It’s said that to successfully eliminate something you must replace it with something new. We see this in the history of systems where people eliminated and replaced part of it long ago. They survived and so are examples of the cycle being applied well. But why is this method applied poorly? Where does it break down? Since there is uncertain ability and low desire to understand changes that might come after eliminating something, whether there is a replacement or not, how should we mitigate the risks that might emerge?

Between eliminating and replacing something there is another way, which I’ll call “elimitigation,” with “elimitigate” being a portmanteau of “eliminate” and “mitigate.”

Let’s look at some examples of elimination done at large scale without replacements in place. What happens next? The obvious examples come from top-down policies.

Take access to food in poor autocratic states. There were a couple models for this. In the USSR, for example, food production was centrally determined and access granted in the form of coupons according to estimated needs.

So when vodka rations were intentionally cut, the effect was not to lower the amount of alcohol people wanted to drink — that was the theoretical outcome, not the actual one. Instead, the result was people started to make their own moonshine, or samogon, using the sugar their coupons allowed them to buy. Because of that, sugar was not available for other uses, including baking and making preserves. The USSR couldn’t elimitigate — their limited resources forced different approaches that had second-order effects.

Another problem emerged when there were shortages of food shortly after the USSR’s collapse. Part of the problem was that prices couldn’t fluctuate as needed, for example by charging different prices for fresh or old bread. As the deputy director of the Moscow Retail Bread Committee said at the time, “[w]e don’t want to have any negative social consequences connected with old bread; for example, if some people could afford fresh bread and others could not.” Just after the fall of the USSR the government couldn’t elimitigate — their social theory forced different approaches that had second-order effects.

In Saddam Hussein’s Iraq, food rationing preserved social order and obedience, given that families would lose their ration cards in certain situations. Ration cards gave inexpensive access to sugar, flour, rice, cooking fat, soap, washing detergent, milk, lentils and beans without regard to religion or social class (though high-ranking party members had many more benefits). “A conversation among high-level Iraqi officials in 1994 suggests the critical importance of the rations system in controlling the population through strategies of ‘policing’ within families. Government control of the ration card system was thought to be decisive in implementing policies against military desertions. According to one official, the regime was able to take advantage of obligations between family members as a potential deserter would not want to be held responsible for his family losing their ration card, resulting in their starvation.”

After the American invasion, the ration card system, which was often previously said to work well, started to beak down. A system that had maintained wide-scale access to staples also kept different factions equal and dependent on the government. Could this system have been maintained until something better was there to replace it? The Americans could have but didn’t elimitigate — they opposed the ration card system and didn’t want or plan to run the country’s infrastructure.

Elimitigate: Eliminate, but Mitigate

Elimitigation aims to change the current situation, avoid creating chaos or generating new unexpected problems, while having something in place to mitigate the risk of an unknown outcome incurred.

If you remove a high-power force (say in Iraq) without either preserving or replacing its institutions, you expose the population to chaos. Institutions that provide a value that are suddenly eliminated lead to unpredictable changes.

Here are two items, increasingly discussed as future changes, whose elimination would affect many.

Driving. Much discussion of autonomous vehicles (AVs) over the last few years has centered around the dangers of human drivers and vehicles, currently just in the US responsible for over 37,000 deaths per year. Humans make mistakes, they drive intoxicated, they drive while tired. All things an AV would never do, the argument goes.

The same argument claims that if human-driven vehicles were eliminated and AVs became dominant, traffic deaths could be reduced to close to zero. You’d have to be crazy (or dangerously sentimental) to want to drive.

This argument for AVs is one for optimization and a more sophisticated system. But this optimization comes at a cost of systemic risk. Things can be better off as long as they work perfectly, but when parts of the system break down, they break down in a big way. But humans, human-driven cars, and the accidents they create operate differently. What could possibly be done to worsen outcomes in the current system? Not much, barring odd and unlikely policy changes such as removing penalties for driving while intoxicated, removing mandates for seat belts, speed limits, or other safety measures.

You can’t easily shock a system that is really made up of millions of individual actors all mostly trying to preserve their own lives. But you can shock a system when there are few points of influence that impact many others.

You can shock a system like a fleet of autonomous vehicles. Target access to the GPS system, or hack a fleet’s communications software or fuel efficiency,  and outcomes can change at scale, meaning a shock — sudden increase in accidents and deaths — is within bounds. Longer-term, if human-driven cars were eliminated and replaced with AVs, how do you go back and reverse the decision? The infrastructure and knowledge has already been lost.

Should a per-mile fee be added to AV rides and set aside for the purpose of dealing with shocks? Proponents of AVs do not elimitigate — it is not in their business models or interest to do so.

Football. There is a small, but growing, movement against American football on grounds of health. The movement started based on understanding the effects of brain injuries caused by playing football. Many players develop Chronic traumatic encephalopathy (CTE). CTE results in everything from loss of mental capacity to suicidal behavior. So far this has not been a technology question, solved by better helmets. Some observers, like Malcolm Gladwell, call for outright college football bans.

But what would happen if someone could eliminate football, especially college and high school football? Proponents of a football ban have nothing to replace it with. Until they do have something to replace football (and that is unlikely) there would be more new problems arising from football’s ban. What social issues will arise from the loss of regular massive and largely positive gatherings of community members, fans, alumni, parents, and friends.  Proponents of banning football do not elimitigate — they have no idea what they would replace it with.

Considerations

• When deciding what to eliminate, people often assume best case scenarios without regard to second-order effects. The mitigation part of elimitigation requires setting something aside to deal with uncertain outcomes.
• Elimitigation is a trade-off between options. Doing one thing vs another. Doing something vs nothing. Causing a change now and potentially dealing with consequences later.
• Elimitigation is not practiced where the actors do not live. Actions at a distance, be they dramatic interventions and policy changes that work in theory make more sense when the actors do not suffer the consequences of their actions.