https://www.sciencenews.org/article/nuclear-weapons-tests-comeback-threats.
Calls to restart nuclear weapons tests stir dismay and debate among scientists
A U.S. return to underground detonations would have wide-ranging implications
Some see the ability to test as a necessity for a world in which nuclear weapons are a rising threat. “We are seeing an environment in which the autocrats are increasingly relying on nuclear weapons to threaten and coerce their adversaries,” says Robert Peters, a research fellow at the Heritage Foundation, a conservative think tank. “If you’re in an acute crisis or conflict in which your adversary is threatening to employ nuclear weapons, you don’t want to limit the options of the president to get you out of that crisis.” Testing, and the signal it sends to an adversary, he argues, should be such an option.
12 comments:
Anyone who has ever designed and engineered a complex system knows that testing is absolutely essential. The notion that you can test some of the individual components through a combination of experiments and simulations and confidently confirm the function of a complex system is pure political hogwash. It gets even worse as components age out and are replaced by new designs with new materials. It's impossible to determine the result of every complex interaction, let alone what those interactions even are. This is like Boeing designing a new jet and putting it into service without any flight testing. Moreover, the testing moratorium has failed to prevent other nations from developing nukes. It's a politically expedient policy that has long since outlived it's usefulness.
But with LANL and LLNL will be turning into AI labs we will not need to test. We will have a ChatGPU just for the labs. If there is any issue we will just ask it questions and there will be no beed to test. In fact we can will not have to worry about who we hire either since the new AI labs will fundamentally change the way we do science and shift to the expertise to the AI which can do it better and faster.
With all of the risk averse culture and "improved" safety protocols, could we actually do a test these days? Did we do a good job of documenting how to do it since all of the expertise in that area has either retired or expired.
2:10 -- Boeing jets do have successful maiden flights. In fact so did the Space Shuttle and the Soviet shuttle "Buran", the moon landing, etc. Also there are other complex projects like the Webb space telescope, or the Titan lander.
It is both unnecessary and impossible.
It is both unnecessary and impossible.
4/12/2025 7:22 PM
AI can do simulated testing far better than a real test ever could. It is good thing that the labs are now AI labs. The world of science completely changed in 2022, it is completely it is unrecognizable now. All money should now be shifted to to large language models in AI. All of it. That will make the Chinese and Russian the most fearful.
When LLNL signed the contract with IBM which started the ASCI project, I asked one of the pioneers of big computing at the Lab what he thought of the Stockpile Stewardship Teraflop (SST). He told me he didn't trust it. He then told me of a design that was successfully tested and then in the following year, that same design was tested again and was a fizzle. It turned out that a vendor supplying a component had changed it's chemical composition without informing the lab. That is something that simulation, even with the highly touted AI wont reveal. In sports they look at the roster of a team and predict who should win, but you still have to play the game.
But AI will change everything
The safety culture is just a response to the post-Cold War environment of having no actual mission or purpose. Lab management is very adept at following the money. Once NNSA makes it clear what they need to do to get their award fee, safety culture will be quickly dropped and the management will turn on a dime. It's already happening with efficiency efforts, where before the motive was to be as inefficient as possible.
My gosh, you are so smart. I never realized that. I guess the possibility of one successful first test proves that no testing is ever required, right?
One might suppose that our adversaries would also test if we did, and the increased confidence in the weapons on all sides might make their use more likely.
The entire premise of the SSP is that testing is unnecessary. Well, if that's true, than other countries shouldn't care whether the US tests or not because testing gains us nothing. You can't have it both ways. Either the SSP is a fraud and testing is necessary or the SSP has been successful and therefore resumed testing is a non-event.
Post a Comment