Normal” Rates in an Abnormal Debt Economy The Wall Street pundit class insists interest rates have simply “returned to normal.” This is a glib phrase masquerading as analysis. The United States carries an extraordinary public and private debt burden, accumulated during the zero-rate era. In such an economy, a given policy rate does not have its old meaning. Higher rates do not merely restrain marginal speculation or cool excess demand. They raise the cost of servicing an enormous inherited debt stock as maturities roll over—hitting households, businesses, commercial property owners, and above all the federal government. The thesis is simple: extreme debt levels are the forcing function that requires a structural adjustment in the economy. That adjustment is not a policy error in itself. It is the necessary consequence of borrowing too much, for too long, on the assumption that cheap capital was a permanent entitlement. The correction of a debt-dependent system is painful precisely because it exposes activities, asset prices, and fiscal commitments that could survive only under artificially suppressed borrowing costs. In this respect, the adjustment is normal. Pretending otherwise is not realism; it is denial. But there is a crucial distinction between allowing an overdue adjustment to occur and driving it recklessly with monetary policy that refuses to account for changing debt sensitivity. Rate-sensitive sectors already exhibit recessionary conditions. Housing affordability has been devastated by high mortgage rates; residential construction remains constrained; commercial real estate faces persistent refinancing pressure; consumer durables are burdened by costly credit; and small businesses confront tighter bank lending alongside higher debt-service costs. The federal fiscal position compounds the problem. Higher yields raise interest expense, which widens the deficit, which requires more Treasury issuance, which can sustain upward pressure on yields. This feedback loop is not theoretical. It is elementary arithmetic. Edward Gibbon understood that great systems seldom collapse because of one dramatic event. They decay through the cumulative effects of fiscal strain, institutional complacency, and a governing class that mistakes temporary endurance for permanent strength. The United States is not Rome, and historical analogies should not be abused. But Gibbon’s central warning remains relevant: accumulated obligations eventually narrow a state’s room for error. The Fed’s latest hike suggests it has learned little from that constraint. Kevin Warsh appears less interested in monetary theory, credit transmission, or the lagged effect of tightening than in mechanically following the emotional churn of prediction markets. Markets are useful signals; they are not a substitute for judgment. If markets anticipate at least three further hikes, that is not proof those hikes are wise. It may instead be evidence that policy credibility has become confused with policy inertia. The consequences will emerge through weakening credit creation, refinancing failures, deteriorating property markets, and a fiscal burden that becomes increasingly difficult to finance. By then, the pundits who called this “normal” will again wonder why no one saw it coming.

Normally higher interest rates put negative pressure on an economy. But @jvisserlabs believes higher rates won’t have that effect now because of the AI industry. Do you agree with him?

// Normalized Low-Rank Adaptation (NoRA) // They propose a one-line change to LoRA that costs nothing and improves convergence, stability and forgetting. LoRA initializes the up-projection to zero, which means early optimization is governed almost entirely by the down-projection. This observation tells you where to regularize. NoRA normalizes the down-projection matrices during training. The authors also show the same normalization applied once at initialization improves standard LoRA without repeating it through training, which is the cheaper of the two options. The benefits hold across pretraining, supervised fine-tuning and reinforcement learning. Faster convergence, better final performance, more stable training, and less catastrophic forgetting. It adds no trainable parameters and no inference-time computation, which is what makes it broadly applicable rather than another specialized LoRA variant. Paper:

Normal people think Bitcoiners are obsessed. We are obsessed. For good reason. We understand the PROBLEM at hand. You have spent 11 consecutive years strategically accumulating Marriott points so you can someday upgrade to a courtyard-facing room in Tampa. I am attempting to escape monetary debasement. We are not the same.