Hacker News Reader

Comment by gojomo | Hacker News Reader

It's good to see Acton quoted; he'd been so silent since his donation-to/association-with Signal that I was beginning to wonder if behind the scenes, there'd been a falling out.

However, these "enrichment features" from Signal over the last 2 years are less than what competitors like Telegram put out in 2 months.

Really basic things remain undone. Like being able to carry your contacts/message-history forward across planned device-upgrades on iOS – https://github.com/signalapp/Signal-iOS/issues/2542 – 4.5 years after it was "on the roadmap".

Many of Signal's novel cryptographic innovations, while cool, depend on trusting Intel SGX: both as a technology & in Intel's stewardship of the master keys/attestation-chains. Many cryptographers don't believe SGX will be technologically reliable anytime soon, and much of the world will probably see US-based Intel Corp similar to how the US sees China-based Huawei.

View on HN

I second this. I like Signal and appreciate all the efforts people put in. But I don’t really care how Signal added cute stickers or optimized image showing. I just need a privacy app and keep my data. The only reason I didn’t recommend Signal to my non-tech friends was not being able to migrate/export chat history on iOS. This seems to be such a basic (not necessarily easy) thing to do but they just didn’t do it.

> But I don’t really care how Signal added cute stickers or optimized image showing.

That's maybe true for privacy enthusiasts. You severely underestimate the need for these cutesy features when targeting mass market though. I claim stickers are much more important for that than data exporting.

> cutesy features

Personally, I don't care much for features supporting cyber-grooming. What's the point of using cutesy features to attract users who don't care about security, just so someone else can _securely_ cyber-groom them?

Are you somehow equating stickers and grooming? That takes quite a leap of imagination.

Stickers aren't targeting young audiences in any sense. It's very weird to draw such conclusions just if you're not using them yourself (if that's the case).

Can you elaborate? I'm not totally sure what you're defining "cyber-grooming" as here.

The point is: if those users don't use Signal, you're forced to use another IM when you want to communicate with them.

There isn't much point in having a secure chat application if you want all your chats to be portable and exportable across devices and operating system upgrades. You may as well just use Telegram and opt for 1:1 secret chats for when you actually want security.

Signal offers this backup/restore function on Android – where about 60-70% of their users are, if I understand the Wired stats correctly.

So user-controlled portability doesn't seems fatal to the "point" of having a secure chat app, there.

My workaround, last time I upgraded my iOS device, was to screenshot all the old conversations I wanted to retain. Pictures come over just fine! Did Signal forcing me to do that make me more secure?

I think they just hate iCloud. (Perhaps while they trust Intel, they distrust Apple? It'd be great to know why.)

Well, I don't trust iCloud with my device backups, either. There are other things, in other apps on my phone, more sensitive than my Signal logs.

But it seems Signal is holding my chat histories hostage because they don't trust me. And it seems their ultimate plan for this will require me to use some Signal-run, Intel-SGX-mediated 'trustable cloud' for my contacts & messages. No, thanks!

A backup method that just copies it _somewhere else on the same device_ is not really a very useful backup. It involves manually moving files around if you want a real backup, which feels very antiquated in 2020. Why isn't there some way to automatically store my (encrypted!) backup via a cloud backup solution, whether that's Google Drive, Dropbox, or just allowing it to be backed up as part of an Android backup?

Doesn't copying device data to another device, such as a laptop or desktop, do that? I mean... it's not cloud based, but I _thought_ it was a full local backup.

Can you do that on Android? Other than shutting the phone down and `dd`ing the correct partition?

> There isn't much point in having a secure chat application if you want all your chats to be portable and exportable across devices and operating system upgrades

Sorry but I don't understand this sentence. Why?

Like, I get that it's challenging from a technical standpoint, but I absolutely don't see why it would be pointless?

Riot/Matrix does it: it's far from mature/polished—plenty of room for improvement—but it uses double-ratchet, and it functions.

Storing conversations is introducing a massive security hole. One single person in a group can compromise every message from every user.

Secure messages shouldn’t persist and if they do they aren’t secure. Being able to export, upload, and restore an entire conversation history is the opposite of secure.

There is no point in using a secure messaging system if your entire conversation history can be downloaded the minute a single person in your group loses physical security on their device. Just use FB Messenger.

> Secure messages shouldn’t persist

This is a nice, but ultimately utopic, non-sensical, impossible security requirement. I take your point that anything persistent is insecure, but anything received is potentially persistent.

There's no such thing as perfect security, there are only levels. Even a system with some hypothetical form of perfect plausible deniability designed into the receiving end could be compromised via transit metadata.

Saying persistence makes secure messaging pointless is pure hyperbole. Encryption protects against a broad range of threat vectors, some simple, others extremely challenging. There's value in every measure.

Your comment only makes sense if we assume your security requirements match everyone else’s under all circumstances.

This is not likely to be the case.

> Storing conversations is introducing a massive security hole.

Sure, but that could be up to every sender to decide whether they want their messages to be view once/view for 24hrs/store forever.

Could this be because of a non-compete clause agreed on during the Whatsapp acquisition by Facebook. I heard Whatsapp team was around 50 people, wouldn't be surprising Zuckerberg wanted some extra assurances.

> However, these "enrichment features" from Signal over the last 2 years are less than what competitors like Telegram put out in 2 months.

It should be noted that most of these were released at breakneck speed in the past couple of months. It feels like they spent most of the past two years hiring and training new people, probably stabilising the code bases a bit, and that that's now starting to pay off.

"Many cryptographers don't believe SGX will be technologically reliable anytime soon"

E.g. for online backups, SGX will only handle rate limiting. If you're using sufficiently secure password (+80 bits), Argon2 will provide robust enough security for your login and not even a backdoor in SGX matters.

this popped up a few days ago. Could become an interesting alternative ode day. It's p2p based (no blockchain nonsense):

> Session is an open-source, public-key-based secure messaging application which uses a set of decentralized storage servers and an onion routing protocol to send end-to-end encrypted messages with minimal exposure of user metadata. It does this while also providing common features of mainstream messaging applications

https://news.ycombinator.com/item?id=22306140

whitepaper: https://getsession.org/wp-content/uploads/2020/02/Session-Wh...

Kind of unfortunate that Matrix is probably not focussing on making the protocol and apps easy to get on and use Matrix/Riot as a WhatsApp, Signal replacement for end users (person to person comm). It seems to be just for teams and orgs (Slack, Discord).

RiotX (the next gen of Riot) is a Matrix client from the core team, designed to go up against WhatsApp/FB Messenger/iMessage by default... and then expand into Slack/Discord style use cases for powerusers who want that. It's currently in the final stages of beta on Android: https://play.google.com/store/apps/details?id=im.vector.riot...

Meanwhile there are lots of other promising Matrix clients emerging which are aiming for person-to-person comms - e.g. http://dittochat.org, https://pattle.im, https://fluffy.chat etc.

The fact that there exist other clients that are specifically "built with user experience in mind" (Ditto) or built "to be a Matrix app that everyone can use" (Pattle) sometimes makes me wonder why that would not also be the goal of the official client?

I think RiotX is supposed to replace Riot eventually, at least the Android client https://github.com/vector-im/riotX-android. Didn't find a link for an iOS client though.

Signal is on route to add contact/data recovery:

https://www.signal.org/blog/secure-value-recovery/

... they also cite this as a bottleneck for adding non-phone identifiers. It seems like they are putting the infrastructure in place to fix a lot of the remaining usability issues that people have with them.

'Secure Value Recovery' that will rely on their newfangled, Intel®-SGX-dependent trustworthy cloud key services.

Why do they need me to trust Intel, or novel crypto/software that might be deployed in... when exactly? Why can't they let me just trust the same secure backups that all my other sensitive data/communications are in?

SGX that, as a user, you can't even verify is in use.

I really don't know why they're focusing on things like this. It does nothing to increase evidence-based trust, so it's no better than saying "we added a rate limiter, and delete keys after X attempts" (which I'd be totally fine with, and they could probably put in place in practically no time). Either way you're still relying on them doing what they said they're doing.

Reply on news.ycombinator.com