Practical cryptography - the key size problem: PGP after years

The size of cryptographic keys belongs to aspects that are quite easily dealt with in theory but that have to be carefully weighted in the practical cryptography. Particularly, current as well as predicted advances in technology must be taken into account as these dictate lower bounds for the key size to ensure reasonable security even with respect to specialized machines. On the other hand too large key size, while secure in theory, brings some implementation problems especially if speed and available storage are important or limited. The key size problem has also some relation to a purely practical problem, namely replacing a cryptosystem whose security level is no more satisfying. Particularly, if a cryptosystem still provides enough strength with respect to known analytical attacks, but advances in technology begin to make brute-force type of attacks possible, some steps need to be done to retain proper level of security. Quite obvious solution, i.e. to simply replace the old cryptosystem by another one is not always the best or even possible one as to devise a good cryptographic algorithm and especially to provide sufficient arguments on its security usually needs considerable time. But from the practical point of view even actual replacement of all installments of the old cryptosystem by new ones with minimal interruption of operations of related applications itself represents challenging problem especially in the case of cryptosystems that are of widespread use (e.g. are formal or defacto standards). Considerably better solution seems to be the one where cryptographic algorithm used offers necessary flexibility with respect to the key size, i.e. the essential architecture of the algorithm remains the same even if the size of its key may be increased.