The immune system at its core is basically rolling dice like crazy for new random RNA source code that is compiled into molecule hardware that will hopefully work against the invasor somehow. It even has a mechanism to deposit samples of the invasor were the freshly rolled candidates are let loose. Apparently it's buying into the TDD paradigm. The dice are probably not rolled to try completely random RNA sequences with a D4 for each base, more like mashups of the existing library from previous infections. Different immune systems can come up with different solutions to the same virus and even the same solution can be stumbled upon very early or reached very late, only hours before EOL. The content of the existing library likely plays a decisive role and even more does the general bandwidth of mashup attempts. This apparently shrinks drastically from children to elderly. Maybe an evolutionary tradeoff to compensate for the weak library children have, maybe a side effect of the existing library.

(source: mostly stuff I read linked from the hn page, hopefully without excessive misreading)