Do you want to see spam and hate speech off the Fediverse? Think there must be a better way than continuing to click Mute and Block all the time?

On this episode of Libre Lounge, Chris and Serge talk about the work they're doing to protect the Fediverse from Spam, Scams and Online Harasment.

@librelounge I haven't listened to the episode, yet, but this has been bothering me for a while, since the mute/block approach means that there's a world of difference between the community established members see and the trash newcomers are going to see, which isn't exactly the best foot forward.

@librelounge Now listened, it seems like a good start, but avoids the broadcast nature of networks: Hate speech does damage without a vulnerable person receiving it.
I once looked at using the Scuttlebutt network as a proxy for appreciation and recursively collect/scale follows/blocks to guess who's worth following or blocking.
I like the premise because it uses real priorities, doesn't require a new system layer, and newcomers can use it. Is something similar feasible for ActivityPub?

@jcolag This account is actually the account for @emacsen and @cwebber who may have different opinions.

@jcolag @librelounge

We can't stop someone from emailing someone else, or sending them a text message. Similarly what two people do between their private instances is beyond the scope of what we can address.

You said that you'd like to see Secure Scuttlebutt networks here, but I don't see what Scuttlebutt could add in this scenario. Public SSB posts are public.

Can you elaborate on what you think SSB brings to the table around the area of abuse/hate speech?

@emacsen @librelounge I probably edited that comment beyond recognition. I meant that I looked at "recursive follows and blocks" in the CONTEXT of SSB, not that I necessarily prefer it.
What I was asking about was the feasibility of using follow/block/like information for each user to calculate what public posts are worth reading.
That is, if I follow A and A blocks B and follows C, there's a better chance of my wanting to block B than C. C's behavior would also be interesting, but less so.

@emacsen @librelounge Digging into the applications of that kind of recursive calculation, if many people, especially people you like, block an account, you can detect that those targets are maybe not worth listening to, spammers, creeps, or just low-quality accounts. No guarantee, because you might like a whole bunch of fascists and bots, but a decent guess for most people.

@jcolag @librelounge I agree with "hate speech does damage without a vulnerable person receiving it." Maybe better phrasing is probably "reducing abuse" (particularly directed abuse) rather than "stopping hate speech", the latter of which is a claim nobody can fully prevent and we can't either.

It does provide a system by which people performing hate speech may be stuck on their own islands, at least; not perfect, better than what we have now though by a wide margin.

@jcolag @librelounge
Can't a spammer grind "stamps" before spamming? It might be hard to grind stamps without revealing your identity, even the way you type could reveal your identity.

Anyways, the idea seems better as the current one. We will see what it becomes, but I currently support the idea (my opinion can still change though).

@librelounge @rmw @jcolag @emacsen Stamps are just *one* mechanism of adding "friction" to interactions you don't have an established relationship with.

That said, yes, you could grind stamps. Grinding stamps still means that you're wasting resources to send unwanted messages, which is better than present, which is free. (Plus, you can *still combine* with present mechanisms; we aren't throwing anything away.)


@librelounge @rmw @jcolag @emacsen But, grinding stamps may turn out to be a waste of your resources... it depends on how you get them.


- If they're tradeable, then they're more valuable, and you're literally wasting money (and enriching the person you're trying to antagonize) to send them.
- Alternately, there's the solve-a-puzzle mechanism to get stamps. This mechanism doesn't mean the person gets enriched, but it does mean that grinding is more worthless

@librelounge @rmw @jcolag @emacsen For the solve-a-puzzle mechanism, consider if it worked like this (what I call "ephemeral postage stamps" or "ghostage stamps"):

- Every user can specify what stamps they accept
- By default, every user's instance runs a "postage stamp dispensor"
- You can get stamps by yes, grinding against some puzzles to get them
- But they expire after say, a month
- What if you hoard a lot to unleash a spam wave? User can "restart" postage dispenser, they're worthless now

@cwebber @librelounge @rmw @emacsen Three questions, here:
1/ How does this work for public areas, flooding the timeline with spam or hate?
2/ Would anyone really accept a micropayment system? Seems like we've been promised them forever and nothing has worked.
3/ If widely adopted, what prevents centralized spam-houses from going wherever there's cheap power and burning heaps of fuel, like cryptocurrency mining already does? And would that solution disrupt real messages?

@jcolag @cwebber

1. Public timeline is addressed in other parts of my paper, specifically MultiBox. I think Chris took that proposal and put it into thiers too. Did you happen to read the whitepapers?

2. We never mentioned micropayments. I think money here is a potentially bad idea.

3. If stamps are limited in time and value and can be nullified, it's not going to stop them, but it will make it a pointless excercise.

@emacsen @cwebber Do you have a link to the whitepapers? I haven't seen them.
And certainly excuse the term "micropayment" if it doesn't fit, and maybe it's a lack of vision on my part, but I can't see an approach to electronic stamps that isn't a micropayment service, especially where the recipient gets some benefit.

@jcolag @cwebber The whitepapers are linked in the show notes, from the original post:

I'll reply to your stamps questions in a second reply (but I think the last one I'll do on this thread)

@jcolag @cwebber @librelounge

There's some confusion, maybe because of my statment about "getting paid" for spam. That was a bit tongue in cheek. The only benefit anyone recieved in buying a stamp is message delivery. When you send me a letter or postcard, the stamp you pay is a delivery fee.

Stamps in our proposal are just delivery fees, paid by a sender. The difference is that the recipient (and or server operator) gets them, and not some third party (ala the post office).


@jcolag @cwebber @librelounge

Stamps are an abstract concept that stand in for any number of things. That could be Hashcash (which we mention explicitly in the whitepaper), or it could be some theoretical work you have to do ala Folding@Home, or maybe it really is some kind of currency payment.

But a Stamp itself is not a form of currency, as its value and longevity are set by the recipient on both the dimensions of work and its relative value.


@jcolag @cwebber @librelounge If you (the recipient) say that a Stamp is worth one CPU minute, you can then turn around and say it's worth two CPU minutes.

You can also say that sending a message not costs 3 stamps, where it previously only cost 1.

You can also say that a stamp expires in a month. Or you can say "All stamps that existed before right now are null and void".


@jcolag @cwebber @librelounge

I could hoard/collect stamps, but I can only use them to send you messages so long as you say so.

As for trading, why would someone pay me for something that you can instantly declare has no value?

You *could* ask for money in exchange for Stamps on your own site in theory. We didn't specify a mechanism for that, but it's possible.

But going back in the other direction (stamps for money), I just don't see that happening.


@emacsen @cwebber @librelounge I hope I'm not just arguing semantics, but at least to me, that sounds like a micropayment system, but with features that make communication unreliable, but (based on how e-mail seems to have gone) still gives an advantage to spambots that mass-send garbage for a queue of clients as soon as tokens are generated.
I'll leave it at that until I've gotten through the whitepapers, since it's likely I don't understand the use case or am hung up on semantics.


Yes, a spammer could generate stamps beforehand, but it's not in their best interest to do so, for a few reasons:

1. The entire idea of a stamp is that it's a cost. It costs you something to get one. That might be CPU time, or money, or something. If the cost you expend is higher than your percieved benefit, you're losing.


@rmw @jcolag

2. One of the ways to address this problem is simply to have stamps expire. Maybe a stamp is only good for a month. If stamps are returned, that's a good deal (and addresses environmental concerns) but going back to make more might be a pain.

3. One might decide that the cost of a stamp changes over time, or that the cost of sending changes. Maybe on Day 1, a message costs 1 stamp, but if there's a storm of stuff, sending a message might cost 4 stamps.


@rmw @jcolag

Ultimately Stamps are part of an overall design. Stamps are just one component. They don't guarantee that the recipeint sees the message.

For spammers and scammers, this makes the prospect of sending spam expensive. If the cost of spamming is so high that the entire venture becomes unprofitable, it'll stop.

For harassers, I don't think that *only* Stamps will be enough. We'll need to pair it with other tools, but like OcapPub, they can work together nicely.

Sign in to participate in the conversation

For people who care about, support, or build Free, Libre, and Open Source Software (FLOSS).