How to respond to low probability events
When is a corner case too remote to not respond against?
We had an attempt at severe disruption of the US government by an angry mob storming into the Capitol. They were, depending on the source, let through or handled with kid gloves by the cops responsible for safety. Officials cowered in gas masks and ran for safety as armed insurrectionists ran free through the halls.
So the question is, what next? What should we do now?
There are multiple calls to action to understand why this happened. Not the presidential instigation or the riots. That part makes all too much sense. If you truly believed that the election was stolen from you, then a riot is a rather understandable, even sensible, response. Fighting injustice with violence is a time honoured tradition after all - more than a few countries started that way. But the question is on why the police were not able to adequately protect the Capitol. Tyler asks in a column:
...There are also reports of Molotov cocktails and pipe bombs in or near Congress. You simply have to secure the physical building if you are to have credible security at all. I am very familiar with these entrances, and even an outnumbered force has the ability to keep out intruders, if it has the will to do so. In contrast, the Secret Service has a long history of agents using their bodies to block or shield presidents from threats of violence. Does Congress as a whole (which is harder to replace than a single president) deserve any less?
The unstated assumption seems to be that the response to a riot needs to be strong and it is also the sensible and appropriate response.
One change that sensible people have suggested is to change the makeup or duty of the cops who were handling the riot. They weren't trigger happy as they usually are, and this time around that annoyed those on the left. That's because if you compare the actions to the BLM protests, there's a clear discordance. And that maybe because one seemed legitimate while the other didn't (seemed, due to all sorts of media outlets pushing agendas, entitlement syndrome, and maybe a dollop of racism), but it definitely is not just.
Arguably though the change should be that more protests needs to be handled the way the Capitol riots were, with a minimum of bloodshed and broken bones, rather than the opposite lesson of needing an even firmer hand. Would you have preferred yesterday to be more stringent with a body count like with the BLM protests? Or would you rather have seen the BLM protests be more like yesterday's?
We all need to enforce more decision making under constraints. When libertarians get annoyed at government overreach it's not just because they're inefficient, it's also because the costs of a larger government increases, but rarely decreases. How many programs were sunset in the last decade? "We need more policing" is not the answer, "we need better policing" is. Though they sound extremely similar, they're polar opposites in how they're put into practice.
We've seen this cost come up again and again in the past. We pay it in travel, in housing, in education, in healthcare, everywhere. Attempts to reduce the severity of cases with tiny probabilities raise the average cost for everyone.
We’ve seen this play out even in the pentagon war games. This should be the scariest lesson of all. Red teams regularly kick blue team asses by not playing by the same rules. This is despite the enormous imbalance in resources and the incredible array of weapons available. But it still doesn't help win against all eventualities. There will always remain the corner cases of false flag operations and cleverly disguised barges ramming into large tankers that are unsolvable by the "bigger is better" mentality.
If a change is made to, say, add further guards to the building, with stricter rules for entry and exit, better barricades, a few assault vehicles on standby, with the national guard riot readiness made high, will that solve the problem? We will definitely be taking on a lot of added pain for sure. The cost is visible and the benefits intangible.
There's a reason we obligingly take off our shoes and go through the TSA security theater when we fly. It's because of Richard Reid. But it's not because it's safer - TSA checks fail 70 to 94% of the time. The libertarian inside me wants to see the cost-benefit trade-off of any intervention to be calculated rationally, not emotionally.
If we change our behaviour to guard against riots in the Capitol by overreacting, that will only create more security theater. You can add barbed wire and gates and more metal detectors and more on duty police officers and arm them all to the teeth. But it's also worth remembering that this was a symbolic failure, not a real failure. There were a sum total of 4 casualties and property destruction. I don't think it's worth adding yet another multi billion dollar boondoggle. Please let’s be careful in what we ask for.
Fact is that it's difficult to secure against corner scenarios on a consistent basis in any system. You just end up adding layer after layer on top of organisations who are ill equipped for any type of scaling. When you throw resources at an organisation the impetus is rarely to become more efficient. It's to get bigger.
And sure you will get better at stopping this particular type of threat. (Well, probably, but TSA doesn't fill me with hope.) But rarely are these things reorganised or removed. And that creates a horrible precedent. It's how we end up with regulatory creep, an unwieldy state apparatus, and ever increasing "run" budgets. We need to try and be smarter about this.
We pay the prices of complexity and bureaucracy not because we're idiots. It's because with the best of intentions when we try to solve problems as they come up, these are naturally arising consequences.
This is not a caution against reaction, or even overreaction. It's a caution against reactions taken without taking the complexity costs into consideration, against doing things that look good or tough or strong without actually analysing the benefit of doing so.