I would posit that in the past we had a bias to over-classified things and today we bias towards under-classifying things.<p>In the Cold War, when the Soviets got their hands on a few of the latest semiconductors there was a limit to what they could exploit from them, as the massive supporting infrastructure and sunk cost capital required was simply unavailable.<p>Today, decisive military advantage(albeit perishable) can be gained by recombining/reformulating existing commercial off the shelf technologies(albeit mixed with some dedicated .MIL R&D).<p>Back then, a few of our best CPUs in the hands of our adversaries would cause limited damage.<p>Today, a few of our best conversations in the hands of our adversaries could cause significant damage.<p>I’m not one for suggesting we classify everything, that would be silly.<p>But I do think an easy argument can be made for the growing challenges around counter-intelligence.<p>The only solution I see is building, measuring learning faster and out-competing adversaries.<p>Preferably at the cost of the fewest lives lost and least national treasure consumed.