Nobody designing a financial system today would invent credit cards. The Western world uses credit cards because replacing legacy systems is expensive. China doesn't use credit cards. They skipped straight from cash to WeChat Pay. Skipping straight to the newest technology when you're playing catch-up is called leapfrogging.

A world-class military takes decades to create. The United States' oldest active aircraft carrier was commissioned in 1975. For reference, the Microsoft Windows operating system was released in 1985. The backbone of NATO's armed forces was designed for a world before autonomous drones and machine learning.

The United States dominates at modern warfare. Developed in WWII, modern warfare combines tanks, aircraft, artillery and mechanized[1] infantry to advance faster than the enemy can coordinate a response.

Modern warfare is expensive—and not just because of all that heavy machinery. Modern warfare delegates important decisions to the smallest unit capable of making them. Officers must be smart and they must be trained. Training officers to fight a modern war is hard. It takes a long time. There's constant turnover. It's a human resources nightmare. You can't just throw money at the problem.

Soon it will be possible to throw machine learning at the problem instead.

At the center of [China's] public discussions is a new and little-understood concept called “intelligentization (智能化),” which represents a new goal for the PLA’s progress in modernization…. Chinese theorists’ discussions about intelligentization overwhelmingly call for highly centralized decision-making structures. These strategists want operational commanders advised by advanced algorithms to perfectly direct intelligent swarms of autonomous battle systems to achieve campaign objectives. Chinese theorists believe this approach will consolidate command responsibility onto a few generals who can remain safely away from the frontlines of the battlefield, which is antithetical to the modern concept of mission command.

Schrodinger’s Military? Challenges for China’s Military Modernization Ambitions

AI-centric postmodern warfare has advantages over human-centric modern warfare.

  • Human communication is a bottleneck for large organizations. Computer command systems can coordinate perfectly and instantly. Human beings cannot.
  • It's easier to mass-produce computers than human specialists.
  • AI-centric warfare is on the winning side of a ratchet. AI capabilities advance while human capabilities remain constrained by biology. Whenever an AI system gets better than human beings at a specific task it remains that way permanently.

Most importantly, AI-centric command is the only viable method for commanding swarms of unmanned aerial vehicles.

Unmanned aerial vehicles (UAVs) are smaller and cheaper than piloted aircraft. A UAV can be remote controlled or it can be autonomous. Remote controlling a UAV takes a lot of bandwidth because the UAV must send back its sensory information to mission command. This works fine when you're controlling a handful of Predator drones. Remote control will not work when you're controlling a swarm of 10,000 small UAVs against a peer adversary. Direct communication is fragile and there isn't enough bandwidth in the radio spectrum for indirect transmission. UAVs swarms must be autonomous.

The disadvantage of postmodern warfare is that centralized computer-controlled systems are fragile in a different way. If critical systems get compromised (or just fail in an unexpected way) then the entire war machine breaks. I think the advantages are worth the risks. It's not like our critical infrastructure isn't already vulnerable to cyberattack. Moreover, distributed fault-tolerant architectures can help mitigate the risks.

Western military theorists claim that today's autonomous systems are not ready to command the battlefield. This is true but it's also beside the point. China is building its military with forward compatability in mind. Software advances faster than hardware. By investing in autonomous battle systems today, China can continuously update to the newest AI as machine learning advances.


  1. In this context, "mechanized infantry" refers to wheeled [edit: and tracked] vehicles, not power armor and battlemechs. ↩︎

57

25 comments, sorted by Highlighting new comments since Today at 10:21 PM
New Comment

At the moment, centralized human command-and-control norms make large militaries similarly fragile. They've gotten around this in recent decades by letting smaller, more autonomous parts take over when there are enough dead bodies, but this does not last long. The same problems in policy elsewhere continue within militaries. While elements of the military have seen this problem (it's an old problem best explained by the writing of David Hackworth, Erwin Rommel, and B.H Liddell-Hart), they can't really change it from the inside, so there's been a shift toward retirees who've started consulting firms with an implicit aim of changing culture from the outside.

Progress in technology will yield the same problems that it did in the 60s (when militaries had access to better radio communication, they used it to try to enforce tighter control on their units while avoiding the ground- commanders literally flew in helicopters instead of having their boots on the ground): militaries will attempt to use that technology for more central control, which will continue making its systems more rigid and vulnerable to a sudden catastrophic failure.

One of the most important decisions in war is when to stop.  Humans evolved fear to solve this problem; there's a point at which soldiers will de-escalate the conflict (i.e. flee the battlefield rather than stay and die).  However, signalling fear makes you a target so people don't discuss it candidly.  I am concerned that military leaders may, in the calm of the office, design AI that has no provisions for de-escalating conflict; this seems very likely to lead to nuclear war.

We seem to be in the midst of a trend away from direct confrontation and toward sowing discord. We want to be able to spread chaos in the ranks of our opponents, and have that intervention be perceived as a failure in our opponents' ability to enforce discipline or coordinate effectively. As a concrete example, Russia may interfere with American social media websites in order to promote violent mysogyny or racism, provoking attacks by Americans on Americans. Americans may then blame the perpetrators, our own citizens, and our own culture for these outcomes, even if Russian interference was a necessary condition for many of these attacks.

I could imagine that AI will enhance the ability of militaries and intelligence organizations to execute such attacks in other ways. For example, we could imagine a whole field of "bug design," in which AI systems are created and sold with deliberately built-in bugs that nonetheless appear accidental, and cause the AI to act up at some pre-determined point in the future. AI might not just help programmers write code - it might help them write buggy code that nevertheless appears bug-free. The outcome can be framed as an accident, the developers of the weaponized bug may be impossible to identify, and it becomes very difficult to know what a proportional response looks like.

Another example would be a country imposing healthy eating incentives on its own population, while subsidizing the export of tasty and unhealthy foods to its adversaries. Then the population in the other country blames itself, or its own economic structure, for the spread of obesity.

I feel OK about posting this here, because I think that on net, the risk of spreading new conceptual ideas to an attacker is lower than the value of a population considering how to defend itself against such attacks on its sense of responsibility. The attack can be done by a small number of specialists. Defending against such an attack will require a heightened level of awareness at the population level, and that requires free and open discussion.

This sounds to me a lot more like "next-generation modern warfare"; postmodern warfare evokes Wizeman's Lethal Theory (pdf), edited for brevity:

This essay belongs to a larger investigation of the ways in which contemporary military theorists are conceptualizing the urban domain. What are the terms they are using to think about cities? What does the language employed by the military to describe the city tell us about the relationship between organized violence and the production of space? What does this language tell us about the military as an institution? Not least important is the question of the role of theory in all these operations. ...

There is a considerable overlap among the theoretical texts considered “essential” by military academies and architectural schools. Indeed, the reading lists of contemporary military institutions include Deleuze, Guattari, and Debord, as well as more contemporary writings on urbanism, psychology, cybernetics, and postcolonial and poststructuralist theory ... the discourses which shaped thinking in various academic fields toward the end of the 20th century have been employed for the reinvigoration of warfare.

See also the blog post Nakatomi Space; 'firehose of falsehood' propoganda tactics, etc.

I haven't explicitly modeled out odds of war with China in the coming years, in any particular timeframe. Some rationalist-adjacent spheres on Twitter are talking about it, though. In terms of certainty, it definitely isn't in the "China has shut down transportation out of Wuhan" levels of alarm; but it might be "mysterious disease in Wuhan, WHO claims not airborne" levels of alarm.

I'd expect our government to be approximately as competent in preparing for and succeeding at this task as they were at preparing for and eliminating COVID. (A look at our government's actions[albeit from a China-sympathetic American] suggests general incoherence.)

If someone with greater domain expertise than me has looked at this, I'd be interested in an in-depth dive.

Can you give some examples of who in the "rationalist-adjacent spheres" are discussing it?

Computer command systems can coordinate perfectly and instantly.

It is important to note that this is not strictly speaking true. Computer command systems can coordinate far more accurately and far faster than humans in many cases; this is neither perfect nor instant. (And fundamentally cannot be in a battlefield environment - the Byzentine Generals problem rules out the former and a finite speed of light the latter.)

This distinction reminds me of the battles in Ender's Game.

As I recall, Ender was the overall commander, but he delegated control of different parts of the fleet to various other people, as most modern militaries do.

The bugs fought as a hive mind, and responded almost instantly across the entire battlefield, which made it challenging for the humans to keep up in large-scale battles.

Ender's Game is about the transition from classical warfare to modern warfare. The formation-based strategies are copied from when European armies stood in a line with muskets. The later bits where Ender delegates decision making to his officers come from modern warfare doctrine.

Yeah, I read that bit in the article when Bret Deveraux linked to it, and I winced hard at this confidence that China's approach of wanting to use lots of AI was obviously was a bad idea. 

Modern warfare is expensive

The Taliban who didn't spend much money and still beat the US army because the US army is mismanaged enough that it can't win a battle against a much worse funded enemy.

Or...the US army had the impossible task of defeating the Taliban without harming the general population, despite a completely porous boundary between Talib and ordinary citizen.

Modern armies aren't just technologically and managerially modern, they are wielded by modern states, and modern states usually aren't ideologically comitted to total warfare.

The US army did poison the drinking water of the general population with uranium (shooting depleted urianium at the sources of their drinking water). The idea that they were avoiding harming the general population is inaccurate. Is always interesting when one reads about how the usage of the uranium mutation produces problems for US veterans without discussing what it actually does in the countries it's used. 

The US did reward commanders for killing a lot of people and thus got them to kill people even when the blowback wasn't worth it from a military perspective. 

They classified Afghan military units as being functional to forge their own statistics and mislead themselves with their bad statistics about what was going on in Afghanistan. Bureaucracies that delude themselves about what goes on aren't effective. 

The idea that they were avoiding harming the general population is inaccurate.

That's not what I said. I said that it was what they were tasked with.

I do agree that there was that expecation but I don't believe that's it's why they didn't win. 

There are a lot of things about winning hearts&minds that are not about waging total war.

There are a lot of things about winning hearts&minds that are not about waging total war

That's a huge understatement. The point was that the US has the technical resources to wage total war against much weaker countries, but doesn't have the political will. And that isn't a management problem.

I was talking about the US being unable to win the war. To the extend that you make a point that's unrelated to that question like whether or not they can wage total war, it makes sense to refer back to my first claim.

It's easier to visualize if you try to work out the hierarchy of software agents you might use for this.  
 

First, most of the bigger drones will probably some kind of land vehicle, whether a legged infantry or a robot on tracks.  This is for obvious range and power reasons - a walking or rolling robot can carry far more weapons and armor than anything in the air.  And in a battlespace where everyone on the enemy side has computer controlled aim, flying drones without armor will likely only survive for mere seconds of exposure.  

So at the bottom level the drones need to be able to plan and locomote to a given location on the battlefield, or report that they are unable to reach a particular location.  (Due to inaccessibility or damage to that particular unit - robotic units obviously won't stop fighting when damaged)

At a slightly higher level you have an agent that coordinates small "units" of drones to accomplish a mission from a finite set of trained "missions".

Missions might be things like "clear this structure of enemy fighters".

The agent at these 2 layers have been trained with collectively millions of years, with the red team agents controlling simulated enemy drones or simulated human bodies.  So the red team will likely be more combat effective at being a human than typical actual human soldiers.  So the trained policy of these agents will assume their opponent is doing the best that is possible.  

We don't know what the trained policy would look like but I suspect it involves a lot of careful control of exposure angles, and various unfair strategies.  

The layer above the bottom agent doesn't know how to perceive an environment, or how to locomote, or ballistics - it queries the lower level agent whenever it needs to know if a proposed action is feasible.  

Then the layer above this agent handles the battle itself, creating units of action of appropriate size and assigning missions.  This is where human generals coordinate.  They might choose a section of city and drag a box over it, ordering that the layer 3 agent subdivide that city section and clear every building.  

Each agent must query the layer below it to function, exporting these subtasks to an agent specialized in performing them.

Even the "level 1" agent doesn't actually directly locomote, it's similarly tiered internally.

The actual compute hardware is hosted such that many redundant vehicles run a VM hosted copy of the agent they need and the agent a level up.  Agents are perfectly deterministic - given the same input and an RNG seed they will always issue the same 'orders'.  This makes redundancy possible - multiple vehicles in parallel can run a perfect model of their 'commander' agent a level up, such that enemy fire destroying the vehicle that hosts a 'commander' will not degrade capabilities for even 1 frame.  

(each 'frame', several 'commanders' broadcast their orders, taking into account information from prior frames.  Each command is identical to all the others so the 'orders' must match or the majority will be used.  So at any given time there are 3-5 sets of 'orders' being broadcast to all agents in this subswarm.  So an incoming shot that blows up a commander, or jams it's communication leaves plenty of redundant copies of 'orders')

I think you're grossly underestimating the following effects/issues:

1. How do multiple redundant commanders ensure that they reliably have the same information, much less in a battlefield environment? Our best efforts still ended up with Bysantine faults on the space shuttle, and that was carefully designed wired connections... (see also Murphy Was an Optimist, which describes a 4-way split due to a failed diode).
2. How do commanders broadcast information in a manner that isn't also broadcasting their location to enemies? (Honestly, the least important of these issues, and I was tempted not to include this lest you respond to this point and only this point.)
3. If many vehicles are constantly recieving enough information to make higher level decisions, how do you prevent a compromised vehicle from also leaking said state to the enemy? Note the number of known attacks against TPMs, and note that homomorphic encryption is many orders of magnitude away from being feasible here. (And worse, requires a serial speedup in many cases to be feasible.)
4. If many vehicles have the deterministic agent algorithm, how do you prevent a compromised vehicle from leaking said algorithm in a manner the enemy can use for adversarial attacks of various sorts? Same notes as 3.
5. "Each agent must query the layer below it to function, exporting these subtasks to an agent specialized in performing them." What you're describing runs into exponential blowup in the number of queries in some cases. (For a simple example, note that sliding-block puzzles are PSPACE-complete, and consider what happens when each bottom agent is a single block that has to be feasibility-queried as to if it can move.) Normally, I'd just say "sure, but you're unlikely to run into those cases", however combat is rather necessarily adversarial.

The OpenAI 5 DOTA2 bot beating professionals received a lot of press. A random team who got ten wins against said bot, not so much. Beware glass jaws.

> in a battlespace where everyone on the enemy side has computer controlled aim, flying drones without armor will likely only survive for mere seconds of exposure.  

In a battlespace where everyone on the enemy side has computer controlled aim, flying drones with armor will likely only survive for mere seconds of exposure. It may be better to have smaller drones, or more maneuverable drones, or quieter drones, or simply more drones, over more armored drones. (Or it may not. The point is it's not as clearcut as you seem to make it out to be.)

(You may wish to look at discussions of battleships, and particularly battleship armor, versus missiles. And battleships are far less weight-constrained than fliers...)

So note I do work on embedded systems IRL, and have implemented many, many variations of messaging pipeline.  It is true I have not implemented one this complex, but I don't see any showstoppers.

  1. This is how SpaceX does it right now.  In summary, it's fine to have some of the "commanders" miss entire frames as "commanders" are stateless.  Their algorithm is f([observations_this_frame|consensus_calculated_values_last_frame]).  Resynchronizing when entire subnets get cut off for multiple frames and then reconnected is tricky, but straightforward.  (a variation of the iterative algorithms you use for sensor fusion can fuse 2 belief spaces, aka 2 maps where each subnet has a different consensus view of a shared area of the state space)
  2. It does, there is not a way to broadcast information that doesn't reveal your position.
  3. Please define TPM.  Message payloads are fixed length and encrypted with a shared key.  I don't really see an issue with the enemy gaining some information because ultimately they need to have more vehicles armed with guns or they are going to lose, information does not provide much advantage.
  4. Thermite, battery backed keystores.  And the vehicles don't have the actual source for the algorithms used to develop it, just binaries and neural network files.  Assuming that the enemy can bypass the self destruct and exploit a weakness in the chip to access the key in the battery backed keystore, they just have binaries.  This doesn't give them the ability to use the technology themselves*.  Moreover, the agents are using near optimal policy.  A near optimal policy agent has nothing to exploit - they are not going to make any significant mistakes in battle you can learn about.
  5. Nothing like this.  The "commander" agent's guessing from prior experience optimal configurations to put it's troops.  The "subordinate" agent it is querying runs on the same hardware node.  So these requests are obviously IPC using shared memory.  And the commander makes a finite number of "informed" guesses, gets from the subordinate which "plans" are impossible, and selects the best plan from the remaining (with possibly some optimizing searches in nearby state space to the current best plans).  This will select a plan chosen from the set of { winning battle configurations in the current situation | possible according to subordinate } that is the best of a finite number of local maxima.

         I am not sure your "glass jaw" point.  OpenAI is a research startup with a prototype agent.  It can't be expected to be flawless just because it uses AI techniques.  Nor do I expect these military drones to not have entire real life battles where they lose to a software bug.  The difference is that the bug can be patched, records of the battle can be reviewed and learned from, and the next set of drones can learn from the fallen as directly as if any experience that was uploaded happened directly to them.

           At the end there I am assuming you end up with varying sizes of land vehicle because they can carry hundreds of kilograms of armor/weapons.  Flying drones do not have even in the same order of magnitude the payload capacity.  So you end up with what are basically battles of attrition between land vehicles of various scales, where flying drones are instantly shot out of the sky and are used for information gathering. (a legged robot that can climb open doors and climb stairs I am classifying as a land vehicle). 

 Maybe it would go differently and end up being a war between artillery units at maximum range with swarms of flying drones used as spotters.

*I think this is obvious, but for every piece of binary or neural network deployed in the actual machine in the field, there is a vast set of simulators and testing tools and visualization interfaces that were needed to meaningfully work on such technology.  This 'backend' is 99% of what you need to build these systems.  If you don't already have your own backend you can't develop and deploy your own drones.

> In summary, it's fine to have some of the "commanders" miss entire frames as "commanders" are stateless.

Having a single commander miss an update? Sure. That's not really the problem. The problem is cases like "half of the commanders got update A and half didn't, which then results in a two-way split of the commanders, which then results in agents splitting into two halves semi-randomly based on which way the majority fell of the subset of commanders that they can see". You really should look up testing of distributed databases, because these sort of split-brain scenarios are analogous there.

You're also currently falling afoul of the CAP theorem I believe ( https://en.wikipedia.org/wiki/CAP_theorem  ). Note that "commanders receiving the same set of observations_this_frame" is equivalent to a distributed database with all observers adding observations and all commanders seeing a consistent view of this database...

> Resynchronizing when entire subnets get cut off for multiple frames and then reconnected is tricky, but straightforward

Again, you really should look up testing of distributed databases. One particularly interesting scenario is asymmetric failures. That is, A can send to B but not vice versa.

> This is how SpaceX does it right now.  

Yep. Consensus among multiple redundant computations is also how the space shuttle operated (although the details are somewhat different for the space shuttle of course). It's not perfect, but it's a fairly decent approach so long as failures are rare enough that multiple simultaneous failures are rare, and you are not in an adversarial environment.

> "commanders" are stateless.

Commanders cannot be stateless unless either a) they do not retain memories of previously-observed world state or b) they are included in the world data every frame.

The former results in demonstrably suboptimal behavior (there's a reason why humans have object permanence :-) ), and the latter requires that all agents be able to receive a full commander state in a single frame. (This in turn would result in your commander memory capacity being restricted by communication bandwidth.)

> Please define TPM.

My apologies. Trusted Platform Module. It's a secure cryptoprocessor standard (and also a term for implementations of said standard, such as the ones on most x86 chips). It's one of the more common attempts at secure computing, and has been attacked a fair few times as a result. (With more than a few successful attacks of various sorts.)

> Message payloads are fixed length and encrypted with a shared key.  

Note that as soon as an attacker learns said shared key you're now broadcasting all of your sensory information for all agents (as you have to in order to allow commanders to work as described) to the enemy. This is generally considered a Bad Idea (although you seem to disagree with this point, see below).

If you assume that information is not valuable, and that all agents broadcasting their position at all times is never harmful, I can see how this scheme of just broadcasting everything all the time can be useful. I disagree with the premise, however.

> I don't really see an issue with the enemy gaining some information because ultimately they need to have more vehicles armed with guns or they are going to lose, information does not provide much advantage.

This is incorrect from a game-theoretical point of view. A single example is matching pennies ( https://en.wikipedia.org/wiki/Matching_pennies ). Normally the second person can at most break even on average. However, they can always win if they knew what the first person's penny is...

(As to how this maps to warfare? You might try playing a few games of Stratego ( https://en.wikipedia.org/wiki/Stratego ). Much of the game is the same sort of "reinforcements are moving to A and B but I don't know which is the actual reinforcement and which is a feint quite yet" decisions.)

For a slightly more concrete example: I'll quite happily play you a game of chess where I start without a queen and you have no knowledge of where any of my pieces are. (That is, you try a move, and if it's legal that's the move that's taken. You do get knowledge of which pieces you have captured.)

Starting without a queen is a significant disadvantage in chess, so according to your assertion this should be easy to win for you. (A random online stockfish engine gives +1260 centipawns, and compare e.g. here https://chesscomputer.tumblr.com/post/98632536555/using-the-stockfish-position-evaluation-score-to where even +500 centipawns gives a win % of >90%.)

> Moreover, the agents are using near optimal policy.  

This is an interesting assertion. Do you have any citations showing that self-training results in near optimal policies in complex environments? In particular policies that remain near optimal in adversarial cases?

> This will select a plan chosen from the set of { winning battle configurations in the current situation | possible according to subordinate } that is the best of a finite number of local maxima.

There are a fair few optimization problems that you can run into that are proven to not have a polynomial-time approximation scheme unless P=NP (e.g. set covering).

As you can construct physical scenarios that map to said problems, this means that either:

1. You've solved P=NP, or are assuming that it is constructively proven that P=NP before this scenario happens.
2. You're requiring that your commanders potentially do exp(num agents) work per timestep.
3. Your agents are not using near-optimal policies.
4. Your agents are using quantum computing, and using an algorithm within BQP ( https://en.wikipedia.org/wiki/BQP ).
5. You are unaware of this result.

2 is typically physically impossible for more than a few agents, for the same reasons that symmetric-key cryptography can be secure.

> The difference is that the bug can be patched

...assuming you haven't already lost the war as a result. If the other side manages to take out 50% of your drones at once when you were previously at par, that's a problem. Saying "we'll get better for next time" only works if you're still in a winnable position.

The main issue I am drawing attention to here when I talk about glass jaws is correlated failure modes in adversarial environments, in particular ones that can result in system-level catastrophic failure. If 10% of your drones fail on average, you can plan around that. If your critical holdpoints are suddenly all catastrophically lost due to the same silly bug or edge case, not so much.

> And the vehicles don't have the actual source for the algorithms used to develop it, just binaries and neural network files.

This doesn't actually matter all that much for adversarial attacks. Look at the cat and mouse game that is Denuvo for the binary side of things (and note that one of the major techniques used now is to run important code on a remote trusted server, which is not something that the techniques you describe does), and note that if you have a (trainable) neural network you can differentiate it, which is all you need for adversarial attacks.

(Actually, you don't even need that. See e.g. https://arxiv.org/abs/2003.04884 )

And yes, there are countermeasures. But there are also countercountermeasures, and so on. It's a cat-and-mouse game, not a clearcut 'defender wins'.

It might be interesting to discuss this in a more interactive format, such as on https://discord.gg/GVkQF2Wn .  You do know some stuff, I know some stuff, and we seem to be talking past each other.  Fundamentally I think these problems are solvable.

(1) merger of conflicting world spaces is possible.  or if this turns out to be too complex to implement, you deterministically pick one network to be the primary one, and have it load from the subordinate network the current observations.  

(2) If commanders need more memory than the communications channel, they must exchange deltas.  These deltas are the (common observations, made as input from the subordinate platforms, and state metadata).  This is how a complex simulation like age of empires worked on a modem link.https://www.gamedeveloper.com/programming/1500-archers-on-a-28-8-network-programming-in-age-of-empires-and-beyond

(3) Free air laser links is one technology that would at least somewhat obscure the source of the signaling (laser light will probably reflect in a detectable way around corners but it won't go through solid objects) and is capable of tends of gigabits per second of bandwidth, enough to satisfy some of your concerns.

It might be interesting to discuss this in a more interactive format, such as on https://discord.gg/GVkQF2Wn .  

I'm not the Mailman, but I'm up there. I tend to write out a sketch, then go back and ponder it a while, massaging it into some semblance of order, deleting/modifying arguments that in retrospect don't work, and inflating it out into a quasi-coherent post in the process. This takes a fair bit of time. It works well in an asynchronous context. It does not work well in a synchronous context. In my experience, when I attempt to discuss in a synchronous context I end up with one of the following two things (or both!):

1. I state arguments or views that are insufficiently thought-out and that are obviously incorrect/inconsistent in retrospect, or are misleading/confusing/weaker than they should be.
2. I end up with essentially just a forum discussion that happens to be on Discord. Walls of text and all.

The 2nd would be fine, but this then runs into another issue:

Much of the reason why I am on a website like this is so that people can follow arguments / point out issues with my views / etc. Partly for the later benefit of others following my chains of logic. Partly for the later benefit of others when they can refer back to my chains of logic. Partly for the later benefit of myself, when someone down the line sees an old comment of mine and replies with something I hadn't thought of. And partly for the later benefit of myself, when I can refer back to my chains of logic.

Discord does not achieve these.

Someone searching this site does not see a Discord conversation. If you (or whoever owns the room, rather) close the Discord room, then the information is lost. (Or if e.g. Discord decides 6m down the line to start dropping old conversation history, etc, etc.) You can, somewhat awkwardly, archive a Discord conversation. And, say, post it on this site. But that's now just a forum conversation with extra steps (not to mention that it's now associated with the person who posted the transcript, not the people in the transcript. If you post the transcript and someone relies to a comment of mine in it, I don't get a notification.).

(2) If commanders need more memory than the communications channel, they must exchange deltas.

You previously stated that '"commanders" are stateless.'. Do commanders have state here?

If commanders are stateless, this technique does not work as they have nothing to base the deltas on.

If commanders are stateful, and are maintaining a worldstate by deterministically applying deltas... you're right back to CAP theorem limitations. Pick at least one of a) diverging worldstates in the presence of network partitions, b) the (bad) assumption of no network partitions, and/or c) arbitrarily-long stalls in the presence of network partitions.

(AoE falls under c) here, if you're wondering. In networked gaming, sacrificing consistency in the presence of a network partition is called a desync and is a Bad Thing(TM).)

(1) merger of conflicting world spaces is possible.

Sure. This is just eventual consistency, restated. With all of the wrinkles that eventually-consistent distributed databases have.

To go back to your game example for a moment. You've got a 2-on-1 match - AB versus X.

A, B and X each have an army.
X's army wins ties. So in a fight between X and A or X and B, X wins. But in a fight between X and AB, X loses.
X's army is faster. If both side's bases are destroyed, X wins, as X destroys AB's bases first.
X has three options: defend, or attack through one of two chokepoints - path P, that A has info on, and path Q, that B has info on.
Ditto, A and B each have three options: defend P, defend Q, or attack.
X went eco-heavy, and so AB will lose if they neither manage to destroy X's base nor destroy X's army.

This is essentially a coordination game of sorts, with an additional wrinkle that neither A nor B has the full picture of what's going on. 

In the presence of reliable prompt communication between A and B, this is a reliable win for AB. AB relay information on P and Q to each other, and both check paths P and Q. If P or Q has X's army, A and B send both armies there and destroys it. Otherwise, X's army must be defending and A and B send both armies to X's base, destroying it.

Now let's say that instead the network connection is lost between A and B. A checks and sees that P does not have X's army, but has no info on Q. This means that A must either send its army to Q, or X. But which one? It could be either. Say it sends it to X.

B checks and sees that Q does have X's army, but has no info on P (not that it matters in this case). This means that B must send its army to Q, so it does.

Some time later, the network comes back online. A and B consolidate their world-state just fine. Annddd... A's army is attacking instead of defending Q, and B's army and then their bases are dying in the meantime, and they lose. X has a 50% chance of winning in this simply by making a random choice between X and P, if communication is disrupted at said critical point. (Or X and Q. Either works.)

Eventual consistency is not enough.
 

(3) Free air laser links is one technology that would at least somewhat obscure the source of the signaling (laser light will probably reflect in a detectable way around corners but it won't go through solid objects) and is capable of tends of gigabits per second of bandwidth, enough to satisfy some of your concerns.

Laser links are wonderful when a) they are through clear air and b) you have a stable alignment between the two endpoints. They fall apart (in the sense of hilariously low channel capacity) in the presence of turbulence / fog / rain / snow / smog / dust / smoke / physical obstructions, or when the endpoints are changing alignment. For many applications this is fine. For rapidly-moving drones in a battlefield environment, where if the enemy knows that they can disrupt you at a critical moment with a smokebomb they absolutely will, not so much. (To an extent you can compensate for some of this by upping the laser power and adding more elaborate beam tracking of various sorts... but a) a small flying craft doesn't exactly have spare energy or mass, and b) you now have enough scattering that the beam is easily detectable.) (Note that it's not just attenuation that's the issue. It's also dispersion.)

(You can kind of get away with unstable but predictable and smooth alignment, e.g. tracking satellites. But the equipment for this is not exactly easily fitted on a small drone, and drone movement in a battlefield environment is not exactly smooth.)

Oh, and laser links are also point-to-point, which increases effective latency compared to a broadcast system (as there's a limited number of transmitters and receivers on any one craft, the current leader cannot directly receive updates from everyone, even if there's the bandwidth available. It has to be bounced/consolidated through relays, which adds latency).

So TLW, at the end of the day, all your objections are in the form of "this method isn't perfect" or "this method will have issues that are fundamental theorems".  

And you're right.  I'm taking the perspective of, having built smaller scale versions of networked control systems, using a slightly lossy interface and an atomic state update mechanism, "we can make this work".  

I guess that's the delta here.  Everything you say as an objection is correct.  It's just not sufficient

At the end of the day, we're talking about a collection of vehicles.  Each is somewhere between the size of a main battle tank and a human sized machine that can open doors and climb stairs.  All likely use fuel cells for power.  All have racks of compute boards, likely arm based SOCs, likely using TPUs or on-die coprocessors.   Hosted on these boards is a software stack.  It is very complex but at a simple level it does :

perception -> state state representation -> potential action set -> H(potential action set) -> max(H) -> actuators.  

That H, how it evaluates a potential action, takes into account (estimates of loss, isActionAllowed, gain_estimate(mission_parameters), gain_estimate(orders)).

It will not take an action if not allowed.  (example, if weapons disabled it will not plan to use them).  It will avoid actions with predicted loss unless the gain is high enough.  (example it won't normally jump out a window but if $HIGH_VALUE_TARGET is escaping around the corner, the machine should and will jump out a window, firing in midair before it is mission killed on impact, when the heuristic is tuned right)

So each machine is fighting on it's own, able to kill enemy fighters on it's own, assassinate VIPs, avoid firing on civilians, unless the reward is high enough [it will fire through civilians if the predicted gain is set high enough.  These machines are of course amoral and human operators setting "accomplish at all costs" for a goal's priority will cause many casualties].

The coordination layer is small in data, except for maybe map updates.  Basically the "commanders" are nodes that run in every machine, they all share software components where the actual functional block is 'stateless' as mentioned.  Just because there is a database with cached state and you send (delta, hash) each frame in no way invalidates this design.  What stateless means is that the "commander" gets (data from last frame, new information) and will make a decision based only on the arguments.  At an OS level this is just a binary running in it's own process space that after each frame, it's own memory is in the same state it started in.  [it wrote the outputs to shared memory, having read the inputs from read only memory]

  This is necessary if you want to have multiple computer redundancy, or software you can even debug.  FYI I actually do this, this part's present day.  

Anyways in situations where the "commander" doesn't work for any of the reasons you mention...this doesn't change a whole lot.  Each machine is now just fighting on it's own or in a smaller group for a while.  They still have their last orders.  

If comm losses are common and you have a much larger network, the form you issue orders in - that limits the autonomy of ever smaller subunits - might be a little interesting. 

I think I have updated a little bit.  From thinking about this problem, I do agree that you need the software stacks to be highly robust to network link losses, breaking into smaller units, momentary rejoins not sufficient to send map updates, and so on.  This would be a lot of effort and would take years of architecture iteration and testing.  There are some amusing bugs you might get, such as one small subunit having seen an enemy fighter sneak by in the past, then when the units resync with each other, fail to report this because the sync algorithm flushes anything not relevant to the immediate present state and objectives.  

FYI re: your footnote- mechanized infantry would have tracked vehicles organically attached to their formation. The term for an equivalent unit with wheeled vehicles would be motorized infantry.