Hacker News Reader

Comment by breput | Hacker News Reader

Matthew Green does a pretty good job picking apart PGP's issues, although he completely fails at suggesting alternatives and also completely ignores non-email use cases.

https://blog.cryptographyengineering.com/2014/08/13/whats-ma...

View on HN

There isn't "an" alternative to PGP, because the idea of a single tool like PGP that covers all the use cases PGP attempts to cover has been revealed to be bankrupt. PGP does a bad job at practically everything it's applied to, because different problem domains ask different things from their cryptography. Backup tools want deduplication. Secure messaging wants relentless forward secrecy. Package signing systems want short, simple signatures --- meanwhile, messaging systems want authenticated key exchanges without signatures. Sometimes you want non-repudiation, but often you want the opposite.

To put this in perspective, it's a little like someone invented a bad balanced binary tree before anybody else came up with a hash table or a radix tree or a heap, and then a weird subculture formed around that balanced binary tree based on the idea that you should only ever use that tree algorithm, even though someone subsequently came up with red-black trees that were superior in every way to the original.

it's a little like someone invented a bad balanced binary tree

It's a little like that culturally but it's almost entirely unlike that technically - a suboptimal data structure or algo tend to be just suboptimal-but-functional whereas bad cryptography and bad cryptography engineering often fail catastrophically. I know you know this, of course! But for one thing, someone used an iffy analogy on the internet, etc. For another, PGP people love misusing exactly this sort of analogy.

I'd use the analogy of "someone came up with C and then people came up with high level languages and even low-level languages that provided memory safety and..." but then we'd just be in another debate about how mired in the 1980s we should remain. :)

Have you written anything on proposed solutions? That would be an interesting read if you have.

I am so tired of crypto experts criticizing PGP without suggesting an alternative. So much has been written, so many soapboxes have been climbed on, and yet there seems to be no good replacement.

I deeply respect Philip Zimmermann for creating "pretty good privacy" rather than trying for "perfect privacy". PGP is exactly that: pretty good. Not great, not perfect, but pretty good indeed. And it's there. And it works. It's a compromise, which works well for many people's requirements. Oh, and did I mention that it EXISTS?

I use PGP every day. My private keys are stored on Yubikeys, which is supported. I have offline backups of those, which is supported, too. I can encrypt my backups for multiple keys, sign lists of hashes of files to verify integrity, and people can send me private E-mail.

None of this works perfectly, but it does work, and (not being a teenager anymore) I appreciate the fact that PGP has worked since 1991 or so, and I can reasonably expect it to work for the rest of my lifetime, unlike much modern software, which while being incredibly fashionable, seems to flare out a couple of years later.

This article present some alternatives to PGP. https://latacora.micro.blog/2019/07/16/the-pgp-problem.html

With attacks on encryption and privacy by governments and big companies, we need more tools such as PGP.

The user should hold the keys, not a government or company.

Technical aspects are generally secondary, and should improve, but we shouldn’t dismiss good approaches due to implementation details.

You're absolutely correct. But unless the tools are designed so the average user can easily manage their own keys, it's basically PGP again.

I think U2F/WebAuthn dongles actually could solve this problem but there are all sorts of new problems now like "how do I use this with my iPhone and also with my PC" or "what happens when lose my (physical) keychain with my dongle".

Not necessarily an endorsement (although I do own one), OnlyKey lets you store and use PGP keys in this way as well as U2F, OTP, etc.

This is an especially funny thing to say when you compare the number of daily users Signal --- itself a niche cryptosystem --- has to PGP.

The Signal protocol is very well designed but the implementation requires a telephone number (I know it is coming). That's a step removed from PGP which can be completely offline.

That's interesting and all, but my point is just that you'll have to stop and think about how many orders of magnitude separate the userbase sizes.

A simpler way to make the same point would be that relative to modern cryptography, to a first approximation, nobody uses PGP.

Maybe only the right people use PGP.

I honestly have no trouble believing that's what PGP's userbase believes about themselves, which is absolutely part of the problem.

Safe communication should be available to everybody. Given how many professional engineers botch PGP at some point, it isn't exactly a great option for people like journalists who may need to depend on it with their lives.

"Don't worry, these few thousand techies are the only ones that need secure communication" is sheer arrogance.

That's fine, but even the "right" people need to interact with the "unwashed masses" sometimes. If your cryptosystem isn't readily accessible to everyone, it's kind of useless.

A good point, but one addition: nobody directly uses PGP.

While Debian and other package signing continues to use PGP (and yes, Debian may move away, and signify/minisign is preferable) there are lots of “infrastructure” uses.

Funny term, “niche cryptosystem”

It’s like a frog, which is about to beat a frog with a wand, frog says, come on, I am already a frog!

OpenPGP implementations suck for usability. All the other messaging implementations also suck for usability.

Having obscure arguments about the underlying cryptography is fun and all, but doesn't really make the world a better place.

Reply on news.ycombinator.com