Are encryptions like PGP really impossible to break? Such that if I was to encrypt a document with PGP not even organisations like the CIA could break it?

2GET!

This question always makes me wonder too, how can one gain more security if everyone can see how the said security works? How can you guarantee that your contributors and volunteers aren't really saboteurs?

The only way to break a PGP encryption or any protection of any kind is if the password over it is weak.

THe encryption itself is hard to break or rather will take immense amount of time to break through. Look at the encryption that PGP provides as that 7ft thick vault door. In addition to that PGP also provides you with a key to open this vault door and you can decide what sort of key you want it to be.

Now all the hacker needs to do, instead of breaking the encryption, he just insteads finds the correct key to open it. Well this is how I understood it. Am I correct?

PGP reminds me of the time when the FBI/CIA wanted to ban PGP from public use because it stopped them from snoopping ppl's email.

Nothing is impossible to break. AFAIK though, very few have successfully cracked PGP (or it's equivalent, OpenPGP).

I use PGP (actually GNUPG), but mainly to digitally sign emails and other files for authenticity. It's there to verify I am who I really say I am. Very rarely have I used it for encryption.

> Are encryptions like PGP really impossible to break? Such that if I was to encrypt a document with PGP not even organisations like the CIA could break it?

This seems like a simple and straightforward question, but it is not. Nothing involving cryptography ever is. This will be a bit long-winded, bu bear with me.

So. First things first. All encryption is vulnerable to the brute-force attack, where you just try every possible key until you find the correct one. (I lie here, the one-time pad is not vulnerable to this, but it is far too inconvenient to be used for anything but secret spy work.) So, assuming there are no shortcuts (Finding such shortcuts would be what cryptanalysts try to do when trying to break a cipher), the question becomes: How long does it take to try every key? For a symmetric cipher of 128 or 256 bit key length, the answer is: Longer than the age of the universe. It doesn't really matter how much processing power you throw at it, or how fast computers get, the problem is still so overwhelmingly large, it's not going to be broken that way.

However, there is always the possibility of finding that shortcut. And the government agency you want to consider here is not the CIA, but the NSA. And they by far the largest employers of cryptographers, and they are incredibly secretive. It is well known that they knew of vulnerabilities in DES, an older cipher not much in use these days, long before anybody else - during the design phase of it, they suggested changes to the algorithm to make is stronger against this attack that only they knew of, and never explained why these changes should be made. A decade or so later, non-NSA cryptographers discovered the attack, and noticed DES was already strengthened against it. This may seem strange from an agency dedicated to breaking crypto, but they actually have two tasks: Break the ciphers of others, and strengthen US ciphers.

So in short, we don't know if the NSA has discovered attacks against certain popular ciphers in use today in things like PGP. Even with all their immense computing power, they can't break them by brute force. But they may not need to use brute force.

Another matter is that PGP uses both symmetric and assymetric (public-key) ciphers. This latter bunch is a lot shakier. There are a lot better ways to break them than brute force. Therefore, you've got to use much larger key lengths. 2048 bits, 4096 bits, and such. These are a lot more susceptible to finding clever shortcuts. The NSA may have much better methods for breaking these than the rest of the world does. Once again, we don't know.

In summary: PGP is absolutely secure against anybody but the NSA. Against the NSA, we don't know. If you're worried, use as long keys as you can.

I should also mention that all this assumes that people would actually try to break your crypto, instead of just using any number of side channels to get your data: Bugging your room, your computer, bribing the person you sent your mail to, beating you up until you give them the password...

Also:

> PGP reminds me of the time when the FBI/CIA wanted to ban PGP from public use because it stopped them from snoopping ppl's email.

That never happened. You're thinking of the export ban on heavy encryption, which was both silly and served no useful purpose.

Heh, yes... anybody remember the 40-bit "crippled" SSL-enabled browsers of yesteryear?

These days, I enjoy 256-bit AES encryption when checking my web-based email in Firefox. That's better than what most sites provides.

>>2
I am shocked that no one has commented on this one.

But to answer your question: Yes, it is easier to crack a system if you know how it works. However, if the system is hidden, you are trusting the author to both be honest, and that there are no missed flaws. An open system is forced to be much more robust to stand up to inspection by people all over the world. That and if the system is widespread, people can figure out how it works just by examining what their computer is doing during the process. They are annoyingly persistent that way.

As for the saboteurs, well, we just hope there aren't any evil geniuses out there. But I'd rather have a system where people look at it and say, "I don't know how to break this" as opposed to a secret one which I'm just trusting to work.

In the end, of course, it doesn't matter, because someone will just call in and ask the secretary for the password, which the manager taped on his monitor because he couldn't remember it.

Not just that... but also look at the figures as to what programs seem to get the most hits.

Apache Web server is by far, the most popular web server in use today. 60% of all web servers run it. Microsoft IIS comes second, with about 25%.

Apache also happens to be open source, under the "Apache-2.0" licence. IIS on the other hand, is closed source.

If being open source was a security risk, then surely we'd see more worms & attacks targetted at Apache no? Well, IIS admins have much more to worry about than us Apache folk. We just get to sit back and laugh when some poor fool tries fiddling with CMD.EXE on a Unix/Linux server.

So just because you're open source, doesn't necessarily make it any easier to attack. ;-)