Should one private company control infrastructure critical to a nation's war effort?
No single company or state should hold a kill switch over a nation's war effort. Ukraine needed multiple satellite providers bound by enforceable contracts, not one founder's discretion.
Ben-Gurion names the real failure: Kyiv cannot fight a war it does not fully control, and Starlink throttling proves the point. Hayek and Friedman agree the danger is concentration itself, not who sits in Musk's chair; Friedman notes the Reuters report that the White House also pushed to throttle the service, denied by Musk, Trump and Rubio, meaning a state would not have fixed this either. Ostrom turns that diagnosis into a design: her fieldwork on fisheries and irrigation systems found durable systems rely on multiple overlapping providers, not one owner or one ministry. Hirschman pushes the timeline back further: Ukraine accepted an irreversible dependency on StarShield in wartime with no alternative lined up, which is the real error, not the 2026 throttling itself.
The split is over what replaces discretion. Ben-Gurion wants the state to take the switch back, while Hayek, Friedman, Ostrom and Roosevelt all warn that just moves the same danger into a ministry or a committee.
Confidence summary: The council holds high confidence that concentrated control over war-critical infrastructure is dangerous in principle; it holds much lower confidence, and open disagreement, about which institutional fix actually removes that danger rather than relocating it.
1. The core argument
The Guardian's framing puts a face on the danger: Musk, one man, deciding whether Ukrainian drones fly. The council's sharper claim is that the face is a distraction. The 2026 throttling episode, where Reuters alleges the White House itself requested the restriction, a request Musk, Trump and Rubio all deny, shows the vulnerability was never Musk's personality. It was the architecture: one provider, one node, no alternative, no enforceable contract specifying what happens during a Russian offensive. Nationalising StarShield would not fix that; it would hand the same unilateral switch to a different actor, this time a ministry answerable to different pressures but no more structurally constrained. The real failure, on this reading, happened earlier than the blackout itself: in the decision to build an entire war effort on a single, irreversible point of dependency with no exit and no binding wartime terms attached.
2. How each member frames it
David Ben-Gurion reads this through the Altalena crisis of 1948, when he ordered the IDF to sink a rival arms shipment rather than accept two armed forces inside one state. His standard is blunt: a state at war cannot outsource its own command function, full stop. He would accept the cost of a slower, state-built alternative over any arrangement where Kyiv must ask permission to fight. He does not dispute that state control can be abused; he simply ranks a flawed sovereign switch above a foreign-owned one.
Friedrich Hayek does not defend Musk so much as refuse Ben-Gurion's remedy. Drawing on his 1944 argument against central planning, that no single authority can aggregate the dispersed knowledge a market does, he treats nationalisation as a category error: it swaps a Texas boardroom for a Kyiv ministry without solving the underlying problem of one actor holding total control. His limit: he offers no account of what happens when the market itself produces exactly one dominant supplier, which is precisely Ukraine's situation.
Milton Friedman sharpens Hayek's point using the Reuters detail the others treat more cautiously: if the White House really did push for the throttling, then the danger was two concentrated powers, private and state, colluding over a third country's war, not private power alone. His 1962 warning that concentrated power is never neutralised by good intentions cuts both ways, against Musk and against Washington. He is candid that this leaves the practical question unanswered: concentration is diagnosed, not cured.
Elinor Ostrom moves from diagnosis to design, drawing on her fieldwork across fisheries and irrigation commons on six continents, where durable governance always rested on multiple overlapping authorities rather than one owner. Her polycentric fix, OneWeb, EU satellites, national systems, bound by pre-agreed wartime rules, is less a critique of Musk than an indictment of Ukraine's procurement choice. Her candid limit: polycentric systems take years to build and Ukraine needed coverage in 2022, not 2028.
Albert O. Hirschman locates the error earlier still. His 1970 argument that organisations decay without a credible exit option reframes the entire episode: the 2026 throttling was not the failure, accepting a single irreversible supplier was. He holds irreversible commitments to a far higher burden of justification than reversible ones, and judges Kyiv to have crossed that line in wartime without building the alternative that would have disciplined StarShield's behaviour from the start.
Eleanor Roosevelt presses past Hirschman's procurement critique to the legal vacuum beneath it. Having fought in 1948 to make the UDHR enforceable rather than aspirational, she treats a founder's private assurance not to throttle a warzone as exactly the kind of empty declaration she spent her career opposing. Her fix is contractual, not institutional: penalty-backed obligations specified before the crisis, since goodwill is not a governance structure.
3. Where the council agrees
The least obvious agreement is that Musk's personal conduct is almost beside the point. Every member, including those most alarmed by the throttling, treats the 2026 episode as a symptom of a prior design failure rather than a verdict on one man's character. That reframes the Guardian's central claim: the danger is not that Musk is uniquely untrustworthy but that any single controller, public or private, occupying that position would pose the same risk. The council also converges on rejecting both of the obvious fixes: pure nationalisation (Ben-Gurion's instinct) and pure continuation of private discretion are both judged inadequate, since neither addresses concentration itself. Finally, all six treat the Reuters allegation of White House involvement as significant regardless of its truth, because it demonstrates that state involvement does not automatically produce accountability either.
4. Where the council splits
The real fracture is over what replaces Musk's discretion. Ben-Gurion wants the state, specifically Ukraine's own state, to reclaim the switch, treating sovereign control as a baseline good even if imperfectly exercised; the Altalena precedent for him settles the matter. Hayek, Friedman, Ostrom and Roosevelt all reject this, arguing a nationalised system merely relocates the single point of failure into a ministry or committee with no better structural guarantee against abuse, and Friedman's reading of the Reuters report, that the White House itself sought the throttling, is offered as evidence state control is no safer. Both sides have real ground: Ben-Gurion is right that redundancy and contracts take years Ukraine may not have; his critics are right that a sovereign switch is still a single switch.
5. For a policymaker to decide on
The choice is between speed and structure. A policymaker can restore state control over Ukraine's battlefield connectivity now, accepting Ben-Gurion's logic that sovereignty cannot wait, even though this likely means a slower, costlier, domestically-built system vulnerable to its own single-point failures. Or they can commit to Ostrom's polycentric model, multiple providers bound by penalty-backed wartime contracts, which better survives any one actor's defection but cannot be built before the next crisis arrives. No member resolves which risk a state at war should accept first.