Hacker News Reader

Comment by dreamcompiler | Hacker News Reader

Sandia National Labs spends huge amounts of time, energy, and money making sure US nuclear weapons turn into bricks if they are damaged or tampered with, while also being ultra-reliable if the President orders their use.

It's a level of safety and reliability engineering that exists nowhere else; you simply cannot comprehend it unless you've worked there.

It wasn't always this way but it's been true since the 1970s. The accident described in the article was one of several that created Sandia's engineering culture.

If you somehow got your hands on a modern nuclear weapon and you had a team of PhD engineers, a year of time, and the best tools in the world, you might be able to get the radioactive material out, but you absolutely would not make it explode.

View on HN

That last sounds wildly implausible. Are you saying they booby trap it to set off the high explosive if someone tampers with it (making it a booby trapped dirty bomb)? Because otherwise there's really nothing stopping some dude with an angle grinder from getting the 'radioactive material' out in a day or two.

Generally this sounds nice, but I don't really believe it without some evidence of actually motivated red teams testing the system.

When I said "you absolutely would not make it explode" I meant a nuclear explosion; I thought that was clear but perhaps not. I said you might be able to extract the radioactive materials by using brute force. You might also be able to extract the high explosives, but so what? There are far easier ways to obtain high explosive materials. The fact that you could conceivably extract the radioactive materials is why the weapons are very well guarded. The point is that if you somehow stole a weapon, you wouldn't be able to use it as designed.

The engineering process for these things is highly adversarial. Layers and layers of red teams trying to break them in the conceptual phase, the design phase, the build phase, and the deployment phase. Plus coding extremely detailed computer models of the weapons and their usage scenarios and trying to make things fail in simulation. On the fastest supercomputers in the world.

https://www.newswise.com/articles/sandia-weapons-program-mee...

https://www.hpcwire.com/2020/10/01/hpe-strikes-deal-to-provi...

You said

> If you somehow got your hands on a modern nuclear weapon and you had a team of PhD engineers, a year of time, and the best tools in the world, you might be able to get the radioactive material out, but you absolutely would not make it explode.

I'm saying that there's no "might" about extracting the radioactive materials. Nor do you need "a team of PhD engineers, a year of time, and the best tools in the world". An angle grinder and a couple days should suffice.

As for making it explode, that may indeed be very difficult, but I am not highly confident given the military's general attitude and track record. Hopefully my cynicism is unfounded, but supercomputer simulations primarily designed to ensure that aging weapons still function doesn't really help with security.

The US military does not design US nuclear weapons or their safety systems. Of course military requirements are factored in, but design responsibilities and final authority for use are in civilian hands, by law. That's why the Department of Energy is a separate entity from the Department of Defense.

Obviously the military could make mistakes when the weapons are in their custody, but civilian designs make sure those mistakes don't cause an unplanned detonation. Nor can a rogue military unit arbitrarily decide to start WW III. If they tried, the weapons would turn into bricks.

This is all fascinating (and comforting).

I'm curious about non-US nuclear weapons and how many of these safety measures are present in those weapons.

Reply on news.ycombinator.com