The uncomfortable lesson is that the United States still has capital, science, and strategic intent, but in too many sectors it has lost the industrial middle layer that converts all three into usable capacity. That middle layer is exactly where China’s system has become most powerful.
The governance gap is what makes me skeptical the dashboard will actually land. CEI% and ME are well-constructed, but no single principal owns the aggregate number. DOE has the stockpile, Commerce runs BIS, DOD owns DPA Title III. A metric without a named fiduciary becomes a publication, not a lever. That is the same accountability fragmentation that made corporate Scope 3 disclosure slow to mature. Everyone discloses, no one is on the hook for the trajectory. Worth proposing a single Senate-confirmed officer who owns both numbers and reports annually under oath.
Overall, I can see why the judges liked each of these so much, the two released are both excellent. Because I wrote recently about transformers, I noticed a gap in that portion. In fact, supply could be scaled quickly—if the design were standardized. India has lots of excess capacity we could tap if utilities were forced to adopt a standard design instead of waiting for custom ones. It would have other tradeoffs—do all the other parts they need work with the standard and how much of a TCO hit is there over the expected life of the unit from not customizing it—but there are scaling options to deploy that utilities are ignoring.
The scoring methodology is useful for congressional testimony. Less useful for predicting which dependencies actually break under stress.
Here's the problem: a chokepoint score measures market structure at rest. It doesn't measure what happens when the chokepoint activates. Rare earth separation capacity scores high on concentration — China controls roughly 85-90% of processing. But the actual timeline to build competitive separation capacity outside China is 7-12 years minimum, not because of market structure, but because the workforce, the tailings management expertise, and the co-located chemical supply chains are in Jiangxi, not Wyoming. Similarly, pharmaceutical API dependencies don't score identically to semiconductor packaging dependencies, even if both show 70%+ China concentration. APIs degrade and require continuous production. Chips can be stockpiled. A scoring system that collapses both to the same vulnerability number has lost the physical information that matters. The deeper issue: substitutability scores tend to be assessed at the product level, not the process level. Western analysts consistently find alternative suppliers for the end product while missing that the alternative supplier's upstream inputs are also Chinese. Vietnam assembles semiconductors. But the lead frames, the bonding wire, the mold compound — trace those back.
The re-routing is real. The decoupling is not. I have argued in 「The Great Re-Routing」 that friend-shoring creates the appearance of diversification while the physical dependency migrates one layer deeper in the stack, invisible to customs data and, apparently, to most scoring frameworks.
A chokepoint map that doesn't show the upstream of the upstream is a map of the shoreline, not the ocean floor.
Wouldn't it be wonderful if all this effort could be directed to feeding, housing and the health of the world rather than its destruction? Just asking for a friend....
The uncomfortable lesson is that the United States still has capital, science, and strategic intent, but in too many sectors it has lost the industrial middle layer that converts all three into usable capacity. That middle layer is exactly where China’s system has become most powerful.
The governance gap is what makes me skeptical the dashboard will actually land. CEI% and ME are well-constructed, but no single principal owns the aggregate number. DOE has the stockpile, Commerce runs BIS, DOD owns DPA Title III. A metric without a named fiduciary becomes a publication, not a lever. That is the same accountability fragmentation that made corporate Scope 3 disclosure slow to mature. Everyone discloses, no one is on the hook for the trajectory. Worth proposing a single Senate-confirmed officer who owns both numbers and reports annually under oath.
Overall, I can see why the judges liked each of these so much, the two released are both excellent. Because I wrote recently about transformers, I noticed a gap in that portion. In fact, supply could be scaled quickly—if the design were standardized. India has lots of excess capacity we could tap if utilities were forced to adopt a standard design instead of waiting for custom ones. It would have other tradeoffs—do all the other parts they need work with the standard and how much of a TCO hit is there over the expected life of the unit from not customizing it—but there are scaling options to deploy that utilities are ignoring.
The scoring methodology is useful for congressional testimony. Less useful for predicting which dependencies actually break under stress.
Here's the problem: a chokepoint score measures market structure at rest. It doesn't measure what happens when the chokepoint activates. Rare earth separation capacity scores high on concentration — China controls roughly 85-90% of processing. But the actual timeline to build competitive separation capacity outside China is 7-12 years minimum, not because of market structure, but because the workforce, the tailings management expertise, and the co-located chemical supply chains are in Jiangxi, not Wyoming. Similarly, pharmaceutical API dependencies don't score identically to semiconductor packaging dependencies, even if both show 70%+ China concentration. APIs degrade and require continuous production. Chips can be stockpiled. A scoring system that collapses both to the same vulnerability number has lost the physical information that matters. The deeper issue: substitutability scores tend to be assessed at the product level, not the process level. Western analysts consistently find alternative suppliers for the end product while missing that the alternative supplier's upstream inputs are also Chinese. Vietnam assembles semiconductors. But the lead frames, the bonding wire, the mold compound — trace those back.
The re-routing is real. The decoupling is not. I have argued in 「The Great Re-Routing」 that friend-shoring creates the appearance of diversification while the physical dependency migrates one layer deeper in the stack, invisible to customs data and, apparently, to most scoring frameworks.
A chokepoint map that doesn't show the upstream of the upstream is a map of the shoreline, not the ocean floor.
Wouldn't it be wonderful if all this effort could be directed to feeding, housing and the health of the world rather than its destruction? Just asking for a friend....