Yes, the rifleman’s day really is over. Errors abound on many levels of analysis and argumentation on this topic; I’ll respond to some here in stream-of-consciousness fashion.

So here’s the first: “Humans will always need to take and hold ground.”

War is not the taking and holding of ground. It is the organized use of violence to impose political will on the enemy. Taking ground is merely one of the means of war, not an end. We did not need to conquer the Japanese mainland to bring them to their knees in WW2. And where war does involve taking ground, robots are well suited to the task.

Already in Ukraine, near-front casualty evacuations and logistics are primarily conducted via drones, ground is sometimes held by drones, and occasionally assaults are conducted entirely by drone. Where the adversarial element is stronger, the progress in automation is slower but nonetheless inexorable. Drones replacing humans on the front is inevitable - at least, insofar as such a front even exists. It’s not a given in war that taking and holding ground is even necessary to begin with.

In fact, throughout history even large ground campaigns often did not involve direct fighting, and where they did, the object was often not to seize a piece of ground (except in relation to where the enemy army then posted itself). See B.H. Liddell Hart’s work, Strategy: The Indirect Approach. The Byzantine general Belisarius often maneuvered in such a way as to merely threaten the use of force, discouraging his enemy from continuing on campaign. Napoleon sought decisive battles, but with his object to shatter the enemy army and national will, not to seize fortresses or cities.

In naval warfare, we no longer need to come face to face to exchange cannon fire, let alone commence ship boarding. And in an air war, pilots do not board each other’s aircraft or need to capture each other’s runways. What happened at sea and in the sky will transpire on the earth as well. It has merely taken longer, since the earth is a denser, slower-moving fluid than air or water, and therefore provides more opportunity for concealment and cover. As ships and aircraft became over-the-horizon assets strike assets, so shall the individual soldier on land.

A geometric analysis yields much the same parallel. As each side’s basic weaponry begins to eclipse the other’s territory, human maneuver becomes irrational. Assume both you and your enemy have 200km of max effective range, and you want to hit something 50km deep into enemy territory - of course you would balk at orders to proceed to within 50km of the “front” yourself if you could stand off from 150km and achieve similar effects. In fact, as you advance, you increase your exposure to enemy teams superlinearly (every time you advance 10km, you add exposure to a widening circular sector of 10 more km across), assuming the front is perpendicular to your line of advance and the enemy has such 200km teams evenly distributed through their depth. When you advance and are spotted, you light up on the enemy’s digital Common Operating Picture (“COP”, like Ukraine’s Delta system), and every bloodthirsty drone team within range will try their hand at hitting you. There is no point at which you break through the enemy’s line and can begin a turning action - you are simply lost deeper in a sea of distributed drone teams the further you go. (As an aside, we may compare this to the experience that the Mongols faced trying to penetrate into central Europe after sweeping through eastern Europe: every hill all the way to the Atlantic was topped with a small but nearly impenetrable castle bristling with crossbows and its contingent of knights ready to sally out and harry the invader, with ready mutual support from neighboring fortresses - only now, imagine there is no need to sally, and every castle within a couple hundred kilometers can also engage in simultaneous attacks with over-the-horizon crossbows) Under such conditions, the “front” as a linear concentration of direct action forces dissolves into a distributed kill web, the so-called “gray zone” that has been gradually forming in Ukraine and will continue to deepen. What happens when the gray zone is an entire nation, an entire combat theatre, or the entire world?

Another change in land warfare that parallels past change in naval warfare is the increasing dispersion at all echelons. This is not merely due to increased destructive payload deliverable to a given area by a given team, which would drive dispersion from a defensive standpoint (standoff, on the other hand, is indeed driven by defensive considerations, as explained in the previous paragraph). Actually, the small drone team does not by itself deliver more destructive energy upon a target than a legacy infantry fireteam or artillery crew. Rather, dispersion is driven by the evaporation of offensive restrictions: the increased ability to coalesce fires at range, organically or on command, together with their already 1-2 orders of magnitude greater range and enabling technology (sensors, COP, AI, delegated authority, satellite comms, etc.) mean that small independent drone teams naturally disperse widely given pre-existing defensive considerations. This historical trend, seen at least since gunpowder overtook cold steel, is now accelerating in runaway fashion. The logical extension is that the human soldiers that make up drone teams will eventually distribute throughout the entire territory available after considerations of a noncombat nature (for example, jurisdictional or geographic) are accounted for. What room for riflemen or tanks is there in such an amorphous mode of warfare? How in the face of all the above considerations can massed infantry or armor even safely reach a collection point to begin transit to an assault against an ephemeral foe?

Returning to the larger debate, what I must say here is that those betting on the rifleman as opposed to the drone are like those who stood in the way of the aircraft carrier in favor of the battleship. The debate has reached its fever pitch, and like in the prior debate, the evident crescendo is itself a leading indicator that the debate is ended. By the time the issue enters public consciousness, it is already decided. Drones and robotics in general are just the physical instantiation of what’s happening in the information space (AI), and those who think either is all hype or a bubble need simply zoom out further on the graph. Once change goes asymptotic, as it already has, an apparent bubble on a long enough timescale is doomed to be lost in the volatility. Those who already possess accurate signal about the issue may confidently leverage information arbitrage in casting their votes, and only if particularly altruistic will bother about the exhausting task of convincing the stubborn old guard. Similarly, by the time the common man realizes a stock is a good buy, he is already too late, and should have bought in years ago. Similarly, the best time to train to be an AI researcher was a decade ago. It is not an accident, but rather the result of a decades-long strategy that the Chinese now have a stranglehold on drone supply chains.

Another pushback: “The ongoing Iran war is a demonstration of the inability of standoff weaponry to achieve victory.”

It is just the opposite. Even Iran’s limited drone scaling (as compared to Ukraine, Russia, or China) has proven to make them enough of a porcupine that we have had to take a step back. From their perspective, they are waging a very effective campaign with what they have, notwithstanding their foolishness in not preparing more (they could have had hundreds of thousands of long-range drones ready, and could have staged some of these to attack data centers and the power grid in the USA). Iran’s limited war goals would have been even better met if they had dissolved their entire legacy military years ago, and focused entirely on standoff weaponry (ballistics and drones). Even in the event of a US invasion and occupation by outdated grunts, short-range FPV drones could easily inflict enough damage to evaporate US public and political support for military action. We are seeing how even the weakened and dispersed Islamist insurgents and militias are able to inflict heavy losses on Israeli conventional forces with fiber optic FPV drones. With Iran’s geography and large population, if they prepared they could hit occupying forces with tens or even hundreds of thousands of FPV drones daily. And if one believes they would lose anyway, as they probably would, they certainly would not perform better by attempting to fight the US with rifles and tanks.

If we decided that we did need to invade, the most effective means would be with a drone-equipped force. Soldiers may still need to carry firearms to protect themselves when they voluntarily move in proximity to masses of people, as when occupying large cities without displacing or slaughtering the populace. But to respond in kind to hostile military or insurgent elements hitting them from tens of kilometers away, the firearms would be useless. Instead, small and flexible formations with organic sensing (radar, acoustics) would need to rapidly respond in kind by finding, fixing, and finishing the enemy with unmanned systems. They would need their own organic interceptor drones as well, to mitigate damage from the enemy. Eventually, all this will take place beyond the human ability to process in real time, requiring every soldier or at least small team to have organic automated counter-drone defenses, like how the Trophy system works for a tank against incoming projectiles - but this is a topic for another essay.

Another objection: “There’s a counter to every new military technology.”

On some level, sure, but the best-known counters to drones are more of the same, both to intercept the enemy’s drones and to attack him directly. Was there some counter to gunpowder weaponry that allowed the knight to continue his reign over the medieval battlefield? No, the knights are all gone, never to return. Their horses, swords, lances, and armor are all gone. At best “cavalry” continues as a loose analogy, for example in reference to mechanized/armored forces or “air cavalry”. The “infantryman” is likewise doomed to be relegated to history, not as the fundamental unit of combat power carrying his equipment on his back and engaging the enemy in direct fires, but continuing only as an analogy or an anachronism.

The most commonly proposed counters to drones typically involve some form of directed energy. Well, why isn’t such weaponry being used widely? The only forms of directed energy being actually commonly employed on the modern battlefield are simple jamming and spoofing. But their effectiveness (which even at peak was never enough to arrest the development and scaling of robotic warfare) is declining with the advent of drones controlled by fiber optic, satellite, or fully autonomous means. The last of those is a permanent escape hatch against jamming now that Pandora’s box has been opened. As for electronic attack meant to cause physical damage to onboard electronics of the drones, this is rarely ever known to be used, since the discouraging economics related to it are so easy to see playing out (expensive to implement across one’s force, but easy to counter with cheap shielding). As for lasers, they are being employed to a limited extent and do offer a decent solution to defending against one or two drones at a time, but also face economic limitations, and are quite helpless against a drone “swarm” attacking in large numbers from multiple angles simultaneously, which must all be engaged in a very limited range and timespan.

What about some wonder weapon at some secret US lab? We haven’t seen it yet, and I doubt that, if it’s out there, it’s cheap and small enough to equip our entire nation’s military with it in the next couple years. Or perhaps it’s some wonder-weapon of an entirely different category - maybe the NSA or some other agency can push a button and shut off China’s power grid or internet overnight. If such capability exists, it’s an argument against us having a conventional military ready for peer-level warfare at all. We would need neither drones nor rifles if we’re really that far ahead. Of course, AI is regarded on some level as such a cure-all weapon, but it’s not there yet. We’d better maintain a military until some hypothetical future where AI alone is the only weapon a human needs.

Another: “If drones are so decisive, why is Ukraine a stalemate?”

As in the case of the Iran war, this makes the opposite point the skeptic expects. Ukraine should not have been able to survive, much less perhaps begin to even win. Human maneuver on both sides has largely ended, and drone technology and doctrine are still unfolding. The next step is to resume maneuver via drones, as various units (Khartia, Azov, etc.) are experimenting with. The rifleman is dead, so what replaces him? Whoever first understands and scales massed combined intra-drone arms and drone maneuver warfare may actually break the tactical and operational deadlock, if the strategic drone campaign doesn’t end the war first.

A final objection: “The experience of war in Ukraine does not generalize. Combined arms maneuver warfare reigns supreme; therefore drones are just one more tool in that combined force.”

Yes, combined arms maneuver, where possible, should produce cleaner victories than the largely attritional warfare seen in Ukraine today. However, the combining of arms and the maneuvering of those arms can be done entirely by drone. Drones can fill every niche and every domain - reconnaissance, armor, standoff fires, direct fires, sea/air/land/subsurface/space, etc. Why on earth would you expose an infantry company or special operations teams by sending them into enemy territory to conduct an ambush when you could have cheap drones do the same? If a drone can fly hundreds of kilometers, what do you gain by pushing the humans launching them forward by ten or twenty kilometers? Drones are already laying minefields, conducting reconnaissance, sitting on roads in low power mode waiting to ambush, and executing logistical and industrial sabotage at a scale humans could never have hoped to accomplish.

Currently, a Ukrainian “mid-range” drone battalion is responsible for achieving effects at a similar depth, about 30-300km, to that reserved for an army corps (two orders of magnitude more personnel) in legacy US doctrine. But unlike the soldiers controlling HIMARS and other centrally directed systems (often now nearly useless due to ubiquitous GPS denial) that the corps used for such effects, the small Ukrainian drone teams that comprise the mid-range battalion are capable of almost fully independent, decentralized operation.

Small Ukrainian teams consistently stop vastly superior NATO naval and land forces dead in their tracks in exercises, and that’s just today; there’s hardly anything left you can do to further refine a rifle or an artillery shell, but robotics and autonomy are just getting started. Yes, the Ukrainians and Russians are deadlocked in attritional warfare, but an opponent with time to plan and scale in peace would not deliberately leverage robotics to enter such a terrible state of affairs. That is, there is nothing fundamental about drones that leads to an attrition warfare bias.

Limitations like bad weather or different terrain are also at worst speed bumps to drone warfare. In the long run, where they haven’t already, machine sensors will of course exceed human senses, not merely in their fidelity but in their resilience to the elements as well. Machine locomotion will handle rough terrain better than any human. Small machines will squeeze in tighter spaces. Machines that do not sleep will dig out and hold fighting positions at lower energetic cost. Those who point to temporary limitations are like those who point to some current deficiency in AI, such as hallucination or need for cooling and powering large amounts of compute - how many people are even aware that AI is already improving itself? Aware that NVIDIA’s latest and greatest chip was designed by AI? How many actual practitioners, who can now do what they never could have hoped to before, still downplay drones or AI? How many naysayers have looked at actual data and drawn out the trendlines? In 2023 Ukraine conducted roughly a hundred long-range strikes a month; two years later it conducts over seven thousand, at ranges out to two thousand kilometers - a seventy-fold scaling, in wartime. Across Ukrainian, Russian, and Western media there is a new fixation with the potential downfall of Russia almost entirely due to the scaling of these effective long-range strikes against industry.

As mentioned in my previous article, the scale this year across all domains of drone is absolutely staggering: 20-40 million drones employed in total this year by Ukraine and Russia, and China likely producing the equivalent of 100 million drones in components and actual finished units. Each of those nations’ militaries is integrating drones all the way down to the fireteam level, replacing swathes of their military structure as quickly as they can. And of course, direction from senior officials of the Department of War is crystal clear that we need every combat soldier to have access to drones, although they haven’t comprehended yet what kind of scale and requirements (range in particular) that notion implies, or how it will revolutionize doctrine rather than change it incrementally.