Hacker News Reader

Comment by voidmain | Hacker News Reader

I don't think the problem is federation. The problem is media where participants have nothing at stake, so that abusers can't be made to pay the costs of their abuse. Changing this isn't incompatible with anonymity - for example you could require anonymous participants to post bonds in a cryptocurrency. But e-mail spam, for example, would never have been a problem if the recipient of an e-mail could push a button and destroy $1 of the sender's money. Getting the incentives right for a social network is undoubtedly harder, but you can't even start until people have something to lose.

View on HN

Yeah, Bill Gates proposed that solution to spam. Another one is having people register with proven credentials of their legal name and physical address.

Yet Hacker News isn't using these.

Because shadow banning, for example, works. Having someone's voice lost while they don't know about it (keep living in their bubble) works.

The lesson here: don't assume that your solution is the only solution.

Usenet, e-mail, and forums have taught us already that the solution to this problem is advanced, silent filtering (which have been innovated in the anti-spam arms race [1]) essentially akin to shadow banning. The enforcement of law also tends to help.

[1] RBLs, Bayesian filtering, and the already mentioned shadow banning

Shadow banning doesn't work against any adversary that's more sophisticated than an annoying person on the internet. As you observe, email spam filters are basically just "shadow bans" and spammers aren't confused by them at all, they just send themselves test messages.

Spam filters work for lots of reasons but not because spammers can't figure out if they're caught or not.

> Because shadow banning, for example, works. Having someone's voice lost while they don't know about it (keep living in their bubble) works.

Shadow banning is evil and intransparent. It may work, but at what cost?

Why evil? Or more specifically, why any more evil than any other form of punishment with regards to online communication? If one bans a user as opposed to just shadow ban, the end result is the user either gives up or just makes another account. In rare instances they reform, but usually it's not a reform, it's a change in methodology to avoid whatever condition got them banned in the first place. This isn't rehabilitation, it's learning not to get caught.

In either case, straight up ban or Shadow ban, someone is silenced as a punishment. In either case, the user likely isn't learning their lesson. In either case, the tools are ripe for abuse.

What is the exceptional cost of shadow bans besides them being opaque?

It is interesting for me that shadow banning can be evil (or not). Shadow banning is comparable to who hears bad words and ignores them, replying only with instinctive filler replies ("yeah", "hmm", "so?" and so on). Of course as a moderation measure shadow banning is a lot broader and more systematic than filler replies, but unless moderators do not disclose that shadow banning is in effect (without disclosing subjects), it doesn't seem to be a subject of moral judgement or whatever because any user can always choose to ignore you exactly in that fashion.

I don't think that's a fair comparison; the one choosing to shadow ban another isn't simply choosing to ignore the target, they are choosing to disallow others from hearing what the target has to say, and in a way that does not allow the target an opportunity to change and improve themselves.

E2E silent ignore does not work in online conversations where multiple people participate ("online group communication") because it meddles with the context since others do not participate. Try it for yourself: join an active IRC channel and ignore the first few people who chat. You won't be able to understand the conversation anymore. Spam filters work on 1:1 communication, targeted specifically at you. Shadow bans work on 1:1 communication and group communication if they're centralised or server-side. Its more akin to (SMTP) tarpits [1].

[1] https://en.wikipedia.org/wiki/Tarpit_(networking)

I've been using the IRC method since the early 90s; it works fine. Sure, there's some appearance of discontinuity but that's a feature, and allows one to personally guage whether the choice to exclude was a good one, without having to be exposed to the comments of the one ignored.

Shadowbans aren't akin to SMTP tarpits insofar as the target of force is less likely to be a bot. Behind the silence is a human being who is trying to engage others.

I've been using IRC since the early 90s as well, and it doesn't work "fine" when you ignore active members of a community who get quoted and such.

SMTP tarpits are like shadow bans because the perpetrator isn't informed about the ban/tarpit, and its meant to slow them down by wasting their time in their bubble that they are getting "work" done.

If they're active members of the community that are regularly quoted then banning them is unlikely to be the best course of action; I prefer to publicly announce that I'm ignoring someone, and perhaps why, before I do. Banning active and valuable community members should, at least, be public and transparent, lest the community react negatively.

SMTP tarpits are more likely to involve a perpetrator who is operating a bot, and less likely a perpetrator who is personally engaged in the conversation and community.

> they are choosing to disallow others from hearing what the target has to say

HN allows you to turn on shadowbanned posts.

In the several years I've been here, I've seen maybe 3 or 4 shadow banned posts that were of value. In those cases, other users replied to the shadow banned user and told them to message the mods about getting the shadow ban lifted.

Heavy moderation of the tone of message works. In a similar vein, the neutral politics sub on reddit is successful because they ban people who are rude.

I think that if we moderated tone more often, that a lot of the toxic communities wouldn't even form.

Others have pointed out that shadowbanning doesn't really work against sophisticated and coordinated adversaries, and that HN has probably never been attacked by an army. But there's also another problem: someone has to pay for the decision, and the system to make sure that the decision is made fairly, that moderation powers aren't abused, etc. HN is subsidized by YC, and it's not very important in the scheme of things so the injustice of a few bad decisions about bans is not the world's most pressing problem. As networks scale up, it becomes much more expensive to moderate them and much more important to get it right. If abusers of a system pay for their abuse, they can also pay for careful and fair enforcement.

If most people's willingness to pay for non-abusive communications media is $0, then... I expect they will continue to get abuse.

Every solution needs to be robust in the presence of an adversary who is more technically savvy, motivated, organised, numerous and dedicated than an average user. That's why it's really hard. Any tool you give an individual to protect themselves can be turned against them by an army.

That's why white/blacklisting and similar measures will never work. What you need is proper delegation with an incrementally built web of trust.

Instead of your email address being a human-readable public address, it should be a revokable cryptographically unguessable address that's unique to each person to whom you're introduced. Introductions happen via a protocol which generates a new address given to that person, and you then assign a purely local pet name you use to refer to them [1].

Given any generated address, you can trace back the history of introductions that led to it.

This very clearly solves spam problems since it completely upends the traditional approach with a bottom-up solution. I'm not sure anything else could be immune to abuse to the same extent though.

[1] https://en.wikipedia.org/wiki/Petname

I think distributing the work of verifying new accounts as non-spammers to over all existing users is the only way a social network can grow without losing to the spammers or overloading their human moderators, so I like your idea.

But how do you solve the problem of people wanting to join who have no existing contacts in the network to invite them? Most likely they would use some other channel to request an invitation from a stranger and hope that they'll grant them access.

If that happens, then it dilutes the power of the web of trust. A spammer could manually create a bunch of accounts in the same way, and then use them to grow botnets at the edges of the web of trust. If only a few bots are used to spread spam, it would take a while to accurately determine which part of the network is controlled by the spammer.

But spamming has no power in this system. You can collect a huge list of addresses, but as soon as you abuse them once they would be revoked. One-time spamming isn't a viable business strategy.

The bootstrap problem is indeed challenging with this approach. Normally you could connect with people you know in real life using an NFC or QR code exchange, but social networks have created an expectation of being able to find people by searching an open directory. That's a spammers dream. As soon as you can send a friend request that's just reintroducing open addressing.

I haven't spent much time thinking about this since I rarely use social media, but my first thought is that you voluntarily join some groups and find people that way, ie. High school groups, etc. Perhaps friend recommendations by randomly asking existing members could be used to very the validity of newcomers.

> I'm not sure anything else could be immune to abuse to the same extent though.

There are multiple solutions that work if you design the system that way from the outset (which is the same issue yours has). One was suggested above. You can have a human-readable public address but the first time a message is sent the sender has to commit $1 (or however much is necessary as a deterrent). If the recipient approves the message you get it back right away, if they ignore it you get it back in 30 days, if they mark you as spam you lose the money (and you know this and know that sending them more messages will be expensive).

Then if you send a message (including a reply) to someone it automatically (but revokably) whitelists them, and whitelisted peers don't have to commit money to send to you.

This also works for mailing lists, because then you subscribe by sending a request message, which whitelists the mailing list, and then they don't have to hold a huge reserve as long as they only send messages to people who actually asked for them.

But none of that works for email because it wasn't implemented that way originally and it's hard to change it now.

> You can have a human-readable public address but the first time a message is sent the sender has to commit $1 (or however much is necessary as a deterrent).

Requiring payments destroys the anonymity though, that's why I don't like it.

> But none of that works for email because it wasn't implemented that way originally and it's hard to change it now.

Actually, my system can be adapted to email but you have to give up typing in email addresses manually. Most people don't do that anyway, so no big loss.

> Requiring payments destroys the anonymity though, that's why I don't like it.

Isn't that what all of this Bitcoin and Ethereum is supposed to be good for?

> Actually, my system can be adapted to email but you have to give up typing in email addresses manually. Most people don't do that anyway, so no big loss.

Don't they? People put email addresses on business cards and billboards and things like that. That's where you would need people to upgrade -- the fix would be to use QR codes but then you need everybody to support that (and understand what to do with it).

Of course, people who have their own domains have long been doing something similar but with human readable names. You have example.com so you can create foo@example.com, bar@example.com, a hundred names for a hundred services.

It would also be possible to do this with subdomains. So if you were bob@gmail.com, gmail could theoretically let you create aliases like foo@bob.gmail.com and use it for your business with Foo Corp.

They do let you do something similar by ignoring dots in the address, so john.smith@gmail.com is automatically an alias for johnsmith@gmail.com, but if you use dots in different places for different activities then when one of them starts getting spammed you discontinue it and add a rule to trash everything sent to that alias. The inconvenience there is it's harder to remember what you used each variation for (and which ones you've already used) when it's time to rotate.

The quantity of physical spam I get through my letterbox suggests charging doesn't cure spam -- although I admit it reduces it.

Metafilter charges a one-time $5 fee to post. It also has extensive human moderation. It’s worked well for them but they’ve never tried to scale up.

By some informal measures, Metafilter has stupidly high S/N ratios.

For those interested, see:

https://old.reddit.com/r/dredmorbius/comments/3hp41w/trackin...

Yeah, that's certainly one way to improve the signal/noise ratio of a social networking site, but it simply can't scale to the size of say, Facebook or Twitter. Human moderation is simply too slow and expensive to be done on an 'industrial' scale, and charging money will always significantly limit a site or service's audience.

Then again, perhaps the unfortunate truth is that something on the scale of today's mainstream social networks is simply unworkable. Perhaps at a certain size a community site simply falls apart, and nothing short of human level AI will be able to keep it intact.

Immunity, impunity, and ignorance are superpowers enabling some good, and far more harm.

Costs do not have to entail financial transactions, though that's one option. Devising a hard-to-establish reputation, which can be lost on signs of clear anti-social behaviour, is one approach. Accounting for the inevitability of Sybil attacks is a challenge.

It's not so much that the problem is federation than that the solution is quite clearly not federation alone.

Tools, policies, values, coordination & cooperation, and the syones to create, specify, and use these, are necessary.

I was originally thinking that putting this in monetary terms would cause the rich to have an inordinate advantage for free speech. Then I realized that the larger their speech reach, maybe the larger their financial penalty (taking the example of spam). But then in that case, it still simply becomes a case of who can afford it. It's like defending against a DDOS attack, it depends on who has the bigger pipe.

Agree that getting the mechanics for social networks would be hard. Not sure if financial mechanisms are the answer (not saying you're saying that). It puts a price on speech. There has to be a better way.

I'm not sure how that makes sense. Wouldn't that just mean that trolls could attack targets financially in addition to silencing them?

Presumably it would still require manual human moderation to ensure that a slew of reports isn't a coordinated attack.

But also, presumably fewer trolls would exist if you had to put up a "$x" bond before signing up in the first place.

So fewer trolls, but the ones who remain will be trolling the system and bankrupting people. Perfect.

Is my spam filter allowed to push that button?

Reply on news.ycombinator.com