On drones

On drones

Some say that UAVs can replace tactical aircraft, and that, being cheaper, can be produced in greater numbers. But UAVs have greater logistical burden when compared to even relatively complex manned fighter, as UAV support crews need to eat, and maintenance isn’t simple either. More C-130s are required, more parts and technicians are also required for both UAV and support mechanisms. Neither can UAVs do ground attack missions, and combat-capable UAVs aren’t going to be cheap even when counting only production costs. And while I did write an article about (in)ability of UAVs to replace manned fighters, I hope for this article to be more detailed and adress some issues I have not touched in my previous article.

UAVs were first used in Spanish-American war in 1898 when journalist William Eddy took hundreds of photographs from camera suspended under a kite. While not strictly “UAV”, in third century BC Chinese used kites to help with triangulating distance required for tunnel dug under the walls. First heavier than air unmanned aircraft flew in 1896, over Potomac river. In World War II US military attempted to develop a radio-controlled bomber, codename Aphrodite; none of prototypes were successful and project was scrapped. First successful military use of UAVs was in Vietnam, with Ryan 147 Lightning Bug flying 3.435 reconnaissance sorties during Vietnam war. In 1972, modified Lightning Bug was used to launch a missile against simulated SAM. In 1982, Israel successfuly used UAVs in reconnaissance role in Lebanon, and US Navy acquired UAVs from Israel for use in Desert Storm. But are modern UAVs useful? I’m going to take a look at following missions: air superiority, ground attack and reconnaissance.

Complet required for two unarmed Shadow 200 UAVs costs 36 million USD, and UAVs themselves cost 275.000 USD. Single unarmed Predator costs 17 million USD flyaway, and 3.200 USD per hour to operate, without counting control systems. Reaper costs 17 million USD flyaway and 3.600 USD per hour; group of 4 plus control equipment costs 129 million USD. Global Whale (offical name Global Hawk) reconnaissance UAV costs 141 million USD plus 17 million USD per year; while SR-71 costs 261 million USD, having in mind sortie rate issues described later, it is entirely possible that SR-71 may provide more sorties per 1 billion USD spent on aircraft acquisition. Single X-45A, a new US UCAV, is estimated to cost 25 million USD flyaway at empty weight of 3.630 kg. As final version, X-45C, weights 19.000 kg, flyaway cost will likely be 131 million USD. Carrier capable X-47B is expected to weight 6.350 kg empty; flyaway cost will likely be around 100 million USD. Thus UCAVs will cost 25-130 million USD and weight 3,6 to 19 tons, resulting in costs of 6.800 to 15.700 USD per kg; compare to 16-18 million USD and 5,7 to 11,32 tons for well-designed manned fighters, resulting in costs of 1.400 to 3.200 USD per kg, and 30 million USD for F-16A, resulting in cost of 4.240 USD per kg. As X-45A is unlikely to be used for combat, especially for air-to-air, operational UCAVs will be as heavy or heavier, and far costlier than my proposed manned fighters. European UCAV project, nEUROn, despite only being capable of ground attack missions, weights 4.900 kg empty and costs 32,5 million USD, a cost of 6.633 USD per kg. Air superiority UCAVs will cost even more than given costs because of requirement for pulling high g maneuvers, which will put strain on airframe, and especially on fragile electronic components – it is likely that cost of typical air superiority UCAV – if these ever appear – will be 30 to 200 million USD flyaway. None of these costs take into account cost of control systems, such as control unit and communications hardware.

UAVs are notoriously unreliable. While UCAV replacing 4th generation aircraft will cost as much as – or more than – fighter it is replacing, UAVs loss rate is higher than that of manned fighter – even if UCAVs cost turns out far less than that of modern fighter aircraft, in the long run it will be far more, and will support effectively smaller force due to high maintenance requirements and inability to build up numbers caused by huge loss rate. Class A mishap (loss of aircraft) rate per 100.000 hours was 4,1 for F-16, 6,8 for U2, 20 for Predator, 88 for Global Hawk and 191 for Shadow. 2005 Congressional Research Service report indicates that UAV is 100 times more likely to succumb to failure than manned aircraft; considering US DoD history of misreportation, it is possible that figures cited underestimate UAV loss rate (CRS report also cites F-22 loss rate as being 6 per 100.000 hours).

Claim is that UAVs do not require a pilot, and that this reduces costs even more. This is incorrect: UAVs require pilots, except these are not in aircraft itself. Further, while 4 JAS-39 require 4 pilots and 40 maintenance personnell, each Reaper 4-drone CAP requires at least 171 personnell, including 13 pilots. This does not account for other support personnell: drones operating in Pakistan are dependant on US intelligence community as well as tens of thousands of troops stationed in Afghanistan; without these troops, they would not have bases to operate from. And in serious air war, people are just as much in danger on the ground as in the air, if not more so; result is that drones put more people at risk, not less. Even in permissive airspace, they are always used alongside ground forces, while in any kind of defended airspace they require manned fighter escort to operate.

Reaper can only withstand 2 g maneuvers, and high angle maneuvers can lead to connection loss, resulting in a crash. Maximum payload is 1/5 of A-10s and it has nothing comparable to massive GAU-8 cannon. It can also loiter between 18 and 40 hours, depending on payload.

There is still around 2 second delay in transmitting the data, which means that operator’s OODA loop (observe-orient-decide-act) is delayed by at least four seconds relative to events he is reacting to. As real-world air combat is visual range combat, this delay makes UCAV completely helpless in air-to-air combat. Autonomous UCAV is not yet an option for air-to-air combat, and it will not be until development of full artificial intelligence because it has to process too much data and make decisions that are hard even for human to make; they also very often misinterpret commands. UCAVs also have limited situational awareness when compared to manned fighters, with most modern UCAVs only being able to look forward.

Even with best sensors, data received by operator is never perfect, and operator’s situational awareness is far less than that of pilot of a modern 4,5-th generation fighter aircraft. Thus, Observation part of OODA loop is imperfect, which automatically means that Orient, Decide and Act parts are flawed as they are based on incorrect or incomplete information. This imperfection can potentially snowball through the loop until pilot becomes completely disconnected from reality – for politicians, it already happened.

Simple fact is that autonomous UCAVs do not have adaptability required for air-to-air combat whereas remotely-piloted UCAVs fly in their own time zone. End result is that UCAVs are incapable of air-to-air combat as well as close air support, while being as costly as fighters that can be made with modern technology if discipline in design process is exercised and loading of aircraft with unnecessary technology is avoided. In fact, in 2003 Predator engaged Iraqi fighter with Stinger missile, attempting to defend itself; it was easily shot down. And despite entire loose-no-pilot-if-drone-is-shot-down thing, when time came for attack on Libya, it was manned aircraft that cleared the way for drones, not the opposite – drones were only deployed after numerous SEAD, DEAD and air superiority missions have already been flown by manned aircraft.

While people harp about UAVs ability to take G forces, simple changes to pilot’s position in manned fighter would enable him to take very high g forces for prolonged periods of time:


With 14 g capability at Mach 0,5, modern BVR missiles would require 896 g to hit it; with 30 g limit of most missiles, Pk would be 3,3% (compared to 3,9% for BVR missiles against modern fighter aircraft with 12 g capability), which would fall well below 1% if ECM is counted in. For WVR missiles, Pk would be 12% even before countermeasures are accounted for.

Modern jet engines however can only withstand g forces up to 12 g, and designing jet engines that can withstand significantly higher g limits would be very expensive, and serve to make air-to-air UCAVs even more expensive.

Drones are no better in ground attack roles. While 17 million USD Reaper can carry 1,1 tons of weapons, 16 million USD A-10 can carry 7,25 tons of weapons. Delay in transmitting sensory data and control inputs means that UAV flying at low altitude is at very large danger of getting shot down or simply crashing into the ground, and lack of armor combined with complex electronics makes them vulnerable to even small arms fire. This means that UAVs are unable to carry out Close Air Support missions, and no drones carry anything like massive GAU-8 cannon, being forced instead to use 60.000 USD missiles. This means that it would take 3-5 Reapers to approach CAS capability of one A-10, even when “quantity is quality by itself” is accounted for.

According to test reports, Reaper’s sensors have problems with tracking vehicles, and are incapable of identifying dismounted persons. Because of it, UAVs typically operate below 3-5 kilometers if shoulder-fired rockets are not the problem. In April 2011, Predator’s inability to discriminate Marines (in full battle combat equipment) from Taliban led to death of two Marines. Due to power limitations, Reaper’s Synthetic Aperture Radar does not allow it to perform all-weather combat missions. Even if sensors were better than on manned aircraft – which will make drone far more costly than manned aircraft – manned aircraft pilot would still have more contextual cues that would allow him to separate combatants from non-combatants.

Only advantage UAVs have in ground attack roles are comparably slow maximum and loiter speeds, but control lag and vulnerability prevent them from coming low enough to exploit these capabilities. While unarmed Predator and Reaper can loiter for up to 40 hours, that time is reduced to 14-16 hours when carrying external ammunition. It is still several times longer than even that of A-10, which can loiter for four hours. U-2 can fly for 12 hours. However, whereas A-10 can fly 23-64 hours per week, each Reaper can fly, at best, 29,5 hours per week. End result is that 1 billion USD will buy 63 A-10 flying 1449 – 4032 hours per week, or 28 Reapers flying 826 hours per week. In presence of air defenses, Reaper would require manned aircraft escort, removing all of its supposed advantages.

Even in low-threat conflicts where they don’t need to worry about being shot down, they hamper rather than help efforts. Out of all people they killed in Pakistan so far, only 1,5% could be identified as militants; rest just happened to be in wrong place at the wrong time – all young males in vicinity of identified militant are assumed to be valid targets by US Government – while at least 5% were children. In June 2009, drone targeted Taliban leader Baitullah Mehsud on funeral: 60 civilians were killed, and Mehsud escaped – such slaughter would have infantry unit court-martialled. Even when attempts are made to identify terrorist leaders from the air, they are based on unreliable data, such as target person missing a limb. Areas that hold most militants, such as North Waziristan, are intelligence black hub because resentment of government in Islamabad is such that any suspected governmental agents are immediately executed – and drones are not helping things.

Not unknown are double-tap attacks, in which drone attacks medical services attempting to help survivors of the first attack. This drone-performed bloodshed has secured Al-Quaeda an influx of personnell and created intense hatred for US in rural areas. These strikes were also illegal, infringing on Pakistani sovereignity – possibly with acceptance of Pakistani government. Attacks also harm Pakistani government’s efforts against terrorists by creating a credibility gap, being difficult to explain for a democratically elected government.

us drone

Predator UAV was in fact purely reconnaissance aircraft in design, and was only armed to kill Osama Bin Laden – in the end, it was deemed too unreliable, and infantry was sent. But bombing is not a solution to what is, in essence, economical and political problem. Under tribal code, family members of civillians killed by drone strike are required to seek revenge; thus each civillian killed brings several new recruits to Taliban. Families in rural areas often have 10 sons; at 98,5% mistake rate, this means 600 new enemies created for each Taliban killed. Even when not killing people, drones create psychological pressure by threat of death.

Further, tribal cultures value bravery, courage and honesty; yet drone strikes build picture of United States as cowardly and unthrustworthy. Considering that war, especially guerilla war, is primarly conflict of morals, and that one with moral high ground is more likely to win, drone strikes are immensly detrimental to US case. In fact, majority of Pakistanis equally despise extremist groups and US drone strikes. Pakistani military does not like drone strikes either as they make it subservient to US military and US interests.

Even when drones do kill terrorist leaders, they leave terrorist cells intact. University of Chicago study has shown that, in 300 leadership assassination attempts, it worked in only 17% of cases. In many cases, terrorist cells whose leaders have not been assasinated actually declined faster than those whose leaders have been assasinated. Leaders aren’t unreplaceable either (noone is): an estimated 10 people have been described by CIA as “Al-Quaeda’s No.3 official” in past decade. Killing targets from the air also prevents any intel being collected, either through interrogation or recovery of documents, thus dealing more damage to counter-insurgency than to insurgency. But there is incentive for US to kill terrorists and not capture them: prisons are already overburdened with domestic and foreign criminals (and “criminals”) alike, especially since closure of Guantanamo Bay.

There is a permanent psychological pressure from drones, which keeps people from sending children to schools, and severely limits all outdoor daily activities. Children who have grown up under such pressure will have psychological problems, including a strong desire for revenge. In Yemen, notional membership of Al-Quaeda has risen by 300% since 2009, despite drone strikes.

In short, while classic counter-terrorism approach is to drain terrorists’ pool of recruits and increase support for central government, drones are draining support for central government in Pakistan and drumming up support for terrorists. That much was even stated by David Kilcullen, former counter-insurgency adviser for general Petraeus.

As for anyone talking about precision missiles, they should read this piece by Vietnam War veteran. Hellfire missiles used by drones are massively destructive, and single hit in house can destroy not only house but entire neighbourhood. As Jeffrey Adicott said, “Killing from that far above, there is always “oops” factor.” And this applies to both civilians killed by drones in Pakistan and US and allied soldiers that will be killed when F-35s start delivering their “close support”. In November 2011, US drone strike killed 24 Pakistani soldiers.

Yet some still think about removing “man in the loop”, allowing drones to autonomously shoot at targets. But given that even human-controlled drones kill 50 civilians for each militant, this is definetly not smart idea.

Drones do not comply with international law. While their use by military in shooting identified enemy combatants is legal, CIAs usage against “suspected” militants – who are usually civilians – is not. Further, just-war principles hold that it is better to put one’s own combatants than enemy’s noncombatants at risk – drones are doing the reverse.

But regardless of UAVs usefulness in ground attack role, strategic bombing (UAVs are CAS-incapable) never won any war, or any battle. Luftwaffe devastated Malta with bombers – failure to invade island with ground troops cost Germany any possibility of success in African campaign. USAAF devastated German cities, under wide definition of “enemy combatant” which included people working in factories – just as many people drones kill are not fighters, even if they do support militant groups – and it simply prolonged the war.

There are several more problems that are connected to basic nature of UAVs:

First, they are liable to be hacked and hijacked. That has already happened, when students and staff of one school have hacked into Homeland Security UAV spying on everyday lives of US citizens. There were also reports that Taliban have hacked into video feed of US UAV’s in Afghanistan, and Iraqi insurgents have hacked US drones – in fact, hundreds of hours of drone video footage has been found on laptops of Iraqi insurgents. If Taliban and civilians can do that, there is no telling what military hackers could do – there is no need for datalink codes as drones aren’t hard to hack even without these. Virus could take control of drones by falsifying input signals and order it to do basically anything.

Second, uplink used to control them can be jammed. This renders drone mostly useless and very vulnerable. While drone can be preprogrammed, it is incapable of adapting to a quickly changing situation, and it cannot adapt to any situation not programmed into it. Even in environments with minimal electronic noise and no jamming systems present, it is not unusual for operators to loose connection with drone for some time, but it can also be permanent; in one example, drone used by Irish peacekeepers in Chad decided to go home after connection loss; naturally, it failed to make it to Ireland. In second example, US UCAV had to be shot down to prevent it from leaving Afghan airspace. Since they fly in proximity to target, UAV has to have high signal-to-noise ratio to counter both jamming and natural noise; this makes UAV easily detectable regardless of any stealth measures (result is that UAVs such as Sentinel, X-47B and nEUROn are flying oxymorons). Further, signals using more bandwidth are at greater risk of being jammed, and Global Hawk reconaissance UAV uses five times more bandwidth than entire US military did in the Gulf War I. Thus, both broadband jamming (jamming wide frequency range) and sweep jamming (dynamic jamming signal frequency) are highly effective against UAV, as they don’t need to deny entire bandwidth to make UAV inoperable. Jamming does not even need to be intentional: already-mentioned electronic noise, a growing hazard on ever-more-technological battlefield, can easily interfere with UAVs uplinks.

Third, nature of UAVs requires them to be micro-managed by central command. This micro-management has been shown to seriously degrade effectiveness of military forces in any conflict.

Four, it is easier for humans to get emotionally and morally disconnected. Gadaffi’s fighter pilots defected to Malta when ordered to kill their own citizens. At climax of 1991 coup, USSR Air Force was ordered by hardline Communists to bomb Moscow – they refused. But after suspected terrorist Anwar Al-Awlaki – who was never charged for any crimes, and it is unknown if he ever killed anybody – was killed by a drone, his 16-year-old son, who had no connection to father’s terrorist activities and was normally taken care of by his grandfather Nassar al-Awlaki, was also targeted and killed while at outdoor restaurant. Question of “clean” murders with drones is similar to one raised in Star Trek: The Original Series episode “A Taste of Armageddon”. In it, two planets fought a “clean” war – casualties were determined by computer projections, and people killed in painless disintegration chambers. Aside from deaths, no physical damage was done; executions were so hygienic that many psychological effects did not exist. As result, war has been going on for 500 years; only once that system was destroyed did the war end. This is reality of drone pilots, who kill people from their office. Yet drone pilots are still human, and majority of them still suffer psychological consequences of murder. While remote control may reduce issue at moment of action, making them more likely to press the trigger, it can not shield them from consequences of the action; in fact, removal from battlefield may make psychological consequences even worse.

Five, they remove pilots from battlefield. One of main points in Boyd’s OODA loop is to reduce one’s own friction to minimum, and increase opponent’s friction to maximum. But technology, while often helpful, can stand in way of clear decision-making; humans present on battlefield itself have better view of situation immediately around their aircraft as no sensor can replace human eye in terms of resolution and fast observation yet, and there is no delay until data arrive; this immediately translates into better ability to decide quickly on course of action and to carry it through. Thus any manned aircraft will easily get inside UAVs OODA loop and shoot it down, even if we assume that UAV is more maneuverable by classic metrics (turn rate, roll rate, turn radius).

While advocates dream of AI drones, human brain performs 38.000 trillion (38*10^15) operations per second and has memory capacity of 3.584 terabytes (3,6*10^12 bytes). Biggest supercomputers do not achieve even 1% of these levels, despite taking up enormous space.

There is, however, one mission which drones can perform: reconnaissance. Such drones do not have to be fast or highly maneuverable, keeping them cheap and thus able to perform tactical reconnaissance. They could be assigned to individual infantry units, helping them avoid ambushes, and could fill holes in data gathered by satellites and high-speed reconnaissance aircraft such as SR-71. Ability to loiter for up to 40 hours above battlefield can be put to very good use in such missions, however their vulnerability would limit usefulness in such missions when in possible combat zone where enemy has shoulder-fired SAMs, necessitating use of survivable manned aircraft such as A-10. Even drones used in US to catch illegal immigrants and drug traffickers, and thus not operating under altitude restrictions, have proven less effective than manned aircraft, and high accident rate impeded operations. While 6 Reapers purchased subsequently helped catch 5.103 illegal transits, they had to fly 10.000 hours – a cost of 7.055 USD per immigrant or drug smuggler caught. At the same time, manned FLIR-equipped Cessna which cost 1,2 million USD to operate caught 6.500 to 8.000 illegal transits, costing 150 – 185 USD per person caught; well below 3% of UAVs costs.

40-year-old manned U2 reconnaissance aircraft is still more capable in its role than any drone. It has better sensors, is less vulnerable and can react to threats faster than drone. However, it is far more expensive and being manned is less expendable than some drones.

Still, manned reconaissance missions – regardless of type of aircraft used – provide higher resolution data than drones, and are more adaptable to various scenarios.

As seen above, drones are neither effective, efficient or moral. But who needs results when high technology is soo sexy? High technology was always appealing to imperialist colonial powers, as it held potential to make unjust wars less bloody for colonial power, and thus less controversial within populace itself – drones are basically remote-control imperialism. But this never made it more effective, and overreliance on technology was often followed by costly failures. At the same time, Western societies are reliant on high technology, which makes it appealing to general populace – thus large number of sci-fi shows and movies. Claims that drones can replace manned aircraft are in domain of science fiction, which is why many people want to believe them, as are the claims that drones allow for casualty-free warfare at either side. But these claims do have effect of making drone usage more acceptable to public at large, even though untrue.

Another reason why drones are so appealing to weapons contractors is that they are very unreliable even when compared to flying technological circus that is F-22. Therefore they hold the promise of keeping money flowing to military industry. In US, FY2013 bodget propsal held 5,8 billion USD for drone purchases, and companies that produce them donated more than 2,4 million USD to members of the Congress.

Further, United States have written a rulebook on drone usage, and it suggests that violating one country’s territory with unmanned aircraft is no violation at all. But in the end, UAVs are repeat of ICBM story: despite all predictions, ICBMs and cruise missiles have supplemented, not supplanted, manned bombers. Submarines did not make surface ships obsolete, even if they did make them extremely vulnerable, and capital ships other than aircraft carriers have made their return in form of large missile cruisers, succeeding battleships and battlecruisers in their surface combatant roles.

Drone use has a lot to thank to Margaret Tatcher. After slaughter of workers she ordered, it was almost certain that she will not be reelected. But she used war with Argentina over Falklands – islands that she earlier described as useless pieces of rock, and in fact reduced UK forces guarding them, an all-too-clear invitation for Argentine junta to invade them – to save her political career.

In the end, drones will only be useful for SAM exporters, giving them unprecedented Pk for ther surface-to-air missiles.

4 thoughts on “On drones

  1. Do you mind if I quote a couple of your posts as long as I provide credit and sources back to your website? My blog site is in the exact same area of interest as yours and my visitors would genuinely benefit from some of the information you present here. Please let me know if this ok with you. Thanks a lot!


  2. Hi Picard,

    Thank you for this post, it is very instructive in order to understand the mistakes made when always trusting ”newer technologies”.

    Hower, if we assume the following : in 2050 for example,

    UCAV could have an artificial intelligence that is good enough to be autonomous ( hence no need for easily jammable large bandwith communication ), but let’s say not intelligent enough for dogfight or close air support, just for strike/bombing/reconnaissance missions,

    as well as not being more expensive than manned aircraft, operating cost included (more like Gripen than F-22 though..), and also as reliable…

    then they could perform SEAD/DEAD missions on their own, provided that :

    they use only passive radar,
    they have low IR signature by flying subsonic,
    low radar signature ”””’stealth””” ( ‘flying wing’ )
    and flying at night
    using only long range passive or semi passive radar homing missile ( so they would be larger than today, and carry more payload, looking a lot like F-117…)

    and therefore it is still good for a nation to put some money into UCAV programs (like neuron/taranis…) in order to stay up to date with the technology, just in case this become a possibility in the future ? ( but not spending billions and billions like the US..)

    ( and in the wildest dreams, on could even have an -affordable- FLX-like aircraft with an AI piloting it.. in several decades ! )

    What do you think ?

    Thank you !


    1. Yes, AI could be used for some more dangerous yet not as complex missions, and in fact using UCAVs for such missions is a good idea. Even now, UAVs can be used as expendable “baits” to get SAMs to reveal their position and facilitate strikes from manned aircraft. But I’m not so sure about operating cost, a UCAV with capabilities of manned aircraft will logically be just as expensive as said manned aircraft. It would still have the advantage of not endangering the pilot. However, more complex missions such as air superiority and especially close air support still require manned aircraft to be undertaken effectively.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s