With 2025 behind us, we can look back on the year and figure out who deserves that most prestigious of Naval Gazing awards, the William D Brown Memorial Award for the biggest naval screwup that didn't kill anyone.
Compared to last year, our crop of entries is slightly down, although one navy in particular seems determined to win a third award on the trot:
- USS Harry S Truman ran into a tanker, then dropped one Super Hornet over the side while evading missile attack before losing a second a week later when the arresting wire broke
- USS Nimitz, for losing a Super Hornet and a helicopter on her final cruise
- USS New Orleans, attempting the classic route to Brown glory by setting the ship on fire
- The US Navy, specifically SecNav Phelan and SecDef Hegseth, for their mishandling of the Constellation program
Fortunately for the free world, none of these were sufficient to claim the prize, or even the coveted status of runner-up. The latter goes to the People's Liberation Army Navy and the Chinese Coast Guard for some truly astonishing shiphandling during their harassment of Filipino vessels in the South China Sea. They claim nobody was killed, preserving their eligibility,1 and for the first time, the incident was captured on video:
But they clearly lacked the necessary will to win, which could have been demonstrated by sinking one or more of the ships involved in the collision, and thus were narrowly beaten out by North Korea, for the truly impressive feat of capsizing a destroyer while trying to launch it. Even though it's not clear how much they wanted the victory, as they chose to refloat the ship and are continuing efforts to put it into service, the judges believe this demonstrates the excellence that the Brown Award stands for. Well done, North Korea, and we hope this is merely the first of many victories.
Past Winners:
- 2018: The Royal Norwegian Navy and the crew of Helge Ingstad for playing in traffic with an oil tanker
- 2019: On account of nobody reaching the required standard of excellence in 2019 alone, this Award has been given as a lifetime achievement award to Vern Clark, Arthur Cebrowski, Wayne Hughes and Donald Rumsfeld.
- 2020: The USN and the crew of Bonhomme Richard, for using fire instead of an oil tanker to write off a ship
- 2021: The Iranian Navy and the crew of the replenishment ship Kharg, although style points are docked for copying the previous year's winner
- 2022: Evergreen Marine, for breaking the actual William D Brown's record for biggest ship hard aground in Chesapeake Bay
- 2023: The USN and VP-4, for turning a perfectly good P-8 into a very bad boat
- 2024: The USN and the USS Gettysburg, for shooting down one of their own planes in the Red Sea
1 It's not entirely certain that this claim is true. The Filipinos are claiming that there were 2 dead as a result of the collision, which if true would disqualify it from contention. I'm not an expert in sorting out these kind of claims, so I'm going with official figures. ⇑
Comments
Being interested in failure modes generally, these are always some of my favorite posts, but this sentence in the summary accident report really piqued my interest... "The second Super Hornet made an evasive maneuver and was set to eject, but the second SM-2 was deactivated before impact" sounds like a great topic for an article. From the that that "It explodes off our left on the surface of the water", "deactivated" presumably means something more like "told to ignore its target" but even that leaves a lot of questions. Do humans or machines make that decision? (Given how little reaction time there must have been over those distances at those speeds.) I assume there's an encrypted command sent, and that there's nothing you can do about jamming (because the obvious alternative would mean that jamming would make all your missiles ineffective), and cryptography covers spoofing? Do we know whether that's post-quantum cryptography yet, just in case some power already has an effective quantum computer, or is everyone still assuming no-one does? If the command is more of a "divert" than a "don't explode" (because even a non-explosive collision is likely to be problematic at those speeds), is there any sophisticated attempt to not accidentally hit other targets [including ground-based ones], or does it just hit what it hits? Or were the pilots' impressions incorrect, and the missile was actually commanded to explode off-target (more like what you see when a rocket launch goes wrong, say)? So many questions raised by one short sentence!
Keep up the good work!
Oh, Bean, please do not investigate on 1997, eh...
BTW, the Chinese coastguardman trying to put out the fender to prevent the collision is hilarious...
enh:
As if they'd tell.
But AES would probably be good enough.
An aside on jamming, prompted by enh - nothing can really “defeat” RDF triangulation, right? Like, no matter how complex the opponent jammer is, even a basic scanned arena will passively provide azimuth - the problem is getting range and/or trying to ferret out signal from noise for targeting purposes.
Calling off the attack would have to be human. As for how it's done, it depends on what type of SM-2 it is. If it's an old one with semi-active homing, all you have to do is switch off the illuminator. If it's a modern CEC model, then it has its own radar, so you'd need do something else via the datalink. Presumably, there's a "don't shoot down anything" mode you can put it in where it shuts off the radar, although that may not be integrated with the fuze because why would you?
@enh I would not really talk about "post-quantum cryptography". The thing is that the Shor algorithm and similar algorithms (which people usually think of when they are calling about Quantum computers and cryptography) are not mathematically proven, and there is nothing that would even intuitively explain why they work. It now works for extremely small numbers; it worked for factorising 21 in 2012, and last years allegedly after billions spent, someone was capable of factorising 35. And it is extremely easy to be right with small primes. E.g. take the Fourier series:
f(x) = sin(3.1342x + 5.0928) + sin(0.0402x + 2.9998) + sin(2.1456x + 3.6480)
This function is actually positive for any prime number between 2 and 35 and negative for composites, and it has only 6 parameters (Disclaimer: I have coded it in 3 minutes, maybe it is not true).
Actually noone thinks that Shor just made it so it would work for a couple of the first primes, but the problem is that the primes have some internal stochastic structure that usually shows after some variations of the Fourier transform (there is a beautiful video even for non-mathematicians on YT, Three Blue, One Brown). So, considering extreme complexity, a lot of rounding up and usage of the Fourier transform, it is very probable that the Shor Algorithm just hit some stochastic invariant that will be close to small primes, but with large primes used in cryptography, it will be useless, as other already known stochastic invariants.
So it is definitely good that there is a big investment into quantum computing, but even if it worked (and it is now 0.0001% what they would need), there is like e.g. 5% chance that it would make any change in cryptography. It is far more probable that the number theorist would just find another stochastic invariant of primes. On the other hand, if this 5% come true, it would change the world.
@redRover
I would say that this is plausible with one jammer, but the SEAD usually do not work this way. You have multiple planes jamming on the same frequencies, so it is really hard to find out the azimuth, because it will point to the centre between these planes.
"Shor algorithm and similar algorithms (which people usually think of when they are calling about Quantum computers and cryptography) are not mathematically proven"
Exactly what part of the algorithm, or the estimate of its runtime, do you think relies on a conjecture?
@Jacob Manaker
The part when you try to get the frequency out of the distribution. For the current RSA, the precision required is 2**-4097. And the problem is that there is nothing that would limit the variance of the distribution to the precision.
The Shor algorithm is replacing finding primes with finding the prime frequency, for which the Shor algorithm can calculate the probability distribution. And you can run it multiple times, where the algorithm gets with 4/pi**2 probability "close enough" with every run. But to prove that it is not just a stochastic invariant but really a working algorithm, you would need to prove that: After n runs of the algorithm, where we have tried k most probable frequencies, we get the correct frequency with probability p(n,k,(N)). And that N in parentheses is very important, because it will probably be dependent on size/precision.
@SB
I don’t think this is true - a directional antenna (either mechanically rotating a traditional antenna or electronically scanning a phased array) should have enough gain to discriminate each azimuth, not just the center.
To be sure if you have multiple transmitters it becomes harder to triangulates because of the combinatorial explosion, but that seems potentially solvable via either time tracking estimates or signal comparisons between receivers.
(For that matter I think you could do TDA analysis on jammer signals if you know the receiver position with high accuracy and time coordination - instead of knowing where the GPS satellites are and trying to find where the receiver is, you know where the receivers are and have to triangulate the transmitters).
@redRover
I do not know if it can be solved just mechanically, for separating the signals, you would need to run some type of eigenvalue decomposition, and I would guess there are some ways to make this more difficult with polarisation, or some clever phase engineering.