The Architecture of Extraction
The United States tax system is often described as having a progressive structure: those who earn more pay a higher percentage of their income in tax. This is technically accurate with respect to the federal income tax bracket structure. It is deeply misleading with respect to the tax system as a whole. The full architecture of taxation — federal, state, local, transaction-level, property-based, estate-based — is not progressive across the full distribution of wealth and income. It is progressive in the middle and regressive at the extremes: heaviest in cumulative burden on wage earners in the middle class, and significantly lighter on the very wealthy whose income is largely non-wage and whose wealth is structured to minimize transaction-level taxation.
The structure is not primarily the result of a deliberate design to burden the middle class. It is the result of a 110-year accretion of tax policies, each designed to serve some constituency at some moment, that have aggregated into a system whose cumulative effect was never fully designed or evaluated. The income tax was introduced to replace tariff revenue. FICA was introduced to fund social insurance programs. Sales taxes were introduced by states to fund public services. Each transaction-level tax has its own legislative history. The interaction effects — the way these taxes compound against the same economic value as it moves through the economy — are largely undocumented in public discourse and rarely appear in tax policy analysis that focuses on individual tax instruments rather than cumulative burden.
The question is not what percentage of income you pay in tax. The question is how many times the same dollar is taxed before it reaches its final destination — and the answer varies by at least an order of magnitude depending on whether that dollar was earned as wages or inherited as appreciated stock.
The Toll Chain — Tracing a Dollar
Consider a dollar earned by a salaried worker in a state with income tax. The following toll chain traces that dollar through its life cycle:
The cumulative effect of this chain on a dollar earned as wages, deployed, invested, converted to property, and transferred at death bears no meaningful relationship to any single posted tax rate. The dollar that enters the chain at 100% may deliver less than 20 cents to its intended final destination across a 40-year lifecycle — not as the result of any single extortionate rate, but as the compound effect of seven distinct extraction events each of which, individually, seems reasonable.
The Parallel Track
The toll chain described above is the track for wage income. A parallel track exists for capital income — income derived from ownership rather than labor — and that track runs through the same architecture with dramatically different effective rates at almost every node.
A dollar of appreciation in a stock portfolio is not taxed until the stock is sold. If the stock is never sold — if it is held until death and passed to an heir — the gain is never taxed at all, because the heir receives the stock at its current market value with no tax due on the prior appreciation (the "stepped-up basis" documented in CS-005). The FICA toll does not apply to capital income — it applies only to wages. The preferential capital gains rate (maximum 23.8%, including the net investment income tax, versus 37% ordinary income rate) means that a dollar earned by working and a dollar earned by owning are taxed at fundamentally different rates by design. A billionaire whose wealth is concentrated in appreciated stock can live for decades — borrowing against that stock, deducting the interest, converting the loan proceeds into non-taxable spending power — without triggering any income tax event at all.
The Complexity Moat
The toll architecture is sufficiently complex that navigating it optimally requires professional expertise that is itself expensive. Federal tax law runs to more than 70,000 pages. State tax codes add tens of thousands more. The interaction effects between federal and state treatment, between different income categories, between different entity types, between different timing elections — are comprehensible, in their full complexity, only to practitioners who have spent careers specializing in specific niches of the system.
This complexity functions as a moat. The tax optimization strategies available to someone with a $50M estate — GRATs, CLATs, intentionally defective grantor trusts, family limited partnerships, charitable lead trusts, conservation easements — require teams of lawyers and accountants whose fees are themselves economically justified only at certain wealth levels. Below those wealth levels, the optimization is either inaccessible or produces insufficient savings to justify its cost. The toll architecture's parallel track is structurally available to all — the statutory provisions that create preferential treatment for capital income apply to anyone with capital income — but practically accessible only to those who can afford the professional infrastructure to navigate it.
The result is a system in which complexity itself is a regressive feature: the wealthy have access to professional navigation of the architecture's favorable paths; everyone else pays the posted rates. The same statutory framework produces radically different effective tax rates not primarily because rates are different (though they are) but because the capacity to deploy the architecture's optimization mechanisms is unequally distributed.
The Legitimacy Claim
Tax systems require a legitimacy claim — a narrative about why this particular distribution of extraction is just. The American tax system's legitimacy claim has historically rested on progressivity: the rich pay more, the poor pay less, and the system is structured to fund public goods that benefit everyone. This claim is partially accurate, partially contested, and increasingly disconnected from the actual distributional outcomes produced by the full toll architecture.
The progressivity claim is accurate at the federal income tax level, where the top 1% of earners pay roughly 42% of federal income tax. It becomes substantially less accurate when FICA is included (a flat 7.65% on wages up to the income cap, effectively regressive above the cap and inapplicable to capital income). It becomes contestable when state and local taxes are included (many states have regressive overall tax structures when property, sales, and income taxes are combined). It becomes highly problematic when the full toll chain is mapped against actual wealth distributions, because the billionaire class pays a substantially lower effective total tax rate than the professional class — not at the statutory level, but at the level of effective taxes on total economic consumption financed by existing wealth.
The Toll Architecture — Named
The layered structure in which the same unit of economic value is subjected to distinct taxation events at each transaction stage of its existence — generation (income tax, FICA), deployment (sales tax, excise tax), appreciation (capital gains tax), holding (property tax), transfer (gift tax, estate tax) — such that the cumulative burden across the full lifecycle of a dollar bears no meaningful relationship to any single posted tax rate, and varies dramatically based on the income category in which the dollar originates. The Toll Architecture operates through legitimate statutory instruments, each individually defensible. Its cumulative effect is a system in which wage income is taxed at multiple nodes across its lifecycle while capital income, through the combination of preferential rates, deferral mechanisms, and the stepped-up basis at death, passes through the architecture largely intact for those with sufficient capital to purchase professional navigation of its favorable paths. The Toll Architecture is not a conspiracy to extract from the middle class. It is the accretion of 110 years of tax policy decisions, each made in response to specific political pressures at specific moments, whose interaction effects were never designed, were rarely evaluated in aggregate, and have produced a distribution of cumulative tax burden that is substantially less progressive — and in the upper extremes, substantially more regressive — than any individual tax instrument's stated rationale would suggest.