The basics first: what the founders actually said the danger was. James Madison's Federalist No. 10 (1787) doesn't warn against democracy or against concentrated power specifically -- it warns against "faction," his word for any group, majority or minority, pursuing its own interest against the common good and the rights of others. A pure democracy, in his framing, is exactly as vulnerable to faction as a monarchy is -- the mechanism is different, the failure is the same. His proposed fix wasn't picking a side between popular rule and elite restraint; it was an engineering solution -- an "extended republic" with power deliberately split across branches and levels of government, so that capturing the whole system requires corrupting multiple independent points at once, not just one. Three of those specific points are visibly not functioning as designed right now.
Mechanism one: the war power, unused since 1942
Article I, Section 8 of the Constitution gives Congress, not the President, the power to declare war -- a specific, deliberate choice to keep that decision inside the more deliberative, multi-chamber branch rather than one person. Congress's last formal declaration of war was June 4, 1942, against Bulgaria, Hungary, and Romania, the tail end of World War II. Every conflict since -- Korea, Vietnam, the Gulf War, Afghanistan, Iraq, and everything smaller -- has run on other legal mechanisms instead: UN resolutions, presidential authorization, and especially Authorizations for Use of Military Force (AUMFs), broader grants that don't require the same public, individually-voted commitment a declaration does. Congress tried to claw the check back once: the War Powers Resolution of 1973, passed over Nixon's veto after his unauthorized bombing of Cambodia, requires the President to notify Congress within 48 hours of committing forces and caps unauthorized deployment at 60-90 days. It has been "generally resisted or ignored by subsequent presidents" of both parties ever since, per the historical record. The constitutional power still exists. It has simply gone unused, by choice, across roughly a dozen administrations in a row.
Mechanism two: administrative power consolidating around one point
The Constitution's original design spread federal administrative power across many officers, some of them deliberately insulated from at-will presidential removal -- independent agency heads who couldn't be fired simply for reaching a decision the President disagreed with. The Supreme Court's own jurisprudence has moved steadily in the opposite direction over the past fifteen years: Seila Law v. CFPB (2020) struck down removal protections for the director of the Consumer Financial Protection Bureau (CFPB), the agency that polices banks and lenders on behalf of everyday consumers; Collins v. Yellen (2021) did the same for the Federal Housing Finance Agency; Trump v. Slaughter (2025) extended the same logic further, to the Federal Trade Commission (FTC), the agency that enforces antitrust law and consumer-protection rules. Read together, these aren't isolated rulings -- legal analysts describe a consistent trajectory toward what's called "unitary executive theory," the idea that all executive power should sit with one person rather than be distributed. Whether that's constitutionally correct is a genuinely live legal debate; that the actual distribution of power is narrowing back toward a single point, case by case, is not in dispute.
Mechanism three: voters not choosing their representatives, representatives choosing their voters
Free, competitive elections are the mechanism the entire system depends on for legitimacy and self-correction -- the thing that makes a democracy's failures different in kind from a monarchy's. Partisan gerrymandering, drawing district lines specifically to lock in one party's advantage regardless of how the overall vote splits, inverts that: instead of voters choosing representatives, representatives choose which voters they'll answer to. In Rucho v. Common Cause (2019), the Supreme Court ruled that partisan gerrymandering claims are a "political question" federal courts have no authority to police at all -- removing the one federal check that could have constrained it, in either party's favor. Both major parties gerrymander aggressively wherever they hold state legislative power; this isn't a one-side failure, it's a check the Court itself declined to enforce, full stop.
The fourth force: why erosion this slow goes unpoliced
None of the three mechanisms above requires a conspiracy -- they persist because the two systems that would normally build public pressure against them are themselves under-resourced by separate, unrelated market and policy choices. On the education side, the federal government spends roughly $50 per student per year on STEM education and about 50 cents per student on civics -- a hundred-to-one gap that traces to a specific 2010 congressional funding shift, not a lack of public interest in either subject. On the accountability side, 136 local newspapers closed in 2025 alone, more than two a week; the number of US counties with no local news source at all reached 213, up from just 150 in 2005; roughly 50 million Americans now have limited or no access to reliable local news. That collapse is a market failure, not a demand failure -- print classified and display advertising revenue moved to Google and Meta, not away from public interest in local government. Local newspapers were historically the layer that caught a school board vote, a state legislature's district map, or a city council's water decision before it became routine; that layer has been shrinking for twenty years for reasons that have nothing to do with the Constitution. Even where public knowledge does move, it moves reactively, not durably: Annenberg's own 2025 civics survey found 70% of Americans could name all three branches of government, up from 65% the year before -- an increase the researchers themselves attribute to the sheer volume of executive-action and court-case news that year, not to any new investment in civic education. People learn the structure when it's visibly under strain, not as standing literacy -- which means the three mechanisms above have the most room to move exactly when public attention is elsewhere.
Why these three, together, matter more than any one alone. None of these is a coup, and none happened in a single dramatic moment -- that's precisely the point, and precisely what Madison's design didn't anticipate well. A guardrail that gets formally repealed generates opposition. A guardrail that simply goes unused, decade after decade, by mutual convenience of whoever currently holds power, generates none. Congress not exercising the war power, the Court narrowing administrative independence, and the Court declining to check electoral self-selection are three separate branches, three separate mechanisms, moving the same direction at the same time -- toward less distributed, less checked power, exactly the condition Madison's "extended republic" design was built to prevent.
This is the live, present-day version of a mechanism already traced historically on this site -- Rome's Republic didn't fall because one law was repealed; it fell because the unwritten norms restraining ambition eroded one at a time, each one individually defensible, until nothing was left to restrain the last one. A fourth mechanism runs alongside these three, not a decline in use but a design that was unequal from the start -- see how the Senate, the Electoral College, and DC weigh a citizen's vote differently, on purpose, by original design.