Hacker News Reader

I don't think that the author ever disputed that GPG is a highly secure system when functioning, or that it is "the best we have". Certainly it's the best we have, and that highlights the sad state of affairs.

I think a lot of engineers make the mistake that regular people will be so won over by a new technology that they'll learn it as well as the engineers do. That isn't how it works. So far we still haven't gotten regular people to learn programming, though more people program than ever before.

Humans like simple. This is not a flaw. Software can do anything, and it is written for us. We should not have to learn a bunch of arcane commands, skills, and concepts unless there is no technical way around this.

Ideal encryption would have a quick initial setup, and then each message would have a sticky button for "encrypt?". Then whatever is required in the backend does what it needs to do.

Complex software that gets the job done is much, much worse than simple software that gets the job done. Unless things get so oversimplified that there is no way to verify correct operation. I agree that people need to learn about encryption, but there's a middle ground between no knowledge and knowing enough about it to use GPG.

The problem is that security and crypto specifically, is a complex beast, so often it isn't possible to abstract away the details.

You cannot expect to avail of the benefits of crypto without taking some time to understand how different protocols work, how public key vs symmetric key crypto works, etc.

I agree that the situation could be much better than it currently is, but people can't expect to be given everything for free; there mist be some effort from the other side, which I don't see happening today, especially with the widespread disregard for mathematics in the general public.

I have to say I really disagree. Why would someone need to know how different crypto protocols work? Do facebook users know how TCP/IP works? Or are they just given a series of user interactions required to make the computer do what they want? Couldn't we do this with secure communications software? I don't need to know why a certain crypto method is recommended, for example, if I know that a trusted party recommends it (say - Edward Snowden and Jacob Appelbaum).

We could have an open source crypto program with a set of basic modes for use. Maybe "mode 3" is good for basic interaction, while "mode 5" is recommended for whistleblowers but requires another step in the physical world, while modes 1 and 2 are deprecated because they were discovered to be vulnerable or weak.

The above is stuff a regular person can understand. Do they need to know how these modes work? Well, they need to know what input is required, but they don't really need to know how it works behind the scenes if they trust someone else to recommend the basic modes. Then it must certainly be possible to write software that walked a user through how to complete the required information. It could say "plug in your secure USB key now", then "type in your passphrase" and then "remove USB key to continue".

Really, I think engineers have a hard time seeing how easy we could make things. Yes that would all take a ground up rewrite and open source projects don't do well with that. Yes it would be easier if people would just learn, but that's a fantasy and we should understand that.

It must be possible to make a secure communications "wizard". Disregard for mathematics has nothing to do with it. People want to know what to do and they want to use as few steps as possible. Installing a handful of command line utilities and remembering a bunch of commands isn't the answer. One program with a simple graphical interface must be possible.

You've missed out "Use GPG, but perform one of many steps wrongly, and thus face torture or death".

See also "deanonymizing alt.anonymous.messages"

https://ritter.vg/blog-deanonymizing_amm.html

If the potential to be rendered to a blacksite or tortured to death somewhere less polite isn't sufficient motivation for people to be careful, I'm not sure "security" is their biggest problem.

Using X is preferable to death or torture.

That's what I call faint praise.

faint? I'd rather read 100 pages of man than die.

Seems like a meager price to ensure another 70 years of life.

http://en.wikipedia.org/wiki/Damning_with_faint_praise

The point is when "virtually everything" is better than death or torture, it's not saying much when that's how you praise using GPG.

> The old adage is that perfect is the enemy of good enough, but the author is claiming that GPG isn't even good enough

That's something you say while building it, so that you get something working built instead of debating what should built, or instead of adding so much stuff that you never finish. It's not something you say to discourage people from improving something that's already shipped.

I use GPG often and it gets the work done. It might be the author is not happy with the long process of using GPG and the shortcomings of the documentation. I think if GPG is integrated with more applications, it might have a lot more users.

Actually this is not correct. There are now situations (such as in the United States) where merely using GPG puts you on a list of suspicious people. So not only is GPG not "good enough", but using it today makes your life worse than it was before. There are places where using GPG can lead to death or torture, even if the message is simply your grocery list.

In order for GPG (or its replacement) to fulfill the dream, it needs to be the norm rather than the exception. There are some companies who could prod things in that direction. GPG acting alone is absolutely not the future.

Besides, the cryptography isn't even very good. There is no forward secrecy, for example. I believe that email has to change at the protocol level in order to fulfill the promise that GPG (PGP) makes. And with so many interested in preventing the security of email, well, it seems like secure email is a dead end. And that's a real shame, because I love the idea of being able to securely send a message between two people on the Internet, using open source software and without trusting any intermediates.

> There are now situations (such as in the United States) where merely using GPG puts you on a list of suspicious people.

What's fun: using ssh might be similar.