Hacker News Reader

Comment by sliken | Hacker News Reader

"easily replaced by other methods" ... like what?

How would this other method make it hard/expensive/infeasible for a malicious company to produce 100M accounts to cause problems with spamming, DoS, and related issues? Phone numbers are finite, cheap, but not zero cost. If sending a spam email cost $0.01 much of spam would disappear.

How would this other method protect a user's social network without having to upload that meta data in signal? Currently this is handled by a user's address book hosted outside of signal. How many users would you lose if you required some key exchange in person?

Sure it's easy to think about generating a key pair, uploading the public key to a DHT, and allow people to communicate directly without a central server. However that has a fair number of practical issues that would be worse for the average signal user, not the least of which would be much worse battery life and bandwidth charges.

View on HN

>How would this other method make it hard/expensive/infeasible for a malicious company to produce 100M accounts to cause problems with spamming, DoS, and related issues? Phone numbers are finite, cheap, but not zero cost. If sending a spam email cost $0.01 much of spam would disappear.

complex captcha and/or expensive PoW. and an option to use a phone number instead

>How would this other method protect a user's social network without having to upload that meta data in signal? Currently this is handled by a user's address book hosted outside of signal. How many users would you lose if you required some key exchange in person?

store it on the server, encrypted by the client. jesus.

So you generate a handle, maybe even a key pair and upload the public half to a server. Then someone you know joins next week and does the same.

How do you find each other?

If it's feasible for an end user, say 30 seconds of a phone's CPU, you can be sure that a spammer could generate millions.

> How do you find each other?

You don't, because that's something that's fundamentally impossible to do in a privacy-preserving way - no way around it.

> If it's feasible for an end user, say 30 seconds of a phone's CPU, you can be sure that a spammer could generate millions.

You're picking overly conservative numbers for "feasible". Choose parameters such that it takes something like half an hour on a phone. Inconvenient? Yes, but privacy is inconvenient in general.

And yes, a spammer will still be able to generate lots of these fake identities, but now the cost is much higher, while the expected profit is roughly the same - and that's how you defeat spam.

Well there are alternatives to signal. Some use p2p, some use tor, some use a federated system of servers. None have the networking effect to be as secure and as popular as signal.

So if signal was privacy preserving, didn't show or tell contacts about others using the service, do you think it would be as popular? Or do you think we'd be discussing a different popular e2e encrypted app and complaining about the very features that made it popular?

It sounds like you want an app that sacrifices usability for a promise of security. That's fine, there are options. Of course, some are honeypots run by the US government, and you'd need to really really have an unusual threat model to warrant such care, but you do you.

>How do you find each other?

by exchanging public keys over an established communication channel

>If it's feasible for an end user, say 30 seconds of a phone's CPU, you can be sure that a spammer could generate millions.

yeah, that's an inherent limitation of PoW, so it has to be an option rather than a requirement. then people who don't care about privacy can do the SMS thing, and those who do can opt to do the 30 minute PoW.

but this is irrelevant to Signal. it's not the phone number verification itself that people have problem with. if the only purpose it served was to prevent attacks on the service by ratelimiting it to one account per one phone number, that would be largely fine. the problem is that your phone number is your Signal identity, which is utterly moronic for a "privacy-first" service, and the reasoning and excuses for this stink to high heaven.

> by exchanging public keys over an established communication channel

I get the appeal, but seems like requiring out of band key exchange will add significant friction and make signal much less popular compared to the competition. I've got a few dozen signal contacts, only a few of them close enough to do a key exchange. If I only had a few, not sure I'd use signal.

Signal has been resisting making special cases for the few, and I get it. Why spend developer time on some feature that's not going to benefit the majority of users?

> the problem is that your phone number is your Signal identity, which is utterly moronic for a "privacy-first" service

I can't think of anything that would work as well and result in so many (40M monthly) people communicating securely. They do seem to be considering their options to help with this, in particular: https://signal.org/blog/secure-value-recovery/

Allowing social networks, surviving the death of your phone, frictionless on boarding, account recovery, etc all interact in complex ways.

Personally I like that there's not a contact database/social graph inside signal.

>requiring out of band key exchange will add significant friction

how do you exchange phone numbers and email addresses with people?

>and make signal much less popular compared to the competition

is popularity among the (COMPLETELY) tech-illiterate worth giving up privacy for?

>I can't think of anything that would work as well and result in so many (40M monthly) people communicating securely.

and virtually everyone who had ever used the internet has an email, despite its apparently cumbersome sign-up and contact discovery procedures.

>surviving the death of your phone, ... , account recovery

are less likely if your account is tied to the phone number or email. they can be lost, they can be stolen, they can be seized.

> How do you find each other?

I don't want to find people on Signal, or be found by them. I want a way to securely communicate with people I've decided I trust enough to be contactable by them - that is, people I've met and sufficiently liked face to face.

I’ve always thought it is dumb for the same reason i think WhatsApp is dumb. In Latin America, WhatsApp exists as a way for people to message each other without paying per text, which is very much still a thing down there. In the US, we get all of our everything unlimited, so we don’t need a wacky app to send texts without paying more money.

But why then do you even need a phone number to use WhatsApp? If it is a texting and calling service that is used to circumvent carrier fees, and is almost exclusively used on WiFi (because data is even more expensive than texts), why not just let them make an account and not use the phone number? That way they don’t have to pay a secondary company to use WhatsApp…

Simply limit incoming traffic to users you've whitelisted yourself. Problem solved.

Let me pay $5 to get a private handle.

How do you pay them without revealing your identity?

Ask mullvad

Monero?

Sounds good, and it could help fund signal without getting advertisers involved.

So you pay $5 on a credit card, which connects you to said handle, much like buying a SIM. How is that any better?

Someone in your social network joins signal next week, how would they find you?

I don't want my social network to find me. I will give them a handle if I do.

I know which of my coworkers use signal, despite them not telling me themselves. That seems wrong.

Buying things anonymously is a solvable problem. Gift cards, cash, etc. Let people resell your $5 sign up card.

Well signal is designed for secure communications for most people. Not designed for anonymous communication. So a key piece of that is easy to use, nice on boarding process, and being useful to new users in that first minute or two.

So the reason signal is so useful is because so many people use it, which depends on things like contacts showing up immediately when you join.

Seems like it's a common theme that people want a messaging client with a huge number of users, but don't want the features that made it popular in the first place.

Mullvad shows that it is possible to receive anonymous payments (eg by non refundable cash by mail)

Signal could use payments (eg via Monero) as spam prevention and revenue stream.

This would not damage any current feature/user

Good news, they are free

https://community.signalusers.org/t/usernames-in-signal/9157...

This is excellent news! Thanks for the link.

Threema only requires an anonymous ID (and has done so for many years), so somehow it must be manageable.

https://hn.algolia.com/?query=Threema&sort=byPopularity&type...

Don't they store IDs on the server?

Not sure what you’re getting at?

Right, but they charge, which connects to your identity just like a sim does.

It’a one-time charge through the app store and unrelated to the Threema ID. You pay for the app, not for the ID. Threema doesn’t receive any PII about you.

Ask them how many users choose to sync their contacts :). I did.

It is as simple as: I dont care. Signal requires a personally identifiable information while the keys can be extracted using rubber-hose cryptoanalysis ( https://imgs.xkcd.com/comics/security.png ) once you know ---> WHO <--- to beat. This is not secure. Sorry.

I understand the whole concept behind it. I understand why they want a phone number. BUT. This is not secure (oh, really, my phone number is hashed, how long do i need to brute force a few numbers in range of 0-9, a few seconds on my laptop, maybe?)

On the other side, I can use a simple email/irc/counterstrike chat to be MORE secure. Why? As it doesnt require any PII from me.

So why bother? Yes, sure, I can use it as a nice chat interface, I can use it as another way beside ~20 chat applications I need to use to contact people I know ( https://imgs.xkcd.com/comics/chat_systems.png ). But dont sell me security. With requiring PII they broke it.

I'll have my 70 year old parents on another continent install counterstrike so we can text each other. Thanks for the suggestion.

Your 70 year old parents don't care about security of communication. Whatever http(without 's') chat is quite enough for them. They were talking on analog, unencrypted line for whole their life. Why should they care now? PRISM anyone ( https://en.wikipedia.org/wiki/PRISM )?

Are you suggesting that people who aren't devoted to the cause of secure communication are not allowed to derive any benefits from secure communication?

That is also not true either. My 80 year old mother uses Signal because she talks property management and money related issues to people. She does not trust WhatsApp because it is owned by Facebook (Which is weird because she uses Facebook all the time). For her, knowing that cell phone company employees can grab text messages is enough for her to switch to something that is secure.

So this is the crux of the disagreement. You believe that you either need maximum possible security or none at all, so when you need to be secure, it doesn’t matter if you have to use an inconvenient solution, because it’s that important. Whereas in reality there are a ton of less severe threat models that can be countered with more user-friendly methods of communication. Advertising companies, ISPs, local cops, etc won’t be able to read your Signal messages or find out who you’ve been talking to, and that can be important to many people. NSA or equivalent probably can, if they’re interested in you already, but that’s a concern for far less.

A big mistake you make is to think that someone would be powerful enough to find you personally on Signal, but would fail to find you on an insecure chat in a video game.

If you are not interesting for the NSA, giving your phone number to Signal is fine. If you are... run. The counterstrike chat won't hide you.

Reply on news.ycombinator.com