I've been using Riot daily for several years, and this 1.0 version seems to work really well. Like any major UI change it's going to take me a while to get used to the slightly different mechanics. Overall it looks more modern than the prior versions.
currently we're aiming for a web of trust between your own devices, and then transitively trusting other's devices once you've verified that user.
however, a GPG-style web of trust which lets transitive trust span over users is a bit questionable: all it needs is one party to be sloppy with keysigning and the whole things falls apart. It also leaks metadata in terms of who knows who like a sieve. So we're going to see how far we can get without it :)
I think I mentioned this before, but I think you should allow for the case where all participants in the virtual room physically meet. Think conference, or a meeting on business premises. They should be able to swap keys with each other efficiently, perhaps using NFC, and without having to have O(n^2) interactions. Yes it may not be super-secure but you can assume the security is already taken care of by the physical security of the meeting.
yeah, this is a good example where a temporary(?) web-of-trust could work. in retrospect, i think the cross-signing proposal (https://github.com/uhoreg/matrix-doc/blob/cross-signing2/pro...) has some thoughts about how transitive trust between users could work, but i'm failing to find it now.
Looks like lots of nice improvements! Glad to see this!
Once there is a way for me to have assigned moderators on my instance able to click / tap to see the ip addy, user agent, associated instance, of users in the room, with the ability to ban them and add a note in an admin/moderator panel..
with the added ability to open list of bans, reasons, and add or remove instances, ip addys, subnets, cidrs, hostnames..
Then I will be bringing a few hundred users to use this instead or wisechat and realchat.
Most popular chat platforms don’t allow admins to see IP addresses or user agents. That’s considered sensitive information these days. There are still niche platforms that behave more like IRC, but they’re pretty rare.
I am guessing the popular chat platforms you refer to are mainly used in walled gardens or have other censoring / moderating options. In my experience chat systems that are open to new registerations are a ripe target for 4chat like trolls - it's only a matter of time before they come and make all your users wish you could see everyone's ips.
even if you can't see them, having an option to ban / block ips, subnets, cidrs, is one of the most important lines of defense if you have an open system. At least in my experiences.
from a privacy perspective we really really don't want to leak client IPs and UAs to other users, even if they're moderators. in a disaster the server admin can get at this, however, and there is a /whois API (similar to IRC's) which server admins can use, but none of the clients hook up UI for it.
I love the privacy thing. With some chat clients I stopped using them, or turned off features to protect ips being leaked from each room participant to all the others in the room. It's important for safety and security.
I believe this is what would cause us to force the use of a stun/turn server if we switched to web sockets. I have seen nefarious users pull and use ip info from chat rooms in many ways for blackmail and worse.
I've also watched some of these wicked people use other tricks like share a picture in the chat room - then check their server logs to see which ips pulled the jpg request - then use that info to hack everyone's home routers, pull their pics, put them up..
So yes I get the importance.
However I can assure you that times are coming where people are going to need better blocking tools with these chat systems.
The main reason I have not started testing with the riot / matrix with my groups of users is becuase I would feel bad for those other people who run instances - as they would be flooded with a few groups of evil wicked trolls - and since there appears to be little moderation ability - it would quickly ruin the chats of all who allowed any user sign ups.
They are coming, it's just that matrix has not been pointed out as a ripe target yet. They will spam, for fun and profit. They will send bots, they will send dozens of humans to do it. Then they will post stuff to make everyone ill - nazi stuff, child prn stuff, all the things you don't want in your system and your other users will leave if they see it.
Whatever you do to ban them, they will return. If it's possible to come back with a new ip and a new email addy - they will be back with six identities before you finish banning the last one. They will register names that look like other's in your system. Heaven forbid you have not blocked blank space ascii/unicode characters..
When these groups of people find that it frustrates you to have your system turned into 4chan, then the real fun begins. Even if you can see their ips it won't matter, they have access to thousands of subnets.
Anyhow, I hope more moderation tools come to riot / matrix / etc - I would be fine with giving my appointed moderators the ability to click to block subnet and have the server do it even if the moderators could not actually see the ip of the individual.
The blocking is most important. Some of my mods look at ips to try to determine cases where someone is suspected of being an imposter and things like that - but that's not as needed. All of my moderators have made agreements with have about ips and personal info during the training process and get how important we take privacy.
I would think this is going to be extremely important with the federated options and such.
Anything we can do to make it easier to block bad users I am all in.
Perhaps we could have options to let moderators see hostnames or something instead? maybe even some kind of hashed code..
thing is we currently use mods seeing ips to determine imposters, and to determine when someone comes into the chat room with multiple usernames at the same time. This is not 100% solution sure, and it's legacy tools we are talking about - but it has been immensely helpful many times.
Some of the worst trolls will come in with a dozen usernames at the same time. When you kick one out of the room - the other 11 can still troll.
Being able to see an ip of different users can give clues as to whether or not other suspicious behavior is indicating more nefarious stuff.
So far being able to ban via ip subnets and cidr has been the most effective method for stemming most of the spam bots and abusive trolls we have encountered.
We do get some smart ones that log in with a dozen different ips from a vpn service - and that does make things much more difficult - it that does come at a cost to the troll as well - so it's worth doing.
I think the instance / server admin may need to view ips to unban if it's an issue - but maybe the moderators just need to see a numeric hashed representation so we can see that sally45 and billy 29 are not in the room from the same connection? Or hostnames or some other thing.
I am open to all ideas on this front. I have spent hundreds of hours dealing with banning chat room users and unbanning people on occasion.
So far the rudimentary moderation tools of the old realchat from 15 years ago are better than the modern chat systems we have tried out (rocket chat, riot, etc) - hopefully this will change soon.
Congrats to the Riot.im folks. The UI/UX was one of the only things that we found was holding a number of organisations we work with back from using it. Combined with Matrix, it's a wonderful tool.
16 comments
[ 3.4 ms ] story [ 51.1 ms ] threadhowever, a GPG-style web of trust which lets transitive trust span over users is a bit questionable: all it needs is one party to be sloppy with keysigning and the whole things falls apart. It also leaks metadata in terms of who knows who like a sieve. So we're going to see how far we can get without it :)
I've raised a bug to track: https://github.com/matrix-org/matrix-doc/issues/1886 - thanks!
Once there is a way for me to have assigned moderators on my instance able to click / tap to see the ip addy, user agent, associated instance, of users in the room, with the ability to ban them and add a note in an admin/moderator panel..
with the added ability to open list of bans, reasons, and add or remove instances, ip addys, subnets, cidrs, hostnames..
Then I will be bringing a few hundred users to use this instead or wisechat and realchat.
Looking forward to using this one day!
even if you can't see them, having an option to ban / block ips, subnets, cidrs, is one of the most important lines of defense if you have an open system. At least in my experiences.
Please see some additional info in this regard I just posted below here -> https://news.ycombinator.com/item?id=19171614
I believe this is what would cause us to force the use of a stun/turn server if we switched to web sockets. I have seen nefarious users pull and use ip info from chat rooms in many ways for blackmail and worse.
I've also watched some of these wicked people use other tricks like share a picture in the chat room - then check their server logs to see which ips pulled the jpg request - then use that info to hack everyone's home routers, pull their pics, put them up..
So yes I get the importance.
However I can assure you that times are coming where people are going to need better blocking tools with these chat systems.
The main reason I have not started testing with the riot / matrix with my groups of users is becuase I would feel bad for those other people who run instances - as they would be flooded with a few groups of evil wicked trolls - and since there appears to be little moderation ability - it would quickly ruin the chats of all who allowed any user sign ups.
They are coming, it's just that matrix has not been pointed out as a ripe target yet. They will spam, for fun and profit. They will send bots, they will send dozens of humans to do it. Then they will post stuff to make everyone ill - nazi stuff, child prn stuff, all the things you don't want in your system and your other users will leave if they see it.
Whatever you do to ban them, they will return. If it's possible to come back with a new ip and a new email addy - they will be back with six identities before you finish banning the last one. They will register names that look like other's in your system. Heaven forbid you have not blocked blank space ascii/unicode characters..
When these groups of people find that it frustrates you to have your system turned into 4chan, then the real fun begins. Even if you can see their ips it won't matter, they have access to thousands of subnets.
Anyhow, I hope more moderation tools come to riot / matrix / etc - I would be fine with giving my appointed moderators the ability to click to block subnet and have the server do it even if the moderators could not actually see the ip of the individual.
The blocking is most important. Some of my mods look at ips to try to determine cases where someone is suspected of being an imposter and things like that - but that's not as needed. All of my moderators have made agreements with have about ips and personal info during the training process and get how important we take privacy.
I would think this is going to be extremely important with the federated options and such.
Perhaps we could have options to let moderators see hostnames or something instead? maybe even some kind of hashed code..
thing is we currently use mods seeing ips to determine imposters, and to determine when someone comes into the chat room with multiple usernames at the same time. This is not 100% solution sure, and it's legacy tools we are talking about - but it has been immensely helpful many times.
Some of the worst trolls will come in with a dozen usernames at the same time. When you kick one out of the room - the other 11 can still troll.
Being able to see an ip of different users can give clues as to whether or not other suspicious behavior is indicating more nefarious stuff.
So far being able to ban via ip subnets and cidr has been the most effective method for stemming most of the spam bots and abusive trolls we have encountered.
We do get some smart ones that log in with a dozen different ips from a vpn service - and that does make things much more difficult - it that does come at a cost to the troll as well - so it's worth doing.
I think the instance / server admin may need to view ips to unban if it's an issue - but maybe the moderators just need to see a numeric hashed representation so we can see that sally45 and billy 29 are not in the room from the same connection? Or hostnames or some other thing.
I am open to all ideas on this front. I have spent hundreds of hours dealing with banning chat room users and unbanning people on occasion.
So far the rudimentary moderation tools of the old realchat from 15 years ago are better than the modern chat systems we have tried out (rocket chat, riot, etc) - hopefully this will change soon.