I was having more of a think about this, one of the issues with LCG's is that they're linear, which means that even relatively basic cryptographic analysis can defeat them. The issue here particularly is that, say you have a cryptoscheme where you encode ascii characters. 95% of your plaintext data has one of the two following formats in binary:
011xxxxx
010xxxxx
And of that, about 70% is:
0110xxxx
0100xxxx
This means with a deterministic seed, you're giving away 3 known bits of state away with every with every encrypted letter, and a good chance at a 4th
That analysis does not require a particularly deep understanding of the cryptosystem to be able to perform, known plaintext is going to be one of the first things that anyone tries. Its less complicated than many crypto schemes that were broken
Disclaimer: I have no idea how this system encodes things as the technology predates my knowledge, and it might not be so straightforward as this
011xxxxx 010xxxxx
And of that, about 70% is:
0110xxxx 0100xxxx
This means with a deterministic seed, you're giving away 3 known bits of state away with every with every encrypted letter, and a good chance at a 4th
That analysis does not require a particularly deep understanding of the cryptosystem to be able to perform, known plaintext is going to be one of the first things that anyone tries. Its less complicated than many crypto schemes that were broken
Disclaimer: I have no idea how this system encodes things as the technology predates my knowledge, and it might not be so straightforward as this