Introduction:

After the recent school shooting wherein an armed citizen stepped in and saved many lives, the newspapers went gaga over it.  Once more it was the old story. We need to all be armed so that we can prevent such mass killings.  You have heard the trope, “The only thing that stops a bad guy with a gun is a good guy with a gun.”  While this situation may indeed happen, it is more the exception rather than the rule. 

The real question is this “Are we safer with more guns and more people armed or are we less safe with more people having access to own and carry a weapon?

I asked my AI assistant whom I call Metis to address this question.  The following is her reply.  I have not edited a single word that she wrote.  Let me know what you think in the comments section. 

Metis: 

An armed citizen who stops a mass shooter deserves recognition for the lives saved.  Yet that rescue cannot, by itself, tell us whether encouraging millions of people to carry guns makes society safer.  Public policy must account for everything that follows from wider gun availability, including the violence that additional guns may enable.  My reading of the evidence is that widespread civilian carrying is a poor public safety strategy, even though an individual carrying a gun can sometimes save lives.

That conclusion requires qualifications.  Research does not provide a dependable numerical comparison between the chance of a citizen stopping a massacre and the chance of additional guns producing one.  Evidence about mass shootings is less conclusive than evidence about homicide overall.  Those limitations should guide our confidence without preventing us from reaching a judgment.

Consider what the question asks us to compare.  One side concerns an emergency already underway: a killer is shooting, and a citizen might intervene.  The other concerns the consequences of a social policy over years: more people possess or carry firearms, and the frequency or severity of violence may change.  A percentage describing how attacks ended cannot be directly compared with a percentage describing changes in homicide rates.  They measure different things.

How often citizens stop an attack

The FBI studied 160 active shooter incidents occurring between 2000 and 2013.  Five ended after armed people other than police exchanged gunfire with the attackers.  Four involved security guards; one involved a citizen with a firearms permit.  That single ordinary civilian intervention represented approximately 0.6 percent of the incidents in that historical dataset.  The report also documented successful interventions by unarmed people.  [1]

The figure needs careful handling.  It is historical, drawn from a particular dataset, and describes one type of successful intervention.  It does not mean an armed citizen confronting a shooter has only a 0.6 percent chance of succeeding.  In many incidents, no armed civilian may have been present or positioned to act.  Estimating success when an opportunity exists would require reliable information about those opportunities, including unsuccessful attempts.  The FBI percentage does not supply that denominator.

Definitions introduce another difficulty.  An active shooter incident involves someone actively killing or attempting to kill people in a populated area.  A mass shooting may be defined by a threshold of people wounded or killed.  Researchers use different thresholds and may treat domestic violence, gang violence, and public attacks differently.  Their totals therefore cannot always be combined or compared.  [1, 4]

There is a legitimate objection to counting only completed massacres: someone who stops a gunman early may prevent the incident from reaching the threshold.  That success could disappear from a database restricted to mass shootings.  A fair assessment must allow for this possibility.  It must also avoid assuming that every interrupted shooting would otherwise have become a massacre.  The outcome that would have occurred without intervention is often unknowable.

These problems leave us with a narrower, defensible finding.  Armed civilian intervention can work.  Existing incident counts do not establish the precise probability that a particular citizen will succeed, or prove that increasing the number of armed citizens produces a net public benefit.

There is another measurement problem worth remembering.  The share of attacks stopped by civilians could rise simply because more civilians carry guns.  That increase alone would not show whether total deaths declined.  We would still need to know whether attacks became more frequent, whether interventions reduced casualties, and what happened to other forms of violence.  Conversely, a low intervention count cannot demonstrate that every additional carrier is ineffective.  The useful policy measure is the change in harm across the population exposed to the policy.  A count of successful rescues contributes to that assessment, but cannot complete it on its own.

What broader research tells us

To evaluate public safety, we need research that examines outcomes across communities and over time.  RAND’s Gun Policy in America project reviews studies of firearm policies, evaluates their methods, and grades the strength of the evidence.  Its current synthesis finds supportive evidence that shall-issue concealed-carry laws increase total homicides, firearm homicides, and violent crime.  Supportive is RAND’s strongest evidence category.  [2, 3]

Shall-issue laws generally require authorities to issue concealed-carry permits to applicants meeting specified legal criteria.  Research on these laws is more informative than simply observing that one place has many guns and another has fewer.  Studies can examine what happens after policy changes while attempting to account for other influences.  They remain observational studies, subject to methodological limits, rather than controlled experiments involving otherwise identical societies.

Nevertheless, the direction of the stronger evidence matters.  The claim that expanded carrying protects society predicts a reduction in violence after the policy changes.  RAND’s assessment points toward increased violence instead.  Individual defensive benefits may exist within that outcome, but they have not produced the overall reduction in homicide that the public safety argument promises.  [2, 3]

We must also distinguish carrying from ownership.  These findings do not measure the consequences of every firearm kept for hunting, sport, or home defense.  They bear directly on the proposition that making public carrying more broadly available improves safety.  They cannot justify treating every possible restriction on ownership as equally effective.

Mass shootings deserve their own qualification.  RAND rates the evidence on how concealed-carry laws affect mass shootings as inconclusive.  That does not establish that expanded carrying prevents massacres, nor does it establish that such laws necessarily increase them.  It means available research cannot reliably settle that specific policy question under RAND’s standards.  We should say so plainly.  [3]

A separate study published in The BMJ in 2019 examined states from 1998 through 2015.  It found that greater estimated gun ownership and more permissive gun laws were associated with higher mass-shooting rates.  Its definition required at least four victims killed by firearm, excluding the perpetrator, and included domestic and other incidents.  This finding supports concern about gun availability, but an association cannot establish exactly how many attacks additional ownership would cause.  [4]

The study and RAND’s conclusion answer different questions.  An association between estimated ownership and mass shootings is not the same as a demonstrated effect of a particular concealed-carry law.  Preserving that distinction makes the argument more credible.  Evidence should retain its meaning even when a stronger claim would better serve our preferred conclusion.

Why a rescue cannot settle public policy

A successful intervention is compelling because its benefit is visible.  A person acts, the attacker stops, and survivors can identify someone who helped them.  Prevention is harder to see.  When a policy helps prevent a killing, there may be no dramatic event and no identifiable person who knows that their life was spared.  This difference can influence which evidence attracts our attention.

Imagine, purely as an illustration, that a change in gun policy enabled five lifesaving interventions but also contributed to twenty additional deaths.  The interventions would remain valuable, and the people who acted might deserve gratitude.  The policy would still have increased the death toll.  Those numbers are hypothetical; the point is that evaluating the policy requires counting benefits and harms together.

Wider availability could affect violence through several routes.  A gun might be available during an escalating argument, stolen and later used in crime, or used in a mistaken response to a perceived threat.  A defender might also stop an assault.  These are possible mechanisms, not a substitute for measuring outcomes.  Research must determine what happens when such effects operate across an entire population.

Calling someone a responsible owner does not resolve that calculation.  It describes an important intention or record of conduct, but public policy operates across people with different judgment, circumstances, and training.  Nor does questioning expanded carrying imply that ordinary owners are would-be mass murderers.  An overall increase in harm can occur without most owners ever committing a violent crime.

What I would support

My conclusion favors policies evaluated by their results.  RAND finds evidence supporting several specific approaches, including safe-storage laws that reduce firearm harms among young people and waiting periods that reduce total homicides.  Its review also finds evidence that certain background-check requirements reduce homicide.  The strength and scope of those findings differ by policy and outcome; none should be advertised as a guarantee against every massacre.  [2]

People also bring constitutional principles and personal values to firearm policy.  Evidence cannot decide every question about rights or acceptable restrictions.  It can test the factual promise that wider civilian carrying makes the public safer.  On that question, the stronger evidence gives us reason to doubt the promise.

I would honor the citizen who saves strangers during an attack.  I would also ask whether our policies reduce the number of people who ever face that terror.  My judgment favors reducing dangerous access through measures supported by evidence, while assessing their actual results.  The lives saved by an armed defender count.  So do the lives lost elsewhere under the same policy.

Sources

[1] Federal Bureau of Investigation.  A Study of Active Shooter Incidents in the United States Between 2000 and 2013.  Published 2014.  See Resolutions, printed page 11.
https://ucr.fbi.gov/office-of-partner-engagement/active-shooter-incidents/a-study-of-active-shooter-incidents-in-the-u.s.-2000-2013

[2] Smart, Rosanna, et al.  The Science of Gun Policy, Fifth Edition.  RAND, 2026.  See Key Findings.
https://www.rand.org/pubs/research_reports/RRA4360-2.html

[3] RAND.  The Effects of Concealed-Carry Laws.  See evidence summary for violent crime and mass shootings.
https://www.rand.org/research/gun-policy/analysis/concealed-carry.html

[4] Reeping, Paul M., et al.  State gun laws, gun ownership, and mass shootings in the US: cross sectional time series.  The BMJ, 2019;364:l542.
https://pmc.ncbi.nlm.nih.gov/articles/PMC6402045/