People, ideas, machines VI: the War Diaries of Alanbrooke, Chief of the Imperial General Staff
A largely unknown British hero, another case study in the unrecognised simplicities of high performance (and relevant to those thinking about nuclear war, AI-alignment etc)
‘People, ideas, machines — in that order!’ Colonel Boyd.
‘Just like all British governments, they will act more or less in a hand to mouth way on the spur of the moment, but they will not think out and adopt a steady policy.’ Earl Cromer, 1896.
‘Every wind that blew swung us like a weathercock. As I was to find out, planned strategy was not Winston’s strong card. He preferred to work by intuition and impulse… God knows where we would be without him, but God knows where we shall go with him.’
Alanbrooke on Churchill, 1941.
‘By general verdict, fearless, formidable, articulate, and in the end convincing.’
Churchill on Alanbrooke
‘It is a most fascinating pastime to follow a great man’s thoughts.’ Pushkin.
[UPDATED 2/2, scroll to 9/8/42]
I've written a few things about the unrecognised simplicities of high performance. E.g In 2016-19 I wrote about ARPA-PARC (and inter alia how lessons from this success could avoid EU procurement law killing people in the next big crisis which it did in 2020). In 2021 I wrote about Lee Kuan Yew. In 2022 I wrote about management at Amazon.
Here is another in this series.
Alanbrooke is one of Britain’s great heroes though not nearly so famous as Nelson or Wellington. Most involved in politics talk a lot about ‘strategy’ but know little or nothing about this crucial strategist of WW2. Other WW2 generals such as Montgomery are much more famous. Alanbrooke was promoted by Churchill to Chief of the Imperial General Staff (CIGS) in the dark days of November 1941 as the Nazis approached Moscow. His nickname was ‘Colonel Shrapnel’. His formidable character burst with energy yet he also held himself in incredible control — true leaders, he thought, had to preserve and project self-control (to a degree that would be regarded as ‘unhealthy’ now). He repeatedly deflated Churchill and others, sometimes snapping a pencil between his fingers as he said ‘I flatly disagree’. His relationship with Churchill was stormy. Alanbrooke deeply admired him but the responsibility fell on Alanbrooke, more than anybody, of dealing with Churchill’s flaws and the dangers they risked. In replacing Dill with Alanbrooke, Churchill wanted someone more vigorous to prosecute the war and he got it.
When I thump the table and push my face towards him what does he do? Thumps the table harder and glares back at me. I know these Brookes — stiff-necked Ulstermen and there’s no one worse to deal with than that!
It’s greatly to Churchill’s credit that he appointed him and stuck with him despite extreme disagreements and passions exploding amid the incredible tension of world events and their vast stakes, but it’s also much to his discredit that his memoirs vastly underplayed Alanbrooke’s contributions to victory.
In Alanbrooke’s diaries are reflections on the most basic and important questions of politics and war.
How to define aims?
How to make strategy?
How are aims, ways and means — and strategy, operations, and tactics — related?
How do politics and war affect each other? (E.g the huge political pressure for a western front long before it could be properly supported.)
How much do personalities, ideas and institutions shape crucial events?
How do/can democracies fight closed undemocratic societies?
How to cope with the poor understanding of specialist subjects among MPs? (NB. this was a serious problem 1939-45 when the Commons was incomparably better equipped to discuss military issues than now.)
How do committees work, especially the Chiefs of Staff Committee and the famous committee known as Cabinet (his sometimes despairing descriptions of Cabinet discussions are priceless and remind me vividly of watching Cabinet in 2020)?
What are the iron laws of bureaucracies, even when stakes are existential?
How to think about military/security spending as insurance during peace, what happens if you fail to pay this ‘insurance premium’?
How do alliances work, what causes tensions, what eases them?
What are the unrecognised simplicities of high performance?
The decline and fall of countries and empires.
Alanbrooke was one of those extraordinary people thrown up by war to senior roles who then almost always disappear when the war ends and discussions at the top revert to the norm — avoiding hard questions. It’s as if normal governments can’t deal with such people unless the stakes are existential. General Groves was another such extraordinary character who ran the Manhattan Project and was pushed out of the Pentagon after 1945 for being ‘too difficult’.
After the war Alanbrooke was given some honours and titles but struggled for cash. This hero, who had made such huge contributions to victory, had to sell his house, move into the gardener’s cottage, and sell his beloved bird books for cash. This seems, like the relative funding of Turing and von Neumann for computer research after 1945, to say something profound about Britain’s rapid decline after 1945.
While children and students are told they study history ‘to learn from the mistakes’, one of the most fascinating and striking things about our world is how little learning there is from the greatest of errors. You can read analyses of deterring Prussia/Germany in Whitehall that are practically identical and indistinguishable from 1866, 1870, pre-1914, and the 1930s. You can read history after history of war after war. Human nature and the dynamics of large organisations don’t change so the same problems recur from the start of written history, and nobody can find a way of creating institutions that surmount these problems for long. You may reshape the Prussian General Staff and then reshape the map of Europe but before you know it, you’ve gone from the Elder Moltke working with Bismarck in triumph to his nephew imploding in disaster. One minute Bill Gates; the next, Steve Ballmer. Everything has its time of growth and decay. This means that we stumble into disaster after disaster where the details change but the fundamental patterns don’t. Covid and Ukraine are just the latest examples.
On one hand, we can see abstract principles of high performance a) recur constantly in written history we can all read, b) are extremely simple and do not require high intelligence to understand, and c) when deployed are frequently shocking, even world-changing, as well as sometimes bringing power, wealth and fame to those who deploy them.
But on the other, these principles remain essentially unrecognised by roughly 100% of institutions of all kinds, private and public. Instead, roughly all normal large organisations actually optimise for the opposite principles and promote those who embody these anti-principles. If there is some occasional high performing blip (such as PARC or the Vaccine Taskforce), these normal organisations, and particularly the middle managers within them, will move swiftly to close them and push away those responsible as far and as fast as possible.
This reliable feature of our world has many effects. One of them is that government systems are essentially ‘programmed’ to be slow and inefficient in updating. All institutions optimise for certain things based on incentives and culture. Large established organisations almost always optimise for ‘protect established power networks’, not ‘update useful information even if it disrupts established power networks’. And this means that the old parties, old bureaucracies and old political media are necessarily constantly surprised by events far beyond what they need to be — beyond the inevitable surprises generated by the uncomputable complexity of the world. The most valuable information is and will be almost always at the edge. Elite self-deception was critical to the context of WW2 and looking over the centuries it seems only safe to assume it’s a permanent state of affairs, at least without revolutionary experimentation with institutions that optimise differently and will (we should assume) in their own ways be as, or even more, dangerous. (E.g Increasing the extent to which institutions optimise for better people will also increase the dangers inherent in better people. There is no free Bismarck-scale-winning-without-dangers-of-Bismarck-scale-reward-hacking lunch — at least, not until the alignment problem is solved!)
Is learning in politics/war doomed? No. Just extremely hard, unlikely and it won’t last longer than a few decades at most — when the odd people are replaced, the new ones very likely mean normal performance, and the crucial learning from wars mostly die with the generation. If you study these recurring patterns and have some measure of self-reflection and are willing to make sacrifices (to be or to do?), you can bend history briefly, knowing that your opponents almost definitely could read the same crucial stories as you but almost definitely won’t apply them. This is why I call them the unrecognised simplicities of high performance (pinching the phrase from Charlie Munger on investing).
But there are always costs to choosing ‘to do’ over ‘to be’, and the currency is fear, hatred, jealousy, revenge, poverty, social failure and so on. Colonel Boyd was a hate figure to many and died in poverty. If it were easy, the world would look very different! This is why so many talented people now prefer startups to large companies or government institutions. Startups are a way to do something interesting and important (to do) while often dodging the crippling social and financial penalties of attacking large organisations. It’s safer and better paid to be Steve Jobs than Colonel Boyd. Unfortunately, while startups can solve many problems and could solve more government problems they cannot solve the most important such as intelligence services, WMD and so on.
The goal of these notes is to look for further examples of the unrecognised simplicities seen repeatedly in history and evidence relating to the questions above. I think these diaries will be valuable for others thinking now about existential dangers. Even when people of the stature of FDR, Churchill and Alanbrooke are engaged in a problem as important as *trying to defeat Hitler before he acquires nuclear weapons*, certain types of problem recur and it seems wise and prudent to assume that similar problems will recur for those dealing now with, for example, nuclear escalation or ‘how to develop a realistic plan for AI-alignment so AIs don’t create extreme destruction’. Such problems have crucial hard technological aspects but running in the background are all-too-human psychological aspects combined with all-too-predictable bureaucratic dynamics that suppress sensible thinking, put exactly the wrong people in charge of critical things, empower the most self-interested power-seeking people and so on.
Keep reading with a 7-day free trial
Subscribe to Dominic Cummings substack to keep reading this post and get 7 days of free access to the full post archives.