it should say so. Or if they called for some leader's heads, it should say so. Then they could avoid doing the same in the future and we could all agree with or criticize the rationale.

As it is, it's likely similar to "a big neural network said so" which tells the users nothing. Facebook's free to run themselves this way but it's sure dickish. Imagine if the government or your boss made rules that way. "Drivers will be ticketed if our models say so."

View on HN

One of the problems with this is opening up to criticism, liability, and further argumentation.

If you say "We removed them because of their political views," then you get criticized for being openly partisan.

If you say "We removed them because their comments were inciting violence," you'll get arguments about how their comments didn't actually tell people to do violent things (for example, in the US I'm sure for the near future, a lot of people will be arguing very loudly whether the specific phrases "fight like hell" and "trial by combat" are metaphorical rhetorical devices or literal instructions to take violent actions.)

People will start finding and pointing out situations where you made different decisions in near-identical circumstances, and criticize you for that too.

It's better for users if the platform's transparent.

But the platform has no incentive to do so. Less than 1% of the time will users say "Okay you told us why this person was banned, we now see the ban's fair, and we'll shut up and go home." Instead, 99% of the time they'll turn your explanation into another reason to be enraged at you, or argue you need to reconsider, or even sue you because that line of reasoning makes you liable.

From the platform's point of view, it's usually better to let the reasoning behind a ban be a question mark.

> "a big neural network said so"

This is part of it too. With ML-driven bans or other actions, it's possible that nobody understands what criteria it uses to make decisions.

> One of the problems with this is opening up to criticism, liability, and further argumentation.

That's not a problem; it's the point.

> It's better for users if the platform's transparent.

Exactly.

> or even sue you because that line of reasoning makes you liable.

Also the point. Corporations are not people, and their reasoning should be assumed liability-worthy unless shown otherwise.

It also leads bad actors to game the system.

That's unfortunate. Not the victim's problem, though.

"bad people might behave better to we can't ban them easily" is a pretty poor argument. Zuckerberg has $50,000,000,000; he can afford to hire some people to do the job.

To be scrupulously fair, I think the concern is more "bad people might behave just as bad, but in ways that technically don't fit whatever criteria we banned the last group for". Or, less directly, "bad people might find good people who behave in ways that technically do fit those criteria, and try to get them banned".

Looking at their financials, it looks like they’ve massively increased headcount to battle these “integrity” problems. It at least looks like they’ve invested a ton of money. From the outside, I am not sure it’s been effective, but to be fair, have there been as many issues since they started investing significantly around 2018? Most of the major problems look like it happened before. (Not to excuse that time, but if they are fixing issues then it changes the future projections while still not letting them off the hook for past problems)

Bad actors don't need to game the system, they either:

* Get their government cronies/buddies to skew the system in their favor.

* Use their business relationship with Facebook to skew the system in their favor.

* Own the system from the get-go.

There are many worse people in the world than this. I should have been more clear in saying malicious and maliciously motivated actors.

> I should have been more clear in saying malicious and maliciously motivated actors.

The actors einpoklum refers to are malicious and maliciously motivated. Do you perhaps mean unsubtle in their malice and lacking in significant public credulity? (Eg, white supremacists.) Because while there are plenty of other bad actors, any 'solution' that doesn't address the bad actors einpoklum describes... isn't.

One of the problems with this is opening up to criticism, liability, and further argumentation.

Too bad. They must do it anyway. None of those are valid excuses for removing all accountability from the moderation process.

> Instead, 99% of the time they'll turn your explanation into another reason to be enraged at you, or argue you need to reconsider, or even sue you because that line of reasoning makes you liable.

Social networks seem to optimize for serving content that engages, and that typically means news with a negative sentiment. So, giving a ban with a reason would engage people for longer, to serve them more ads, and make the audience more profitable.

> Instead, 99% of the time they'll turn your explanation into another reason to be enraged at you, or argue you need to reconsider, or even sue you because that line of reasoning makes you liable. //

Which is 'fine', let them be enraged. Transparency shows it to be fair and removes the ability to reasonably argue otherwise. Then at least one can decide if the platform is right for you, and indeed of the rules are anti-democratic we can move for changing them.

'I'm so enraged, $platform said that inciting violence wasn't allowed, then I incited violence and got censored', whose going to rally around that flag?

It's Facebook's platform, they can remove anyone they want without explanation or transparency. The sooner people realise this, the sooner people will stop using Facebook - which in my opinion is a good thing.

The SWP could instead have their own website (which they do) and the ability to subscribe to their newsletters via email (they may have this already). Further to this they could setup their own forum on their own website, again they may already have this). This involves far fewer third parties to shut them down.

Why anyone would rely on Facebook when there are so many other options is just a sign of laziness or ignorance.

'Why would anyone go to parliament to talk, there are so many other buildings to talk in'.

I don't believe you are really that clueless about online social networks.

Is is kinda hilarious that the SWP is reliant on the granddaddy of surveillance capitalism.

Obviously they feel the need to go where the eyeballs are, where they can target ads to the fertile soil of ignorance.

Many claim that FB can do what they want. Not so, they must respond to legislation wherever they operate (and make money). Just like Rupert they have huge influence, but they can be brought to heel if there is public will to do so.

We should create a brightline, even if approximate and requiring later adjustment, to determine which companies are too big and must be governed by a different set of rules. Normally we would consider it unremarkable if a restaurant quits a customer, or if a law firm refused a client altogether.

On the other hand we consider it disaster if Amazon quits you as a vendor, or if your only local hospital refuses you. Perhaps any firm handling more than $1M yearly revenue shouldn’t be able to randomly quit customers.

I strongly disagree. Government control should be as minimal as possible, otherwise it helps a pitfall of potential tyranny. The people need to "vote" by using platforms that they respect the governance of. If they don't like how/who the leadership of Facebook is moderating, they should go to another platform; there does need to be better data and network portability laws/policy, so users are mobile and can leave any platform easily. This still doesn't solve the filter bubbles that get created in digital communities.

Any sufficiently-dominant corporation is indistinguishable from a government.

Then perhaps we should address these too big corporations issues. It appears, if no big social media corporation existed, and instead there were a bunch of smaller players, we could argue that the customers will make a good decision.

Perhaps, the government should be involved into it (in form of perhaps adjusting anti-trust laws)?

Citations needed. Strong statement with zero points argued to back it up.

Edit to add: downvotes don't count as argument points.

No, they don't, but that's what you get around here. If there's a bigger hive of intellectual douchebaggery in this sector of the galaxy, I haven't found it yet. Still, the good usually outweighs the bad by far.

To be fair to my downvoters, it would be more accurate to say, "Any sufficiently dominant corporation is indistinguishable from a monarchy." With the US government, at least, you have the right to vote, along with some constitutional protections including various avenues of appeal. When Google terminates your account for no good reason, on the other hand, there are fewer options. There's not much you can do but post a cri de coeur on Twitter and hope HN or some other intercessory spirit notices it and calls the King's attention to your plight.

I get what you're saying, however your "vote" is by not using Google - I understand that's easier said than done when they're conveniently the best search engine, etc. Laws are certainly behind as well, there should be due process since a single internet account at a company can be a critical fabric to our life.

Exactly, I'm pretty curious what they said? Were they being dishonest? Hate speech? threatening violence?

Sadly par for the course. Games do this to avoid bots improving against detection. Dating apps do it for safety. This is certainly more egregious because it’s about basic speech but by no means is the lack of disclosure new.

And it makes sense. It's like SEO. It's an adversarial environment without a perfect answer, so keeping the reasons hidden truly does help most everyone. It just sucks that it leaves the false positives hanging out to dry with no explanation and no recourse. And this has all gotten to a scale where those false positives are a huge number of people, even if it's a minuscule fraction.

> "Drivers will be ticketed if our models say so."

Safe driving is more than just obeying road signs. You might follow all the rules exactly as they're written in the book but still be a reckless driver. As you point out feedback is the key.

I've driven a car in countries where the traffic officers place obscure road signs to trick drivers into speeding so they can make a quick buck on a bribe.

Sometimes you might break laws without being aware of them. A model could suggest you do something differently before you cross the line?

Reply on news.ycombinator.com