Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Missile defense is NP-complete

smu160.github.io

Post reply on HN
Print view View original

91–100 of 446 posts

Re: Missile defense is NP-complete

#91

Post by btown »

neutronicus wrote: post #54
myrmidon wrote: post #29

Earlier quoted context omitted.

This is absolutely true, but there is a strong counterpoint: You also learn the limits of your own systems and how to operate them most effectively yourself (and better than adversaries can, too). Just to pick a recent example: Russian air defense in the early stages of the Ukraine war was dismal (more specifically: defense against big, slow drones like Bayraktar), despite having sufficient AA capability "on paper"--…

"Data moats" are a problem for military tech, too, I guess.

One very interesting instance of the "military data moat" is Ukraine's annotated database of drone footage, perhaps the first of its scale from live engagements [0]:

> They can now draw on an enormous pool of real warfare information. Last year alone, Ukrainian drones recorded around 820,000 verified strikes against Russian targets... Meanwhile, the country’s Avengers AI platform detects upwards of 12,000 enemy targets every week. Developers can now access these sources and the data that they gather to train their systems on the movements of a real Russian turtle tank or a camouflaged Lancet launcher.

> “Ukraine currently possesses a unique body of battlefield data unmatched anywhere in the world,” recently appointed Defence Minister Mykhailo Fedorov said in a statement. “This includes millions of annotated frames collected during tens of thousands of combat drone missions.”

With the latency and offline constraints of battlefield technology, smaller models, trained with better data, may prove to have a significant edge. But it's still early days on how data like this might prove advantageous in other environments.

[0] https://resiliencemedia.co/how-ukraine-is-transforming-its-b... (unconfirmed source, this is not an endorsement)

Re: Missile defense is NP-complete

#93

Post by adampunk »

energy123 wrote: post #44

What is the steady state? Assume you have two competent superpowers, both researching missile offense and defense, over the next 1000 years. What are the asymptotics of the interception rate from 0 to 1000?

https://en.wikipedia.org/wiki/The_bomber_will_always_get_thr...

That’s the steady state. Interceptors are expensive; missiles are (relatively) cheap. There’s no sine wave or cat and mouse game. If you’re trying to defend against a peer, missile defense loses.

Re: Missile defense is NP-complete

#94

Post by wat10000 »

gos9 wrote: post #23
busterarm wrote: post #5

Earlier quoted context omitted.

Careful. Preemption takes many forms, some of them many would find unpalatable.

Unpalatable preemption is generally better than reentry vehicles coming down your chimney.

A thought experiment: would the world be a better place if the US had preemptively attacked the USSR in the 50s or early 60s when it was possible to do without more than “get[ting] our hair mussed” as General Turgidson put it?

Re: Missile defense is NP-complete

#95

Post by keybored »

lejalv wrote: post #53
1234letshaveatw wrote: post #21

Earlier quoted context omitted.

That seems like an acceptable trade off to get some real world experience with what works and what doesn't with regards to massed drones and swarming. There is a lot we can learn in this conflict with relatively low stakes

Stakes for whom? >100 kids got murdered the first day of this "low stakes" war

“Iranian kids may die... but that’s a prize I’m willing to pay.”

Re: Missile defense is NP-complete

#96

Post by jandrewrogers »

jmyeet wrote: post #76
jsw97 wrote: post #8

The author explains that this problem is actually adversarial, in the sense that the attacker gets to observe defenses and allocate warheads and decoys accordingly. Thinking of our current circumstances, this suggests another cost of war: our offensive capabilities, as well as our defensive capabilities become more observable. Our adversaries are studying our strengths and weaknesses in Iran, and they will have a muc…

In strategic circles, this was a common thought in the 12 day war: Iran was essentially mapping and testing defenses. As evidence of this, the US was forced to hastily move THAAD ground station radar from South Korea because Iran destroyed a bunch of them in the Gulf [1][2]. Bear in mind there aren't many of these and they cost half a billion dollars each . Further evidence of this is how quickly it happened. Iran mo…

> Iran destroyed a bunch of them

If by "a bunch" you mean one.

Re: Missile defense is NP-complete

#97

Post by sgtsteaks »

jmyeet wrote: post #45

Pardon the pun but this is an arms race and the defenders are going to lose. There are broadly five classes of missiles (one isn't a missile per se): 1. Ballistic. These are traditional rockets, basically. While rockets are designed to reach orbit or leave the Earth, a ballistic missile basically goes straight up and comes down. The higher it goes, the further away it can get because of the ballistic trajectory and t…

You cannot shoot a drone down with a shotgun. The range on a shotgun is nothing.

Do you know that it actually fires bb's out in a cone shape? If you aim a shotgun up in the air, you are not taking out any drones.

Look up the video of the drone hitting the hotel in Bahrain to get an idea of the speed and altitude.

Re: Missile defense is NP-complete

#98

Post by energy123 »

adampunk wrote: post #93
energy123 wrote: post #44

What is the steady state? Assume you have two competent superpowers, both researching missile offense and defense, over the next 1000 years. What are the asymptotics of the interception rate from 0 to 1000?

https://en.wikipedia.org/wiki/The_bomber_will_always_get_thr... That’s the steady state. Interceptors are expensive; missiles are (relatively) cheap. There’s no sine wave or cat and mouse game. If you’re trying to defend against a peer, missile defense loses.

I'm talking about interception probability, not the relative cost. I get that interceptors will probably be more expensive indefinitely (unless we start putting lasers into orbit to get around atmosphere, or something unexpected like that).

Re: Missile defense is NP-complete

#99

Post by ceejayoz »

energy123 wrote: post #56
myrmidon wrote: post #29

Earlier quoted context omitted.

This is absolutely true, but there is a strong counterpoint: You also learn the limits of your own systems and how to operate them most effectively yourself (and better than adversaries can, too). Just to pick a recent example: Russian air defense in the early stages of the Ukraine war was dismal (more specifically: defense against big, slow drones like Bayraktar), despite having sufficient AA capability "on paper"--…

Practise is good, but exhaustion is bad. Russia is getting exhausted, which is why their influence collapsed in Syria, Azerbaijan and Armenia, allowing the US to overtake those vacuums. The US in WW2 staged their 20th century by letting others (China, South East Asia and the British/Soviets) get exhausted first. This was more an accident of geography rather than US grand strategy, but it worked all the same.

Except this looks likely to exhaust the US/Israel alliance, if it continues long, leaving China in the "US in/after WWII" spot in the analogy.

Re: Missile defense is NP-complete

#100

Post by roysting »

myrmidon wrote: post #29
jsw97 wrote: post #8

The author explains that this problem is actually adversarial, in the sense that the attacker gets to observe defenses and allocate warheads and decoys accordingly. Thinking of our current circumstances, this suggests another cost of war: our offensive capabilities, as well as our defensive capabilities become more observable. Our adversaries are studying our strengths and weaknesses in Iran, and they will have a muc…

This is absolutely true, but there is a strong counterpoint: You also learn the limits of your own systems and how to operate them most effectively yourself (and better than adversaries can, too). Just to pick a recent example: Russian air defense in the early stages of the Ukraine war was dismal (more specifically: defense against big, slow drones like Bayraktar), despite having sufficient AA capability "on paper"--…

There is no amount of math that can make up for the lopsided dynamic of hypersonic missiles. The only reason the “iron/gold dome” con job was even plausible to plunder trillions in U.S. Monopoly money was because missiles were crude, slow, and not MIRVed or had decoys at one time. That was a long time ago though.

MIT Prof. Emeritus, Theodore Postol, has been trying to warn about this basic, mathematically proved fraud for many years now. However between the indifference because the party was still in high swing and the plundering was making people rich who could pay professional lobbyists/liars, very few people were paying attention or really cared, even though it’s clear fraud and just a false confidence; as is the objective of a con job, which comes from “confidence trick”.

There are several lectures he gives and more recent appearances on various YouTube channels where he clearly describes the inherent fraud in “missile defense”.

Here’s the synopsis; it’s like trying to prevent sand from hitting you once someone has thrown a fist full of dry sand at you.

It’s basically just the end game in a long history of American snake oil salesmen turned missile defense salesmen. You get useless junk, they run off with your wealth.

Post reply on HN