7 ms·

Given that Bluesky makes it this easy to download data, it's quite alarming that the graph of who blocks & mutes who is fully public and easy to extract into a

by btown 3y ago

Given that Bluesky makes it this easy to download data, it's quite alarming that the graph of who blocks & mutes who is fully public and easy to extract into a database: https://atproto.com/lexicons/app-bsky-graph#appbskygraphgetblocks https://atproto.com/lexicons/app-bsky-graph#appbskygraphgetb... On Twitter, blocking a toxic user does not notify them - while they can query the block status of one profile at a time, they can never get a full list of people who block them. But it would be trivial to create a Bluesky app view that provides this inverted index. And some people would be inclined to use the list of people who block them as a "target list" of people whose views differ from them, to share with their networks as prospects for targeted harassment that may even cross into real-life violence. (The fact that critics of the infamous Ki*f*ms forum have been swatted - and that I am even now reticent to type the full name - is just the tip of the iceberg of potential dangers here.) I hope that Bluesky comes up with a better mechanism here - it's tough to do in a federated system, but research like https://eprint.iacr.org/2021/1577.pdf https://eprint.iacr.org/2021/1577.pdf may be helpful.

capableweb 3y ago
It's a tough line to walk. On one hand, if you participate in a system where almost everything is inherently public (the web/internet, since anyone can screenshot anything and publish wherever they want), it's hardly unexpected that information that was once public, can remain public forever. On the other hand, people have some sort of expectation that the data they publicly post online to remain in some sort of semi-private state. Bluesky app might invite only for now, but the underlying protocol and technology makes everything very public, forever, and makes it trivial to cache locally (for good or bad purposes).
- btown 3y ago
  There's a vast difference between "public" and "actively surfaced." Indeed, blocking someone on Twitter would not prevent them from finding your public profile in an incognito window and seeing your posts - it would simply prevent your posts from being easily and automatically accessible to them in their feed. In practice, this tends to reduce conflict. My concern with Bluesky is that it makes it very possible for tools to make block information easily and automatically accessible - in fact, it would allow a bad actor to create a service that shows a feed of just content from people who want to block you from seeing it. That's a recipe for disaster.
  - detaro 3y ago
    I think that anyone can see who-blocked-who is the bigger issue, because that's usually not discoverable and allows a) a lot more large-scale analysis and b) when it comes to "celebrities" of various kinds, their "fanbase" is quite likely a bigger source of harassment etc overall.
  - capableweb 3y ago
    Yeah, you're right, there is a big difference. I think one of the changes they're trying to push through is that since you can't really block someone from viewing your content, if they want to, they're turning it around a bit. So don't think of blocks as "You won't be able to see my content" but more like "I won't have to see anything of you anymore, and you can't interact with me", which is more accurate on how blocks actually works, even if platforms tell you otherwise.
brundolf 3y ago
(Saying this as an excited bluesky user and someone who really really hopes all of this succeeds long-term) I think possibly their biggest challenge ahead will be making parts of it non-public. Private mutes/blocks, having some analogue of Circles for whitelisting post viewers, etc. Having a truly open and public database - especially once signups no longer require invite codes - is going to mean a cambrian explosion of tools and clients the likes of which we've never seen before in social media (which is already sorta happening even with closed signups). But that might include malicious apps that take advantage of that same openness to stalk, spam, and harass (especially given it's coinciding with a huge leap in AI technology). It might be AT Protocol's biggest test.
VancouverMan 3y ago
Maybe it'd be better to just not have any blocking/muting built into such a system at all. If User A doesn't want to see posts from User B, that's fine. User A can have his client filter them out locally, prior to when they'd otherwise be displayed. Nobody else has to know this is happening. I don't think that User A should be permitted to prevent User B from replying to User A's posts, which in turn prevents User C (or all other users) from discovering what User B thinks about whatever User A posted. User A trying to prevent his otherwise-public posts from being visible to User B seems pointless to me, as User B could log out, or use another account that hasn't been blocked, or ask somebody else who hasn't been blocked to screenshot it, or use some other way around it.
- zimpenfish 3y ago
  > User A can have his client filter them out locally We've been trying that with email and spam for the last 3 decades and I've got to tell you the results are not good. > prior to when they'd otherwise be displayed. A has to download the harassment from B just to then filter it locally? Seems like you're putting the onus on A here to do the work. What happens if B sends enough to DOS A's link? Or sends CSAM and now A is committing a crime just by downloading the content? > as User B could log out, or use another account that hasn't been blocked Which increases the friction on B if they want to harass A - this is good! We want to make harassment as friction-full as possible!
  - Gigachad 3y ago
    It works pretty well on email. When I mark a user as spam or set up a filter rule, all of their content goes direct to spam or even deleted if I want. Doesn't bother me at all. Email has also dealt with the spam issue mostly. If you blast me with too many emails, you get put on a spammer blacklist and servers will just drop the content without delivering it. It also isn't a crime to be unwillingly sent illegal content which you immediately delete.
    - mort96 3y ago
      Email has not solved the spam issue. I regularly get obvious spam email in my inbox, and legit mail still goes in the junk folder from time to time – with gmail. Plus, the automated spam filtering systems make it practically impossible to host your own e-mail, and any self-hosted email will have even worse issues with inbound spam than gmail does. Spam is at the core of what makes email a terrible example of a federated system, along with absolutely terrible client compatibility issues.
sennight 3y ago
[flagged]
ChrisMarshallNY 3y ago
I'm writing a fairly basic social media app, right now. The target demographic is very security/privacy-sensitive, so we have discussions about stuff like this, all the time. If you block someone, they have no idea, except that you disappear from their radar. Even that, only came about, after a long discussion. Before, blocking someone simply meant that we could not see them, but they could see us. My location is something that I can force off (meaning that no one can find me in location searches). Even when it is visible, it is "fuzzed," so that the location anyone (other than you) can see, is a random one, within a 10Km square. All visibility/security is enforced at the server. If possible, in SQL, so it never even makes it into the server code. Having API access buys you nothing.
- gpm 3y ago
  > Even when it is visible, it is "fuzzed," so that the location anyone (other than you) can see, is a random one, within a 10Km square. I hope the implementation of this isn't that I can start sampling your location repeatedly and average them out to find out where you live?
  - ChrisMarshallNY 3y ago
    Oh, sure, but it would take a while. The system is not Fort Knox. If someone wants to find one of the members that badly, then there are actually easier ways, outside of the system. If we decide that it is a concern, I already know how I'd deal with it. I would add a "governor," to slow the process down. Depending on how the location is updated, that could be enough to render it worthless. People can't even see the "fuzzed" locations, unless they are registered members of the community. The locator would have to go through a registered account, and those don't come too easily. However, if someone is that concerned, they would either hide their location entirely, or not join the community. The app is all about people finding each other.
    - eat_veggies 3y ago
      The standard solution for dating apps is to snap the location to the nearest x kilometer grid point, so that you are losing information rather than simply adding noise https://robertheaton.com/2018/07/09/how-tinder-keeps-your-location-a-bit-private/ https://robertheaton.com/2018/07/09/how-tinder-keeps-your-lo...
gfody 3y ago
isn't the whole point of making this a decentralized protocol to be able to have nuanced access policies that you can trust are actually enforced? if I'm participating in this protocol and someone requests my blocklist shouldn't that request get routed to my node which would respond according to my preferences which might be some choice like ignore them, answer them, lie to them, ask me, etc. depending if the request came from a contact, stranger, friend, foe, corporation, or whatever?
anigbrowl 3y ago
No, this is quite a good thing, although it will lead to new kinds of information warfare.
m3kw9 3y ago
The blocked user would know most of the time if they were active in viewing the account that blocked. It’s not a problem, or that they’d eventually know.
chillbill 3y ago
You’re being too dramatic about it… it’s not that big of a deal.
riedel 3y ago
Don't know how much it is wort, but also GDPR applies. So you need a legal basis for processing (at least data from EU citizens). That could well be journalism or scientific research, but for anything else it probably should be compatible with the expectation of the users. Fines have been typically rare and low for 'illegal' processing of twitter data. However, also blue sky and any other federating entity could face a legal risk if they do not protect that data reasonably against misuse.
oneeyedpigeon 3y ago
I think it's useful to have different networks with different rules - that way, we learn about what works and what doesn't. Twitter still has the behaviour you discuss here with people screenshotting pages showing someone blocked them, and publicising them. I'd like a social network that's much more public, to compare it's results with twitter, etc. I'm thinking non-anonymous identities and fully public records of votes, etc.

by btown 3y ago

capableweb 3y ago

It's a tough line to walk. On one hand, if you participate in a system where almost everything is inherently public (the web/internet, since anyone can screenshot anything and publish wherever they want), it's hardly unexpected that information that was once public, can remain public forever. On the other hand, people have some sort of expectation that the data they publicly post online to remain in some sort of semi-private state. Bluesky app might invite only for now, but the underlying protocol and technology makes everything very public, forever, and makes it trivial to cache locally (for good or bad purposes).

btown 3y ago
There's a vast difference between "public" and "actively surfaced." Indeed, blocking someone on Twitter would not prevent them from finding your public profile in an incognito window and seeing your posts - it would simply prevent your posts from being easily and automatically accessible to them in their feed. In practice, this tends to reduce conflict. My concern with Bluesky is that it makes it very possible for tools to make block information easily and automatically accessible - in fact, it would allow a bad actor to create a service that shows a feed of just content from people who want to block you from seeing it. That's a recipe for disaster.
- detaro 3y ago
  I think that anyone can see who-blocked-who is the bigger issue, because that's usually not discoverable and allows a) a lot more large-scale analysis and b) when it comes to "celebrities" of various kinds, their "fanbase" is quite likely a bigger source of harassment etc overall.
- capableweb 3y ago
  Yeah, you're right, there is a big difference. I think one of the changes they're trying to push through is that since you can't really block someone from viewing your content, if they want to, they're turning it around a bit. So don't think of blocks as "You won't be able to see my content" but more like "I won't have to see anything of you anymore, and you can't interact with me", which is more accurate on how blocks actually works, even if platforms tell you otherwise.

brundolf 3y ago

(Saying this as an excited bluesky user and someone who really really hopes all of this succeeds long-term) I think possibly their biggest challenge ahead will be making parts of it non-public. Private mutes/blocks, having some analogue of Circles for whitelisting post viewers, etc. Having a truly open and public database - especially once signups no longer require invite codes - is going to mean a cambrian explosion of tools and clients the likes of which we've never seen before in social media (which is already sorta happening even with closed signups). But that might include malicious apps that take advantage of that same openness to stalk, spam, and harass (especially given it's coinciding with a huge leap in AI technology). It might be AT Protocol's biggest test.

VancouverMan 3y ago

Maybe it'd be better to just not have any blocking/muting built into such a system at all. If User A doesn't want to see posts from User B, that's fine. User A can have his client filter them out locally, prior to when they'd otherwise be displayed. Nobody else has to know this is happening. I don't think that User A should be permitted to prevent User B from replying to User A's posts, which in turn prevents User C (or all other users) from discovering what User B thinks about whatever User A posted. User A trying to prevent his otherwise-public posts from being visible to User B seems pointless to me, as User B could log out, or use another account that hasn't been blocked, or ask somebody else who hasn't been blocked to screenshot it, or use some other way around it.

zimpenfish 3y ago
> User A can have his client filter them out locally We've been trying that with email and spam for the last 3 decades and I've got to tell you the results are not good. > prior to when they'd otherwise be displayed. A has to download the harassment from B just to then filter it locally? Seems like you're putting the onus on A here to do the work. What happens if B sends enough to DOS A's link? Or sends CSAM and now A is committing a crime just by downloading the content? > as User B could log out, or use another account that hasn't been blocked Which increases the friction on B if they want to harass A - this is good! We want to make harassment as friction-full as possible!
- Gigachad 3y ago
  It works pretty well on email. When I mark a user as spam or set up a filter rule, all of their content goes direct to spam or even deleted if I want. Doesn't bother me at all. Email has also dealt with the spam issue mostly. If you blast me with too many emails, you get put on a spammer blacklist and servers will just drop the content without delivering it. It also isn't a crime to be unwillingly sent illegal content which you immediately delete.
  - mort96 3y ago
    Email has not solved the spam issue. I regularly get obvious spam email in my inbox, and legit mail still goes in the junk folder from time to time – with gmail. Plus, the automated spam filtering systems make it practically impossible to host your own e-mail, and any self-hosted email will have even worse issues with inbound spam than gmail does. Spam is at the core of what makes email a terrible example of a federated system, along with absolutely terrible client compatibility issues.

sennight 3y ago

[flagged]

ChrisMarshallNY 3y ago

I'm writing a fairly basic social media app, right now. The target demographic is very security/privacy-sensitive, so we have discussions about stuff like this, all the time. If you block someone, they have no idea, except that you disappear from their radar. Even that, only came about, after a long discussion. Before, blocking someone simply meant that we could not see them, but they could see us. My location is something that I can force off (meaning that no one can find me in location searches). Even when it is visible, it is "fuzzed," so that the location anyone (other than you) can see, is a random one, within a 10Km square. All visibility/security is enforced at the server. If possible, in SQL, so it never even makes it into the server code. Having API access buys you nothing.

gpm 3y ago
> Even when it is visible, it is "fuzzed," so that the location anyone (other than you) can see, is a random one, within a 10Km square. I hope the implementation of this isn't that I can start sampling your location repeatedly and average them out to find out where you live?
- ChrisMarshallNY 3y ago
  Oh, sure, but it would take a while. The system is not Fort Knox. If someone wants to find one of the members that badly, then there are actually easier ways, outside of the system. If we decide that it is a concern, I already know how I'd deal with it. I would add a "governor," to slow the process down. Depending on how the location is updated, that could be enough to render it worthless. People can't even see the "fuzzed" locations, unless they are registered members of the community. The locator would have to go through a registered account, and those don't come too easily. However, if someone is that concerned, they would either hide their location entirely, or not join the community. The app is all about people finding each other.
  - eat_veggies 3y ago
    The standard solution for dating apps is to snap the location to the nearest x kilometer grid point, so that you are losing information rather than simply adding noise https://robertheaton.com/2018/07/09/how-tinder-keeps-your-location-a-bit-private/ https://robertheaton.com/2018/07/09/how-tinder-keeps-your-lo...

gfody 3y ago

isn't the whole point of making this a decentralized protocol to be able to have nuanced access policies that you can trust are actually enforced? if I'm participating in this protocol and someone requests my blocklist shouldn't that request get routed to my node which would respond according to my preferences which might be some choice like ignore them, answer them, lie to them, ask me, etc. depending if the request came from a contact, stranger, friend, foe, corporation, or whatever?

anigbrowl 3y ago

No, this is quite a good thing, although it will lead to new kinds of information warfare.

m3kw9 3y ago

The blocked user would know most of the time if they were active in viewing the account that blocked. It’s not a problem, or that they’d eventually know.

chillbill 3y ago

You’re being too dramatic about it… it’s not that big of a deal.

riedel 3y ago

Don't know how much it is wort, but also GDPR applies. So you need a legal basis for processing (at least data from EU citizens). That could well be journalism or scientific research, but for anything else it probably should be compatible with the expectation of the users. Fines have been typically rare and low for 'illegal' processing of twitter data. However, also blue sky and any other federating entity could face a legal risk if they do not protect that data reasonably against misuse.

oneeyedpigeon 3y ago

I think it's useful to have different networks with different rules - that way, we learn about what works and what doesn't. Twitter still has the behaviour you discuss here with people screenshotting pages showing someone blocked them, and publicising them. I'd like a social network that's much more public, to compare it's results with twitter, etc. I'm thinking non-anonymous identities and fully public records of votes, etc.