I'm not saying diversity is good, I'm saying that the community's preference for a single strand of work does a lot to explain why real crypto systems are based on a very small number of problems.

But, if these problems were found to have solutions that resulted in practical attacks, I don't think we would have as much difficulty finding a new problem to hang crypto on. Or at least not as hard as this article makes out by suggesting there are only 2 mathematical problems underpinning all of public key crypto.