Pretty Good Privacy was proposed by Zimmermann in the early 1990s. Zimmermann managed to combine RSA, IDEA-or-3DES and the idea of digital signatures into a whole, and give it an interface that was seamless to a normal user of the Internet. His motive was a desire to allow anyone to be able to use encryption, since at the time only companies with large computing resources could use RSA, and the symmetric systems still had the key-distribution problem.
The first part of the solution is to encode the main message using IDEA (or 3DES). This is quick and anyone's computer can cope with the calculations. To avoid the key problem, the key is encoded using RSA. The receiver can now get the key securely, and then decrypt the IDEA/3DES message easily and with no more key-distribution hassles.
Another advantage is that digital signatures can be incorporated. One problem with using the Internet is that it is difficult to be sure if an e-mail is who it says it is from. For example, if I get a e-mail from Paul asking me what time we arranged to meet at the pub, how can I be sure if he sent it, or if his mother sent it in an attempt to lure me into telling her our plans? In a physical letter, there is Paul's handwritten signature, but that can't happen on the computer.
Diffie and Hellman had provided the solution years before, and Zimmermann added it to PGP. This solution used RSA. Paul has encoded his message using my public key, so that only I can decode it. But he can also encode a message using his private key. This can be decoded by anyone with Paul's public key (like the inverse of the normal system). This is pointless for normal e-mails, since everyone has access to Paul's public key, but for signatures it proves to me that Paul sent it, since he is the only one who knows his private key, and so is the only person who can encode a message that can be decoded by his public key.
PGP performs all these functions automatically, and so can be used by anyone using a computer to send information regardless of computing power or computing knowledge. This is good for the cryptographers, who are now ahead in the age-long battle against the cryptanalysists. But it causes a new battle.