Searching for Submarine Safety

After the story of the OceanGate submersible disappearance overtook social media, it has now become clear that the Titan suffered a catastrophic failure, resulting in a rapid implosion and the death of all onboard. In the aftermath, criticisms of the CEO, Stockton Rush, and his submarine were ubiquitous. The viewport was rated for only 1300 meters depth when the Titanic sits at 4000 meters. The use of carbon fiber made the sub susceptible to the stresses of repeated dives. The thing was controlled by a Logitech video game controller for goodness sake! But Rush, when he was alive, would shoot back that traditional industry players were “try[ing] to use a safety argument to stop innovations.”

The thing is, he may have a point. In his book “Searching for Safety,” political scientist Aaron Wildavsky argues that sometimes our attempts at safety make things less safe. Making one thing safer, often results in another thing becoming less safe. There are trade offs. For example, regulations on vaccine development are necessary to prevent massive public health dangers. But it also lengthens the process and makes it more expensive. In the meantime people suffer and die without the vaccine, and a larger percentage of people won’t be able to afford it when it is available, mitigating the effect of herd immunity. Replacing an unsafe technology with a safer one has trade-offs as well. The harms of air pollution from fossil fuel burning alone are enough to justify replacing it. That should make us all feel good about putting solar panels on our roofs, but in 2020 111 workers died installing those panels.

That doesn’t mean we shouldn’t have safety standards, or that we shouldn’t replace unsafe technologies. But it does mean that we have to have some way of dealing with those trade offs. How much medical regulation balances protecting against unwanted side effects with the harm that comes from delaying treatment or prevention?

Wildavsky concludes that the problem arises when we end up with what he calls “trials without errors.” When, in the name of safety, nothing can be done unless it can be guaranteed to be perfectly safe, then nothing will ever get done because such a guarantee is impossible. Since we cannot know perfectly well beforehand what will happen with a new innovation, the only option is to maximize safety through trial and error learning.

But here OceanGate has a problem. They didn’t really seem to care about learning. When one of their own engineers sounded the alarm as warning signs of safety issues began to rear their ugly heads, OceanGate  stayed the course. Now that those safety risks have manifested, there is no CEO left to learn from them. Despite the rhetoric about safety stifling innovation, Rush fell into the same pattern of trials without errors. The Titan had to work perfectly every time. We now know what happens when it doesn't.

But there is nothing special about deep sea submarines that makes safety and innovation mutually exclusive. On the contrary, there was plenty of room for trial and error in OceanGate’s innovations. Three innovations come up the most when discussing their impacts on safety: the cylindrical shape, the carbon fiber material, and the use of off-the-shelf components. None of these innovations preclude safety testing and gradual scale up.

A 3d model of the Titan submersible during a dive.

A cylinder shape isn’t as strong as a sphere because the extreme pressures of the ocean floor are distributed unevenly across the shape. But it can fit more people and is much easier to construct. Trade offs. But just because a sphere has safety advantages doesn’t make a cylinder unsafe. Simulations combined with uncrewed test dives could have demonstrated the safety of the innovative shape. Instead, OceanGate claimed to have done testing with University of Washington Applied Physics Lab, but only for shallow depth submersibles. 

Similarly, carbon fiber is relatively inexpensive, light, and strong. But as trade offs, carbon fiber becomes weaker over time under high loads, and can lose up to 12% of its tensile strength in saltwater. How do you know if this new material is good enough? More trial and error. Simulating and testing carbon fiber to understand how it degrades after several dives is hardly an onerous bar to clear for the payoff of “disrupting” a whole industry. And the whole issue with off-the-shelf components is that they aren’t designed for the extreme environment of the ocean floor. Again, there isn’t anything inherently unsafe, but the safety of off-the-shelf components should have been demonstrated through trials just like anything else.

Perhaps most irresponsible was trying to implement all three of these innovations at once. When there are warning signs, this makes it harder to tell which innovation is failing. Since they didn’t know beforehand which innovations would work and which wouldn’t, it also multiplies the chance of failure. Instead of just one, now there are three things. And if any one of them fails, so does the whole submarine!

And who says Rush had to implement all three innovations before he could start making money? He could have, instead, started with whichever innovation had the potential biggest impact (perhaps off-the-shelf components?). After testing that innovation, OceanGate could have sent passengers down in a submarine that was otherwise using tried and true designs and materials. In the meantime, the profits from those trips could be reinvested into testing the next innovation. So on and so forth. The end result of such a trial and error approach is still an innovative submarine that disrupts the industry, but does so without killing anyone.

Instead, every innovation Rush made absolutely had to work. He plunged blindly ahead. Now we know what happens when they don’t work. And now those same “industry players” that he decried for stifling innovation are still operating while his innovations are at the bottom of the ocean along with the people, including a teenager, who they killed. In the case of submersibles, the costs of trials pale in comparison to the costs of errors, paid for in the lives of passengers.