Gun policy needs a “Decision Support System”

One thing I always say, before arguments begin, is that no two people with the same givens can come to a different conclusion with logic. If we’re going to have rational discussions, and we define “rational” to be adjacent to “logical”, we have a duty to set ideology aside and establish the givens, and determine the truth of the case, before we ever get into discussing the plan. Anything less is an exercise in denial.

We all know that “denial” is not just a river in Egypt. But the funny thing about denial in the gun policy debate, is that that very river in Egypt can show us a path to making actual sense of the gun policy debate, and coming to an actual resolution. This is wild, bear with me. You’ll like where it goes. And where it goes, by the end, is a comprehensive national gun policy proposal that fixes everything that can possibly be fixed, and tells us exactly how many lives we’re going to save.

The Nile DSS

When I was pursuing my graduate degree in Environmental Fluid Mechanics and Water Resource Management at Georgia Tech, we spent quite a bit of time talking about the Nile, because Georgia Tech has a special connection to it. Dr. Aris Georgakakos ran the program at the time and taught several of my classes. He’s a genius. Undergrad degree from the National Technical University of Athens Greece, multiple graduate degrees from MIT, and current head of the Georgia Water Resources Institute. Every bit the expert at this stuff, and his fingerprints are all up and down the Nile River. 

The problem: Egypt and Sudan are 100% dependent on the water in the Nile. They have no other source of water whatsoever. But the total Nile watershed covers Tanzania, Kenya, Ethiopia, Uganda, Eritrea, and the Congo, so the things they do with the water upstream of Egypt affect Egypt, and Egypt has an air force, with attack bombers. 

The stories Dr. Georgakakos told us of the early multinational policy discussions about this issue were wild. Imagine a bunch of diplomats sitting in a room, without a tremendous amount of science being discussed, trying to hammer out an international water policy agreement. Imagine Kenya decides they’re going to take X gallons of water out of Lake Victoria for irrigation purposes. Imagine Egypt says, “You can’t do that, that will impact our water supply a tremendous amount!” Kenya says, “No it won’t, it will only impact your water supply a tiny amount!” They disagree on the givens. Voices get raised. They can’t come to a resolution. Eventually the discussion ends by them literally throwing chairs at each other over the negotiating table. 

I’m old, and heard this story two decades ago, but I definitely recall Dr. G saying he personally witnessed chairs being thrown.

What did these policy wonks need, to avoid their negotiations breaking down into a game of projectile furniture? They needed a scientific tool to answer a specific question. “If Kenya takes X gallons out of Lake Victoria, that will impact Egypt in what way?” They needed more answers too. How many gallons should be let out of which dams at which times of year? How many kilowatt-hours would be lost in power generation if water spilled over a spillway instead of making it through a turbine? How much rain should we expect during next year’s monsoon season, and how much of that rain is going to turn into runoff? How much irrigation water do we actually need, and what will irrigation restrictions do to food production? These are complicated questions, and all impact each other, and in the end they all boil down to water policy. 

The multinational water negotiators of the Nile needed a “Decision Support System”, a DSS, which could make predictions of the future based on a given water policy, so they’d all agree on the givens before they started negotiating the policy. And that’s what Georgia Tech gave them. 

There’s also a funny story about how GT beat out the United States Geological Survey on this task by building a better climate model with artificial neural networks in Visual Basic instead of complex physical modeling on Cray supercomputers, but you’ll need to buy me a beer to hear that one.

Here’s an image of the Nile Decision Support Tool (DST), a blended network of DSS tools for different regions on the same backbone:

This is one of the key features of water management of the Nile, and an example for the world going forward as our water resources become more strained. It avoided innumerable projectile chairs. It may have possibly avoided projectile rockets and bombs.

And when I look across the space of the United States Gun Debate, all I see are chairs being thrown, with some foreboding talk of military intervention.

Throwing chairs

When Elizabeth Warren posted her 32-point gun policy plan on Medium, she claimed that her goal was to reduce gun deaths by 80%. A quote:

In 2017, almost 40,000 people died from guns in the United States. My goal as President, and our goal as a society, will be to reduce that number by 80%.

But if Warren were to magically end all gun homicide in the United States tomorrow, which is impossible, that number would only be reduced by around 33%! Two-thirds of gun deaths are suicides, and she only had one bullet in her plan that addressed suicide. 

This is an obvious mathematical lie. What should gun-positive people do when they read this? What should people who understand basic arithmetic do when they read this? Throw chairs? That’s certainly what the media seem to be doing, and the social media soldiers in the twitterverse seem to be doing. We don’t get to a resolution about anything by throwing chairs. We need to pivot away from denial, and look to the Nile. We need a Gun DSS.

Michael Siegel and his research team at Boston University gave us the closest reasonable approximation to a Gun DSS backbone in the spring of 2019. He and his group have done a lot of research about gun violence in the United States, but this was the first time I’ve heard of a study that looked specifically at the efficacy of policy changes on firearm suicide and homicide rates. It was a longitudinal study, looking at changes over time based on when certain laws were passed or repealed. It was multivariate, and controlled for the other major factors that impact these rates, such as black population rate, poverty rate, unemployment rate, per capita alcohol consumption, and such. The study seems, by my read, to be solid. Let’s look at the study’s findings:

• “Assault weapon” bans have no effect.
• Magazine capacity bans have no effect.
• None of the gun laws they modeled affected the suicide numbers at all.
• Limiting handgun purchase to age 21 and over has no effect.
• Trafficking prohibitions (restrictions on buying with the intent to sell) have no effect.
• Junk gun laws (prohibiting handguns that fail to meet certain requirements) have no effect.
• Stand-your-ground laws have no effect, positive or negative.
• Permitless carry laws have no effect, positive or negative.

There were only three laws that had any effect whatsoever. 

• Universal background checks, either through required background checks for all sales or through a firearm purchase permit, reduced gun homicides by 14.9% and had no effect on suicide.
• Prohibiting those convicted of a violent misdemeanor from buying a handgun reduced gun homicides by 18.1%, and had no effect on suicide.
• Shall-issue laws, which ensure that law enforcement officers can’t discriminate when issuing concealed carry permits, increased gun homicides by 9.0% and had no effect on suicide.

I’m not completely clear after reading it that the effects of the laws were truly modeled to be independent of each other. There may be some overlap between the reduction from universal background checks and the reduction from violent misdemeanor laws, for instance. But let’s presume they’re independent and move forward with some mathematics, to check Elizabeth Warren’s efficacy claims, or the claims of others in the anti-gun space. 

Gun DSS

First, we presume that the sorts of rate changes from Siegel’s work are predictive. This is a presumption that may not hold true, because correlative studies sometimes don’t show causality, particularly across cultures. The culture of Alabama is very different than the culture from California. This presumption is the most generous possible case for the gun-control side. 

Twelve states already have some form of universal background check: California, Colorado, Connecticut, Delaware, Hawaii, Illinois, Massachusetts, New Jersey, New York, Oregon, Rhode Island, and Washington. These states would not be affected at all by a nationwide universal background check, so they get no gains by a federal law. The effectiveness of a nationwide law would only be realized on the states that don’t already have this law. Evaluating the effectiveness of this law, or the other two laws with identified efficacy, must be limited to the states it would impact. To determine the impacts, we need a table of firearm homicides by state, cross referenced with their current state of law. OSD has now officially done this for you. Took me about two hours, and here are the results:

A true DSS would be far more complicated than a simple spreadsheet, but based on Siegel’s team’s research at Boston University the results of a true DSS will probably look something very close to this.

I chose to use 2017 numbers from the CDC’s Wonder Database, which are easily accessible. We have 14,322 gun homicides in this sample set. Some state records were missing from the database, so were omitted in the analysis. I also chose to use the “state of law” as shown in the Siegel study, without back-checking to determine whether that’s current.

So (again, using the most generous possible case for the gun control side) let’s simply grant that moving to some kind of universal background check or license to buy system would avert around 1,600 firearm homicides per year, nationwide. This would be a gun homicide reduction of 11%, or a gun death reduction of 4%. Please note, that this doesn’t need to be tied to a gun registration database. The results could be realized purely with a one-time voluntary NICS check to get a slip of paper to use during peer to peer sales. 

Let’s also grant that preventing perpetrators of violent misdemeanors from owning handguns would avert around 2175 gun homicides per year. This would be a gun homicide reduction of 15%, or a gun death rate reduction of 5%. I would like to note here, that the sample size of states with this law is relatively small, and the only two large ones are New York and California, so there’s probably some concern over the confidence of this result.

And finally let’s grant that unilaterally repealing shall-issue concealed carry laws, and moving to either may-issue laws or constitutional carry would eliminate 1015 homicides per year. This is a reduction of 7% of gun homicides, or 3% of gun deaths. It may not make sense to some that going from shall-issue concealed carry permitting to “no license required to carry” laws would reduce gun homicides, but that is literally what Siegel’s research predicts. 

This is what our decision support system tells us, based on Siegel’s research, and it exposes Warren’s claims for what they are. She is exaggerating the impact of what she proposes by a factor of twenty. And we all throw chairs.

“Common sense gun reform”

Common sense should mean mathematically informed compromise. You figure out what the juice actually is, and then the collective policy makers decide if the juice is worth the squeeze. If the federal government wanted to step in and lay out a mathematically correct compromise position that overrode state firearms laws, this is what that position should look like:

1. Overturn all state prohibitions on magazine size, state-specific prohibitions on types of firearms and accessories owned, and state age restrictions on purchase. Federally bar “assault weapon” bans. Give everyone their AR-15s and their drum magazines back, even in California.
2. Implement a national stand-your-ground law overriding state laws on appropriate firearm engagement.
3. Allow states to choose among either may-issue or constitutional carry categories of firearm concealed carry law. Many shall-issue states today would probably go constitutional carry.
4. Implement a universal background check system, by allowing peer-to-peer buyers to print their own NICS check to show to peer-to-peer sellers. 
5. Prohibit perpetrators of violent misdemeanors from passing the background check.

Number four and five are going to be hard to swallow for the “shall not be infringed” crowd. Number one and two are going to be hard to swallow for the gun control crowd. This is a hard pill to swallow for everyone, but that’s what actual compromise looks like. And if you’re going to wield the federal government to fix state gun law, you need to wield it to fix them all, going both ways, to better match reality.

We should understand in advance, that the efficacy of this proposal will reduce gun homicides by some number less than 5000 victims per year, which is significant but not tremendous, in a country of 330 million people with 100 million gun owners. And the proposal will do nothing to affect the primary mode of “gun death” in the country, which is suicide. The net death rate reduction from this proposal would only be around 1 per 100,000 per year, but it is literally the best possible death rate reduction from any policy proposal on the table.

That’s the juice. Science has spoken. We have the answer. 

Is the juice worth the squeeze? For anyone? Or do we keep throwing chairs?