Signal is experiencing technical difficulties
status.signal.org
status.signal.org
A single tweet 3 hours AFTER the outage has started is just not acceptable [1]
[1] https://twitter.com/signalapp/status/1442354759009247232
Edit: Yes I know signal is free and funded by donations. I've donated multiple times. I still don't think 'free' is a a valid reason for not informing people sooner. It's just unprofessional. This response from MrDresden is on point: https://news.ycombinator.com/item?id=28667509
I don't think we have the right to hold them to high standards because you get what you pay for. Signal has always been a bad choice for messaging in practice. In theory it's really secure and all, in practice, they have security issues that don't get fixed and no real reason to provide a quality service.
Completely agree with this point. Unless you are paying for the service (by being the product or via donations) you have no particular right to get upset with the creators of said service.
> "What do you expect?"
Professionalism and an understanding of the impact this will have on Signal's users. Some will be cut off from necessary communication channels while others might literally have their life depended on the service working.
This very delayed response from Signal, coupled with some of the stranger behaviour of the last year and inability to host my own servers is really pushing me towards looking at Matrix as an alternative.
And for the record, I've donated to Signal for the last 3 years.
If your life literally depends on a service, make sure to pay for that service. Also, make sure to have a backup.
These individuals may not have the means or the option of paying for the service (either due to being blocked or by potentially exposing them selves to said regime if they would).
I know this point has been made repeatedly, but it's important to beat the drum, since it's recommended for journalists and dissidents, and that this "feature" will quite possibly get you killed if you're a dissident in a totalitarian system with a very simple two-part process:
1. All it takes is ONE of your fellow dissidents to turn snitch (or be tortured, or have their phone cracked) and disclose all their contacts.
2. Burners are NOT anonymous. Everything from location analysis to CCTV footage from the physical place you bought the SIM will give you away.
Signal is an obvious honeypot. Moxie's total refusal to even allow the option to use throwaway identifiers while burning precious development time on silly things like emoji support makes their purpose very evident.
Sure, use Signal to send cat pics. But stay far, far, far away if your life or freedom is at risk.
But that doesn't change the fact that there are non technical people who are using Signal in places where they believe it will protect them.
And I am not cynical enough to be ready to say that they are not being protected in some way by using Signal rather than say email/telegram/whatsapp/facebook etc.
But the points you raise stand, and I agree. I would never use Signal for things that are highly sensitive.
But we are tech literate individuals. We are a wast minority.
>For governments, federated organisations & regulated industries
One set of examples is online services and apps like Facebook, GMail, WhatsApp and such (although you could argue whether your life actually depends on any of these).
Another set of examples are things like transportation infrastructure, fire-fighting services, etc.
- I pay for mail service from Google.
- I don't depend on whatsapp. I have multiple alternatives, including SMS.
- I still have Facebook and LinkedIn accounts that I never actually use.
- I have multiple fire extinguishers, because I don't depend on fire fighters to arrive on time, they're a last resort.
My biggest dependency, by far, is the power grid. I really need to look into battery storage for my solar. Because my life literally depends on that... Ooh. And the supply chain for groceries I guess.
I don't think matrix is there yet as a good replacement, at least for my use case. It's just not mature enough.
I know, chicken and egg problem, but I don't want to be debugging my primary form of communication. It's too important.
After all there is a well funded foundation backing the development and operations of the platform.
And the same people would also complain about alternative apps finding ways to monetize while still offering the service for free and keeping it reliable.
How spoiled are we? :)
If we want people to switch from Whatsapp to Signal, "it's free lol" really isn't a response. From an end-user perspective, Whatsapp is free too [1], and generally wouldn't have issues with downtime. The fact is that to many people, Signal is the "new" thing, and any friction getting started with it is likely to make some people say "fuck this, I'm going back to Whatsapp."
This isn't a project by some random dev on Github that suddenly blew up, Signal has a foundation and millions of users, and if I'm going to convince my parents to use it, the sales pitch cannot be "it's like Whatsapp but doesn't work all the time, but hey it's got freedom or something".
[1] I know about data harvesting and whatnot, I'm talking nominally.
like, imagine if gmail announced they were shutting down...
but its free so you cant complain /s
Signal could easily start offering named, verified accounts as a paid service for say, businesses, to have secure texting to their customers. Heck, named personal accounts for this would be ideal to. This is a service which has direct value to me (I already pay for an email domain for the same sorts of reasons).
Signal's product is the sort of offering which ticks so many security textboxes that it would be a no brainer, and getting even 1 big enterprise pushing it out to their customers would guarantee adoption.
If not this, could another reason might be that they're still a small team and this, along with the other changes that would presumably need to come along with it (accounts that aren't tied to a phone number), are what are holding them back.
No, integrating a get rich quick for the founders shit coin into their app is what is holding them back.
Signal aren't allowing accounts not tied to phone numbers because it currently serves a very useful spam mitigation problem for them. There's no real alternative there that's going to solve it without more work on the business side. Technically the phone number requirement right now means they do charge for "named" accounts, they just don't get paid for them - the phone providers do, since in many countries even prepaid phones require proof of identity to activate.
It seems to me that all the alternatives are either free and suffer from some of the same issues Signal does, or don't have all these features.
An outage confirmation is nice to have before I debug things on my end or ask other people to check things, so I can understand the frustration to an extent (even if it seems overblown).
But it doesn't need to be a tweet for me. These people need sleep and/or could be spending their time on fixing the problem like you say, so just get a status page right? Everybody happy. (Just not one of those political ones like Amazon and co have, showing greens until someone hits the outage button manually.)
Most people I know that started using Signal a few months ago are now back to using Whatsapp. Privacy is a nice feature to advertise, but when both your UX and your reliability suck, it's not a good combo.
EDIT: A tweet has emerged 3 hours after the outage started.
I haven't tried Signal on a budget grade Android phone though. I have a few Galaxy S4s I use for various odd projects, but most "modern" apps run fairly badly on those.
I'm also "looking forward" to a weekend or two's worth of frustration failing to get Signal for Linux desktop running on my PinePhone when it arrives... ;-)
And all that with most features of Whatsapp missing(on my last test). Currently I don't see a reason to go from Whatsapp to Signal. Encryption is the same and Moxie see's the Signal-Network as his own where no other client has to connect. So the trust in him is better than in Facebook, but not as good as it could be.
What do you find "awful" about it's usability?
Whatsapp sucks for me because of an unusable desktop client. Their backups system is extremely slow. I've found that messages get sent and received slowest on whatsapp when comparing with other messaging apps even if the gap is almost inconsequential. There's a woeful lack of features that I love on telegram that make the app very barebones and dull. You cannot edit messages. The replying UX is a bit irritating. There's tons of stuff but this is a small gist
https://faq.whatsapp.com/general/download-and-installation/a...
Instead of silently failing, I wish they would pop up a notification that says "Your version of Signal is now out-of-date, and you will not receive new messages until you update."
But, more importantly for gnicholas's point, Whatsapp gives you a warning days/weeks ahead about the need to update the app. My grandmother is almost never connected to a Wi-Fi network, so her apps don't usually update, and she hardly uses any app other than Whatsapp anyway. She got the warning one time and gave me a call immediately, so I walked her through about what to do (i.e. yes grandma, click the update button). If it silently fails, like gnicholas says, and if this happens often, then I can't imagine using Signal with her.
[1]: https://faq.whatsapp.com/general/download-and-installation/a...
BTW, I got messages from both that I have to update. No silent failing at all.
Who's paying for all their infrastructure anyways? What's the plan Moxie?
The size of the network will grow with time and using it obviously contributes to its growth.
Honestly I might be moving off to I2P instead.
Now the founders could of course have idealistic goals in mind but we might never know that. Realistically there is no market giving that currency a value and they're asking for literally thousands of dollars just to support the onion routed network.
That means that _you_ are going back to WhatsApp, then, since when you want to talk to them - that's what you'll use too.
The best thing is definitely having two devices, one you can trust and one you can abuse.
However, I don’t think the UX is bad, so far it has been sufficient for me.
Overall i would say its reliable but not communicating out issues is a really fast way to lose user trust.
Fully agree on the UX but not the reliability. WhatsApp used to be famous in Switzerland for its unreliability. On big European football games and Christmas / New Year WhatsApp was regularely down for 3 hours and more.
How do you know this is not intentional silencing by the govt to prevent terrorist attacks?
Whoever is keeping a Whatsapp Account nowadays shits on the privacy of all their addressbook contacts and can't be helped, obviously. It may not be a friend.
Brian Acton then used that money to bankroll Signal, which unlike WhatsApp doesn't seem to have a clear path to profitability in the long term.
Whatsapp is actually an amazing app, now that I think about it.
Do you feel dumb now too?
The real question here isn't about which system is technologically superior (though HN gravitates to that question), it's which one is technologically superior taking into account network effects. Briar and Session may be great, and I wish everyone used them, but at present at least where I live it'd be very tough going. Getting people to use Signal isn't all that hard, the network is pretty robust in my communities (university city).
- federation (while P2P might be interesting for some, it definitely isn't something for everyone - yet, see below)
- scalable from private messaging up to thousands of users ((soon) with Spaces and threading)
- unlimited multi device with synchronization
- prospectively P2P with offline usage (Pinecones)
- open standard enabling competition
I haven't been very successful convincing people to join. Some tried but were immediately put off by the complicated UI. It takes too much explaining and hand holding.
Those constantly in-your-face "session verification" prompts are bewildering and useless. And then it's impossible to disable spell checking in the Element client.
I hope it continues to get better.
Being engineers, we did figure everything out, but I couldn't see any non-technical users going through the headache, particularly when Signal is just "install and you're done".
beeperhq is doing this for I believe $10/month (even iMessage). Free, reliable bridges would be the special sauce for Matrix IMHO.
* it can't save photos to the camera roll
* notifications are hit and miss
Blame your Android distribution https://github.com/signalapp/Signal-Android/issues/5085#issu...
(The setting is allowing Signal to run without "intelligent" battery optimization. Will also have to see if it affects battery life.)
It's not an excuse, but I've seen other people on iOS have similar problems.
Both of these issues do not exist on Android versions of Signal.
heck for years you could not even select multiple photos to share at once from file picker
I vivily remember how Signal didn't work for like 8 hours because it was during night in US and the responsible person was sleeping, so these 3 hours outage is nothing with this profesionally managed client/network
that last drop was when they started to nag me with mandatory PIN code through half screen, then we just left with extended family, F that
If you're on Android, check out https://dontkillmyapp.com/
Saving pictures also works fine.
1. Click Photo.
2. Click Share Sheet.
3. Click save photo.
I would like to say to each his own but the nature of communication apps makes them useless if others don't also use it. The typical response I get for this is some form of whataboutism. I will just use whatever else I find for now until they figure out a way to monetize it and care about user needs (such as not mandating phone numbers and being able to register the same number on multiple devices so I can use it on iPhone for example without stopping android usage) or some other app like threema will gain popularity like Signal.
What does one need to actually contribute code to Signal?
Lots of people are trying to contribute, but Signal is more and more becoming a black box, I say this as a t-shirt wearing (Open Whisper Systems) advocate, Signal is falling out of favor with many "OG" advocates.
This is why I would wanna see user-hosted instances of the Signal server [0] plugging in to the network and supporting traffic.
I think a zero-config server app install of Signal server would work great for decentralizing the messaging platform too, but unfortunately it's not set up to handle that yet.
... is the article in question. Basically the argument is that federation slows down the possible rate of change. If you have but the one server in a system you can change things whenever you want and then push out any client changes.
The root concept here is that constant protocol and feature change is the most desirable state. Not all of us accept that.
*edit
https://signal.org/blog/the-ecosystem-is-moving/ here is a post from moxie himself about why its centralised not decentralised
Those are called XMPP or Matrix ;)
Does Signal work better on other platforms? (For me, no - I also have it installed on Windows and do not get notifications.)
Searching through your messages by type, really good video messaging, voice messaging, video chats, and the best stickers on any platform.
It doesn't do E2EE by default (you need to manually flip on Secret Chats) and group chats are not E2EE at all, both of which make it HN kryptonite because apparently E2EE everywhere, by default, is the only thing that matters even though there are technical trade-offs to an E2EE-everywhere model. I appreciate I can flip on Secret Chats when I need them but do wish it was an option in group chats.
I selfhost a Matrix homeserver: it's nowhere near ready for primetime and misses a lot of those thoughtful touches less technical people appreciate. If you do want to try to onboard people to Matrix with the Element client (and be very careful because you only get one, maybe two shots with your friends and family before they go back to whatever they were using before) use Mozilla's homeserver for its vastly better performance and option of SSO sign-on.
This is true. I've been on IRC since the mid 90's with ircII and BitchX and even I find Matrix/Element to be tedious to navigate.
There's exactly zero chance I will suggest it to anyone non-technical until they get their clients up to snuff with the competition.
Telegram and Discord are the two best IRC-replacements, both offer easy native bot support and tools to manage bigger group chats (100+ members).
For 1-on-1 it's all the same to me. Telegram is still the easiest, since I can have the same chat history on multiple devices easily.
No it doesn’t. It’s owned by the 2nd largest surveillance capitalist; it shares a back-end with Facebook (so at a whim they can use the metadata around your chats to update their advertising models of you and your contacts); IIRC it leaks your engagement to FB (via web link previews).
-End-to-end encrypted by default (WA yes, Telegram No)
-End-to-end encrypted group chats (WA yes, Telegram No)
-End-to-end encrypted group video calls (WA yes, Telegram No)
-End-to-end encrypted desktop chats (WA yes, Telegram No)
-Client-side encrypted cloud backups (WA yes, Telegram No)
-Cross-platform end-to-end encrypted chats (WA yes, Telegram No)
It's a bit hard to make a case for Telegram's good intentions when they lack the ability to actually deploy usable, ubiquitous end-to-end encryption. They could even pay people like Moxie to implement it for them, which begs the question, why haven't they. Durov absolutely has the money.
Yes WA and FB has access to your communications metadata, but so does Telegram. And every entity that hacks Telegram.
Pavel Durov isn't a magical entity of good will and intentions, he's a Russian guy with sure, decent eye towards good user experience, but the cost for that is too great, considering he's never hired a competent cryptographer to even see if the features the app provides could be implemented in a privacy preserving way. I'd love to be able to defend Telegram's design choices, but the more they spend their time adding "sleek animations" and other bs features, the clearer it becomes it's just another social media platform with same privacy features as Facebook Messenger:
-Not end-to-end encrypted by default (Telegram Yes, FBM yes)
-No end-to-end encrypted group chats (Telegram Yes, FBM yes)
-Opt-in 1:1 end-to-end encrypted chats on mobile (Telegram Yes, FBM yes)
In that comparison, WA comes at the top. But then again, it's a false dichotomy as Signal puts incremental privacy features with no real down-sides.
-Open source clients (Signal yes, WA no)
-Reproducible builds (Signal yes, WA no)
-User-managed groups (Signal yes, WA no)
(And no, Telegram being open source with reproducible builds isn't really worth anything until the verification confirms that it's really using proper E2EE. Currently Telegram's source only confirms that it's not E2EE by default etc.)
As for the surveillance capitalistic aspect, Telegram is only as good as their word. FB wasn't surveillance capitalistic when it started back in 2005. Telegram isn't slaughtering the piglet before it's time to cash in, but they can get hacked at any point, and the hundreds of billions of messages that sit on their server, effectively unencrypted (storing the database key in server's memory is the same as "not encrypted"), Telegram isn't private even if Durov never wanted to use that data. The fact is experts agree such data is a liability, a toxic asset, that nobody can protect forever. Durov isn't as security researcher patching zero-days faster than NSA, GCHQ, Fancy Bear, Unit8200, CCA et. al. find them. Most probably Durov wouldn't even realize a rootkit has been sneaking data out of his server for the past 8 years.
Compared to these risks, WA has been ahead of Telegram ever since they implemented Signal protocol in 2016.
I still think the WA privacy story is more suspect than Telegram’s at this point.
Paying lip service is not the same as practicing what you preach. I'm not saying Mark Zuckerberg is on your side. I'm saying WA team managed to implement Signal Protocol before Zuckerberg et. al. realized what was happening wrt. WA's data aggregation capabilities. Also, Durov isn't called the Mark Zuckerberg of Russia for nothing. Also, it's not like VKontakte users Durov made his fortune with, had too much privacy. Why is Telegram suddenly a magical fountain of privacy when on tech level it's worse than WhatsApp?
"the user has no control over the keys"
Extraordinary claims require extraordinary proof.
"everything goes via their servers"
Kind of the central point of centralized messengers?
Also, with Telegram there is a guarantee 100% of group chats, and 100% of Win/Linux desktop chats go through the server in the form that allows the server to spy on the content. All Telegram clients also leaks the metadata that you specifically enabled E2EE with some person, i.e. you disclose your desire to hide data from all third parties. That type of metadata is extremely valuable, and something WA -- due to ubiquitous E2EE -- is unable to collect.
"I still think the WA privacy story is more suspect than Telegram’s at this point."
Your opinion doesn't override that of Matthew Green, or Bruce Schneier who have explicitly advised to NOT use Telegram. I have never seen a security expert recommend Telegram, and don't expect you to provide evidence on the contrary, either.
I didn't say it was. I said the Telegram _story_ was _better_ than WA's at this point.
> when on tech level it's worse than WhatsApp?
I'm not arguing that Telegram has a better or worse technical model; I'm arguing that the WA business model and senior leadership and share structure, which controls its tech, creates a _story_ which is less trustworthy.
I take your points about WA appearing technically better.
What I don't want is a corporation behind my chat app whose business is in monetizing my data. If Durov can keep Telegram out of that business I will be pleased.
E2EE is the bare minimum for security these days, as so carefully explained by the message you're replying to.
"If someone were to come to me and offer me E2EE by default without those features I would say no thank you."
This is such a loaded and unthoughtful comment. It assumes you have to make a choice. You don't, at least per technical reasoning. The only situation where you can't have E2EE is massive groups, and it's obvious those have no expectation of privacy. Anything else, I can explain to you on a technical level why it can be done in a privacy preserving way, and the only reason its not, is because Telegram's team lacks the know-how. Signal has already shown they can pull off pretty much any feature in a privacy preserving way.
So the question is not "what features do I need", but "should all my features be actual features". A feature that doesn't protect your privacy by definition can be used to spy on you, and I'm sure we both agree that is not a feature.
So ask yourself
Group chat with end-to-end encryption (Signal) or Group chat without end-to-end encryption (Telegram).
Which one do you choose? The answer is obvious.
>What I don't want is a corporation behind my chat app
Also, you should know Telegram is not a non-profit like Signal. Telegram is a limited liability _company_. It's a for-profit entity, and the LLC only means, the owners are "legally responsible for its debts only to the extent of the amount of capital they invested".
When Signal provides these features, please feel free to ping me back here but until then, the proof is in the pudding and the Signal pudding doesn't taste like the Telegram pudding.
As Signal's client is open source and its server is whenever they feel like it, I'm sure they'd also value your contribution to prove out your point.
>"Actual multi-session clients across devices."
Not available with E2EE
>"No Electron."
Sure, because you need all 32GB of RAM but not your human right to privacy. Also, this is an argument against Signal's client, not excuse why you can't have E2EE on non-electron native client.
>"Truck loads of features and thoughtful touches everywhere."
So, eye candy. Not very strong argument.
>"Searching through your messages"
Only exact messages. Gets incredibly slow once you go past few weeks. Client-side searchers are as fast as your phone, i.e., fast.
>"video chats"
Not E2EE for groups
>"best stickers"
1:1 match between my Signal and Telegram sticker packs (50+ packs)
>"even though there are technical trade-offs to an E2EE-everywhere model"
The problem with this BS claim, is you can't point to a single trade-off. Everything you mentioned can be done with E2EE.
>"but do wish it was an option in group chats."
So the question is, why aren't you demanding it?
>"When Signal provides these features, please feel free to ping me back here but until then"
Again, which features? Please provide an actual, exact list so we know when to ping you.
>"the proof is in the pudding and the Signal pudding doesn't taste like the Telegram pudding."
The problem when you add a ton of sugar into a pot of porridge made from shit, is now you have a ton of sugar with an even coating of shit in it. No matter how much eye candy you glue on top of insecure design and spaghetti code that is Telegram, the fundamental truth is, it was never empowering you, only its owners who are collecting as much data about you as FB does, if not more.
EDIT: Fixed lack of E2EE on video calls to mean group video calls. Apologies if this looks like moving the goal post, not my intention.
>>"video chats"
> Not E2EE
...isn't true. Please see https://core.telegram.org/api/end-to-end/video-calls
It's not my intention to be aggressive towards you or anyone else. It's the argument (or lack of arguments) I'm attacking, not the person.
Facebook Messenger, Hangouts, Google Chat, and above all SMS / iMessage. Whatsapp by all accounts is the most popular (as much as I personally detest it).
I know I know, not HN-fare exactly, but they're easy, popular, and work. That's all that most of our's social network actually cares about. There's a lot of discussion in this thread about detailed encryption/security policy, but while I am a privacy nut, none of my non-IT friends/family could possibly care less... And they're not wrong. This convoluted encryption which severely limits multi device usability Andb frequently has louse ui is out of proportion with other comms methods such as email also used. While I may stand up for principles of privacy, my family isn't wrong asking why should their exchange of recipes or birthdays wishes be any harder than necessary.
Given you’re calling out WhatsApp specifically as a service you detest I’m going to assume you rank it last of all these options. Why is this? WhatsApp is E2EE, has a simple UI (at least IMO) and doesn’t store messages centrally. If I claimed to be a privacy nut I’d put WA above any of these.
for what little it's worth: Facebook Messenger, Hangouts, Google Chat, let alone messengers of yore like ICQ, MSN, AIM, etc all allow me to create a userID & password on computer and use them from any device of my choosing, as well as change/transfer/obsolete devices seamlessly.
Whatsapp is bound to my phone number and device, and its experience with multi-device is.... actively hostile. My preference is to not type messages on a 1.5" keyboard, but rather on any number of ergonomic keyboards and devices I use around the house and work. Whatsapp is simply prohibitive for my usage, and for positively no actual (as opposed to theoretical/principled) benefit whatsoever. I've pulled what little hair I have trying to use it, but the 27th time it locked my account for using too many devices (Oh noes! I have a phone, tablet AND a laptop? Crazy me:), I gave up.
(privacy wise, also, my phone number is one of the more private and static pieces of identifying information I own; I guard it carefully and I have no idea why I need to expose it to everybody I want to communicate with, as opposed to anonymous and disposable userID or even email).
I did not mean to turn this into anti-whatsapp rant though; Many other E2EE-focus services are similar. Whatsapp just happens to be wildly popular and, for my use-case, completely and actively impractical :-/
Shiny stickers and broader user market vs privacy.
Almost constant uptime due to being owned by a multi million/billion dollar company vs a group that mainly runs on donations.
Etc etc
I really hope it's the former, instead of the latter.
Whatsapp is a non starter for me because the company that owns and operates it is not trust worthy. Very profitable, and probably making some of the people here great returns on their investments...but it's not trust worthy, nor ethical.
Beyond that, there are free XMPP servers, and you can even run you own. Clients like Conversations, blabber.im are all wonderful. I've been using it with friends and my wife for years without issues, albeit I have my own server.
There is also a Jitsi app available for mobile devices, but it is optional.
But.
In a thread about a Signal outage with a request for an alternative "casual messaging" app - I really don't think open source video video meetings is a relevant answer...
Perhaps Telegram but I recall reading some commentary on their crypto that made me hesitant.
Conversations is a good XMPP messenger, and it supports crypto extensions, OMEMO, etc.
WhatsApp is arguably more secure than many of the alternatives. Just look at the repo for Signal. For privacy it probably isn't, when you're in the US, that is.
But WA has been working on interesting proof of concepts for wider deployment such as the client-side encrypted backups. Those are actually fantastic when properly implemented. There's of course shitty aspects with WA about those like the 64-bit passwords, but, if that's a feature Signal thanks to possible collaboration can properly deploy later, I'm all for it.
There's a difference between Signal being categorically better for pretty much everything when compared to WA, but, WA isn't the worst option, even when considering its owned by FB.
The one part I see people struggle with is transferring keys across devices which is slightly convoluted (your keys are not stored on the matrix home server--for obvious reasons) but there are wizards to help you install your keys to a new device and for most use cases with phones completely unnecessary since you will always be using the same device.
-No end-to-end encryption for groups (on any platform)
-No end-to-end encrypted desktop chats (Win/Linux)
-No cross-platform end-to-end encrypted chats (iOS)
-No end-to-end encrypted group (video) calls
Great, for Durov if he wants your data, and all entities hacking their server for your data. Not that great for you.
They also meddled with Russian elections this year, blocking opposition bots, but allowing govt bots.
Telegram is not E2E encrypted by default, so the other comment makes sense.
I'm not sure of a good alternative as simple as Signal
If my message is over two weeks old, I'd prefer it not to be on anyone's server. I can keep a local backup if I wish to do so.
It doesn't have to be this way, as WA's planned feature of client-side encrypted cloud backups[1] shows, disproving Durov's implied claim that Telegram must have access to messages to provide such feature.
[1] https://engineering.fb.com/2021/09/10/security/whatsapp-e2ee...
And Win/Linux desktop chats are groups are not E2EE even if you want, so it's not just about it not being default, it's about the complete lack of E2EE for those.
"but this makes the distributed app model sooo much more user friendly."
It does not. It forces you to either drop E2EE, or whip out your phone hundreds of times a day.
"All your history is instantly available on the web client for example."
Which is a privacy nightmare, web-based crypto is never safe, not even when it supports E2EE. https://tonyarcieri.com/whats-wrong-with-webcrypto
"Try that with Signal."
You obviously can't because a) it's a security risk as per above and b) you don't need to as you can have native Signal client on your phone and your laptop and your desktop. Try having just one E2EE chat with Telegram across all three devices. You can't, even if you want to.
"The web client works if your phone is out of battery. Try that with WhatsApp."
Yeah, WA's system is shit, but at least they're working on a native desktop client (that also works on tablets), and that too, will feature E2EE for everything. Can't say that for Telegram.
Most of my friends don't care about privacy, I tried to love Signal for 2 year, I recommended it to friends, only to give up and make a fool of myself. But after so long, the fact that I regularly don't get notifications of new messages is inexcusable.
So in my inner cirle, Signal has damaged people's willingness to improve privacy. Thanks.
After reading it, you might want to reconsider Signal again.
It's not that Telegram started being good, it's that the experts felt it was so bad they decided it was for the public good to give pro bono advice on the topic and didn't feel too strongly about advocating wrt the matter ever since, aside occasional social media chuckles when more vulnerabilities in Telegram have been revealed.
Sure, all messaging apps have problems, bugs, and issues, but with some there's clearly more effort in building a foundation that's not accumulating technical debt like there's no tomorrow. If Telegram suddenly wanted to enable E2EE, they'd have to start from the ground up. You can't glue E2EE on top. So when you look into the software engineering aspect, you're inclined to look into how the product was made, and the design rationale. None of that holds water for Telegram. What you get is something that on paper looks exactly like FB Messenger, and a bunch of promises to never abuse the access to the insane amounts of private data that accumulates.
I've been looking to secure messaging for a decade now and Signal's by far the best option right now. Threema is pretty good, and Element's getting there, although it needs more time to mature. You were already using Signal, abandoning it wasn't perhaps the smartest thing to do: Downtime is of course very unfortunate, but I've had more outages with Telegram than with Signal. Also, again, it's easy to fix a broken window, much less easy to rebuild due to rotten foundation.
But after this outage, clearly not.
Unofficial comment: https://community.signalusers.org/t/signal-down-for-anyone-e...
Official comment which came later: https://twitter.com/signalapp/status/1442354759009247232
https://getsession.org/ I doesn’t need phone numbers
"Session is enabled by services provided through the Loki blockchain network. The Loki cryptocurrency is a fundamental part of these services, providing an anonymous way to transfer value between people. A lightweight wallet integrated into the Session app, using keys derived from the users existing keypair, could allow users to quickly and privately transfer value inside Session."
Or is there anything wrong with RCS I should know, like rampant data collection or sth?
RCS has a long lasting history but is driven forward mainly by google during the last years. [1] One mayor flaw is that it comes basically without end-to-end encryption which - again - google announced to add an roll-out slowly. [2] Furthermore its unclear in how far Apple will implement RCS [3]. And last-but-not-least there is lack of some feature we are used to by messenger apps, like avatar picture, status message, live location etc.
With 473 M users worldwide [4] it seems to be everywhere an nowhere. I'm checking from time to time with my contacts, but I've not seen the lock symbol in the Android Messages App on by up-to-date Pixel phone.
[1] https://en.wikipedia.org/wiki/Rich_Communication_Services#Hi... [2] https://www.theverge.com/2020/11/19/21574451/android-rcs-enc... [3] https://appleinsider.com/articles/21/07/20/apple-holds-out-i... [4] https://www.gsma.com/futurenetworks/rcs/
This is just false. You can disagree with moxie all you like and make your case but he has been very clear on the reasons why he hasn't gone that way and the success of his project among non-technical users where so many others have failed suggests he's doing something right.
I will applaud and cheer if you show his concerns on federation aren't really a problem by making a federated, e2e encrypted message app work as well for users as moxie has with his. Seriously. I will cheer you like you won't believe. Go do it! Or do you have an excuse?
I never made you aware of any such limitation, quite the contrary, in fact.
The top selling points of Signal are security and privacy. If you want these things, then there's a lot of things that maintaining complete control of the protocol gives you. From the original "Ecosystem is moving" essay: "By contrast, WhatsApp was able to introduce end-to-end encryption to over a billion users with a single software update". That's the point of centralisation. Control means agility in the face of adversary, which when you're trying to optimize for security and privacy is pretty useful.
The comparison to HTML is therefore wide of the mark, because the whole point of HTML is to be interoperable. A better comparison would be with something like TLS/SSL, and then you can judge how effective the combination of slow-moving protocol evolution and a security focus has worked over the last few decades.
There is no bad choice. There are just compromises that you have to make depending on what you choose, and some of those compromises are more suited to some objectives than others.
The tradeoffs Moxie was thinking of when rejecting federation had more to do with centralizing his power over Signal rather than any kind of technical limitations.
draft-moriarty-tls-oldversions-diediedie (now RFC 8996) only exists because of this. Moxie doesn't need to worry about security problems with a protocol he shipped last century but for TLS we had to actually document that people should stop using that - because of course they are anyway.
In the Web PKI we have in practice even more centralization, in principle it would be possible to operate a public CA that simply isn't trusted by Mozilla, Apple, Microsoft or Google, but in practice you need all of them to have a viable service. The de facto situation is that you need logging with Google (and any other service, but Google is mandatory) and you need a sign-off from Mozilla, or you are dead in the water.
Now, I think the Web PKI is arguably the only global PKI with something approaching public oversight and it's a huge success. But it's not decentralized in the way you mean, not at all.
Also, there are degrees of federation. It doesn't have to be a wild wild west of the sort we see with projects like Mastodon. A closed federation can have strict rules on who can participate, can enforce constraints on auditing and rapid updates, and so on, and still be a federation. In fact, we see some of this in TLS - not everyone can participate, but there are dozens of independent organizations working within an internal governance models. Another example is EFnet, which is a federated chat network which has regulations on who can participate:
http://www.efnet.org/?module=docs&doc=16
Moxie conveniently ignores all of this because it serves his interests to be ignorant of it. It has been repeatedly brought to his attention. Moreover, he has put out legal threats to any project which tries to fork Signal to research these features.
I think what's happened here is that you don't understand the difference between TLS, a network protocol, and the Web PKI, a Public Key Infrastructure and so you've sort of muddled both together, which is revealing on its own.
You've also imagined a fairytale outcome that doesn't match reality.
Moxie's strategy works and yours didn't, and you're here telling everybody you don't agree but reality doesn't care whether you agree with it or not.
I understand the difference, and the condescending snark is not appreciated. I don't think the difference is very important in this case.
It's also revealing, perhaps, that you don't understand that the "Web PKI" is used far and wide in domains outside of the web.
>Moxie's strategy works and yours didn't, and you're here telling everybody you don't agree but reality doesn't care whether you agree with it or not.
Oh, bugger off. The only group who has even made a serious attempt to do it properly is Matrix and one "meh" implementation does not damn an entire design.
As an example, you wrote earlier: > In fact, we see some of this in TLS - not everyone can participate
As written this is nonsense of course. TLS is a series of published standard protocols, everyone can "participate" by implementing one of the protocols, and at their option some or even all optional extensions.
But for the Web PKI things are quite different, only some can participate and under exacting conditions. On one side, Certificate Authorities can only participate via the root trust programmes of key trust stores operated by Mozilla, Microsoft, and Apple, plus Google (for Android and, perhaps sooner or later, Chrome) and Oracle (for the JRE). Participation in these programmes requires obeying the CA/B BRs, but also is at the sole discretion of each programme operator and so their own policies apply†
As subscribers, the Web PKI is only interested in certifying names in the Internet's DNS hierarchy and, to a small extent in the Internet's IP address hierarchy. Other names aren't actually certifiable in the Web PKI, sometimes to the surprise of would-be subscribers (and even, once in a while, CAs)
Finally as a relying party (the role all of us occupy even if in addition to other roles) the Web PKI is only applicable to the Internet, which of course is a lot of things but it isn't everything.
> It's also revealing, perhaps, that you don't understand that the "Web PKI" is used far and wide in domains outside of the web.
That's true, and has been for many years, and yet, there's a reason it's named the Web PKI anyway. Notice the B in CA/B stands for "Browser" even though today these are (except Mozilla) operating system vendors, historically they're the browser vendors. You could imagine alternate worlds where say, IBM, Apple and Sony are the biggest players in CA/B instead of Mozilla, Microsoft and Apple, and it really isn't about the web at all, but they're quite different from our world.
If you don't like the phrase "Web PKI" you could sometimes substitute PKIX, but, ultimately PKIX isn't quite the same thing. PKIX is technical, it explains how to use the X.500 system's certificate format (X.509) to talk about the Internet instead of the X.500 directory, but it makes few substantive policy decisions. Certificates you get from Let's Encrypt are PKIX certificates, but, the ones you mint for testing with self-service tools or that come supplied with many Linux installs are also PKIX certificates and those aren't trustworthy. Certificates from Let's Encrypt are also part of the Web PKI and that's why you trust them.
† These policies can be quite significant, for example Microsoft's programme requires that participants agree to revoke certificates if Microsoft so chooses. In practice Microsoft seems to use this power to kill off certain types of phishing scam that attack Microsoft's products. Mozilla's programme has historically been the origin of a lot of important public policy some of which is now in the CA/B BRs.
Thus, tradeoffs.
I'm not yet aware of any federated approaches that prevent a server from connecting a sender and recipient, but would love to be educated otherwise.