Nestor G Pestelos Jr · Writing

The Only Concrete Escape From Triffin's Dilemma Doesn't Work

Published August 14, 2026

TL;DR

A Socratic AI session correctly diagnosed Triffin's Dilemma's core tension, but the one concrete fix it built, a Bitcoin-pegged stablecoin replacing the dollar, fails on mechanism (see Terra/UST) and on scale (crypto is too small).

I spent a long session with Grok working through Triffin's Dilemma: the old observation that a country issuing the world's reserve currency must send out more dollars than it takes back to supply global demand, and that doing so eventually undermines the currency at home. Grok ran it Socratic, each question building on my last answer, until it paused the thread to ask me to reflect: what had I uncovered? My own answer surprised me: "You can't have a central entity if you want to escape the dilemma." I hadn't been steering toward that conclusion. I just kept answering until I landed on it.

That's a sharp diagnosis. It's also not as settled as it sounds.

The diagnosis holds, and it's contested

The mismatch I landed on is real: "There's a central entity that controls the supply of the world's reserve currency, but it has less control over how these dollars flow around the world." An ECB Executive Board member reached the same root cause from a very different seat. In a 2011 speech, Lorenzo Bini Smaghi argued reserve-currency issuers hold something close to monopoly power, and "over time they tend to take advantage of other countries' high dependence on their domestic money," with nothing in the system forcing the issuer to prioritize long-term stability over short-term domestic incentive.

But Triffin predicted this mismatch would eventually force a crisis of confidence, and it hasn't, for over sixty years past the 1971 collapse of Bretton Woods. A standing rebuttal, not a fringe one, predates this diagnosis by six decades: economists Despres, Kindleberger, and Salant argued in The Economist on February 5, 1966, that the US functions as the world's banker. A contemporaneous Princeton review of their argument quotes it directly: "The United States is no more in deficit when it lends long and borrows short than is a bank when it makes a loan and enters a deposit on its books." "Triffin's original forecasts failed," as one summary of the debate puts it. "The world did not collapse into deflation, and the dollar survived the end of Bretton Woods." Whether it's a crisis waiting to happen is a live, unresolved argument in the actual literature, and neither this session nor this piece settles it.

The same asymmetry cuts both ways for the US: buying cheap imports keeps trade partners supplied with the dollars they need to service debt, giving them a reason not to retaliate - leverage and a trap in the same mechanism.

A review of Barry Eichengreen's Exorbitant Privilege locates a similar trap through a different channel: not trade partners' restraint, but the absence of anything that forces US fiscal discipline. Reading these as two different mechanisms converging on one structure is mine, not a claim the review itself makes. And leverage doesn't run as cleanly as either version implies. When the US raised tariffs through 2025, foreign countries retaliated about half as hard, not not-at-all, and the trade war corroded the dollar's centrality on its own, independent of anyone's retaliation calculus.

The one design this produced, and where it broke

The same Socratic push then tried to build a way out: strip the dollar to domestic-only use, settle global trade in a new currency pegged to Bitcoin for its fixed supply, through a Bitcoin-pegged stablecoin - a token engineered to hold steady value rather than track Bitcoin's own price swings, its supply expanding or contracting against demand, backed by a treasury. Grok asked what that treasury would hold, and how minting and burning could stay anchored to Bitcoin's cap without becoming a new central controller. I didn't have an answer - the exact structure the design was built to remove.

Two separate tests, neither hypothetical, break the design:

The same session also produced an opposite bet: use a Bitcoin reserve as collateral to extend the dollar's reach rather than replace it. This is where Grok stopped asking and started explaining: its own read names the limit directly, that the stablecoin's "primary function would be to extend the dollar's usability in a digital format rather than fundamentally altering its dominance." Triffin's underlying tension, needing enough reserves while keeping sole control, stays intact. And the one US Bitcoin reserve that exists, established by executive order in March 2025, is capitalized only with Bitcoin already seized through criminal and civil forfeiture; the order states plainly that it "shall not be sold." No provision authorizes using it as collateral, and a reserve barred from sale can't defend a peg either.

What held up

Both halves of the diagnosis hold under real evidence, with real cracks on each. The proposed escape fails on the one large-scale attempt at its exact mechanism and on the actual size of the market it would need to replace. A currency basket, an SDR-style system, or any other multipolar exit might fail differently, or not at all. This piece tests only the one design I built with Grok's questions, and it breaks two independent ways before its own fallback runs into a legal wall.

What stayed with me was the shape of that first stretch: the diagnosis, then the design. Grok never told me the answer there. It asked until I found the diagnosis myself, then kept asking until I hit my own wall on the mechanism and admitted it. It never filled the gap for me. Grok's mode changed later in the session: it switched from asking to explaining, and the passage above on extending the dollar's reach is Grok answering, not questioning. But through the diagnosis and the design, the questions never stopped. Naming a problem correctly and building something that fixes it turned out to be two different skills. I only learned that because nothing in that stretch let me skip from one to the other.

Sources