Video of my #MCH2022 talk about #Signal is already available:
> This is a rant about how moving ecosystems are not a good reason for centralizing a crucial service, how stickers are no substitute for a desktop client that does not crash, and how effectively shutting out less popular OS platforms is just not cool.
This is a rant. Take with a grain of salt and perhaps some popcorn.
@feonixrift thank you! It was a blast to give it, and even more of a blast to see all the people lined up to the microphones afterwards!
@rysiek I was fortunate enough to catch both that and your libresilient talk .. which link I'll be fwding around, because we have uses for that. You solved some huge problems there.
@feonixrift thank you very much!
I do hope I did, or at least I offered ideas how to solve them.
Can't wait for it to be deployed in production on websites that need it. It's exciting that after three years in development it reached a point where this seems viable.
@rysiek I feel so fortunate, trying to ramp up projects now, to be able to piggyback on such brilliant tools; as well as feeling myself rather late to the party!
@feonixrift that's exactly how I feel, and so does anyone who starts doing this kind of work.
You are not late. You joined exactly on time, at the right moment, and your contributions will enable others to stand on your shoulders.
@kwj It is expensive. In the 2018 form 990 tax return, Signal Foundation lists the webhosting costs at 458,066 - mind you, before the explosive growth of the last years.
If only there was a more affordable platform than AWS...
The last line was sarcastic. I am unable to grasp the decision to go with AWS. Batshit expensive (assuming that the yearly AWS is well above a million this year), bad for privacy and bad for the overall state of the internet.
@rysiek Great talk! I appreciated the informality, and framing that you gave.
I agree that #Signal is least-worst for some people, and should be recommended for them today---but hopefully not too much longer. For tomorrow, when we're concerned with how /good/ things can become, there are other projects with a better, more community-oriented vision that deserve our development effort. Ultimately Signal's insularity doom it to be superseded, IMO.
@rysiek I watched live from EDT after HOPE wrapped that day, and thought you drove a lot of important convo via your clever “stand in for Moxie” role.
Particularly enjoyed the bit on community over product 👏 and how annoyed the stage manager got at the classic statements-during-questions-time from the audience at the end 🤣
Here’s to you hopefully having catalyzed a family of less central, community-centric, interpretable services 🍻
@mr_z_ro thank you, this is really good to hear.
Honestly it really was a semi-drunken rant that got almost no prep. I am amazed and a bit spooked that it went so far and wide. Had I known, I would have prepared better (and remembered to mention MobileCoin®™, too).
@rysiek that was great, thank you. we agreed here before on just about everything about #signal except wrt to the conclusion and what to do and where to go. i still think we shouldn't compromise long term goals for short term practical and that we should be going to decentralized im right now, not feeding centralized and faux open/free. this is my evaluation though and people in different situations may need to choose differently, totally recognize that.
thank you for advancing the discussion.
@zeh yeah, this is complicated, and we will have different takes on this. Those "short term goals" might mean making sure people at-risk who need to communicate now, today, urgently, have the means to.
All compromises are crap. 🤷♀️
@rysiek yes, agreed. the challenge is precisely to discern when there is the need to compromise on what. that is why it is difficult (and risky) to make a general recommendation...
@rysiek Great talk! But in my opinion decentralization and true open source are most important things for a messaging platform. Signal fails in both of them.
This is why I suggest Telegram for non-technical users. It's centralized, but at least the client is truly open source.
For most I'd suggest Matrix as it ticks all the boxes. It's more complex and clients are not (yet) as polished, but it's the best hope for decentralized, secure and open messaging.
@cos @rysiek I'm sorry for those who have seen me write this a million times already, but please, please, please, don't recommend Telegram over anything: the server is closed source and it is not E2EE by default. No one cares whether the client is open-source if there's no E2EE, the server can do whatever it wants with your plaintext messages!
There are issues with Signal but not over privacy. Anonymity, a bit about open-source, centralization, OK, but they're solid on privacy.
@silmathoron @cos agreed. Telegram is not decentralized (you cannot run your own server that talks to the main servers), and if it was, that would be dangerous since almost all chats are not end to end encrypted.
That is why my 30min talk has a 10min disclaimer at the beginning, clearly stating that as much as Signal needs some criticism, alternatives (including Telegram, mentioned directly) are just absolute crap.
Stop recommending Telegram, you are putting people in danger.
There are very popular things like Slack, Teams and Discord that have no E2EE and server gets all the messages in plaintext. People still use them while knowing this.
For those who want or need E2EE, Matrix or Signal are the best choices. Majority of users don't care about it.
In centralized services it's irrelevant if server code is open or not - nobody can check what is really being run.
1. if your threat model doesn't require E2EE for PMs you missed something (privacy)
2. if the client uses proper E2EE, it does not matter what the server does since they cannot access the content.
In the case of Signal, the server cannot even access any metadata except time, IP, and target.
3. you can actually check that the server runs the open-source code it's supposed to run via SGX attestation (that's what signal does btw: https://signal.org/blog/secure-value-recovery)
2. Yes, that's correct. But having all the metadata in centralized database makes it possible to do some interesting analytics.
3. TIL. Can SGX attestation be faked if the provider wants to do so?
1. sure, most people don't care, but, being here, you should know better, so please don't tell them to use Telegram
2. that is true, though potentially of limited interest with changing IPs, probability of harm seems especially low for non-US citizens and the fact that they don't do it (granted, for now), has been shown in court
3. I don't know enough to answer that
@sotolf @silmathoron @cos "Signal vs. Telegram" is not even a conversation worth having. Telegram effectively *lies* about chats being e2ee (they are not by default, and group chats cannot be e2ee). This leads to people accidentally talking about shit they wouldn't have if they knew the real status of e2ee on Telegram. And Telegram *knows* this confusion happens, and refuses to do anything about it.
This is all you need to know about Telegram. They knowingly and willfully put people in danger.
As the other comment that I made here was about XMPP it would kind of be more close to compare it to that I would say. I won't move over to signal, it's only servers are Hosted in the US, the one country in the west with the worst privacy laws, especially for us "foreigners" I'm also no fan of telegram, I don't know, I haven't found others than XMPP that I really would be willing to say is good, but it's not really that newbie friendly.
I've had some conversations with people involved in it lately and as much as 10y ago XMPP was a joke (I ran several XMPP servers, so I speak from experience), the current push on XEP suites and making sure they are broadly implemented in clients and servers — so as to give users the certainty that whoever they talk to using XMPP will support the features they need — seems to really be making a difference.
Matrix solved this chicken-and-egg problem by having first class bridging features.
I used it first as a silly IRC bouncer, then as Telegram client and later realized the full awesomeness and started multi-bridging communities back together.
@cos @rysiek the #telegram clients are not what many ppl regard as "open source": weird upstream, commits arriving months after an update, spaghetti code, tons of magic numbers instead of enums. you cannot really base your own client on top of that - we did that to some extend with early #deltachat ui - but gave up. also, the server code of #telegram is not open afaik.
@rysiek good rant, good discussion (albeit sadly also a bit time constrained).
That 36C3 presentation has also irritated me quite a bit. Including the (I think pretty unreasonable) wish of the presenter to limit the audience by vetoing CCC's usual approach of publishing recordings of talks. So, discussion was not only cut short after the talk at the venue (as you pointed out), but also clearly and deliberately for the internet at large, after the event.
@rysiek Huh. Apparently the policy about that video has since changed again, the recording was re-published a year after the talk: https://media.ccc.de/v/36c3-11086-the_ecosystem_is_moving
So it seems the video-publishing policy is also moving.
In case anyone wants to learn - or be reminded - what Rysiek is referring to, you might be interested in watching this.
This Mastodon instance is for people interested in technology. Discussions aren't limited to technology, because tech folks shouldn't be limited to technology either!