Pfft. It's because of academics like @matthew_d_green that we still have a strange value for Pi like 3.1415926 rather than a simpler value like 3. We could solve the Pi problem if they weren't letting ideology get in the way.
This is the argument they make. They claim if only cryptographers were doing their job instead of bowing to ideology that we can solve the crypto backdoor problem. That's obviously a silly argument.
2/ Crypto is about making you secure against eavesdroppers, and backdoors is about eavesdropping on your unencrypted data. There is no way to have a backdoor that doesn't make you less secure.
3/ Sure, there might be compromises, "secure enough", so it's not unreasonable to expect some sort of tradeoff. But right now, we are so bad at crypto that we can't even do that.
4/ The state of crypto and software is this: we are really, really bad at this. After 20 years of experience with SSL, for example, we are still releasing new versions to fix the problems in the old versions.
5/ No matter how simple a "backdoor", as soon as you add one, you introduce so much complexity as to add bugs. Maybe in the future we can get good enough at it to make acceptable backdoors, but that's not today.
6/ Moreover, it's not just code, but humans. The problem with crypto is that we aren't solving just math, but issues like usability, so that users don't shoot themselves in the foot. Ozzie's approach is a good first step in solving some of the problems, but is still insufficient.
7/ Lastly, there is still the issue that we aren't solving for United States, but for China, Russia, and Iran, where corrupt police oppress citizens. Assuming a perfect compromise on backdoors that works in the U.S., this will only harm dissidents throughout the world.
8/ Even our own police aren't trustworthy. This is demonstrated by their claims of "going dark" in the "golden age of surveillance". They have vastly more evidence from our electronics than they ever had in the past, and they claim the opposite, because of the THEIR ideology.
