7745 stories
·
34 followers

The Dawn of Automated Warfare: Artificial Intelligence Will Be the Key to Victory in Ukraine—and Elsewhere

2 Shares

When Russia first launched its full-scale invasion of Ukraine, the conflict drew comparisons to wars of the twentieth century. Tanks, armored personnel carriers, and artillery dominated the battlefield, and both sides’ infantry were dug into trenches. We witnessed this old-school style of war when we made our first visit to Ukraine in September 2022. Since then, we have made regular trips to Ukraine, affording us firsthand insight into a monumental transformation: the beginning of a new kind of warfare.

In summer 2023, the commander of Ukraine’s Third Assault Brigade’s Drone Unit, whom we’ll call Fil (not his real name), told us that a new weapon had begun to change the conflict: first-person-view drones. These small, cheap, maneuverable quadcopters transmit real-time footage to their operators and detonate kamikaze-style on their targets. That year, Ukraine flooded the field with thousands of them and Russia soon followed suit. Today, hundreds of thousands of these drones fill the Ukrainian skies.

What began as a war with drones has become a war of drones. Indeed, two years ago, a Ukrainian brigade’s strength was judged mostly by its inventory of Western-supplied tanks, armored personnel carriers, and artillery. Since 2023, however, drones have become the most important weapon on the battlefield. Because of their low cost, speed, and precision, drones have now largely supplanted traditional weaponry, including antitank missiles, mortars, tanks, and even artillery and aircraft. Today, a unit’s power and resilience are dictated by its number of skilled drone operators and its ability to deploy drones at scale. (One of us, Schmidt, has been a longtime investor in defense technology companies, and is currently an investor in companies supplying drones to Ukraine.)

This represents a profound shift in warfare, largely instigated by Ukraine to compensate for its shortfalls in conventional weapons and manpower. In the world’s first drone war, drones determine how battles are won and how soldiers die: Ukrainian drone strikes now account for 90 percent of destroyed Russian tanks and armored vehicles and 80 percent of Russian casualties. They have also made it possible for each side to attack far past the frontlines without having to gain air superiority over the battlefield. Ukraine, for example, hit Russian airbases 5,000 miles from Kyiv in June by smuggling drones across the border and launching them from the beds of trucks.

Russia, for its part, was originally slower to field drones in large numbers. But it has dramatically increased its production of first-person-view drones, as well as those used for strategic bombardment, such as the Iranian-designed Shahed. Today, Moscow matches Kyiv’s extraordinary rate of technological adaptation. It has developed equally capable models, such as the Orlan, which is used for surveillance, and the Lancet, which loiters over a target before exploding on impact.

Because Russia and Ukraine are constantly iterating on hardware, software, and tactics, the war changes at a breathtaking rate. The saturation of drone surveillance, for example, has made nearly all troop movement visible and therefore vulnerable, creating a transparent battlefield: anything that moves near the frontline is struck within a matter of minutes. Drone pilots have become prime targets, and with many traditional weapons rendered obsolete, drones are increasingly fighting other drones. Amid this cycle of innovation, the two sides are inching toward a new frontier: entirely automated warfare.

EYES EVERYWHERE

Surveillance and reconnaissance drones have become so ubiquitous that both Russian and Ukrainian forces scarcely move in the daylight. During a recent visit, we witnessed the motion of a single Russian van, five miles from the frontline, cause a sensation among drone operators, who then destroyed it. To avoid detection, movement near the frontline tends to take place during sunrise and sunset, when neither the daylight video cameras nor night-vision infrared cameras operate properly.

The fight for information advantage is always important in war, but even more so in this one, where it means the ability to form and maintain resilient drone-based sensor networks over the battlefield. If a unit is “blinded”—unable to maintain surveillance drones overhead—it becomes exceedingly vulnerable. For that reason, roughly 3,000 Ukrainian troops work around the clock to operate reconnaissance drones, mostly Chinese-made DJI Mavics, along the entire 750-mile frontline. Ukrainian brigade command centers display as many as 60 of these drone feeds around the clock.

This transparency means that the military maxim “what can be seen can be hit” is truer on today’s battlefield than at any point in history. It is nearly impossible for either side to mass and maneuver forces along the frontline, as troops are now easily spotted forming up for attack. The Russian army has historically relied on its ability to deliver impressive firepower through concentrated tube and rocket artillery fire, but these tactics are useless when any attempt to amass forces is identified within minutes. Russian guns are now widely dispersed, deeply dug in, and operate primarily at night.

Both Russian and Ukrainian forces scarcely move in the daylight.

Fil says his team has expanded from 400 troops in 2023 to over 1,000 now, and he expects it to continue to grow in the coming months. The brigade’s frontline operations are driven by data, of which there is more every day. Fil’s brigade, for example, tracks every engagement, drone mission, and vehicle or piece of equipment hit. That data, in turn, drives decision-making, including over the kinds and quantities of drones to procure. Fil’s team spends nearly $2 million each month on small quadcopters, mostly Mavics, for frontline reconnaissance and more than $500,000 per month for longer-range fixed-wing surveillance drones, such as Sharks or Lelekas, that can be used to see much farther from the frontline.

This is expensive, but one new battle tank costs more than $10 million. A tank was long regarded as the best weapon to defeat another tank; now, a first-person-view drone costing less than $800 is, thanks to its ability to strike with precision and move much faster than any ground vehicle. No armored vehicle—no matter its camouflage or anti-drone barriers—can survive for long on the modern, drone-swept battlefield. As a result, Ukrainian soldiers believe tank-led assaults to be suicidal. Russia still launches them occasionally, but most do not make it to the front line.

In response, Russia has shifted primarily to infantry assaults. It is no surprise, then, that more than 75 percent of Ukrainian drones now target infantry. Because surveillance drones have a difficult time spotting scattered infantry in urban terrain and forests, the Russians are deploying small assault groups, typically consisting of five or six teams of two to three people, to simultaneously attack a concentrated area. In recent months, the Russians have turned to motorcycles to more rapidly cross no man’s land—a shocking contrast to the line of tanks that rolled toward Kyiv in the earliest days of the conflict. When surviving members of a Russian assault group reach a building, they immediately dig in. Gradually, more soldiers join them. Over days or even weeks, the Russians gather their forces until they judge that they have sufficient strength to make the next bound toward Ukrainian positions. According to an officer from Ukraine’s Azov Brigade, Russia’s “cheap infantry”—its disregard for soldiers’ survival—allows for this kind of constant experimentation.

DRONE ON DRONE

Because drones have become so important to almost all battlefield operations, destroying them has become critical. Drone-on-drone battle is now a central part of the war. Last year, an estimated 1,200 Russian surveillance drones were operating behind Ukrainian lines on a given day, so Ukraine built the first drone-based air defense system to fend them off. Its forces began using first-person-view drones to chase down larger, slower, and much more expensive surveillance drones. Russian surveillance drones now fly higher to avoid Ukrainian interception. Still, roughly 80 percent of all surveillance drones that cross the frontline, be they Russian or Ukrainian, are shot down by either interceptors or traditional air defenses. Because of these changes, Russia has dramatically reduced its use of Lancet drones. Instead, it has developed smaller, faster, and camouflaged surveillance drones, including some with rear-facing cameras, that let operators spot and evade pursuing drones. 

Unsurprisingly, drone pilots and their control stations have become prime targets for both sides. Fil’s unit has found that a successful attack on Russian Mavic operators can pause enemy activity for three days. Because pilots have become such a precious resource, integral to defending the infantry, Ukraine is working to relocate as many of them as possible away from the frontlines to integrated remote operations. In an attempt to further decrease the number of forward-deployed soldiers, Ukraine is now working to establish a so-called drone line along its entire frontline—a layered defense corridor six or seven miles wide, made up of obstacles such as ditches, minefields, and razor wire, and hundreds of drone teams that wait at the ready to destroy any targets before they reach Ukrainian positions. Once this barrier is in place, and more drone functions are automated, far fewer troops will need to defend the frontline. Ukraine hopes that this approach will help alleviate its manpower shortfalls and save lives.

Behind the frontlines, tactics are evolving just as fast. On a nightly basis, Russia launches hundreds of Shahed long-range drones at Ukraine, particularly its major cities. Russia increasingly uses sequenced launches and circuitous route planning so that multiple drones arrive at their target simultaneously from different directions; these attacks amount to manually coordinated “drone swarms.” Short of air defense systems, Ukraine has prioritized the development and production of interceptor drones to counter these swarms. Russia also sends dozens of cheaper “dummy” drones—drones without real capabilities—into Ukrainian airspace, forcing air defense radars to reveal their locations. Russian ballistic and cruise missiles then route around the defenses to strike their targets.

ITERATION OR OBLITERATION

The speed of technological adaptation and iteration—or innovation power—is a new measure of combat strength. The key to adaptation is the lightning-fast feedback loop from operator to engineer. The best Ukrainian drone pilots, therefore, are both tactician and technician, able to make modifications and improvements on the fly.

Consequently, the most important progress in drone development is happening at the front. Operators are supported by research and development labs and manufacturing and repair facilities located near the frontlines. Drone teams constantly test and deploy new radios, antennas, and circuit boards; software updates are pushed out on a near-daily basis. To create an effective weapon now requires adapting and iterating against an equally adaptive adversary, resulting in a highly dynamic contest of action and reaction.

Once a new weapon or technique is introduced to the battlefield, it has a limited window of utility before the opponent develops countermeasures. New kinds of drones appear at a rapid rate: two years ago, the Russian Lancet was the most threatening model. Last year, it was the first-person-view drone. Now, strike drones controlled by fiber-optic cables, first fielded by the Russians, have taken hold of the frontline.

Unlike drones that run on standard radio frequency, these quadcopters spool up to 25 miles of fiber-optic cable in their wake, leaving them hard-wired to their operator. Although these drones are slow and limited by the length of their wire, they are impervious to jamming, relay clear images, and can operate outside radio line of sight, which means that they are well suited for hilly and urban terrain. Since they do not emit radio signals, their pilot’s location cannot be identified by electronic means, and they strike with shocking precision.

Fiber-optic drones are effective ambush weapons. Russians fly them across the frontline and park these models on roads or rooftops and wait for passing vehicles. Their high-quality control signal and camera resolution allows them to be maneuvered with pinpoint accuracy into tight areas, such as buildings and bunkers, that normal first-person-view drones, which rely on radio, cannot access. Russia now has elite units of drone pilots using fiber-optic drones stationed along heavily contested parts of the frontline in order to target Ukrainian drone operators, attack enemy supply lines, and ultimately isolate forward units.

Drone innovation is not just about making drones better but also about driving down their cost. Over the course of the past year, both Ukrainian and Russian drone units have replaced the pricey few with the inexpensive plenty. Expensive drones, including the Russian Lancet and the American Switchblade 600, which each cost between $65,000 and $150,000, are being pushed aside in favor of fixed-wing strike drones, such as the Russian Molniya and the Ukrainian Dart, both of which cost less than $3,000. Because Molniya kamikaze drones are so cheap, Russia uses them as a mass strike weapon, sometimes launching 15 at a single target.

For the most part, Ukraine still uses first-person-view drones because they are cheap, relatively easy to use, and readily available. Brigades on the most active fronts consume more than 5,000 of them per month. But because their rate of success in striking a target is low, estimated at no more than ten percent for the average unit, many frontline units favor larger bomber-type drones for their versatility, reusability, and modular configurations. A single Ukrainian-made Vampire hexacopter drone, for example, can drop antitank mines or rain down munitions on enemy infantry, achieving the same effect as dozens of artillery rounds, and with greater precision. And because they are both reusable and have a larger payload than first-person-view drones, bomber drones can saturate the ground with high explosives much more rapidly and at much lower cost. They are also better at targeting infantry when repelling assaults and can collapse buildings with a few accurately placed explosives that would otherwise have taken hundreds of artillery rounds.

Bomber drones can also place mines, a tactic that is quickly becoming one of the most effective ways to halt Russian assaults. Russian units on the attack must use largely predictable routes because of terrain, so Ukrainian forces create a dynamic, mobile minefield by dropping mines in their path. Ukrainian forces then use first-person-view drones to herd Russian vehicles toward the mines. One Ukrainian brigade estimates that 50 percent of enemy vehicle kills in recent months have resulted from drone mining. Ukrainians also use bomber drones to run continuous waves of attacks, similar to artillery bombardments, to keep Russian infantry suppressed, underground, and unable to advance.

Ukraine still employs its legacy systems, such as artillery, to support its drone tactics. When Russian infantry are protecting a valuable target, for instance, Ukrainian troops use artillery to suppress the infantry so that Ukrainian bomber drones can destroy the target without being shot down. Ukrainian troops will also use surveillance drones to ascertain where Russian drone pilots are before shelling those locations. These approaches allow Ukraine to minimize its use of expensive legacy systems that can be difficult to acquire; Russia, by contrast, has less need to adapt its use of these systems because it can afford to expend shells in quantities that Ukraine cannot.

SWARM AND ATTACK

Automating drones with artificial intelligence would solve a variety of problems facing the modern warfighter. A large number of drones are lost to pilot error. And the Ukrainian battlefield is saturated with systems that jam and spoof signals across the electromagnetic spectrum, making it difficult to rely on any technology that requires constant radio connection to a human operator. Thousands of Ukrainian troops operate Mavics all day, every day, a function that could certainly be automated. Collecting and processing surveillance data automatically—ideally from multiple layers of sensors across the frontline—would save hundreds of man-hours a week. And current systems require drone pilots to operate close to the frontline, putting them at risk.

Today, algorithms can augment human control of the battlefield. They reduce error by helping pilots detect, track, approach, and strike targets. AI targeting systems are trained nightly on combat footage to adjust to Russian countermeasures, such as camouflage or decoys. Ukrainian and Western companies are creating software that supports drone pilots even more, by selecting routes, stabilizing flights, navigating to waypoints, recognizing targets, and guiding toward the destination. If these efforts are successful, becoming a drone pilot will require fewer skills and less experience.

In particular, defense firms are keen to develop AI tools that make it easier to carry out the final phase of an attack. The Ukrainian battlefield is a challenge for machine learning because enemy tanks and artillery pieces constantly change appearance with added armor and camouflage. Algorithms also perform poorly at identifying scattered infantry, particularly in dense forests or other complex terrain. AI-assisted target acquisition and terminal guidance have already proved effective even in the face of radio signal jamming. Although the future of fully autonomous drones is unclear, a more autonomous drone strike complex—one that combines reconnaissance and strike drones to identify, track, and hit moving targets—would enormously improve Ukraine’s ability to defend against Russian attacks.

The next phase of the war will be determined by software.

Defense companies are also racing to create AI that can coordinate attacks by multiple drones in an automated drone swarm—the holy grail of drone operations. Today, Ukrainian forces can form a carousel of drones over a target to repeatedly strike at it, but doing so requires multiple pilots and operators. With an automated drone swarm, a single pilot could guide many drones, flying independent routes, to overwhelm defenses and saturate a target.

To pull off such a feat, defense firms will need to develop AI-powered systems that enable drones to communicate automatically—not just with one another but also with a host of sensors. These networks exist, but not at the required scale. And the task is getting harder each day: as the drone-versus-drone war escalates, the quantity of drones deployed in each operation will grow from hundreds to thousands, making their automated coordination increasingly difficult.

Eventually, Ukraine will need its own version of Israel’s Iron Dome air defense network to protect its cities and factories from Russia’s constant drone and missile raids. Of course, Ukraine’s vast size makes this a daunting challenge, but it can begin by shielding its major cities. Greater automation will be key to fending off Russian attacks. Whereas the first phase of the war was defined by hardware, with each side competing to invent new kinds of drones, payload, and munitions, the next phase of the war will be determined by software.

WAR OF FACTORIES

Drones have upended the old ways of war. Military doctrine, tactics, and organization will never be the same. Armies everywhere will need to completely revamp their doctrine and training to reflect the realities of fighting on a drone-swept battlefield. And the best way to prepare for the future of combat is to speak to those fighting this war.

Historians often call World War II a “war of factories.” The same is true for the war in Ukraine today. Ukraine produced more than two million drones in 2024 and plans to make over four million by the end of 2025. Its adversary is also getting better at drone production: last year, Russia was building 300 Shahed drones a month. Now, it can produce 5,000 in the same time frame. The side that consistently builds the most drones is the one most likely to prevail. And it is in the interest of the West, and of the United States in particular, to support the Ukrainian people in their dogged determination to win that fight—not only for Ukraine’s sake but also for its own, so it can learn to reckon with this new reality of war.

Read the whole story
cherjr
1 day ago
reply
48.840867,2.324885
bogorad
6 days ago
reply
Barcelona, Catalonia, Spain
Share this story
Delete

How Humanitarian Aid Feeds War Machines - WSJ

2 Shares
  • Who/What/When/Where/Why: Ms. Barak‑Corren, a Hebrew University law professor, argues in a 2023–2024 opinion piece that international donors should reform humanitarian aid rules in conflicts (Gaza, Somalia, Syria, Ethiopia, Yemen and others) to stop diversion to armed groups and align aid with humanitarian intent.
  • Somalia example: WFP transport contracts are dominated by three clan cartels that reportedly skim 30–50% of cargo, with an estimated one‑eighth of donated food reaching intended households.
  • Syria and Ethiopia examples: In Syria the Assad regime allegedly forced aid conversion at below‑market rates (≈$60m pocketed in 2020); in Ethiopia U.S. USAID found large‑scale theft by the military involving U.N. partners and private mills.
  • Gaza and UNRWA: UNRWA’s large budget and local presence in Gaza effectively provide public services, enabling Hamas to tax aid, place loyalists on payrolls, and divert resources toward tunnels and rockets.
  • Why diversion persists: Moral urgency to save lives, institutional survival of the humanitarian sector, adversaries’ adaptability (fake NGOs, collusion), and fear of donor backlash produce weak accountability.
  • Evidence conditionality can work: A U.S. pause on food aid to Ethiopia in 2023 coincided with reduced diversion and the government accepting QR‑coded tracking.
  • Recommended donor conditions: Enforce nondiversion benchmarks; integrate security (donor‑vetted guards/UN escorts); fund external audits and whistle‑blower legal defense; impose 10‑year sunset clauses; fund fintech and QR tracking innovations.
  • Feasibility and risks of inaction: The U.S., EU and Gulf states provide >70% of humanitarian funding and can impose these measures without changing international law; without reform, aid may bolster autocrats, increase corruption, fuel migration and squander growing budgets.

You may also like

Up Next

CheckboxEmbed code copied to clipboard

[Share on FacebookFacebook](https://www.facebook.com/sharer/sharer.php?u=https://www.wsj.com/video/series/journal-editorial-report/wsj-opinion-the-gaza-aid-crisis-and-the-backlash-against-israel/52F7F4A2-42AC-443F-A09B-02E471B6D146&t=WSJ Opinion: The Gaza Aid Crisis and the Backlash Against Israel "Share on Facebook")

[Share on TwitterTwitter](https://twitter.com/intent/tweet?url=https://www.wsj.com/video/series/journal-editorial-report/wsj-opinion-the-gaza-aid-crisis-and-the-backlash-against-israel/52F7F4A2-42AC-443F-A09B-02E471B6D146&text=WSJ Opinion: The Gaza Aid Crisis and the Backlash Against Israel "Share on Twitter")

Your browser does not support HTML5 video.

0:02

Paused

0:01 / 6:08

Journal Editorial Report: The U.K., France and Canada move to recognize Palestine.

The pictures are heartbreaking: convoys of United Nations-marked trucks inching toward bomb-scarred cities, desperate children clamoring for supplies. These images seem to prove that the international system is, at the very least, trying to help. Yet in every conflict I have studied—Somalia, Afghanistan, Iraq, Syria, Yemen, Sudan, Ethiopia and Gaza—the same trucks double as cash machines for warlords, militias and authoritarian regimes. Aid diversion is a widespread problem in humanitarian operations. Unless the U.S. and other donors rewrite the rules so that aid can’t be separated from accountability, they will keep subsidizing the conflicts they abhor.

Somalia shows how thoroughly diversion can be built into routine. Three clan cartels win most World Food Program transport contracts, skim 30% to 50% of the cargo, and then split the spoils with those who transport the food and those who control the displacement camps. Based on U.N. reports and monitors, my coauthor and I estimate in our study that barely one-eighth of donated food reaches intended households.

In Syria, former President Bashar al-Assad insisted that all aid be converted at a government-set rate to roughly half the market price, enabling the regime to pocket at least $60 million in 2020 alone, while blocking aid from reaching the opposition by designating it as “unsafe.”

In Ethiopia, U.S. Agency for International Development workers discovered that their implementing partner, the U.N.’s World Food Program, was aware of industrial-scale theft by the Ethiopian military but failed to stop it. Large quantities of wheat were diverted to private mills to make flour for the army.

Gaza presents the longest-running case of diverted aid. The U.N. Relief and Works Agency, with a $2.48 billion annual budget in 2024 and 13,000 local staff, effectively provides most public services in Gaza. That frees Hamas to use its resources for tunnels and rockets. In regions under its control, it taxes incoming aid and places loyalists on the payroll. A donor freeze after evidence that Unrwa personnel joined the Oct. 7, 2023, attack lasted only months; services resumed with no vetting reforms because, donors said, “there is no alternative.”

Diversion persists for several reasons. First, the moral considerations are difficult: When officials weigh “people will die tomorrow” against “fighters will grow stronger next month,” tomorrow always wins. Second, people give priority to institutional survival: The humanitarian sector employs some 570,000 people and until 2025 spent about $35 billion a year, and agencies that refuse armed groups’ terms lose access, budgets and jobs. Third, adversaries are able to adapt: Warlords remain in place while aid workers rotate; they open businesses posing as nongovernmental organizations and bill the U.N. for delivering cargo. Add fear of donor backlash and the result is a continuing lack of accountability. Those who break the rules hold more leverage than those sworn to enforce them.

But accountability is still within reach if donors step up and put pressure on humanitarian organizations. When donors finally cut funds because they find the scale of diversions unacceptable, even hard-line actors bend. When Washington paused food aid to Ethiopia in 2023, diversion plunged, and the government grudgingly accepted QR-coded tracking.

The U.S., the European Union and Gulf states combined supply more than 70% of global humanitarian budgets. Together, they could enforce five conditions on every grantee:

Require nondiversion benchmarks. All access fees, escorts and local taxes must be disclosed in advance; one missed benchmark triggers a 12-month funding pause.

Integrate security. Grantees may hire donor-vetted guards or accept U.N. peacekeeper escorts; private deals with militias void the grant.

Build in insurance for whistle-blowers. Two percent of every grant pays for external audits and the legal defense of whistle-blowers.

Create sunset clauses. Missions longer than 10 years shut down unless donors unanimously extend them after a public review.

Fund innovation. Create dedicated funding streams for fintech tracking, QR-coded commodities and other tools that make diversion more difficult.

None of this requires rewriting international law. It’s simply about enabling donors to use the one lever they control: money. Every diverted dollar undermines U.S. counterterrorism goals, fuels migration pressure on Europe, and forces Gulf monarchies to spend twice—first on aid, then on arms. Conditional funding flips those incentives. Regimes that want to look legitimate on the global stage must choose: guns or food, not both. Boosting accountability would also blunt the common critique that Washington preaches good governance while financing bad actors through humanitarian back doors.

Critics say pausing aid is immoral, as civilians would starve. This risk is real but not as definite as how it is often presented. Typically, there isn’t enough data to infer the effect that a pause will have. The June 2023 Ethiopia pause triggered dire warnings of widespread famine. While some people no doubt suffered, the warnings didn’t materialize. Acute food insecurity and child malnutrition went down in 2023 compared with 2022. In Yemen, food-security analyses during the suspension were limited to southern Yemen, where the pause in aid didn’t occur. Acute food insecurity did rise in these areas, which were unaffected by the halt but where multiple other forces were at work: currency collapse, wage arrears, siege tactics.

Against the immediate risk of human suffering, donors must consider that allowing aid diversion could extend that risk for years. A cross-country analysis of 621 leaders in 123 countries from 1960 to 1999 showed that large, unconditional aid inflows help autocrats survive. World Bank data show that unconditional aid correlates with higher corruption and weaker rule of law.

Nothing obligates donors to bankroll the fighters causing the suffering. Setting conditions on aid to prevent diversion aligns humanitarian spending with humanitarian intent.

Climate shocks, urban sieges and pricier grain will likely push humanitarian budgets beyond $50 billion within a decade. Without reform, much of that money will feed armies before children. Worse, each scandal makes future interventions politically toxic, putting civilians in real need at risk of not receiving help. The choice is stark: tighten the taps now or watch the well run dry.

Ms. Barak-Corren is a law professor and the Haim H. Cohn Chair in Human Rights at Hebrew University of Jerusalem.

image

Volunteers arrange parcels as trucks loaded with humanitarian aid await entry to the Gaza Strip, Aug. 6. Photo: str/Agence France-Presse/Getty Images

Read the whole story
cherjr
2 days ago
reply
48.840867,2.324885
bogorad
7 days ago
reply
Barcelona, Catalonia, Spain
Share this story
Delete

Табу и мусор: из жизни полевого социолога

1 Share

Табу и мусор: из жизни полевого социолога

Есть у меня подруга К., она по жизни - полевой социолог, и изучает приятные оппозиционные настроения самый хардкод - полицию, суды, адвокатов. Без шуток очень опасная профессия: недавно за К. пришли объекты изучения.

Я решила записать тут две ее истории истории - так, как слышала от рассказчицы, а прочитать их можно вот тут и вот тут)

-1-.

Как вы понимаете, социологи в полиции для полицейских - это как приезд ревизора. Абсолютно непонятно, чем человек занимается, но явно ничем хорошим для тебя (полицейского) не кончится.

И вот К. проводит наблюдение работы полиции в отделении одного небольшого города. Но в голову К. как назло лезут мысли, что К. обещала приехать к нам на Летнюю школу по антропологии и фольклористике с докладом про поверья и магические запреты внутри полиции, а доклад не очень-то готов... И вот, когда в интервью возникает неловкая пауза, К. выпаливает в лицо дежурному полицейскому:

- А у вас табу какие-нибудь есть?

Полицейский меняется в лице, бросается к папкам, начинает в них рыться с криками:

- Т.А.Б.У... Территориальное....что-то...Управление?!! Было же! где же оно? Поверьте, есть у нас ТАБУ, есть, только вот найти не могу!

НАСТОЯЩИЙ МАТЕРИАЛ (ИНФОРМАЦИЯ) ПРОИЗВЕДЕН, РАСПРОСТРАНЕН И (ИЛИ) НАПРАВЛЕН ИНОСТРАННЫМ АГЕНТОМ АЛЕКСАНДРОЙ СЕРГЕЕВНОЙ АРХИПОВОЙ, СОДЕРЖАЩЕЙСЯ В РЕЕСТРЕ ИНОСТРАННЫХ АГЕНТОВ ЛИБО КАСАЕТСЯ ДЕЯТЕЛЬНОСТИ ИНОСТРАННОГО АГЕНТА АЛЕКСАНДРЫ СЕРГЕЕВНЫ АРХИПОВОЙ, СОДЕРЖАЩЕЙСЯ В РЕЕСТРЕ ИНОСТРАННЫХ АГЕНТОВ 18+

-2-.

Помните ли вы, дорогие читатели, приложение "Социальный мониторинг", которое в 2020 и 2021 изрядно испортило жизнь москвичам? Если ты заболевал ковидом, ты должен был сидеть дома, а для этого устанавливать приложение "Социальный мониторинг", которое следило за каждым твоим шагом и требовало от тебя посылать фотографию из дома в три часа ночи, например. Более того, многие "арестанты" жаловались, что приложение видит их на улице, в то время как ты пьешь чай дома на кухне. Из-за этого капали штрафы.

Так вот, в результате исследования К. с коллегами выяснили, как это приложение появилось. Вдумайтесь: пандемия, надо делать что-то быстро, Собянин был явно сторонником китайского стиля борьбы с ковидом, а на новое приложение нужно тендер проводить и вот это все. При этом нужно было детектировать а) геолокацию, б) передвижение заболевших.

И тут выясняется, что у Департамента информационных технологий уже было в хозяйстве нужное приложение - для отслеживания мусоровозов! Оно позволяло следить за их передвижением (чтобы не простаивали и не тратили вхолостую бензин) и за тем, чтобы они не скапливались в одном месте. Его и переделали в Социальный мониторинг для заболевших ковидом.

Так что для власти мы все немножечко мусор.


У К. теперь есть свой ТГ-канал @pravo_a_la_russe, в котором она делится деталями исследований силовых структур. Если вам это интересно, вам туда.

Read the whole story
cherjr
2 days ago
reply
48.840867,2.324885
Share this story
Delete

Never Work Alone, Even in the Age of AI | Psychology Today United Kingdom

2 Shares

There’s a growing belief that with all the AI agents and tools available today, the first one-person startup unicorn will be born in the next decade. According to experts, it’s only a matter of time before someone builds a company worth over a billion dollars without hiring a single employee—just the founder running the entire operation.

The question is whether—with enough automation—one person could handle everything needed to build a sizable business: coming up with a product idea, building it end-to-end, selling it, supporting customers, and more. But there’s another, similarly important question within the first one: Would anyone actually want to do all of that work alone? And would they stay sane if they tried?

We collaborate not because we’re lazy or inefficient. Working with others tends to bring out the best—and sometimes the worst—in us. Real debates spark the most creative ideas, and team spirit helps us dip into extra energy reserves when the going gets tough. Maybe AI will eventually replace some of these elements as well, so it’s worth asking the question: What do we really gain from seeking help from partners, co-founders, co-authors, and other collaborators?

“My brain is open”

Paul Erdős, one of the most prolific mathematicians in history, showed us just how powerful collaboration can be. Although he was a brilliant thinker, Erdős chose not to work alone. He gave up having a permanent home and instead traveled from one colleague’s home to another, greeting each with his famous line: “My brain is open.” He was always ready to tackle new problems with a new group, and by the time he died in 1996, he had co-authored around 1,500 papers, more than any other mathematician on record.

Erdős’s network spans over 500 different co-authors and links over 130,000 mathematicians. His productivity highlights the key strengths of collaboration: the sharing of ideas, the combination of different skills, and the creative sparks that fly when minds meet. Working together also helps uncover blind spots, challenge assumptions, and build solutions that nobody could reach alone. As mathematician Béla Bollobás once put it, Erdős had “an amazing ability to match problems with people,” connecting the right minds to the right challenges.

Research shows that teams consistently outperform individuals across disciplines, especially when solving complex problems. A Northwestern University study found that, by combining diverse expertise, groups produce more effective and more innovative solutions. Even the slower parts of collaboration, such as explaining ideas or resolving disagreements, often lead to deeper understandings and better outcomes in the end.

This raises the question of whether success is determined less by individual genius and more by where you sit on a network. Sociologist Ronald Burt’s research on “structural holes” suggests that the most innovative people aren’t always the smartest, but rather the best positioned—the ones bridging gaps between disconnected groups. In a study of 17,000 scientists, those who connected previously separate fields produced the highest-impact work, regardless of raw talent. Simply put: Being in the right network multiplies the reach of your ideas.

Need for speed

Collaboration has its limits, too, of course. Networks can become echo chambers where bad ideas spread just as easily as good ones. Group dynamics can also stifle dissent, making radical breakthroughs harder. And when too many people share the credit, participation can drop off. The real skill isn’t just joining networks—it’s knowing when to lean on them, and when to step away and work alone.

So, how do you build a network for maximum success? Or choose to go solo?

Avoiding the costs of coordination is a big win. As teams grow, communication overhead grows even faster. Harvard research shows that larger groups face steep innovation drag simply because it gets harder to move ideas through the system. Solo founders sidestep all that friction.

In fast-moving environments, small teams have real advantages. They don’t need endless meetings or consensus-building. They can pivot quickly and execute clearly. When Instagram founder Kevin Systrom decided to pivot from Burbn to a photo-sharing app, he didn’t have to argue the case—he just did it.

Your next best friend

The rise of advanced AI tools has opened a third path between going solo and building a traditional team. AI now acts as a sounding board that’s available 24/7, extending individual thinking without the coordination costs of human collaboration. Solo founders can keep full decision-making control while tapping AI for fresh perspectives, challenges, and blind-spot detection.

One especially powerful use case is “red teaming,” or testing ideas by deliberately trying to break them. While human collaborators might hesitate to challenge you, AI can be programmed to do it without the ego or status games. It can surface weak points, generate counterarguments, and suggest alternate approaches, acting as a brutally honest, always-available sparring partner.

This process is exactly what legendary chess grandmaster Garry Kasparov called “weak human + machine + better process.” The highest-performing systems will blend human and AI strengths. Research from Stanford’s Institute for Human-Centered AI shows that human-AI partnerships consistently outperform working alone, especially on creative and analytical tasks. AI’s knack for spotting patterns pairs perfectly with human contextual understanding, creating a symbiosis where individuals can think bigger, faster, and smarter without giving up autonomy.

But this new model demands new skills. Collaborating with AI isn’t about teamwork in the traditional sense. It’s about learning to frame problems clearly, question outputs critically, and maintain final judgment. The collaboration of the future won’t revolve around group consensus—it will revolve around sharper thinking, better questions, and smarter decision-making.

As Grace Hopper, computer scientist and rear admiral in the U.S. Navy, famously warned, “The most dangerous phrase in the English language is ‘We’ve always done it this way.’”

Read the whole story
cherjr
3 days ago
reply
48.840867,2.324885
bogorad
11 days ago
reply
Barcelona, Catalonia, Spain
Share this story
Delete

Анекдот дня по итогам голосования за 13 августа 2025

1 Share
Моя девушка била меня по лицу, когда испытывала оргазм. Я не особо переживал, пока не узнал, что она притворяется.
Read the whole story
cherjr
4 days ago
reply
48.840867,2.324885
Share this story
Delete

Is SOC2 the new tech protection racket?

1 Share

Hi! It’s me, Can again. Today, we need to talk about the $4 billion elephant in the secure room: SOC2.


Working at a small tech company has perks. You want to fix a bug? Push a change to production. A user has problems? Connect to the database and look up, and hopefully fix up, their data.

Large firms work differently. When you are a large enterprise, you promise your customers that the product won't change under them or that their data won't be snooped on. It’s why they have large customers, and large sums of money.

So what do you do as a small company when you want to sell your product to a large company? How do you bridge that trust gap? Well, these days, you check your bank balance and then call up Vanta to ask about SOC2.

In related news, Vanta just raised a bunch of money:

Vanta, a startup with software for managing compliance with cybersecurity and privacy standards, said Wednesday that it closed its latest fundraising round at a roughly $4 billion valuation.

The $150 million round, which included funding from CrowdStrike’s venture arm, represents a valuation increase from $2.45 billion last year.

What is SOC2, anyway?

SOC 2 is essentially a way to tell people that you have undergone certain security checks. You hire an auditor, who gives you a huge spreadsheet and tells you to start acting like an adult. You start doing criminal background checks on employees, write up all sorts of policies (all the way from password rotation policies to what happens if your CTO gets stranded in a remote island), tighten up the security of your data, code, and a bunch of other stuff. It’s a lot, and feels like it won’t ever end. You put all that into the spreadsheet over the course of weeks, or even months.

If the auditors like what you've put into the spreadsheet, you get a Type 1 certificate, which is more of an IOU. What you really want is the Type 2 certificate, which requires a 6 month wait. During this "observation period", the auditors continuously check if you've been following your own procedures. Then, if they are convinced, you get the Type 2 certificate as a fancy PDF. You put the SOC2 logo on your website, write a blog post, tell all your prospects you are now a secure company. And, of course, the observation period starts again. The cycle continues.

OK, this is a funny billboard.

So, where does Vanta fit in all this? They make and fill in the spreadsheet for you.1 As "the compliance automation platform", they sit between your company, "the service organization" and the auditors. Auditors look at the spreadsheet created by Vanta and give you a pass/fail score.2

The "tech" angle behind all this is connecting to all your systems like AWS and Github and verifying things are locked down. Vanta also syncs up with your HR systems like Google or Workday to make sure people follow their training schedule, accept their policies, and pass their criminal checks. It's a big value add for Type 1, and a true life-saver for Type 2.3

Now, in a pure business sense, Vanta should be good for the world. If they are lubricating the wheels of commerce by making sure even small companies can cross the enterprise chasm, it's fair for them to get a cut of the marginal transactions they are enabling. After all, that’s why most SaaS companies exist. We all make the world a better place, in our small ways.

There's, however, a cynical view that SOC2 (and other compliances) is essentially now a tax that you have to pay. You do it not because you have to, or because it makes your product more safe and secure, but because the industry undertow makes you do it. With a conservative 20x multiple, Vanta could be making $200M a year. What percent of that is the enterprise value their customers have created that they are capturing and what percent is just a tax?

Linear co-founder (via x)

Problems with SOC2

First, the perverse incentives and the pay for play. The auditors work for you, the client. It's in the auditors' interest to keep you happy, so it doesn't take a lot to convince them you are doing the right thing. The enterprise customers often don't even know who the auditors are before seeing the SOC2 report, because seeing the report requires an NDA. And these customers rarely, if ever, choose their vendors based on their auditors, if they ever read the report at all. Having read a few SOC2 reports for my company's vendors, I can tell you the quality of reports, and the security postures, vary wildly. And, needless to say, no one cares if the product is good but the report is bad.

That’s an excellent title

Second, the entire process is incredibly janky. A shocking amount of the SOC2 process is built on screenshots. If that sounds extremely insecure, you are correct. You take screenshots of your tools. You take screenshots of your JIRA tickets. Whatever you need to “prove,” you take a screenshot, and upload it to Vanta as evidence. The word "evidence" belies a forensic approach, but it's a lot more TSA checkpoint than actual detective work. There’s no chain of custody, authentication, anything. The entire system largely hinges on your auditor’s trust in your screenshots.

Third, the network effects. The entire thing spreads like a virus in the system, in two ways. First, it's via the technical dependencies embedded in the SOC2 process. Even the smallest of SaaS companies run on other SaaS companies' products4. In order to get your auditors' approval for SOC2, you end up having to validate all your vendors' security practices. You can already see where this is going. That, of course, also requires you to get their SOC2 certificate. Turtles all the way down!

via xkcd

And then, more subtly, there's the social pressure. Sure, SOC2 makes your company more secure. But ultimate security is an elusive, platonic ideal. No company with SOC2 is immune to a rogue employee or a dedicated hacker. Things always slip through the cracks.

However, with everyone around you getting SOC2, your investors being investors in various players in the ecosystem like Vanta, the FOMO to "just get SOC2 and be done with it" is immense and hard to argue against. Not only that, it comes earlier and earlier in companies' lifecycles now. Companies without product-market fit or even a sales team spend weeks on SOC2. Is that the best use of anyone's time or are we just watching a very elaborate Kabuki theater?

SOC2 is serious business

It’s Just Business

Look, this is not to say that Vanta is a bad actor. It’s a hugely successful business. And clearly, many people love it. I know lots of great people who work there. And I keep mentioning them as a placeholder for this industry because they are the leaders.

Making it easy to get SOC2 earlier in a company's lifecycle is a great idea. A company that has a security mindset earlier on is going to be a Fort Knox compared to a large company that bolts on a bunch of policies later on. It'll also be more pleasant for everyone, and cheaper. These are undeniably good things. Any time technology makes something hard easier, I err on the side of celebrating that.

Not only that, I know people who have had to get their SOC2 before Vanta and those people would have killed to have the automations and integration tools that Vanta provides. It's a clear value-add for those who need it. It’s a lot more fun to click buttons than to fill in those spreadsheets.

Yet, in this case, I am not convinced the tech industry has become better thanks to our newfound love for SOC2. Vanta is not a villain, but they aren't heroes either. Currently, they are the enablers of a flawed system, not unlike TurboTax and the IRS.

Ask yourselves. Are we all more efficient at sales and go-to market now that we are all SOC2 certified? Are everyone's products safer as a whole? As a customer, have you ever checked if an app you are using is SOC2 certified. As a professional, did seeing SOC2 on your vendors’ security page give you more confidence or did you just feel good about CYA?

What many won't tell you is that the real secret to SOC2 is less about using the right platform and more about finding an agreeable auditor. Like I alluded to before, most SOC2 reports aren't read. If a company wants to use a vendor, and their vendor's report is awful, it won't stop them. But everyone's stuck doing the right thing, even when it's the wrong thing.

Breaking the Screenshot Tax

The irony is thick. Small companies used to sell the dream of being a cowboy: pushing to production, fixing problems directly, moving fast. Then we built tools to help small companies act like big companies so they could sell to big companies. Now those same small companies spend their precious early days filling out compliance spreadsheets instead of finding product-market fit. Where did the fun go?

It is now time to replace SOC2 with something better5; something continuous, programmatic, and discrete. A compliance scheme that monitors security in real-time, just like we monitor our apps. One that updates automatically when you fix a vulnerability, not six months later when the auditors check back in. The good thing is Vanta now has both quite a bit of leverage and expertise on both sides of this market. They have the integrations, the customer base, and now $150 million in fresh capital. They should use their powers (and new coffers) for good. Enterprises won't like giving up their familiar PDFs, but someone has to drag them into the 21st century.

The same ingenuity that created this compliance-industrial complex can tear it down. We've already transformed how we build and deliver software—from waterfall to agile, from shrink-wrap to SaaS, from annual releases to continuous deployment. We monitor our systems in real-time instead of waiting for post-mortems. We scan for vulnerabilities continuously instead of scheduling yearly pen tests. Hell, we even moved from PDF manuals to living documentation.

So why are we still treating compliance like it's 1995? Imagine getting dinged for a security issue you fixed three months ago, just because the auditors haven't caught up. Or passing your SOC2 while harboring vulnerabilities that appeared yesterday. It's time to ship the future of trust, not just automate the past. All that's missing is our collective will to stop paying the tax so I can go back to pushing code to prod.

Thanks for reading Margins by Ranjan Roy and Can Duruk! Subscribe for free to receive new posts and support our work.

1

Unsurprisingly, a hand-woven spreadsheet was also the first version of Vanta.

2

In reality, it’s a lot more an interactive process with a whole lot of back and forth convincing your auditors. On one hand, it’s a lot more human than I imply. On the other, it’s really about who you know.

3

I have used a Vanta competitor, Drata but they were largely the same. I did get a two-week demo of Vanta as well.

4

I have mixed thoughts about it. It’s always good to have a bunch of people who can bootstrap your business but it can definitely hide the real challenges. Not even going into the circular trading nature of increasing ARR of it all, however.

5

Beat you to it.

Read the whole story
cherjr
9 days ago
reply
48.840867,2.324885
Share this story
Delete
Next Page of Stories