If Gun Laws Worked, You Could See It.

Read the overlay from the baseline. Colorado’s magazine cap is the cleanest test, because it predates the pandemic: two years of a small dip, then a climb to +29% by 2017 — more than the nation’s +21% over the same years — and +83% by 2020. Five years after a working law, the outcome it targets is below where it started and stays there: Louisiana’s unhelmeted deaths down 68% (88% at the low point), adult smoking down 14% and still falling, teen fatal crashes down 9% and still falling. Five years after Colorado’s laws began, firearm suicide — the outcome most of them were sold on — is up 4%, and firearm homicide is up 22%, having risen faster than the country’s during the pandemic (peak +82% against +45%) and come down with it. The national line is the control. A law that worked would have pulled Colorado’s line below it. Six laws didn’t; the line never went under the baseline at all.

Tab 4 is the honest hard case, because the assault-weapons-ban line does go down: U.S. firearm homicide fell 43% between 1993 and 2004. So the test is what happened when the ban lapsed. A law responsible for a 43% drop should leave a mark when it disappears, the way Louisiana’s helmet law did going out (unhelmeted deaths up 85%) and coming back (down 88%). Firearm homicide kept falling, another 10% by 2011. The Justice Department’s own evaluation, written for the National Institute of Justice the year the ban expired, put it this way: we cannot clearly credit the ban with any of the nation’s recent drop in gun violence, and a renewed ban’s effect would be small at best and perhaps too small for reliable measurement — because AWs were used in no more than 8% of gun crimes even before the ban. BJS’s own trend report attributes much of the 1990s rise and fall to gun homicide by teens and young adults, mostly with handguns — a category the ban barely reached.

Gun laws have had fifty states and fifty years to leave a mark like that.
The mark isn’t there.

Two 90/10 rules.
One: ninety percent of crime guns never touch the gun counter. One hundred percent of purchase laws do.
Two: ninety percent of crime guns were bought more than three months before the crime. A three-day wait cannot reach them.
Nine times in ten, the restriction lands on a citizen exercising a right of self-defense — and infringes it for nothing.

ATF’s own words, from its 2023 crime-gun assessment: prohibited persons “are unlikely to purchase directly from a licensed federal firearms dealer. Instead, prohibited persons determined to get crime guns acquire them through underground crime gun markets that involve unregulated transactions with acquaintances and illicit ‘street’ sources.” Forty-three percent of prisoners said “off the street.” Another quarter got the gun from a friend or family member. One in ten bought it at a store.

A waiting period, a permit card, a safety course, a magazine cap and an excise tax all operate at the retail counter. The retail counter is where roughly one crime gun in ten was last seen — three years before the crime, in the hands of someone else. The laws are aimed at the one door criminals don’t use. That is not a failure of enforcement. It is a failure of design, and it is why the map below looks the way it does.

Keep the honesty straight, because the other side will. Total firearm deaths are 44,447 — roughly the same as car crashes — but 62% of them are suicides, and drowning (4,201) and lightning and floods are far below guns, not above. The claim here is narrower and true: the thing Colorado has chosen to tax, license, delay and cap kills fewer people by homicide than pills, than stairs, than the drive to work, than the bottle, than suicide by rope or pill — and every one of those bigger killers is met with education, treatment and engineering rather than a permit to exercise a constitutional right.

There is a faint slope in the national picture — a correlation of 0.31, meaning strictness accounts for roughly one-eleventh of the variation. Look at where it comes from. Ten of the twenty-four F-graded states sit below the A-grade median. Two A-graded states, Illinois and Maryland, sit above the F-grade median. The slope is produced almost entirely by five Deep South states that have both loose laws and the nation’s highest homicide rates — for reasons that long predate any gun statute.

So compare like with like. Break the country into its four Census regions and rerun the same test.

Switch the map above to firearm suicide and the F states light up — Wyoming, Montana, Alaska, Idaho. That is the gradient behind every “strict states have fewer gun deaths” chart. It is real, and it is a suicide gradient, tied to rurality, isolation, and firearm ownership. It is not what the public hears when it hears “gun violence,” and it is not what magazine caps, waiting periods, or permit-to-purchase laws are sold as preventing.

Read it straight. New York City and Los Angeles — strict state, strict city — are among the safest large cities in America. Chicago, Baltimore and Washington — equally strict — are among the most dangerous. Houston and Phoenix, in F-graded states, sit near the bottom of the chart with New York. St. Louis, Jackson, Birmingham, New Orleans and Memphis, also F, sit at the top. Strict and loose interleave from end to end. Whatever separates New York from Chicago, or Phoenix from St. Louis, it is not the gun-law grade.

The standard reply is that Chicago’s guns come from Indiana. ATF’s own 2024 trace data says 54% of Illinois crime guns with a known source were bought in Illinois; Indiana supplied 16%. For New York, 79% came from out of state. Both facts are on this page’s source list. Neither explains why New York and Chicago, with similar laws and the same porous borders, are at opposite ends of the chart.

Firearm homicide rose from 2.89 in 2019 to 5.27 in 2022 and fell back to 3.54 in 2024. That is the national curve — the U.S. went 4.59 to 6.66 and back to 4.74 over the same years — and it tracks the pandemic, not any Colorado law date. Firearm suicide, the outcome most of these laws were actually justified by, sat between 10.7 and 12.2 the entire time. Six laws in, the line didn’t move. To be fair to both sides: the post-2023 decline coincides with the waiting-period and age-21 laws and with the nationwide drop, and one year cannot separate them. That is exactly the problem — if the effect were large, one year would.

RAND’s scale runs inconclusive → limited → moderate → supportive. Read the table honestly and two things are true at once. The marquee restrictions — the magazine cap, the SB25-003 rifle regime, the age limit, the red-flag law — are inconclusive for homicide: after decades of study, the evidence cannot say they do anything. And two older, less intrusive policies — background checks and waiting periods — reach moderate, RAND’s second-highest rating, for total homicide. Child-access laws reach supportive for youth outcomes. That is the whole honest picture: the laws with some evidence behind them are the ones Colorado passed a decade ago; the laws it has passed since carry a heavier burden on the right and an emptier evidence file.

One more line from the table that cuts the other way and belongs here: RAND rates shall-issue concealed carry and stand-your-ground laws as supportive for increasing homicide. We disagree with the framing of several of those studies, but you should know the rating exists before someone else tells you.

Now hold the gun-law map up against those five charts. Helmet use in Louisiana went from 56% to 26% to 91% as the law went off and on. Blood lead fell twenty-fold in lockstep with the phase-out schedule. Smoking fell in every decade with every tax and every smoke-free law. These are not subtle effects requiring a regression to detect. They are visible from across the room.

The fair objection is that homicide has more causes than head injury does, so the signal is harder to find. True. But that objection concedes the case: a policy whose effect is too small to see in the population data, after decades and dozens of studies, is a policy whose benefit is unproven — and the cost is not. The cost is paid by every peaceable citizen who waits three days, pays the tax, takes the class, gets the card, and loses the magazine. They pay it for a result nobody can measure.