Follow

I feel slightly embarassed that I ever taught classes on PGP.

@rysiek So many of the cryptographers I've followed over the years, whom I learned about PGP from and through, are now spending time slamming it and calling it a once, now, and always failure.

Maybe I missed a memo someplace.

@drwho @rysiek I'll sit myself in the 'pgp is and has always been a tool suited to particular uses and if folks would understand which mathematical properties were and weren't guaranteed they'd be a lot less peculiar about it' boat. It's probably a pretty lonely one but... so much of the argument both for and against in the wider techsphere seems to consist of 'but humans can't understand numbers and select the crypto properties they want correctly' which :eyeroll: is an education issue.

@feonixrift @drwho usability is crucial, and there is plenty of valid criticism of the OpenPGP ecosystem.

Quite honestly, GnuPG at some point started being a problem, rather than a solution. Hence projects like @sequoiapgp , who are doing some amazing work with making OpenPGP, well, usable.

At the same time, it remains a useful and important tool. People ain't going to move off of e-mail, because there's no-where to move to:sequoia-pgp.org/blog/2021/06/2

@rysiek
That's a great article! This point is one I rarely see mentioned:

there is a fair amount of evidence that people who have a pressing need can use OpenPGP successfully

I'm glad I'm not the only one banging that drum. And thanks for turning me on to Sequoia PGP, which I was unaware of!

@feonixrift @drwho @sequoiapgp

@drwho i keep it as my guilty pleasure. mhhhh, let's extend that subkey, yeah…

@orko I keep it around just in case, but I doubt I'll ever use it for e-mail again.

@drwho is it because in retrospect, PGP was clunky and overall not very secure? or another reason? or is this not at all about the pretty good privacy encryption and something else entirely lol

@root Clunky as fuck? Sure. So much so that the UI shells people built for them suck almost as much.

Not very secure? We could argue that. Threat models and use cases and all that. Anonymity was possible but a hidden feature for at least 20 years so just about nobody used it.

Nope, it's about PGP and GnuPG.

@emacsen @drwho Because it's a usability nightmare, which makes it less secure and also dramatically reduced its ability to have any impact, at best, and at worst it taught people that encryption is hard so they might as well not even try?

@freakazoid @emacsen That's a fair observation. As a turbo nerd sorting function it works quite reliably.

@freakazoid @drwho

I've thought a lot about this over the last 4 years or so, and here are my thoughts in a nutshell (maybe I should write this up):

1. PGP would be simple if it weren't for things like WoT. Without WoT we don't need to worry about signed keys and chains and any of that.

2. We should have just accepted some defaults ala P3P such as approve new keys automatically and made that tunable for extra security by others.

...

@freakazoid @drwho

3. Purists are a real problem. Look at this comment on Hacker News in response to my mobile OS review:

news.ycombinator.com/item?id=2

This commenter would prefer to tell people *not to use mobile phones* than let them use FLOSS OSes with binary drivers.

These attitudes put people in harm's way.

4. People believe that security = complexity. There's a paper on OCAP as implemented in HP that talks about this issue. It's a serious problem, and PGP absolutely fit into that mindset.

@emacsen @drwho I think it's much better to just tell people to be mindful of their use of technology and to realize that nothing is perfect.

@freakazoid @emacsen @drwho Another barrier to GPG adoption was that mainstream email clients (Gmail, Outlook and, yes, even Thunderbird) have supported other useless encryption schemes like S/MIME. GPG plugins and extensions have been around for a long time, but each one came with interoperability or usability issues.

Finally Thunderbird has built-in GPG support (still very immature, and using its own key store rather than the system's).

@codewiz @emacsen @drwho The number of people who use any email client other than the web & mobile client provided by their email provider is vanishingly small, so native support in Thunderbird is irrelevant.

@freakazoid @codewiz @emacsen Yeah.

Largely because it's much, much easier than trying to configure a client on four different devices. Hell, even I do it.

@drwho @freakazoid @emacsen Yes, finally. But they went the way of reimplementing the full-blown OpenPGP spec rather than use GnuPG (or GPGME).

It's probably good enough for a beginner user, but it doesn't support all the key formats and algorithms of GnuPG and requires manually importing/exporting private keys if you use PGP with other applications.

#thunderbird #gpg #pgp #openpgp

@codewiz @freakazoid @emacsen That's...

<sigh>

It traded off one set of UX problems for another.

@drwho @emacsen @freakazoid @codewiz

I don't care about people. GPG is amazing because it's widespread integrated. I have my private key on the yubikey. It is used for email, for my password manager, ssh authenticates over gpg as well. Authentication withe remote machines work by propagating the key exchange back to my client...

I can authenticate everywhere local and remote without my private key ever touching any computer.

And just by piping things through gpg I can secure stuff and read it again without entering a passphrase again. Again, it doesn't matter if this happens locally or 3 ssh hops away.

This is a level of convenience I've not reached with any other tool.

Yes, I also use gpg for email encryption, but most of the time it's authentication and file/credentials encryption and decryption.

The same setup works with macos and windows as client.

Is it easy to set up? For me: yes. For someone who has no clue about linux/bsd: no.

Most critiques about gpg usability seem to assume that it can only be used if it is usable by the average joe. I mean... it can be... look at DeltaMail for exaple. But this is very short sighted. GPG is a versatile toolchain with many uses which hasn't seen any competition as a whole yet.

@sh @drwho @freakazoid @codewiz

I'm going to keep my response short.

> I don't care about people.

If you don't care about people then you misunderstand the deeper need we have for these technologies.

Encryption must always exist in a sea of encrypted data or else the only ones who use it will be suspect and marginalized.

Only by normalizing the use of these technologies do we gain the protections of being in the crowd and having the support of the public.

An ivory tower is vulnerable.

@sh @drwho @freakazoid @codewiz

Of course this doesn't even touch on the moral imperative to actually help people. That's even more important, but is a separate argument altogether.

@emacsen @codewiz @freakazoid @drwho

One is tech, the other is a social problem. I don't think it's gpgs fault to not be the tool that can solve this social problem directly.

I'm tired of people bashing on a great toolchain because it doesn't fit their use case.

GPG is on the level of SSL. But ssl is way more clunky to use and to understand. Nobody bashes on that. Gpgme is a good library and stuff built on it can just work. But nobody seems to care because people are creating this gpg is bad bias.

@sh @emacsen @codewiz @freakazoid

> But ssl is way more clunky to use and to understand. Nobody bashes on that.

Um... I hate to tell you this..

@sh @drwho @freakazoid @codewiz

Okay, maybe another response, because you've conflated many things. You can use your private key for many things aside from email, but that's not what PGP was made to do. It was designed for email, as Phil Zimmerman said himself: philzimmermann.com/EN/essays/W

And then he codified in RFC 1991 (since obsoleted).

We measure PGP's deployment success around email because that was its stated purpose.

@emacsen @codewiz @freakazoid @drwho

You could also argue that email itself has failed. We're approaching the first generation without email accounts. Or one just for legacy things.

@sh @codewiz @freakazoid @drwho

You can successfully argue that email is an obsolete technology but not that it's failed.

Nearly all people who have access to the Internet have email, and the number of people who have email increases over time both in hard numbers and as a percentage of the population.

The number of people who use PGP has largely stagnated or fallen, except for Protonmail users, but their implementation doesn't interoperate well outside of Protonmail.

@emacsen @drwho @freakazoid @codewiz

You can successfully argue that email is an obsolete technology but not that it's failed.

That's fair.

Yes, I think PGP failure in this is not on PGP, but on major email clients refusing to use/integrate it properly. Alternatives like s/mime are even less wide spread.

Email encryption just never has been a priority. I don't think it's fair to blame the library.

@sh @emacsen @drwho @codewiz In order to be an obsolete technology, email would have had to have been obsoleted by something. It has not.

@emacsen @sh @codewiz @freakazoid It seems to work decently well. Before I migrated my domain to Protonmail I did some interop tests (at the time, our mail was hosted at a Google Domain, so the GnuPG workflow involved CnP into a text file) and it was decent.

@drwho @emacsen @sh @codewiz I just wish they weren't so stingy with aliases and they still supported their IMAP gateway.

@freakazoid @emacsen @sh @codewiz I think it depends on what hosting package you get. I've got the Visionary package for virtadpt.net, and while I'm limited on accounts (just seven) I'm pushing 50 aliases on my personal account and still have some room left over.

@sh @codewiz @freakazoid @drwho @emacsen

Not OSS, but symantec provides transparent pgp encryption for outlook/exchange. My company used that to encrypt internal emails (100k employees). This worked flawlessly.

I really with that would be oss.

@sh @drwho @freakazoid @codewiz @emacsen I'm not ready to buy that. What other reliable and distributed messaging system is there that has any kind of penetration and is still being actively maintained?

I think we should stop conflating "not novel" with "obsolete".

@BalooUriza @drwho @freakazoid @codewiz @emacsen

I agree. And I used the term on purpose, because it's equally wong for email as it is for pgp/gpg. Both is in use, both is not "hip". There's bashing on both topics out there. Not completely unfunded, but incomplete. And we should know better than joining the choir. Also xkcd 927.

@sh @BalooUriza @drwho @codewiz @emacsen I could not disagree more with "equally wrong for email as it is for pgp/gpg". PGP is used by at most 1% of Internet users, and that's probably an overestimate.

@sh @BalooUriza @drwho @codewiz @emacsen Also, XKCD 927 is descriptive, not prescriptive. Its lesson is certainly not "you should never create new standards because some existing standard is always adequate for all use cases."

@BalooUriza @sh @drwho @codewiz @emacsen While I also objected to "obsolete" and "failed", if you take a global view I'm not sure even a third of global Internet users has a working email address they actually know. That's why both Facebook and Twitter are usable with only a phone number. SMS has much higher penetration.

@BalooUriza @sh @drwho @codewiz @emacsen There are 4 billion email users according to 99firms.com/blog/how-many-emai . There are more active mobile phones in the world than people according to en.wikipedia.org/wiki/List_of_ . For there to be less penetration than email, that would have to be almost 2 mobile phones per person *on average*.

@freakazoid yeah, and not even all mobile phones have SMS at this point, since having internet service solves the same problem way more flexibly.
@sh @drwho @codewiz @emacsen

@BalooUriza @sh @drwho @codewiz @emacsen There being better alternatives is different from a phone's not supporting SMS. I'm not aware of any mobile phone numbers that are not reachable via SMS. Do you have a reference handy?

@drwho @BalooUriza @sh @codewiz @emacsen It's not uncommon, but I don't think it's common enough for almost 8 billion mobile phone numbers to be in the hands of fewer than 4 billion people.

@BalooUriza @freakazoid @sh @codewiz @emacsen Going back to its origins as a degenerate case of telephony signalling, SMS has incredible penetration. It was only in the early 2000's that USian providers started making it available to customers (at usurious prices, unlike the civilized world).

@emacsen @freakazoid Having many fewer command line options would have been nice. Most of them were completely irrelevant to any everyday use case.

@emacsen So many of the people I learned about it from and through now spend all of their time calling it a failure.

I feel like I got suckered, or I missed a memo or something.

Sign in to participate in the conversation
hackers.town

A bunch of technomancers in the fediverse. Keep it fairly clean please. This arcology is for all who wash up upon it's digital shore.