Hacker News Reader

Comment by naringas | Hacker News Reader

>“We are going to be shocked by the speed, the chaos, the bloodiness, and the friction of a future fight in which this will be playing out,

I think this article's focus is way off

It's more plausible that algorithmic warfare is going to make us unsure as to what is even happening. We are going to be lost, drowned, in conflicting plausible information to the point we won't even know what is going in nor who to believe

View on HN

I agree with you. Since WW2, the trend in military weaponry and intelligence is precision.

- Instead of blowing up a submarine, why not disable the one critical crew member needed for it to launch?

- Instead of leveling a city to stop arms production, why not just infect the control systems of the factories so only plows can be manufactured and not swords?

- Cause a single screw to fall out of a piece of artillery so that it can no longer fire on target?

- Better yet, cause a fire in the iron mine so that gun can't even be manufactured.

To be effective, the affected parties wouldn't even know they were being targeted. They wouldn't even know they were at war. The lowest friction conduit your enemy civilization can proceed on is the one you want them to, and over time that's the one they'll inevitably settle into after becoming frustrated with their inability to make meaningful war on you.

In a perfect scenario, you might not even know you're attacking them either.

The series "The Golden Oecumene" explores this concept. In the story, the entire world's military might has reduced to a single sophisticated computer A.I. and a single man, who is usually cryogenically frozen and only brought out under extreme circumstances for specific operations that might include severing a single synaptic connection in a target using an x-ray laser stationed in orbit -- in order to change the target's mind.

Since WW2, the trend has been the construction of bigger and bigger doomsday weapons.

Why disable one critical crew member needed for a submarine's launch, when you can kill him, the entire crew, and everyone in a 10-mile radius with a single nuclear warhead?

The only thing 'precision' weapons have been found to be any good for is bombing stone-age adversaries, without putting boots on the ground. I see no situation where nuclear superpowers can get into a high-stakes[1] shooting war which will not quickly devolve into a nuclear exchange.

[1] Low-stakes, low-commitment engagements can continue to be performed by finding a stone-age adversary that you can bomb, as a proxy for your enemy. It's great for domestic morale, great for the military-industrial complex, and great for furthering our international ambitions.

> Why disable one critical crew member needed for a submarine's launch, when you can kill him, the entire crew, and everyone in a 10-mile radius with a single nuclear warhead?

Because you don't have enough of them...

You can't win against a nuclear state with 100+ ICBMs unless you strike first, or do a beheading strike.

USSR had really no defensive plan, as was found after the fall. They really bet all on "firing all your guns, before enemy can destroy them"

This only works during a “Cold War” one of the goals of intelligence is to prevent war by either maintaining your balance or superiority over your enemies and once a war breaks out provide intelligence that would grant you an advantage on the battle field.

However if a full scale war ever breaks down it will definitely regress back to leveling whole cities fairly quickly because in the end of the day it’s the only way to beat an enemy.

If the enemy could not sustain low levels of attrition even if key individuals they likely wouldn’t engage in a war in the first place.

However people over estimate how much intelligence should be credited to stopping all out war vs the economics of the connected world even during the Cold War.

Trade stops wars, hence why all recent wars have been civil in nature or ones against isolated actors of little to no economic importance on a global scale.

While a war with China might break out under extreme circumstances it’s highly unlikely when the US public relies on it for anything from their tooth brushes to their iPhones.

War with Russia is also extremely unlikely since half of Europe would freeze to death, they don’t need to nuke Berlin, as shutting off the gas supply would be nearly as devastating.

Russia’s reliance on natural gas to fuel not only their economy but also their geopolitical capita is also why it reacts so aggressively against European incursion eastward that could threaten their monopoly on LPG.

What Russia fears more than 500,000 NATO troops on its border is a Royal Dutch Shell pipeline going form the Caspian Sea through the Black Sea into Europe.

So while space death rays might seem to be effective they aren’t really not are they necessary today to actually prevent conflict between the big players.

With all the death rays in the world unless you’ll be willing to kill off the entire population of your enemies as long as they have enough peasants and rifles they can wage an effective war.

For China to effectively conquer a country they only need enough guns as nukes aside even the US might not have enough conventional weapons to deal with China if they’ll go bat shit crazy and issue a gun, an inflatable raft and a parka to every able man and woman of military age and tell them to march north west till the reach Alaska then start marching south.

There are small towns in Texas with enough conventional weapons for that. Had me till them

What a ridiculous scenario - such a gigantic force marching on foot at human speed would not only die in huge numbers of exposure, law and order / supply chain breakdown - they would also be a perfectly amassed easy-pickings for slaughter via air bombing.

> In a perfect scenario, you might not even know you're attacking them either.

How terrifying. To think that we could have been living in this world for 10 years now and not even realize it.

Remember the SR71 spy plane was classified for 40 years and when it was finally acknowledged after all that time it was still the fastest airplane ever created.

Military technology tends to have a guiding effect to regular technology. I wouldn't be surprised to learn that war has been guided in some capacity by AI for quite some time.

Uh, I had toys in the mid-1980's that were based on the SR-71 blackbird.

https://www.yojoe.com/vehicles/86/nightraven/

There were accurate SR-71 scale model kits by Monogram and Revell that I was putting together with Testor's glue in elementary school. It was not classified for 40 years. From 1959 to 1979 would indicate 20 years, if that.

https://en.wikipedia.org/wiki/Monogram_(company)

I was a small child when I learned about Guinness world records, and first discovered the SR-71 while writing a book report. It was understood that the blackbird had established records of mach 3 in the late 1960's. They even used to have an exhibit of Guinness records in the lobby of The Empire State Building in Manhattan that included this record.

The Guinness record states 1968, and that was in print in the early 1980's. You can probably buy old, used copies of Guinness World Records editions, which prove this.

https://www.amazon.com/Guiness-Book-World-Records-1980/dp/08...

The 1988 edition had a newsprint/paperback quality photo of the plane:

https://www.amazon.com/Guinness-Worlds-Records-Giant-Revised...

Parts of the program were obviously classified, but the aircraft was known to exist and was photographed by the press, unlike the F-117 and B-2, which had not emerged into public until the 1990's.

What is the target of algorithm warfare supposed to be? This concept makes no sense to me.

Submarine warfare? Makes sense. You kill subs and ships and control waters. What is the algorithm seeking to control?

It doesn't even seem like a new kind of warfare, just a decision maker for existing types of warfare. So it basically has rules for land, air, sea, internet warfare and ties everything together?

Is the US President basically going to enter "control Russia" at a command line and the AI plans and executes a cold/hot war? This seems laughable.

Sure, to a certain extent, algorithms are not so different than "strategies" that generals employ in response to enemy movements.

But the speed and execution of these strategies can be improved order of magnitudes as we replace more parts of the system with machines.

One can imagine Sun Tzu's chapter "Weak Points and Strong" discussing various heuristics and strategies to determine a combatant formation's weakest point. With machine algorithms, you could launch a similar probing attack instantaneously, and have the results analyzed and the striking attack coordinated before your enemy even hears the first reports of encounter.

>But the speed and execution of these strategies can be improved order of magnitudes as we replace more parts of the system with machines.

You're going to be limited by military logistics. Things moving at certain speeds across the Earth. An algorithm can't change that until we get to AGI (maybe). I'm sure it can remove some human inefficiencies in organization, but there is a minimum time and you still need an overall goal like deciding to attack Russia.

But besides that, the formations don't really matter. All country vs country warfare is based on long range, fast moving weaponry. Something like MOAB will incinerate the entire formation.

To me it feels like military people trying to find a brand for this particular thing and to justify programs for it by inventing a need.

Even then, for the US, these small gains just don't matter because of our overwhelming might.

Usually folks in the business are concerned about cyber attacks on important warfighting infrastructure (GPS satellites, for example). Cyber attacks against early warning radars, communications infrastructure, civilian power infrastructure and so on. You can imagine a battle fought solely in ‘cyberspace’ over control of critical satellites.

Sometimes electronic warfare is also lumped into this in the form of jamming or spoofing radars.

I think the disinformation you're thinking about matters [cf 0], but being able to project power to the right place at the right time requires a lot of logistical planning and material even if the information can be trusted. There are a lot of military areas where AI can inform traditional bureaucratic processes from planning to command and control [cf 1].

DOD maintains a lot of troops/bases/relationships around the world, and my guess is they want to be optimizing all of them as much as possible and be able to adapt policy to changes. Like any large organization, it's hard for the higher ups to have a good grasp on what's going on at lower echelons. They've had some pretty bad policy decisions handed to them from not being able to explain what was going on (e.g. the disbanding of the Baath party in Iraq). Being able to algorithmically integrate the information they've got and present it to policy makers has been an unsolved problem in its own right without the disinformation stuff.

[0] Whaley,B. (1982). "Toward a General Theory of Deception" https://www.tandfonline.com/doi/abs/10.1080/0140239820843710...

[1] MCDP 1. (1997). "Warfighting" https://www.marines.mil/Portals/1/Publications/MCDP%201%20Wa...

Going to be? We're already there.

Who's to say that we're not in the middle of it now? Lots of little surgical strikes appearing to be russian hackers, heart attacks, ethical lapses and folks caught in unlucky situations?

Meanwhile, in the late-twentieth-century phase of the arms race, the role of unpredictable chance increased. When hours (or days) and miles (or hundreds of miles) separate defeat from victory, and therefore an error of command can be remedied by throwing in reserves, or retreating, or counterattacking, then there is room to reduce the element of chance. But when micromillimeters and nanoseconds determine the outcome, then chance enters like a god of war, deciding victory or defeat; it is magnified and lifted out of the microscopic scale of atomic physics. The fastest, best weapons system comes up against the Heisenberg uncertainty principle, which nothing can overcome, because that principle is a basic property of matter in the Universe. It need not be a computer breakdown in satellite reconnaissance or in missiles whose warheads parry defenses with laser beams; if a series of electronic defensive impulses is even a billionth of a second slow in meeting a similar series of offensive impulses, that is enough for a toss of the dice to decide the outcome of the Final Encounter. Unaware of this state of affairs, the major antagonists of the planet devised two opposite strategies. One can call them the "scalpel" and the "hammer." The constant escalation of pay-load megatonnage was the hammer; the improvement of detection and swift destruction in flight was the scalpel. They also reckoned on the deterrent of the "dead man's revenge": the enemy would realize that even in winning he would perish, since a totally obliterated country would still respond -- automatically and posthumously -- with a strike that would make defeat universal. Such was the direction the arms race was taking, and such was its destination, which no one wanted but no one knew how to avoid. How does the engineer minimize error in a very large, very complex system? He does trial runs to test it; he looks for weak spots, weak links. But there was no way of testing a system designed to wage global nuclear war, a system made up of surface, submarine, air-launched, and satellite missiles, antimissiles, and multiple centers of command and communications, ready to loose gigantic destructive forces in wave on wave of reciprocal atomic strikes. No maneuvers, no computer simulation, could re-create the actual conditions of such a battle. Increasing speed of operation marked each new weapons system, particularly the decision-making function (to strike or not to strike, where, how, with what force held in reserve, at what risk, etc.), and this increasing speed also brought the incalculable factor of chance into play. Lightning-fast systems made lightning-fast mistakes. When a fraction of a second determined the safety or destruction of a region, a great metropolis, an industrial complex, or a large fleet, it was impossible to achieve military certainty. One could even say that victory had ceased to be distinguishable from defeat. In a word, the arms race was heading toward a Pyrrhic situation.

Reply on news.ycombinator.com