Quote · Modern Wisdom
Ex-Military Hacker: The Secret World Of Government Surveillance - Bill Thompson - #1131
Where this was said
Is DOGE Really a Mess?
At 29:31 · chapter starts 25:17
The conversation pivots to DOGE and Elon Musk's government efficiency push. Chris describes his interview with Sam Korkos, the Treasury's CIO, who explained the Frankenstein patchwork of 16 incompatible legacy systems underpinning federal finance. Bill goes further with a firsthand account from his last military posting, advising the general overseeing all U.S. Army offensive cyber development [1] — Bill Thompson "In the U.S. Army's offensive cyber program, generals who didn't spend every dollar of their budget risked being fired. The only metric was …" 27:20 . The budget dysfunction was breathtaking: a general who failed to spend every dollar of his annual allocation risked being fired. By July, if $100 million was still unspent, everyone would scramble to burn through it on arbitrary courses and conferences before September's fiscal year end. The actual mission — meeting the president's National Intelligence Priority Framework requirements — was secondary. As long as the money was spent and the reporting numbers could be fudged to a plausible threshold, it was 'successful.' Bill's view: a general who delivered full mission effects at 70% of budget should be getting promoted, not threatened.
In the U.S. Army's offensive cyber program, generals who didn't spend every dollar of their budget risked being fired. The only metric was execution rate — not outcomes. This is what Bill Thompson watched destroy the efficiency of America's most sensitive national security work.
In the U.S. Army's offensive cyber program, generals who failed to spend their entire annual budget faced being fired, incentivizing waste over efficiency.
In the offensive cyber development unit Bill Thompson advised, approximately 85% of staff were civilians rather than military personnel.
Vulnerability research that once took 3 months to a year — dumping a phone, parsing its code, finding exploits — now takes roughly 3 hours with AI. The math behind modern AI was known since the 1970s; what changed was GPU compute and internet-scale training data.
