The current release candidate will indicate to blocked people that they are blocked while viewing a profile #10433
Comments
|
yeah this "feature" is pretty unacceptable, and designed solely for the benefit of bad actors in the fediverse. |
|
I agree that this "feature" has adverse effects, it doesn't solve or help anyone. |
|
I can see why it was added (reducing confusion means fewer non-bug bug reports for the developers to deal with), but I do agree that telling people when they've been blocked isn't a good idea. I would feel safer if it would just ambiguously error out. I agree that this feature will help blockees and hurt blockers. Edit: Here is the PR: #10420 It's a change to a sensitive bit of the system, so at the time I was concerned that there were no screenshots or anything. |
|
I am really not looking forward to people dunking on other people with screenshots showing that they got blocked, like they do on Twitter "Lol I did this terrible thing and lol they blocked me" [screenshot] |
|
the fact that some profiles don't load and some do is impossible to be
ambiguous about. the only time mastodon will refuse to show you someone's
posts is when you're blocked.
…On Sun, Mar 31, 2019, 10:26 AM Cassolotl ***@***.***> wrote:
I can see why it was added (reducing confusion means fewer non-bug bug
reports for the developers to deal with), but I do agree that telling
people when they've been blocked isn't a good idea. I would feel safer if
it would just ambiguously error out.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#10433 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAORV6LvELvcKCJ0hPUfVBs2WHG08g4Gks5vcMWngaJpZM4cUTrE>
.
|
|
Yeah but we don't have to give the trolls a screenshot |
|
Please revert this change, it doesn't help anyone that we should want to help. |
|
'It isn't perfect' isn't a reason to make it worse. |
A blank profile is still pretty ambiguous as to whether it's a block or if their posts were never there. A harasser can't easily get their kek on if all they have to show for it is an empty profile. A better choice would have been to prevent blockers from showing up in the blockee's search results or the profile column altogether. You know. Completely vanish from their end. Because that's the goddamn point of blocking: the blocker doesn't want to be seen by the blockee, full stop. |
|
I understand the issue this addresses, but another solution should be implemented as this is a huge help for harassers and as I see it will bring bad behavior we know from Twitter |
|
Software shouldn’t lie to its users, and being blocked is important information. This allows users to reflect that they may have done something wrong and/or remind them that further correspondence is not wanted. Besides: There will always be pieces of information that give it away anyhow. |
Plot twist. They never do. Twitter has proven this time and time again. |
|
Adding my support to this. "YOU ARE BLOCKED" is like handing trolls a trophy. |
|
It's already pretty obvious if you're blocked, so this barely changes anything. |
|
I mean a 403 forbidden when trying to follow someone only happens when you are blocked, so it is in essence a block message already. If anything blocking someone should make your profile invisible to theirs and vice versa. |
|
The above arguments are why I initially designed the block feature in a way that would be impossible for the blocked person to tell they have been blocked. But the community insisted on a way to stop the blocked person from performing certain actions, and at that point the cat is out of the bag. If someone wants to favorite or follow and they can't, it's quite obvious what happened when they get a "403 not permitted" error. But not obvious enough that I don't regularly get bug reports from people and have to tell them "you have been blocked". You can't really hide that. That's why i personally only use the mute feature, for example, I don't want anyone to make a big deal about me not wanting to see them. This change had the added benefit of hiding follower/following lists as well as posts from the web UI. Before: blue follow button that says "not permitted" when you click it After: greyed out follow button and "you are blocked" clarified before clicking |
|
I agree with Gargron here, it's already possible to tell if you're blocked, now it's just a bit more intuitive. |
|
This should be opt-in on the blocker side. I don't know that a block indicator was part of what turned Twitter into hellsite, but I do know this was part of the debris falling over the cliff with it. Maybe opted in blockers could provide a reason and possibly an alternate form of contact. Edit regarding the UX for trying to figure out why you get the error: Shadowbans were invented for this purpose. Twitter goes too far with it. I'm currently in the box for...who knows. But a block could just make it look like the blocker hasn't tooted since the block. The blockee sees a profile frozen in time, possibly with a fake bio provided specifically for them. |
The point of a safety feature is to make it harder to get around, if not completely impossible. You know, like how any lock can be inevitably defeated, but that doesn't mean you should remove all the bloody doors. >.> |
|
So making Mastodon "more intuitive" for harassers and abusers is a goal now? |
|
One of the "stop the blocked person from performing certain actions" were types of harassment. I remember this one because people i had blocked were still able to follow me, boost my posts and expose them to all of their terrible followers. Which just brought in more harassment. A 403 message already tells them they are blocked. While mutes dont directly inform a user something has happened, its not hard to figure out when it happens. Knowing you are blocked i dont think has a lot of hard incentives behind any particular behavior. Sure some people might post some screenshots that they were blocked but most will not. I have thousands and thousands of people blocked on twitter and thats large enough for a good sample size. Very few make a big deal about being blocked. The rest i never see or hear from again. Theres the occasional determined harasser but when you have a determined harasser you need stronger tools than blocking anyways. |
|
403 is only used to indicate a block, so maybe use a different error code which means other things too. GitHub is purposeful in using 404 fir permissions errors as well because they feel that obscuring the accuracy of the error helps to reduce the attack surface for people trying to guess at things; maybe Mastodon could serve up a 500 (because instances go down) or 404 (yay troll you won now you can’t see them anymore) or something. Yeah it sucks to have the wrong status code but what sucks even more is a culture of abuse and harassment. Most status codes are to benefit humans; computers only really care about 200, 30x, and everything else. |
|
Smart idiot: Smart enough to find out he's been blocked without explicitly telling him. But most likely he's also smart enough to refrain from killing me. Stupid idiot: Doesn't understand all these 403s and 404s and most likely moves on to next victim due to boredom. But if he would know that he's been blocked because there's an in-his-face red sign, he most likely would come and kill me. Because he's a stupid idiot. |
|
403 errors only appearing when you are blocked isn't a reason to make it easier for people to tell if they are blocked, it is a reason to change that behaviour. This is a consistent message from anti-harassment people, you don't give information to harassers. |
|
Blocks shouldn't give in your face indication that you've been blocked it should be silent and indistiguishable from resource not found. If you block someone your profile should be from their perspective gone. 404 Not Found therefore it is just like the resource is gone. |
|
I think the most sensible solution would be to completely remove blocks and get rid of this hassle. The block feature is finicky anyway. |
This comment has been minimized.
This comment has been minimized.
|
@haleyashleypraesent I disagree. Shadow banning/blocking is a way bigger issue than dedicated harassers being able to see if someone has blocked them. |
@ealgase sorry but if I block someone I don't want them to easily figure out I've blocked them, I want it to look like my profile from their perspective is not found that way they're more likely to give up, my experience on other social networks is if the blockee knows they've been blocked they will often go to make other accounts, if they think my account is gone I don't thitk they're going to waste time making a new account just get around it. |
|
@Gargron I agree that that's a bad scenario and one we want to avoid. That's why I suggested that 410 return a substantively different result from 404. Another potential solution is to have Mastodon compile a list of the last N people who mentioned you, whether or not they've blocked you since - such that blocking you might remove their mentions from your notifications column, but not from the list - and allow the user to block/mute/follow from that list. (This has the added benefit of providing an additional feature for people who get lots of likes/boosts and don't want to cull them from their notifications.) @witcheslive Yes, we are absolutely looking for a technological solution to a social problem here, and we would all, I think, prefer a social solution! Unfortunately, part of the issue is that as federation expands, so does the society associated with it, and it's impossible to prevent bad actors from joining that society. That problem is what the technological approach attempts to solve. |
|
Hm, what if someone harasses/threatens you, but then you get a deluge of other mentions (perhaps you're very popular or are in a lively discussion) and they fall off the end of the list? I agree that would be nice for other purposes though. |
|
HTTP return codes are irrelevant to the ongoing discussion: If you're talking about showing a "minimal UI" (which the thing that sparked this issue was, in a way), you need the JSON to do it, so you won't be returning errors. I think that reverting everything to how I did it, but rewording the "You are blocked" string to "Profile unavailable" would be the most optimal solution. |
|
There is not a perfect solution to this. @SilverWolf32 If you're getting so many notifications that your harasser always scrolls off the list before you get to it, odds are you're not actually noticing the harassment. And if you are, that's a case so far to the edge that we've reached diminishing returns by considering it. @Gargron You don't have to serve JSON to provide the UI I'm suggesting. You just need a template for handling a 410 that uses session variables to populate the "username", "block user", and "mute user" fields. You don't need any data from the server beyond that. Anyway. I have said all I have to say about this. You may take my suggestions or leave them. (Please don't @-mention me, as this will resubscribe me to the thread.) |
|
imo the solution that seems most ideal is the one where you stop serving updates to that account, effectively freezing it in time and making it look inactive rather than invisible or outright telling you "you are blocked". if you consider email, your old inbox messages aren't affected when someone blocks your messages, so the analogous thing would be to retain all visible messages at the time of block and hide all new messages. pro: it makes it effectively undetectable that someone has blocked you unless you manually compare the public page to the in-app profile (with no other side effects). you'd probably want to make the account appear locked or at least cause the follow button to look like "request follow" (since you can technically send a Follow, but will not receive an Accept Follow due to being blocked). you'd also probably want to error out any interactions with existing public objects, or maybe silently discard them idk con: old messages remain visible, though this is no different than opening in new tab. could also be computationally a bit expensive (unless you store a max_id that each person is authorized to see and silently apply that filter?) the simpler, lower-computation solution would be to genericize the block message so that it is indistinguishable from any other error. something like "no posts available" is less aggressive but still obvious that something is not normal. or you could simply revert both this and the "you are blocked" commit so that blocking profiles go back to being empty as if a loading error occurred |
|
"Profile unavailable" sounds like it could be a workable solution...but I would still be worried about "look, they blocked me!" screenshots, since "profile unavailable" right after you harassed them is still pretty likely to be a block. That would also affect the minimal-profile solution, although not as much since there's no punchline text. The last-N-mentioners list sounds like a good idea. Anyone have thoughts on that? |
|
Just silently maintaining a shadow state for every blocked person is an interesting idea. Everything is timestamped, right? Yes, we could probably just store a max ID/timestamp, possibly applying that filter when we get stuff from the database. Does anyone else have concerns about this approach? |
A profile not updating is a common networking issue that you'll find in multiple troubleshooting threads on this GitHub. Furthermore, displaying old posts for the sake of ambiguity may run counter to people's expectations who might want to stop someone from looking at all of their posts, right now. |
|
Oh, those are both very good points. Don't know how I didn't consider the networking-error issue. |
that's the implicit tradeoff being made here, though. this is fundamentally an issue of deniability vs. clarity, with "you are blocked" at 100% clarity and "frozen profile" at near-100% deniability. so the only question is, where along that spectrum should the optimal solution lie? since there is no perfect one. in other words: do you want to deal with people not knowing they are blocked, or people knowing they are blocked? |
I think at some point, we need to accept that people will just share these screenshots. It's not something that's really worth it to prevent. |
|
A good reason to block someone is to keep them from rifiling through your shit. I mean, yeah, right click new window is always there, but having 0 control over at least some basic obstacles between a stalker and your corpus is not a very good idea. |
It is already the case, to some extent. Blocking someone does not prevent you from seeing their toots if you look for them (and that's needed, otherwise the block-then-report workflow wouldn't work), but limits the interactions you can have with them.
Serving a minimalist user info panel for a 410 is tricky because the client may not know stuff like the username at all. Also, without allowing people blocking you in search results, this is not very practical. If we do re-enable search results, that's already one difference in behavior with suspended users (who do not show up in search results).
That would be a useful feature, but only to the extent you know the exact handle of the person you want to block, which might not be practical, especially if you have to deal with multiple people. Also, the current implementation assumes the blocked account is known to the instance, so we would either have to error out if the typed handle isn't known, or change things significantly. Which we could do, but would be a separate issue with a longer term resolution I think. All in all, I think the best solution, at least short term, would be to revert both #10442 and the original PR that spawned this thread, even if I don't like the prior state very much as it was neither a good way to provide plausible denial nor a good way to avoid bogus bug reports 😩 |
Are basic obstacles really worth sacrificing usability for everyone? |
|
How much usability does this really sacrifice, though? If we don't load their posts and instead just show a blank profile, it still lets you block the user back and such, and if you're blocked, you really shouldn't be seeing their posts anyway. |
|
@ealgase I'm talking about like, leaving your statuses before the block visible to someone you blocked, I think that is a really bad idea |
|
Does anyone here opposes to us reverting both PRs, so that things behave like they did before this release candidate? That is, accounts will still show up, pretending they don't have toots, and you won't be able to follow them, but you'll be able to block, mute or report them (but not specific toots since they'll be hidden from you). EDIT: Oops, I hadn't seen it had been reverted already in #10491 |
|
Shadow banning and ghosting. Both are insidious anti-solutions, and not the same as muting at all. Muting makes the offender disappear from YOU - you never hear from them again, and this is good. All of their efforts at stalking or harassing fall on deaf ears, as it should, and without notice, their profile and history still appears when you peruse them, and so does yours, but it's like they are being Uber-ignored by you - again, as it should be. You see none of their posts unless you bother to proactively look for them. The paradigm is the same as it is in firewalling, we know it is best to DROP packets rather than REJECT packets, in most cases, and for goodness sake, don't REJECT packets with a message, that just baits the troll leaving your box or network especially vulnerable to getting pwn3d. If one must insist on taking the easy blanket way out and instituting an actual block (analogous to "null routing" the packets before they even reach the firewall), then it is most certainly not a great idea to say so with a message (go back and read my previous paragraphs above). A 403 is comical. By definition, it is actually saying that you are in fact blocked - Google "403 forbidden", or "HTTP 403", and then read my previous paragraphs. A 404 is a lie... Sort of, and hinges on the unethical, IMO, in the sense that although it is indeed a result of a resource not found, it is but for an artificially manufactured reason. It's like advertising that the eggs in your farms cartons are laid by hens that are free from hormones - and that is a lie, not because it isn't true, but rather, because it's true by virtue of being illegal to inject commercial laying hens with hormones in the first place. Therefore, I recommend that if one were to search for someone who had blocked them, that instead of receiving an intrinsic lie in the form of a 404 error, or a tell tale 403 error, which in fact is actually saying that they're blocked, that we instead hand them a real result - the same result that one would have returned to them on Mastodon were they to search for a non-existent profile... And that result currently is: A splash page that reads, "No results!". i.e., no HTTP Error code included with the result! It's plain, simple, it's not a lie that breaks the notion of HTTP protocol, and gives no indication that the blocked account has in fact been blocked. Here's another reason, and a particularly important one. Almost everyone commenting in this thread seemingly is of the opinion that accounts are only blocked for reasons related to harassment and stalking related offenses. Such is not necessarily the case - you might wish to proactively block someone who is actually a friend, classmate, relative, or neighbor, simply because you feel constrained to post freely in an environment where your profile is readily available to them, perhaps you're engaged in activities or involved in communities where you seek to keep those aspects about yourself private where those relatives, neighbors, classmates, or associates are concerned. i.e., for privacy reasons that are your concern alone... As an example, if you were a whore, it might be prudent to block your mom from seeing any of your ads before you make any postings. Or maybe you just haven't come out of the closet yet and you don't want your dad to know about your same sex partner. There are thousands of use cases. Now, a 403 (and even a 404 error probably) says, "yes the user exists!!!". And then the hunt for Red October begins. Not a pretty picture, when an adult child signs up, seeks out their parents' accounts to preemptively block them from seeing what the child isn't comfortable leaving laying around for their parents to see. Another functional point on blocking someone is important, and gplus was wise to institute this methodology: all posts up to the point in the profile of the blocking party as it were, a/o the day the block went into place, will continue to suggest and indeed appear to the blocked person just the same as it always did, before the point in time that the block went into effect... Like a snapshot in time, with the profile of the blocking party appearing to be forever frozen (unless and until the block is disabled). Finally, we get back to Shadow banning and ghosting. These are system-wide anti-solutions designed to thwart postings by people and bots, by virtue of hiding the fact that all of their posts system-wide have been moderated to dev/null, with the exception of when they themselves try to lookup their posts, in which case they will see those posts and hopefully think that the public can see them too. This type of blocking methodology is commonly Incorporated at places like Craigslist and Reddit. I believe that although it is often a frustratingly effective measure against spammers and haters, these measures typically serve to work against legitimate folks who actually trigger those trip wires accidentally (especially when algorithms are used to institute these measures), and sometimes for a very long time before they discover having been wronged by the admins or robots of the system. So yes, ghosting and shadow banning are different tiers at the system-wide level as opposed to those remedies of blocking and muting at a user to user level. That's my two cents, thank you for taking the time to read and digest this reasoning. |
|
I agree with @tallship that shadowbanning/ghosting are anti-solutions and inherently counter-intuitive, and thus I think #10491 was a bad move. I think the way the current system works assumes the blocker is the good guy and villainizes the blockee (that is the general sentiment I get from comments in this thread as well after briefly skimming it). This happened to me recently. I followed someone whose toots I enjoyed (albeit I knew they had a volatile persona), and one day their toots stopped showing up. I checked their profile and was greeted with "Profile is unavailable." I assumed there was something wrong with their server, so I waited a day, but still same thing. I spent an hour troubleshooting, uninstalling my app, reinstalling, checking my desktop, until I realized "oh this is how Mastodon indicates you have been blocked." It hadn't crossed my mind that I had been blocked because I didn't think there was any reason I should be blocked. I think making a liar of software is bad in general, and I also think software that makes assumptions on its user's ethics is ironically unethical. My 2 cents. |
This is what any tech related people will do. Or they just create shadow accounts on different servers to debug such things because the shout after a broken server. #10491 is a bad decision and if more decision like this are made mastodon will fragment and forks will start to arise. Or because the servers don't know about the blocks just DDoS each other until the network is overloaded. |
|
Again, nobody is entitled to know they have been blocked. Software not giving you that info even if it has it isn't really software “lying to you”. End-users aren't entitled to the full knowledge of the server software when this involves multiple users.
Stop saying things like that, that makes no sense, there is no increase of load in any way because someone blocks you. |
I agree no one is entitled to know they have been blocked. I don't think entitlement was relevant to my point. My point is that showing "Profile is unavailable" is unintuitive to the end user, and if you are tech-inclined, you can end up spending more time than it's worth troubleshooting (funny enough, I found this Github issue trying to troubleshoot 😊). If I had known that I was blocked, I would have left it at that. |
|
The current implementation is a bit weird. It gives you a 100% verifiable way to know you're blocked… but the wording doesn't make it clear that you are blocked. I think it would have been better not to display the info at all (thus leaving you with something like “The user hasn't posted anything”, which is wrong, and may let you think that there is an unrelated issue, much in the same way we have now, but at least wouldn't disclose you you're blocked… idk) |
That's the problem!
Whats the problem with that? Security by obscurity is not an effective solution to a problem. This just encourages fake accounts to stalk people and try getting around it instead of just telling you the truth. Mastodon tries not to be shady or do any of the weird and random stuff other platforms do but this is exactly this: shady and non trustworthy |
I don't think this is a solution. |
I agree it is tricky if you are trying to balance intuitive-ness with curbing abuse by folks who are inclined to make smurf accounts to continue their abuse. My few questions are: do the Mastodon devs believe that most blockees are blocked because they themselves were offenders/harrassers? If no, the solution in place is inadequate/just plain bad, because the user will just end up confused and assume Mastodon is broken. Or like me, spend a bunch of time troubleshooting only to realize they were blocked. If yes, do Mastodon devs believe that most blockees would in turn start creating smurf accounts to continue their harassment? If no, then again the solution in place is bad. They won't know they are blocked, might assume they did nothing wrong, or that Mastodon is broken. If yes, then like @ThibG said, they will eventually find out that they certainly were blocked (maybe by running into this Github issue like I did) and then will find a way to continue their abuse. In the end, I think the current solution can potentially cause more confusion/annoyance to benign users than it does actually stop abuse or harrassment. |
I believe people will start to think that Mastodon is unreliable or just broken. |
|
I'm so confused because would you have really preferred the way it was before? A profile that shows no toots with no explanation for why and a follow button that you can click but that always resets to "follow" because it never goes through? In my view that's way worse in terms of making you believe it's a glitch. Right now you're saying it's broken literally because the message says "Profile unavailable" instead of "Profile unavailable to you because you're blocked". |
|
I know I may be in a minority here based on earlier discussion, but I actually thought the idea to just let a user know that they are blocked, or a "You can't view posts by this user" kind of message, was fine. In my opinion, the benefits outweigh the negatives. The OP of this issue and the two following comments present a false dichotomy. I think saying that the RC in question was "designed solely for the benefit of bad actors in the fediverse" is disingenuous. I sincerely doubt that was the intention of Mastodon devs, and it paints a space where only harassers & those that are harrassed exist. EDIT: fix typo |
No, it creates all sorts of administrative issues completely unrelated to the matter at hand. Woosies complaining about it only serving to notify an abuser to incorporate another method are demonstrating a bit of idiocy. An actual abuser knows this already, and to be sure, blocking is different than being reported. If someone rises to the level of incorporating another avenue (which, by the way, they can absolutely do anyway - Duh!) to harass, it is called STALKING, and on most servers, grounds for an immediate Kick/Ban on most instances, if it's not the policy on the home instance of the person being victimized, then they might well be advised to migrate their account to another instance where it is the policy. Kinda sorta really simple, but some people are just stupid little babies. |
and others
Expected behaviour
Mastodon doesn't enable harassment by not giving information to harassers that have been blocked.
Actual behaviour
Mastodon is going to tell harassers that they have been blocked so that they know to find another avenue of harassment.
Steps to reproduce the problem
Read the release notes.
Specifications
You can see the problem listed here: https://github.com/tootsuite/mastodon/releases/tag/v2.8.0rc1
The text was updated successfully, but these errors were encountered: