From Jonas reading, our ability to predict consequences is weaker than our ability to cause them. That short arm of predictive knowledge means cyber policy can not rely on confidence within forecasting. There are several ways to try to fix this issue. One would be to shift from prediction to precaution. Instead of asking what will happen, we could ask what could go wrong. Another way to fix it would be to build reversibility into infrastructure. If we can not predict what the outcomes could be, we need the ability to undo mistakes quickly and efficiently.