'People, ideas, machines' I: Notes on ‘Winning the Next War’

How do militaries innovate? Why do good ideas stall? What sort of organisations innovate best? Why can't politicians learn? Implications for UKR/Russia?

‘All warfare is based on deception… Know the enemy and know thyself, in a hundred battles you will never be in peril.’ Sun Tzu

‘There is nothing more difficult to carry out, nor more doubtful of success, nor more dangerous to handle, than to initiate a new order of things. For the reformer has enemies in all who profit by the old order, and only lukewarm defenders in all those who would profit by the new order.’ Machiavelli

‘People, ideas, machines — in that order!’ What Colonel Boyd shouted in the Pentagon for years…

Introduction

Below are some notes on Stephen Rosen’s book Winning the Next War: Innovation and the Modern Military. Given UKR, I’ll post notes on this and other books on war/science/management I’ve read over the years. It’s about how military organisations invent, innovate and procure.

Rosen asks some basic questions about military organisations and learning:

• How do military organisations learn, create new ideas about how to fight, and create new capabilities?

• Are they doomed to fight the last war?

• Is it easier to innovate in peacetime or wartime?

• How does technological innovation differ from innovation in ideas about how to fight?

• What is the role of intelligence in military innovation?

• What’s the difference between invention/innovation, and research/development? What kind of organisations and ideas are suitable for which category? E.g what are the pros/cons of having one R&D organisation or does it make more sense to split functions up?

• How and why do bureaucracies stop new ideas? How can common problems be overcome?

• How does feedback happen between people-ideas-machines-bureaucracies?

These questions are directly relevant to a) general problems we have with our institutions for crisis management whether war, pandemic etc, and b) this Ukraine crisis, both short-term and long-term: e.g if western countries want to take defence more seriously, what should they really focus on?

Strong caveat: the book looks at some specific US and UK examples from ~1900-1960s. It does NOT look at, for example:

• the famous example of Germany in the 1920s/30s bringing together tanks, radios and new ideas that together crushed France so quickly in 1940.

• the Manhattan Project or much of the ICBM project under Schriever (though it touches on some background on missiles).

• the combination of precision strike, stealth, GPS, drones and so on that emerged in the 1970s (partly from DARPA and Skunk Works), which I wrote about here and which apparently is directly relevant to Ukraine (because thinking from Ogarkov in the 1970s has apparently been revived in Russia, which I was not aware of until I read a brief a few weeks ago from Bismarck Analysis on Russian modernisation).

• Russian or Chinese innovation.

I started reading Rosen in January as I started thinking about the 2020 ‘integrated review’, a story I should write about. Media duds like the FT’s Philip Stephens have been posting comical takes about how UKR shows … we should have stuck with tank acquisition plans that were ‘world leaders’ in wasteful, disastrous, and dangerous procurement.

Prediction: 1) lessons from UKR will overwhelmingly support the arguments of those who in 2020 argued for radical MoD changes (including taking money from old tank projects that everybody privately admitted were a multi-billion pound disaster) and 2) the correct criticism of the review and connected documents will be seen as a) they did not go nearly far enough, b) the collapse of No10 follow through on defence reform in 2021 was — like the collapse of 2020 plans for planning reform, tax cuts, deregulation, Project Speed, intense focus on R&D and skills etc — a disaster for the country (and a political disaster for the Tory Party).

I wrote a lot about defence innovation/procurement 2016-19. E.g here on the crucial emergence of ‘systems engineering’ and ‘systems management’ in the 1940s-60s and how highly effective processes for managing complex projects and procuring new technologies were then dismantled in the Pentagon and lessons were forgotten. (Also here in 2014.)

A good example recently was the collapse of the Obama website for healthcare. He had to call friends in Silicon Valley to come to DC to fix it because basics have been forgotten and DC couldn’t fix it. We’re used to thinking of technological progress but most people in politics have no sense of how management has deteriorated in many parts of our civilisation and how this connects to how almost everything to do with infrastructure and military procurement takes much longer, is much more expensive, and often fails — and politicians don’t even pretend to try seriously to change the systems that repeatedly fail.

They also pay no political price because all the parties are the same year after year. There is a stable equilibrium where the sort of people who almost always dominate western political parties are the sort of people who have neither the interest in nor ability to change these big complex bureaucracies. Provided they all stick to roughly the same model of political action — i.e focus on the media and signalling to in-group factions, not the public or important problems — the equilibrium holds, even after repeated disasters, because it’s hard for ‘live players’ to break into the political competition.

Also cf. here, where I made predictions about how our EU-law based (Whitehall-goldplated) procurement system would endanger us in a crisis. These predictions came true in covid. SW1 response? Fail. Lie. Don’t learn. Repeat. Wave exclusives at lobby. Lobby ignore because procurement is ‘boring’ or cheer legal actions brought to slow everything down…

In the next few days I’ll blog on nuclear weapons. Next week I’ll blog on The Kill Chain by Brose, copies of which I gave in 2020 to some of those working on the UK ‘integrated review’ and which has been read by many of those now responsible for defence reform. Brose now works at Anduril, an interesting startup that is trying to bring the world of Silicon Valley to the world of defence procurement.

Please leave links to books/articles on these themes with a 25 word or less summary of why it’s important.

I’ve started posting SNIPPETS on UKR here instead of adding to the noise/hysteria of Twitter.

NB. Invention and innovation are often treated synonymously in policy world but most scientists I speak to treat invention as something new and innovation as something incremental. Invention at its best can have 1000X or more impact than ‘innovation’. It can be the difference between, say, Feynman and Deutsch coming up with the first concepts of ‘quantum computers’ and Google improving an important bit of engineering by 10%. Rosen has his own definition of innovation which is not the same.

Quotes below are from Rosen unless specified.

CH1: Thinking about military innovation

‘The problem of military innovation is necessarily a problem of bureaucratic innovation.’

Large bureaucracies are designed not to change, not to innovate.

Weber: the essence of a bureaucracy is routine, repetitive orderly action.

Colonel John Mitchell, British Army, 1839:

Officers enter the army at an age when they are more likely to take up existing opinions than to form their own. They grow up carrying into effect orders and regulations founded on those received opinions; they become in some measure identified with existing views till, in the course of years, the ideas thus gradually imbibed get too firmly rooted to be either shaken or eradicated by the force of argument or reflection. In no profession is the dread of innovation so great as in the army.

Romanticism is inevitable in the military and this can keep old ideas in place too long.

Wars are sporadic so ideas can be untested for many years.

Herbert Simon: Little or nothing is known of the determinants of the rate and direction of inventive activity, of how technical know-how is stored and reproduced by society.

Doesn’t seem like trying to acquire more resources makes bureaucracies innovate.

No good theory exists and studies are contradictory.

Rosen splits into three:

1. Innovation in the operational behaviour of military during peace.

2. Innovation in the operational behaviour of military during war.

3. Technological innovation in the military.

1&2 are social behaviour, 3 is building machines.

21 cases are studied, 10 in detail.

In the first 2, a major innovation is defined as a change in one of the primary combat arms of a service in the way it fights or alternatively, as the creation of a new combat arm. A combat arm is a functional division within the military in which one weapons system dominates the way in which its units fight. E.g infantry, artillery, armour, helicopter aviation. A major innovation involves a change in the concepts of operation of that combat arm — a change in the ideas governing the ways it uses its forces — not a tactical innovation, which is a change in the way individual weapons are applied to the target and environment in battle. A major innovation also involves a change in the relation of that combat arm to other combat arms and a downgrading or abandonment of older ideas of operation. A military innovation may not involve behavioural changes, it may be a new technology.

Peacetime innovation

Normal government bureaucracies have functions they execute daily. An army exists to fight an enemy so does not execute daily. It has to plan for wars that may not happen. It is governed by officers who are refreshed only with the approval of the existing senior leadership.

Defeat is neither necessary nor sufficient to produce innovation. History is full of examples of armies that continued being defeated without innovating: e.g the Tsarist army after the Russo-Japanese war and US army after Vietnam.

Barry Posen re interwar period: there was little internally generated innovation in the RAF, French army and German army. French failed to innovate. Changes in Germany were not genuine innovation. RAF only innovated after civilian leaders intervened.

Neustadt: for a President’s order to be obeyed by the bureaucracy:

1. The President must be clearly involved in the decision.

2. The order must be unambiguous.

3. It must be widely publicised.

4. Those who receive it must have control over everything needed to carry it out.

5. They must have no doubt of his authority to issue the order.

Militaries often don’t do what they’re ordered. e.g JFK tried to push army to work on counter-insurgency. They wouldn’t.

Mavericks dont have the effect claimed. E.g Hap Arnold: Billy Mitchell made the War Department ‘draw more into their shells’ and take ‘if anything a narrower point of view’! Rickover was an extraordinary manager but his influence on nuclear propulsion was exaggerated. Liddell Hart went to Hore-Belisha but this didn’t work.

Best example: the development of anti-aircraft defences in the RAF before WW2. This came after civilian intervention, particularly Sir Thomas Inskip. But the RAF pushed through a shift to Spitfire and Hurricane. Shadow factories converted from making cars. Training reserve pilots. RADAR: first demonstrated 1935. Doctrine of building infrastructure for acquiring and using intel on enemy planes pushed by RAF from 1920s not just civilians. Civilian role exaggerated in many studies focusing on e.g Tizard and Cherwell. RAF was building the infrastructure for telephones, coordinating intel etc. Once radar was demonstrated in 1935 it was developed fast.

G Allison, Essence of Decision: military made up of subunits battling each other for more of what they already have.

But military organisations are political: they are thinking, who should rule, how should its people live? They are more divorced from society. They are preparing for things that may not happen, they are thinking about what war may look like and how to achieve victory. New missions must be defined so officers know what is expected. Control over promotion is the source of power. Innovation comes when senior people have a new idea and change promotions to encourage it.

Nukes so different that military cannot exclude civilians legitimately from decisions.

Wartime innovation

Old and new ideas can be tested.

People die and must be replaced.

E.g long range bombers were sent out without fighter escorts, huge casualty rate, missions cut back until long range fighters available for escorts.

But very hard to see what to learn. How do you define ‘this is working’? Doing something new by definition is a leap in the dark. E.g Gallipoli. Naval guns firing at land turned out to have little effect. It was hard to figure out what lessons to learn from the early failures. Maurice Hankey, secretary of the War Council and first Cabinet Secretary: ‘It looks as though this accursed trench warfare in France had sunk so deep into our military system that all idea of the offensive had been killed.’

‘Continuity of command’? NO. Sometimes stops innovation. E.g Harris wouldn’t change targeting to oil/industry etc after precision improved and even after he was ordered to.

Intel crucial… Sun Tzu starts with ‘Assessments’ and ends with ‘Employment of Secret Agents’.

Success depends on foreknowledge. Foreknowledge cannot be elicited from spirits, nor from gods, nor by analogy with past events, nor from calculations. It must be obtained from men who know the enemy situation… [Only commanders] able to use the most intelligent people as agents are certain to achieve great things. Secret operations are essential in war.

After Poland 1939 the German army undertook a strikingly effective but normal lessons learned exercise.

Jungle warfare — allies figured out JAP didn’t work with big logistical reserve, so if they soaked up initial pressure they could counterattack. It was a normal lessons learned exercise.

When military innovation is required in wartime, however, it is because an inappropriate strategic goal is being pursued, or because the relationship between military operations and that goal has been misunderstood. The old ways of war are employed, but no matter how well, the war is not being won. A new strategic goal must be selected and a new relationship between military operations and that goal must be defined… Until the strategic measure of effectiveness has been redefined, organisational learning relevant to innovation cannot take place… A redefinition of the strategic measure of effectiveness tells the organisation what and how it should be learning from wartime experiences. Until such a redefinition takes place, a wartime military organisation will learn from experiences in terms of existing measures.

E.g How to fight German U-boats? Traditional ideas suggested antisubmarine patrols. When this failed, the answer was more patrols. Why? Partly this was a pre-war commitment to offensive patrols. Partly data and definitions. Convoys were rejected partly because it was thought there were not enough warships to convoy transatlantic shipping. Why? Navy relied on Customs Office data which counted all ships entering and leaving ports, giving a figure of 5000 ships per week, far too many for convoys. But a commander reviewed the data in spring 1917 and realised that only 120-140 big ships per week needed to be protected, which could be convoyed. Convoys began and a terrible problem was overcome. (Lloyd George criticised the navy for this failure.) [How true is this story?! Was there no entity operating then to check such things? How much of the resistance was to do with duff data, how much normal conservatism? This seems to me an example of how if you had an elite data science team with access to information and to senior people, you’d see the problem much earlier. It was not a hard conceptual breakthrough.]

E.g Strategic goal of bombing in WW2 should have switched from population centres to road/rail/industrial networks, as accuracy improved, but this was resisted. This meant that data showing the potential impact of attacks on transport were overlooked and learning didn’t happen.

Rosen (p38) - intel relevant to innovation will probably not be available, if it is then everything will take too long. [NO. He assumes everything must take ages but the convoy problem sounds like it could have been solved much more quickly by a serious OR/DS team. He also assumes a contrast between an organisation that moves fast and an organisation that lets ideas circulate, but actually great organisations do both. a) Let’s assume he’s right in describing the examples he considers. b) It doesn’t have to be so slow. c) It probably will be.]

Tech innovation

Railroads 19th C: causation was demand caused investment caused innovation (measured as patents).

DOD study, Project Hindsight, reviewed 20 major weapon programs: a clear understanding of a DOD need motivated 95% of action. I.e innovation comes from the military stating requirements.

Creveld concluded the opposite: in the 20th century ‘none of the important devices that have transformed war [including airplane, tank, jet engine, radar, helicopter, computer] owed its origin to a doctrinal requirement laid down by people in uniform’.

RAF ignored Frank Whittle’s early work on jet engines.

Arms races can generate competitive innovation. e.g Dreadnoughts.

[He spends some pages on the Cold War but we never really knew the right numbers on budgets etc, it turned out that all sorts of estimates on which studies were based were wrong. These pages did not really convince me of anything other than that it’s hard to make confident conclusions from many of these studies.]

CH2: The shape of wars to come

Officers developed new ideas about how to win future wars. These theories were surprisingly unrelated to close study of potential enemies. Peacetime innovation in US has proceeded remarkably independent of intel re foreign powers.

Perceptions of change in the structure of the international security environment (i.e factors outside the control of the US or hostile powers that constrain and create opportunities for the military) have been the source of such innovation.

Tech revolutions outside the control of the military (e.g airplane) or large changes in the international role of the US (e.g US emergence as a Pacific power) have triggered military innovation.

Major changes took decades. Enemy behaviour sometimes shifted much faster.

How does it happen? Senior respected officers created new operational tasks relevant to the new capability and a new promotion pathway for young officers.

E.g US beat JAP in 1942-3 with carriers laid down in 1920s-30s, NOT shocked into carriers by Pearl Harbor. Nor was it Billy Mitchell’s bombing of a captured German cruiser in 1921. The navy had submitted budget requests earlier. In 1920s US intel did not support the idea that US should fear JAP carriers. Wargames explored the possibilities of carrier forces with bigger and bigger aviation forces than existed at the time. This encouraged their development and also all the thinking about new forms of battle. Wargames encouraged a shift from ‘carrier as adjunct to battleship’ to ‘replacement of battleship’.

E.g Marine Corps shifted from small wars and infantry to the first force capable of amphibious assault on heavily fortified island targets. Driven by taking Philippines and the emergence of US as a Pacific power.

E.g helicopters.

These examples show innovation driven by structural changes in security environment, not by specific capabilities or intentions of potential enemies. Figuring it out came from wargames/simulations. Then a political process that changed the way officers and organisations lived their professional lives and were promoted.

3 Making things happen: the politics of peacetime innovation

Carrier Task Force

Moffett was not an aviator but was made head of the Bureau of Aeronautics in 1921. Died 1933. He created career pathways so that a) non-aviators were brought into the ranks of aviators as navigators, b) junior aviators could be promoted and take senior positions.

Some of the old guard resisted the growing power and promotions of the aviators. Some of the aviators resisted being absorbed by the old system and pushed for a separate force they could run. Moffett resisted both.

Carrier doctrine was developed by many senior aviators who were promoted. The doctrine of multi-carrier task forces was crucial: the massing of offensive and defensive power. It was opposed by traditional officers who saw carriers primarily as ships — not floating airfields — and therefore to be dispersed to minimise vulnerability. This battle was won with publication of PAC-10 in 1943.

The change took ~23 years.

From small wars to amphibious assault

The same pattern is seen with amphibious warfare. It took decades to move from the new ideas of Major Ellis to the initiation of a bureaucratic strategy for the Marine Corps’s promotion process to the first full-scale amphibious exercises in 1940 and the first combat assault in 1943.

When the famous General ‘Howlin’ Mad’ Smith arrived at Naval College in 1920 he found disdain for the marines from senior navy leadership and no sign the problem of amphibious warfare was being seriously studied. ‘Even the bitter lessons the British learned at Gallipoli had little effect on the War College.’ Small wars remained the main occupation of the Marines in the 1920s and 1930s.

General John Russell did for the Marines what Moffett had done for naval aviation: a new promotion path. He created the Fleet Marine Force which ensured the Marines were integrated with the navy and had an organisation for the study of amphibious warfare. Gradually they developed the ideas and equipment, such as landing craft.

The story of helicopters in the army supports the crucial point: the innovation went mainstream when the promotion system changed to support it.

Failure: Royal Navy and carriers

In 1918 the Royal Naval Air Service had ~3,500 planes and 55k men. In 1919 the RN had the first operational carriers in the world. Like America, the RN faced the problem of Japan in the Pacific. Like America, the RN was pushed towards carrier construction by the limits on heavy cruisers imposed by the 1922 Washington Naval Treaty, and both navies circumvented these limits by converting to heavy cruises into carriers. In 1939 Britain had 14 carriers built or in construction versus 8 for America and 11 for Japan. Britain pioneered many new technologies such as angled decks, optical aids for landing.

But when war started Britain failed at carrier warfare. WHY?

1. Little effort to to redefine the essence of naval operations using carriers.

2. No senior officers drew officers from the traditional elite into the innovation with new promotion paths. Officers with combat aviation experience had to leave the navy!

3. Simulations were not used to explore the possibilities for improved aircraft performance. Because it was believed that aircraft could not sink battleships, no money was made available to develop carrier-borne bombers. ‘No exercises were conducted to test what would happen if large numbers of fighters were available for the defense of carriers, so no effort was made to increase the number of aircraft that could be operated from a carrier.’

4. Moffett redefined carriers as mobile airfields. The RN took the opposite view. In 1918 the RN botched the debate over the future of aviation and handed over its planes and people to the RAF. There was a 20 year bureaucratic battle between the RN and RAF over planes on the carriers — training, command etc. It was therefore potential career suicide for a naval officer to pursue aviation.

5. Efforts to shift on the eve of war came too late.

Failure: attempt to create US Army counter-insurgency capability

JFK pushed the army to develop this capability. In 1961 he summoned key people to the Oval Office to give them a push. Got support from Army Chief of Staff General Harold Johnson. But it failed. ‘Little resulted from the intense presidential pressure.’ There was some relabelling of existing doctrine as counterinsurgency [classic bureaucratic resistance move!].

The army was not convinced that there needed to be a fundamental shift of resources, planning, promotion etc towards CI given the situation in the Cold War and the conventional/nuclear Soviet threats in Europe.

The Army Chief of Staff 1964-8 was General Harold Johnson. He agreed on the importance of CI in Vietnam. But he was thwarted ‘by the strength of traditional beliefs about promotion pathways within the officer corps’.

Promotion in the Army had aways been accelerated by command of a combat unit in war. No other service conferred the same legitimacy. Advisory duty on CI was not such a command. Officers thought it detrimental to their careers. Johnson tried to counteract this force, of which he was aware, with a directive formally instructing promotion boards to regard service as an advisor in Vietnam as equivalent of a combat command in consideration for promotion. He tried other incentive changes. This did not work. Perception of the real, rather than formal, criteria for promotion was such that officers did not believe promotion boards would really follow the directive — and they did not! Cynicism was justified!

[This is a really good example of how deep into a bureaucracy you sometimes have to search to find what’s really happening. President orders ‘X is a key priority’… Senior bureaucrat agrees (doesn’t resist)… Bureaucrat with the power to change incentives understands he must change incentives… And manages to do so… And follows up rather than assuming oh that’ll work… But it is still not enough… Because of perceptions of the relative power of implicit ideas versus explicit orders on careers!]

CH4: The British Army and the tank, 1914-18

In peacetime the old people, ideas and machines can stay because reality has not spoken.

In war you face reality and people die/fail and need replacing.

The real problem with the tank was: how to measure its effectiveness and use most effectively?

In 1914 the British Army suffered a general incapacity to learn from experience. The military historian G.F.R. Henderson, a professor of the British Staff College, had served as the director of intelligence in the Boer War. In an essay written before 1914 he described the British military's lack of formal means to collect data and analysis from soldiers with combat experience and with these develop a coherent view of the changing character of war. British soldiers acting as observers had carefully studied and reported on contemporary military operations of foreign armies. The data had revealed crucial points such as the suicidal character of assaults on prepared defences with modern weapons. But there was no institution for learning and spreading crucial ideas and suggestions to create one were howled down.

In the British Army no means existed for collecting, much less analysing the facts and phenomenon of the battlefield and the range. Experience was regarded as the private property of individuals, not as a public asset, to be applied to the benefit of the army as a whole… The suggestion that a branch should be established for the purpose of dealing with strategical and tactical problems involving both technical knowledge and patient study was howled down by the economists. (Henderson, The Science of War, 1905) [This obviously resonates with covid 2020 and the lack of an institution to learn from crises and disseminate lessons.]

General Haig thought that detailed studies would bind the hands of wartime commanders (!). When war started the General Staff was disbanded so officers could take up field commands(!). Intelligence functions were divided between a unit in headquarters in France and a unit of the Imperial General Staff in London. ‘These institutional arrangements did not substantially improve during the course of the war.’

Many individual soldiers accurately predicted crucial aspects of war including the strength of the defensive using artillery. But the army could not ‘prepare itself in peacetime either to fight well under the new conditions or to learn and innovate in wartime. In 1912, the British General Staff argued that a general European war would last at least six months, and that, if it lasted longer, it was preferable to wage war by means of an economic blockade of Germany rather than by raising a mass army.’ (Cf. Gooch, The Plans of War; Hankey’s memo, 26/12/1914)

Kitchener thought by 1909 an Anglo-German war would take 3 years at least and require exhaustion of German manpower. Therefore while Germany, France and Russia exhausted themselves he advocated Britain raising a million man army to throw into the war decisively when they stalled. He defined the problem as attrition. Churchill cast about for amphibious alternatives. Haig continued to hope for a breakthrough and advance.

It took until 1917 before there was broad agreement that the measure of effectiveness was not ‘forward movement of the front line’ but ‘most efficient use of our men to kill their men’.

Then there were still big problems with data collection.

Postwar data showed each major attack actually led to roughly 1:1 casualties between attack/defence, p117. During the war intel on casualties was hopeless. London estimated Germany would run out of men within 6 months and kept rolling forward this 6 month forecast to the end of the war. I.e intel did not indicate a need for innovation.

The basic idea of a tank occurred independently to many people at roughly the same time. E.g Swinton proposed a tank in November 1914 and got funding in April 1915. ‘From first conception to first production required only fifteen months. The technology was not resisted by senior military commanders.’ From the start the tank was supported by the highest ranking officers in the army.

But it was not built at scale to the detriment of other requirements. Its effectiveness was only really demonstrated by the Battle of Cambrai in 1917 and the analysis of this battle was not available until March 1918. By the time Britain realised the true potential effectiveness of the tank and started mass production, ‘the outcome of the war was already largely determined’.

At the tactical and operational level it was necessary to learn to use the tank. This required intellectual and organisational changes but the army was bad at this. E.g no new army-wide infantry training manuals issued by general headquarters from 1914 to 1917. Haig abolished all corps level schools. When a crucial improvement in artillery was made (the creeping barrage) it was confined to that corps for some time. Improvements by Rawlinson in surprise barrages ‘failed to touch the army as a whole’.

People close to the details understood how to use the tank but there was no way to get the senior levels to understand. Generals continued with artillery barrages before tank assaults, which were not only unnecessary but actually counter-productive (e.g. breaking up the land making it harder for them to move).

J.C.F. Fuller was a staff officer with the Tank Corps. He advocated the creation of a position on the Imperial General Staff with responsibility for tank warfare and channels for handling information about tanks similar to those for older weapons. He finally prevailed and had a senior job himself. He wrote ‘Plan 1919’ including how to deal with German countermeasures. Britain now had ‘all of the intellectual bases for blitzkrieg warfare.’ But the war was over.

The core problem was not getting funding for the original idea or even building prototypes. The main problems were:

1. The overall conception of how to win.

2. The measure of effectiveness to use.

3. The problems of intel collection and analysis.

4. The lack of organisational learning about how the tank could be used at scale and other organisational changes this implied.

CH5: New blood for the submarine force

During WW2 US sub force changed from targeting JAP battle fleet to raiding JAP merchant shipping to strangle economy.

Required big changes to officer corps and was effected by radically decentralised command structure with good intel.

Subs built pre-war were not built with Japan in mind (they thought they may be used against UK too!). No new type of sub nor significantly better weapon was introduced in the war. (Modest improvements in torpedo.)

The innovation was entirely in concepts of operation.

Peacetime training embedded some duff ideas and great caution.

Why did subs which trained to fight cautiously against fast moving heavily armed and protected warships struggle to target slow, unarmed merchant ships? HARD TO FIND. Even if they found a target, hard to stick with and engage.

Higher echelons wanted commanders to be more aggressive but very hard to make them be so! And hard to figure out what sort of characters would be aggressive. Just ordering them had very limited effect as they were autonomous.

Firing people was crucial to learning: in 1942 30% of sub commanders were ‘relieved for cause’ (fired), 14% 1943, 14% 1944..

In the first weeks of WW1, Joffre removed 9/21 corps commanders, 5/10 Calvary commanders. But Joffre stayed until 1917. And his replacement Nivelle stuck to same concepts and broke the French army as an instrument of offensive warfare. British army was similar — stability at the top, high turnover below. In US Air Force in Europe WW2, similar story.

Officer corps of pacific sub force similar - some initial changes at top then stability, turnover lower. What changed things? High command decided - patrol commanders had two patrols to prove themselves by sinking ships, then replaced. Many asked to be relieved or collapsed. The turnover did cause change.

There was no great rethinking of submarine war, commanders were just told to sink any ships or they would be relieved! And the success overtook policy makers at the centre because it was not properly analysed in real time. Intel gathering and analysis was patchy, the Atlantic stayed the focus, US did not know what was happening with JAP ship building.

Other decisions might have been made differently if US leaders had understood how effective the sub campaign against the JAP economy was.

CH6: US Strategic Bombing force 1941-45

During WW2 the US Air Force (USAF) developed 1) ability to provide fighter escorts for bombers, 2) ability to analyse enemy to determine targets that most undermined the military machine.

Fighter escorts

Effective long range fighters were thought unfeasible pre-war.

Douhet and Mitchell did not describe clear doctrine for mass bombing.

USAF thought of bombing civilians as illegal and counter to US foreign policy. There was almost no work on the idea of offensive use of long range bombers (a few individuals wrote about it). Bombing was thought of in context of defending US homeland.

B17 first delivered 1937. But not mass produced for offensive operations. In 1938 funds for B17 and other bombers were cut from budgets for 1940/1941. Only 14 B17s delivered by Sep1939.

When FDR set a target of 10k military aircraft in 1938, and 50k in Sep 1940, this paradoxically did NOT favour bomber production. Fighters were smaller, cheaper and easier to build. It took vigorous lobbying by Lovett (one of the famous figures ‘present at the creation’ in Acheson’s memoirs) to reduce FDR’s targets to make more bomber production possible. [Excellent example of how easy it is for centralised targets to have unforeseen damaging effects.]

Intel on enemy military was, thanks to Ultra, much better in WW2 than WW1. But targeting intelligence on the German economy and what really damaged the war machine was very poor at the start, took a long time to improve and was still faulty.

Our target intelligence, the ultimate determinate, the compass on which all the priorities of our strategic bombardment campaign … would depend, was set up only after we were actually at war. (Arnold)

[This implies that Britain also did not develop the necessary system despite the war starting earlier for us.]

The lack of intel reflected a) the defensive role assigned to bombers, b) no money for analysis of foreign industries and relation to war economy. There was therefore no developed measure of effectiveness for long range bombers.

Starting in August 1941 Allies started building rapidly such a system. 4 different entities worked on it, decentralised in competing groups: the Air War Plans Division of the Air Corps (AWPD), the Economics Objectives Unit (EOU) of OSS (London), a civilian-military group the Committee of Operations Analysts (COA) and British Ministry of Economic Warfare (MEW).

MEW did not adapt quickly to war. It did not develop the framework or tools needed 1939-40. The official history (Hinsley) says it ‘did not recognise that the study of those subjects were central to intelligence work on economic problems’. Britain struck at oil in 1940 on simple basis of ‘must be a good idea’ but could not measure whether effective or compare the effort made viz other potential targets. A model of German oil industry was built. The post-ward verdict was that it was ‘highly optimistic’.

AWPD developed the first plan in 8/41. No analytical framework for deciding targets. They chose electrical power, transport and oil systems. Largely accidental output of what info they could get their hands on, e.g some civilians pointed out that US banks had financed lots of German building so would have records and blueprints for power plants etc. Lloyds had info on factories they’d financed.

Little progress in US effort in first year of war, bombers aimed at U-boats because it was such a massive problem. COA created after a request by Hap Arnold 12/42 motivated by the need to have ideas for the Tehran Casablanca conference 1/43. COA analysis of oil disagreed with UK analyses. Instead of resolving the disagreement analytically, it was resolved bureaucratically by splitting the difference and assuming it was true!

EOU developed most comprehensive measure of effectiveness and most sophisticated innovation in targeting over winter 1942-3. The goal was defined as harming Germany war machine before invasion in 1944. Eisenhower rejected EOU’s suggestion of oil capacity because EOU said it would take effect three months after bombing. Then he changed his mind and in fact the bombing had immediate effect (p164). ULTRA immediately showed the attacks were working (though this was an unusually lucky intel success, not normal).

EOU analysis had been wrong, the targeting information had hindered the effectiveness of bombing. Similarly postwar work showed that wartime analysis of attacks on the ball bearing industry overestimated the effects.

ULTRA was valuable in all sorts of ways but Rosen says it revealed ‘surprisingly little’ about the economy and ‘provided very little help’ in assessing the impact of bombing. ULTRA targeted military traffic, not factory traffic (which was by telephone). It showed the value of bombing oil after we did it, but did not play a role in the decision to do it, nor did it provide similar intel for other industries.

Allies based lots of decisions on serial numbers (if you monitor serial numbers you can count the units) but starting in spring 1943 Germany started introducing random gaps. This made many assessments false but we did not realise until after the war. USAF wrongly thought they were winning the war against German air force by attacking its production base.

Arguably the best result of bombing was forcing the shift of roughly a third of German war production to anti-aircraft munitions.

Long range fighters

USAF doctrine before the first US missions in Europe was broadly supportive of the idea of fighter escorts but LRF were not developed. Why? Assumption: a fighter with the range long enough to make it useful for long-range escort would have to be a large, multi engine airplane, and would not be as agile as short range fighters that US bombers would encounter. RAF 1940 thought it axiomatic that short range fighters would be better and LRF couldn’t be built. USAF thought maybe greater defence for bombers plus massed formations to combine firepower might be ok.

The crucial innovation for LRF was the drop fuel tank: simple unpressurised external fuel tanks extended range and could be dropped when enemy fighters were met so they did not slow the aircraft.

They were procured by 1941 and used in 1942 in the Pacific but not Europe. Why? Because in Europe fighters would encounter Germans as soon as they hit the coast so would have to drop the tanks so fast it seemed pointless.

Germany could have made use of drop tanks impossible by sending up German fighters to challenge over the Channel, forcing them all to drop their tanks and depriving them of additional range.

Introduced in Europe July 1943. Because of heavy casualties as often stated? A mix. Heavy casualties in May did cause calls for drop tanks. 28 June 1943 Arnold issued an order: you have six months to get a fighter that can protect our bombers by Jan 1944:

Whether you use an existing type or have to start from scratch is your problem. Get to work because by Jan ’44 I want an escort for all our bombers from UK into Germany.

Drop tanks worked because ‘German fighters were pulled back from France and the Netherlands in order to protect central Germany. This withdrawal was not a gift from Goring, who opposed the shift. It was the result of the professional advice of the Luftwaffe senior officers after American bombers began bombing raids deep into Germany. According to Adolf Galland, the commander of German fighter forces, the strategy of using German fighters far forward for a peripheral defence of Europe was Goring's idea. That strategy, if it had been adhered to, would have prevented the effective American use of drop tanks. Galland and his Luftwaffe colleagues succeeded in reversing Goring’s strategy.’ The withdrawal from France began in autumn 1943.

[It’s unclear from this account whether Goring realised the point about fuel tanks and this is why he wanted to do it, or whether it was a lucky incidental element of a view he had for other reasons. Anybody know?]

Drop tanks were not used because of organisational learning in the sense of defining a new measure of effectiveness. ‘Drop tanks were adopted not because there was an understanding that they stood a better chance of solving the problem, but because something had to be done.’ Casualties plus desperation — we have to bomb for the invasion, so we’ll try anything. Arnold was going to bomb whether there was a new fighter or not. He was told a true LRF was impossible and he thought experience showed drop tanks would not work.

‘Bomber missions were conducted without full escorts, and this had the unanticipated effect of leading the enemy to give up the countermeasures against drop tanks… It was not the shock of combat losses that prompted this innovation but the dynamic combat situation that made innovation possible.’

CONCLUSION

A new measure of effectiveness was important for the tank. Intelligence was bad.

It was essential for the assessment of the new form of submarine warfare, but not of major importance for the initiation or development of that innovation. Intelligence was good but badly distributed.

It was at the heart of performing strategic targeting but not part of initiation. Intelligence was good re military operations but irrelevant to the problem of targeting.

LRF was born of desperation, not analysis.

All 4 of them took so long to develop they only kicked in late, unlike the shift to carrier groups.

Rosen: these examples suggest that ‘peacetime innovation ought to be pursued because wartime innovation is so terribly difficult’.

CH7: What is the enemy building?

In major cases involving the United States Army and Air Force 1930-55 the organisations responsible for R&D did not have good information about German and then Soviet military technology.

Unclassified information can provide insights regarding secrets, for example Paul Rosbaud (scientific advisor to German publishing house Springer Verlag) in the 1930s had access to much German research and he may have been responsible for the famous 1939 Oslo report on German R&D (see below). Scientists and engineers say that in the cases studied in this chapter they were ignorant of foreign military technologies except in so far as they were made aware of them through formal intelligence reports.

Declassified intelligence reveals a pattern of organisational behaviour in which the R&D community in peacetime and war was surprisingly divorced from intelligence about enemy technology, caused by difficulties in collecting and circulating intelligence. It changed in the middle 1950s but many key decisions about new land warfare systems in the interwar period and the first generation of long-range nuclear weapons appear to have been made with surprisingly little intelligence input.

From 1918 through the 1930s the Ordinance Department of the US Army was responsible for new weapons and equipment. It developed an intelligence unit. Its performance was poor. It was almost entirely dependent on the General Staff‘s military attachés ‘who had no technical background and who were not competent to collect or evaluate technical intelligence.’

These attachés had to meet the costs of diplomatic service from their own pockets! From November 1930 to May 1940 there were two technically qualified ordinance attachés in Europe, one in Moscow 1934 to 1940, one in London 1936 to 1940.

For example America lacked a competitive tank engine before World War II but the chief of the Ordinance Department asserted in 1938 that American tank engines were the best in the world. Heavy anti-tank weapons and heavy armour were successfully resisted until 1944. Small calibre anti-tank weapons were justified on the grounds they could penetrate the armour of US tanks but no data on foreign tanks and weapons entered the debate!

The Army’s official history was scathing on the intelligence failure. George Marshall testified in 1947 that prewar military intelligence was confined to what its attachés ‘could learn at dinner, more or less over the coffee cups.’

‘Liaison between the British scientific intelligence establishment and the American army was good in the area of electronics warfare against Germany. Where the US Army was not able to rely on British intelligence, the situation was abysmal… There was no central army intelligence organisation assigned to scientific intelligence at all until early 1943… Eventually a central office was created but it was limited to 3 officers who were not given access to Ultra intelligence until March 1945. This office had no long-term research background or files. It had no technically trained translators.’ Information was scattered in numerous offices. Security restrictions often blocked involvement of civilian expertise. Foreign technical publications were ignored.

Electronics warfare

Office of Scientific Research and Development (OSRD) spent $457 million on contract research to support the military from its creation in June 1941 until November 1945. $128 million was spent on radar and radar countermeasures, more than on any other category of research.

Before World War II US, German and British electronics intel was poor for different reasons. Louis Alvarez, a physicist working at the famous MIT Radiation Lab, said that in 1940-41 America had essentially no information on enemy development in radar. The director of the radar countermeasures program said of March 1942 they had very little information on German radar and ‘no idea at all whether the Japanese had even heard of radar’.

British intel on radar was wrong pre-war and did not recover quickly. Prewar German radar used the Very High Frequency band which had limited range so radar testing in Germany was not detectable in Britain. In autumn 1939 the British military attaché in Oslo received intelligence from an unknown source about German scientific developments which came to be known as the Oslo report. Among the capabilities disclosed was radar for early warning. The report however was not credible to the British military service intelligence branches. Even photographs of radar antennae on the wreck of the German battleship Graf Spee taken in June 1940 were insufficient to convince British military intelligence of the existence of German radar. (Jones stated before his death that he had identified the man who supplied the Oslo report (not the man identified in The Griffin) but refused to identify him on the grounds that he was still alive.)

German intel on radar was wrong pre-war and did not recover quickly. The Germans in summer 1939 flew specially equipped Zeppelins near a British research lab to investigate British radar. They drew the wrong conclusions.

Intelligence improved in Britain and America when the war began. Initial efforts to establish an institution in the defence community to provide intel to those designing and operating new military equipment encountered considerable resistance. RV Jones sought December 1939 to establish a central scientific intelligence section. It was initially opposed by the Navy. After extensive digging, including POW interrogation and inspecting German planes, Jones was able to persuade Churchill‘s advisor, Lindemann, and then Churchill himself to devote necessary men and equipment to search for radio transmissions used by the Germans. Invasion was expected within weeks. Five radar sites were dedicated to the search. They detected German beams and created a new section for the construction of electronic jammers and deception devices.

‘What is noteworthy is that despite the brilliant success of this combined intelligence and development effort, no immediate efforts were made to expand the collection of intelligence concerning German radar.’

Jones set up another group to work on German radar in order to help our bombing missions. The Bruneval raid successfully seized and returned to England a German radar. Jones pieced together a picture of the German defence line and developed offensive tactics and jamming equipment to attack its vulnerabilities. One of these was operational rather than technical: the division of the line into non-overlapping sectors, any one of which could be overwhelmed by streaming large numbers of bombers through it and overloading its capacity to handle enemy aircraft.

‘British intelligence efforts continued in an ad hoc fashion that nevertheless on occasion yielded impressive results… British electronics intelligence collection was not put on a firm organisational footing until December 1943.’

After the war ended Britain captured intact the German air defence command for Denmark and decided to conduct Operation Post-mortem: ‘a full-scale mock Bomber Command raid against the German system complete with the German wartime air defence staff to evaluate the effectiveness of bomber command ECM.’

1. British assumptions about how the system worked were very accurate.

2. The use of chaff caused the Germans to overestimate the size of the raid by a factor of 10.

3. While German equipment incorporated some useful antichaff features, those same devices were unusually vulnerable to jamming so the effectiveness of the combination of chaff and jamming was considerable.

The American government found that interrogations of German anti-aircraft gun crews, the Commandant of the German school for training those crews, and German radar scientists all estimated that ECM reduced German anti-aircraft effectiveness by 70-75%. This figure was roughly confirmed by the number of rounds required to shoot down an aircraft before and after heavy use of ECM.

There are many subtleties. For example many automated German systems could be spoofed by intelligent allied tactics. The Japanese proved relatively immune to deception efforts not because they were sophisticated but because their system was so rigid it tended not to react to anything including deception.

Britain shut down most of its electronics warfare establishment in 1945. After the Berlin crisis, there were discussions about rebuilding it. Despite his wartime service Jones was not used. According to Spycatcher [anybody got other sources on this??], a senior Security Service officer said ‘Jones has been making a play for the job [of chief scientist] but if we let him in, he’ll be wanting to run the place the next day.’ [Chimes with General Groves being pushed out postwar. People like that tend to be unpopular in normal bureaucracies.] Data was in the shortest supply 1945-51 then defections, new flights along Soviet borders to listen to traffic etc improved.

US intel immediately after 1945 was aimed at German tech, not Soviets. Hap Arnold asked von Karman (brilliant Caltech physicist) to survey S&T developments as a guide for air force R&D: ‘What I am interested in is what will be the shape of air power in 5 years, or 10, or 65.’ Von Karman arrived in UK April 1945. His team had >3 million German documents and did many interviews including with von Braun.

Von Karman’s report, Where We Stand (August 45): German radar had with some exceptions lagged that of America and Britain, but Germany had devoted more resources than the Allies to supersonic aerodynamic testing and modelling and had developed ‘arrowhead’ designs for wings that could reach and surpass the speed of sound without becoming unstable. This had an immediate and dramatic effect on American weapons. A Boeing engineer in von Karman’s team immediately prompted a complete redesign of Boeing’s plans which became the B 47, the first swept-wing bomber.

The report also projected:

• supersonic flight

• unmanned aerodynamic systems capable of delivering weapons payloads at ranges up to several thousand miles

• target seeking anti-aircraft missiles

• the need for supersonic offensive systems to penetrate the new anti-aircraft systems

• systems for communication between fighters and ground control stations

• all-weather navigation systems.

[Von Karman was a legendary scientist & for its accuracy and quality this report should be better known in SW1 as an example of what’s possible with ‘horizon scanning’.]

Stalin was kidnapping/recruiting German scientists. [He also had far better intel on US weapons including nuclear than the US realised for decades.]

1945-50 intel on Soviets was in short supply, there was not a big effort on Soviet radar, ECM, missiles, computers, air defence etc.

Vannevar Bush [also crucial to Manhattan] created the Weapons System Evaluation Group (WSEG) in 1948, responsible to the Secretary of Defence’s Research and Development Board and the Joint Chiefs of Staff. Its first task was to evaluate the capability of America to execute its air war plan against the Soviet Union. It had full access to JCS war plans and most (but not all of the most sensitive) intelligence sources. Philip Morse, physicist, had worked on intelligence on submarine warfare in WW2. He was the first WSEG director of research. He wrote that WSEG had shockingly bad intel at first.

Reported February 1950 (WSEG Report No1):

• It cast doubt on whether the B36 would be better at penetrating Soviet air defences than the B29 and the B50 it was to replace.

• Intelligence about the Soviet Union necessary for weapons system evaluation, particularly in ECM, ‘is essentially nil’.

• There existed nothing like the ‘Y’ intercept branch of British wartime intelligence created to supply data to the allies about enemy radar and radio systems. It concluded that ‘the present deficiencies in intelligence … seriously limit [Strategic Air Command’s] capabilities to execute a successful ECM campaign’.

• The intelligence failure extends to an accurate count of the number of fighter aircraft available to the Soviets.

• Because of the intel failures, WSEG judged SAC’s warplane by two criteria: a) assume Soviet air defence capabilities had the best features of allied systems, b) poor, no good early warning etc. Against the good version, WSEG judged that SAC bombers would succeed in dropping nukes against 80% of targets but would lose a third of their forces and would not be able to launch conventional followup strikes in the plan. SAC did not have the logistical strength to deliver their plan.

• SAC should rethink its war plans.

• The ‘grave deficiencies’ in intel be corrected.

CIA report to Truman August 1951 on Soviet ability to attack also shows very little intel on Soviet capabilities.

In the 1950s intel improved. We flew flights. Recruited agents. Intercepted comms. Russians defected.

Nov 1952 NSA created.

Herbert York: the intel available to the Teapot Cmte in 1954 came primarily from German scientists who had been taken to work on Soviet missile programs but were now returning to the west, and defectors. E.g Colonel Tokaev who said Stalin had told him he wanted missiles to reach America as ‘an effective straitjacket for that noisy shopkeeper Harry Truman’.

December 1953 intel report on Soviet rocket work.

Teapot Committee (von Neumann) reported in 1954: it would be possible to produce intercontinental ballistic missiles (ICBMs) by 1960 and deploy enough to deter the Soviets by 1962, that there should be a major crash programme to develop them, and that there was an urgent need for a new type of agency with a different management approach to control the project. Intel doesn’t allow firm conclusions on Soviets but there is evidence of ‘an appreciation of this field on the part of the Soviets and of activity in some important phases of guided missiles which could have as an end objective the development by the Soviets of intercontinental missiles.’ The possibility of them being ahead of us ‘certainly cannot be ruled out’.

The report did not assess Soviet air defences but vN stated his personal view that Soviets would strengthen defences sufficiently in later 1950s to warrant a US missile program of the greatest urgency.

[I wrote about the vN committee and its results, here.]

USAF sponsored two study groups, Vista and Beacon in 1953 and both highlighted the intel problem. USAF recommended new high altitude spy plane, which led to Skunk Works creating the U2.

1955 Killian reported to the President on the intel problem and a month later the government issued a requirement for satellites and other collection tech.

Project Sovoy tried to improve the access of analysts (including outside in RAND) to intel. Limited effect. Nobody at RAND had access to U2.

McNamara: the 1960s buildup of US nukes was bigger than needed to be because of bad intel.

Conclusions

This is a limited survey of 3 cases 1930s-55.

Rosen: 1) ‘US Army and U.S. Air Force strategic weapons development appears to have taken place in a context in which the research and development community did not have access to good intelligence about qualitative technological developments among the actual and potential enemies of the United States… Whatever drove US technological innovation in the cases discussed, it was not intelligence assessments of foreign capabilities.’ This reflected the problems of collecting and distributing intel.

2) Allied electronics warfare is a counterexample. They created new capabilities that reduced effectiveness of German anti-aircraft capabilities by 70-75%. Intelligence played a crucial role.

Rosen does not really explain why we did much better on EW. Was it because of unusuall capable/determined people like RV Jones? Was it because figures in the bureaucracy prioritised it more? Did senior politicians care more? Did senior generals?

CH8: Strategies for Managing Uncertainty

Royal Navy had failed to develop links to science world pre-1914 and therefore did not prepare for subs. In response the CID established 1915 the Board of Invention and Research (BIR) with civilian scientists. But RN refused to make available to the board its own subs for them to examine. RN blocked the production of a hydrophone. BIR was effectively crushed by 1918. Fisher (who had chaired it): ‘We are doomed to exasperation and failure by not being able to overcome the pigheadedness of Departmental Idiots… Never has the Admiralty Executive wholeheartedly supported the scientific and thoroughly practical proposals of the research.’

US Navy got Edison involved. Disaster. Edison: ‘I made about 45 inventions during the war, all perfectly good ones, and they pigeonholed every one of them. The Naval officer resents any interference by civilians.’

In WW2, Bush set up OSRD. He pushed for more influence for scientists. Military does not necessarily know what it needs unless it’s speaking to the people who can explain what may be possible. Scientists working in military labs pre-war were treated as a ‘lower caste’ and those labs generally failed, we need to preserve their independence. ‘Really striking movements away from the crowd will not occur once in 1,000 years in a tight bureaucracy’ (Bush).

Rosen: many of Bush’s own views on missiles etc turned out to be wrong.

1945 Bush:

Basically, research and procurement are incompatible. New developments are upsetting to procurement standards and procurement schedules. A procurement group is under the constant urge to regularise and standardise, particularly when funds are limited… Research, however, is the exploration of the unknown. It is speculative, uncertain. It cannot be standardised. It succeeds, moreover, in virtually direct proportion to its freedom from performance controls, production pressures, and traditional approaches…

To be effective, new devices must be the responsibility of a group of enthusiasts whose attentions are undiluted by other and conflicting responsibilities.

OSRD was a) outside military bureaucracy, b) run and staffed by civilian scientists, c) protected by Presidential charter but it itself hugely prioritised procurement over long-term research. [But this was wartime so inevitable, doesn’t mean a peacetime entity has to do the same, see Conclusions.] It did a good job getting scientists to the front line (Office of Field Services).

2/6/42 report by Dr Darwin to JCS committee: we need physicists involved because they have the best training in thinking about data and uncertainty.

Clear bureaucratic problem that entities resist things that seem to push against their core: e.g USAF resistance to ICBMs, USN resistance to cruise missiles. Bush made the same point — many technologies were neglected pre-WW2.

BUT lots of resistance is because military people don’t think the new ideas have value or will work — scientists are not best placed to persuade them.

Morgenstern & Knorr, 1965: amid such uncertainty, it is fundamentally important to conduct ‘an unremitting search for new military technologies, if only on the assumption that a resourceful or lucky opponent might be capable of making any discovery we are able to discern at the technological frontier.’ Invention and innovation are different, military has done neither well and should listen to scientists more

Randall (Birmingham University) developed a solution for microwave radar, the multicavity magnetron, demonstrated in 1939 and produced orders of magnitude more power than earlier magnetrons in America or Britain. The Tizard mission (1940) brought the multicavity magnetron to America and it was eagerly embraced by NRL. Randall began work on his device after he'd been recruited autumn 1939 into a consortium of scientists coordinated by the director of scientific research of the British Admiralty.

Conclusions

Rosen thinks that 1945-55 technological innovations in the US military were ‘largely the product of long-term development programs conducted within or at the direction of the military. Technological breakthroughs, such as microwave radar and the hydrogen bomb, did change parameters of innovations, but those breakthroughs were themselves products of programs designed explicitly to support the military, not of independent scientific investigations.’

While many errors were inevitably made he does not think that the pathologies of military bureaucracies blocking promising ideas was a major problem.

The USAF and Navy had a strategy viz missiles to manage uncertainty: they researched multiple technologies to the point of procurement but deferred large scale production while other uncertainties resolved themselves.

Overall technological innovation was ‘largely unaffected by the activities of potential enemies’. It was a ‘rather self-contained process in which actions and actors within the military establishment were the main determinants of innovation.’ It resembles peacetime and wartime innovations discussed earlier ‘which were only loosely related to enemy capabilities and intentions.’

‘If this picture is correct, military innovation as a whole is much less bound up with foreign military behaviour or civilian intervention than is ordinarily thought.’

[I think in the area I know best, ICBMs, Rosen’s argument is wrong. Von Neumann, Schriever and others pushed for a new team and new approach precisely because they thought the Pentagon and other bureaucracies could not deliver and they wanted to return to something like Groves’ management of Manhattan. Rosen does not explore ICBMs in detail. I think it’s also not just a question of ‘blocking’, it’s a question of speed. Saying ‘bureaucracy A did not block X’ is consistent with A delaying X so much it makes little progress, people leave to work on other things etc.]

Conclusions

Peacetime innovation has been driven by senior military leaders with traditional credentials reacting not to intelligence about specific capabilities nor the intentions of potential enemies, but to structural changes in the security environment, often using wargames/simulations, who changed the way officers and organisations lived their professional lives and promoted young talent.

‘Wartime innovation, rather than reform, has been most effective when associated with a redefinition of the measures of strategic effectiveness employed by the military organisation, and it has generally been limited by the difficulties connected with wartime learning and organisational change, especially with regard to time constraints.’ E.g the tank in WW1. This redefinition is hard and getting the right information is hard.

Overall wartime innovation seems less important than peacetime in the US.

Occasionally a decentralised capability to experiment in the field drives innovation, e.g electronics warfare in RAF WW2, but normally new concepts and conclusions from intelligence tend to be driven from the top/centre.

‘Technological innovation was not closely linked with either intelligence about the enemy, though such intelligence has been extremely useful when available, or with reliable projections of the cost of utility of alternative technologies. Rather, the problems of choosing new technologies seem to have been best handled when treated as a matter of managing uncertainty.’

Resources have not been the issue. Starting an innovation and developing it to the point it is strategically useful can be done when budgets are modest, even though then full development will be expensive. E.g Carrier transformation, US Marine Corps, RAF air defence all happened when budgets were modest. Counter-insurgency innovation did not happen even when budgets were vast.

Intelligence has been only ‘loosely connected’ in US. Too slow, too tightly held.

Aspects of the security environment outside US control have been more important. E.g emergence of US as a Pacific power.

Simulations/wargames have been more important than intel. E.g carrier war-games 1920s/30s.

Civilian military leaders have not had a major role in peacetime. Sometimes in wartime, e.g Lloyd George influenced convoys.

Civilian intervention in the officer corps can bring political factions into armed forces with disastrous effect, e.g France pre-1914.

Civilian scientists ‘did not play a major policy role in pushing the American military to develop capabilities it otherwise would have neglected’.

We need to:

• Get better intel on foreign leaders/organisations, their goals and perspectives

• Language training

• Imaginative simulations of future war to guid new capabilities

• Need to be able to build new things FAST

---

I will summarise my thoughts on this book after I review a few others on similar topics. I don’t agree with all his conclusions but I think it’s a valuable book.

I’d be very interested in comments from people at the sharp end, especially those of you now serving in MoD, armed forces etc who I know read this blog…

Also what are the very best things written since Rosen on these themes?

If you haven’t read it I highly recommend Alan Kay’s essay about how ARPA-PARC invented the internet and personal computing. Many people now talk about ‘DARPA’ but most of what you read is wrong, often very wrong. E.g DARPA now is different to the 1960s ARPA. In my view the UK should create both because both did valuable things. But the old ARPA was more valuable and if the UK is only going to try one model, it should be the older one that generated more value for the world.

There are three American institutions that were extremely successful and which we should learn from in setting up rough equivalents:

1. The 1960s ARPA.

2. The old JASON.

3. The old Skunk Works under Kelly Johnson.

I started 1 and 2 in 2020. 1 is happening but we don’t know yet how much real diversity will be allowed. 2 was officially agreed the day I left No10, and which I got the PM to sign off literally the last time I saw/spoke to him (no he didn’t know what he was signing) but I don’t know if it is now real or dead. I encouraged discussion about 3, and I know that serving officers have tried to push it, but I assume no progress. If you know re the status of 2 and 3 please get in touch.

If you find this interesting and/or useful, please consider a gift subscription…