Thirteen patterns that decide a battle

Read a battle for who was great and all you get is admiration; nothing is left that you can use. So this site looks at who held which premise, and at which moment what gave way. Do that often enough and the same shapes start appearing across cases. Those shapes are the patterns.

But a pattern is not a template. Later commanders who copied the shape of Cannae without its conditions sank into a quagmire. The use of a pattern is not to ask "which pattern is this situation?" but "are the conditions this pattern needs actually present right now?" Hence every card below carries the conditions under which it does not work.

Line the cases up and, quite apart from the winning patterns, the same failure keeps reappearing: someone reads the conditions correctly, and the structure of the organisation crushes that judgment.

• Kutuzov (Austerlitz) — saw that time was on his side and argued for waiting; a young emperor and a hawkish court overrode him.
• Paullus (Cannae) — wanted to avoid a battle on the plain; a system in which two consuls held the whole army on alternate days cancelled his caution. He died on the field.
• Gravina (Trafalgar) — argued against sailing on account of the weather. As one of two partners he had standing to object, but not the weight to reverse the decision.
• Dumitrescu and Paulus (Stalingrad) — the weakness of the flanks was reported and permission to break out was requested. There was no reserve, the estimate above was wrong, and neither carried.
• Boucicaut and d'Albret (Agincourt) — they brought the lesson of the last defeat, planned to fight dismounted, and raised the option of not fighting at all. But with the king and the Dauphin absent, a council of princes who outranked them would not carry caution. Even the formation was shaped by precedence rather than tactics.
• Gérard (Waterloo) — hearing guns to the west, he urged his superior Grouchy to march to them. The reading was right; the structure of an order that stated no purpose would not carry it.
• Pemberton (Vicksburg) — the department commander said abandon the town and save the army; the president said hold the town. When two superiors say different things, the man on the spot takes a little of each and loses both.
• Ishida Mitsunari (Sekigahara) — the operational plan was sound, but as one of five commissioners he had no authority to guarantee anyone's land. When the man commanding has no standing to command, the army he has gathered is a number and nothing more.
• Rundstedt and Rommel (Normandy) — holding the reserve until the main landing was identified, and stopping a landing at the water's edge, were each militarily sound. But the input to the judgement was manufactured by the enemy, so a correct rule kept returning the answer "not yet".
• Westmoreland (Tet) — he built a measure of progress, sustained it, and threw the offensive back in weeks. But the scoring was not being done on his yardstick. Attrition measures how much the enemy will absorb; the same clock is counting your own limit too.
• Lord Halifax (Dunkirk) — with the army in a pocket and the equipment certain to be lost, he argued for exploring terms. This one did not lose to a structure; it lost to another reading. And the side of a closed option can never be tested.
• The German Crown Prince (Verdun) — pressed early, from the front headquarters, for the offensive to be broken off. He read it correctly. But in a plan with no stopping condition, the burden of proof falls on whoever wants to stop. That is not a matter of logic; it is a matter of what the document contains.

Twelve of the thirteen cases show the same shape. That is not coincidence; it is a pattern. Reading it right and being able to make an organisation act on it are different abilities. And this is the pattern most likely to happen to you.

Every case has a story that became more famous than the events. Each article on this site carries exactly one correction. The correction is not the point — an explanation of the form "one battle decided everything" or "it was bad luck" pushes the things that actually decided it out of view.

• Austerlitz — thousands said to have drowned in the frozen ponds. When the ponds were later drained, far fewer bodies were found.
• Cannae — the idealisation of the double envelopment. Commanders who copied the shape without the conditions sank into a quagmire.
• Midway — the "fateful five minutes". Against the Japanese primary records, what stood on the decks was mainly the defensive fighter patrol, not a spotted strike.
• Trafalgar — "it stopped the invasion". The invasion army had broken camp at Boulogne in late August, before the battle. The order is reversed.
• Stalingrad — "General Winter did it". The ring closed on 23 November, and both armies were fighting in the same winter.
• Agincourt — "the longbow pierced plate armour". Neither recent scholarship nor shooting trials support it. Some arrows got through, but what told was less killing power than pushing men toward the centre and raising the density. And the ratio of forces is itself unsettled among historians, so even the premise of an overwhelming disparity is unproven.
• Waterloo — "a near-run thing". That verdict only holds from inside the battlefield. The Prussians were closing by four in the afternoon, and from that hour the French reserve was already being cut into.
• Vicksburg — "the turning point was Gettysburg". They fell in the same week (3 and 4 July), and it was the second that halved Confederate logistics permanently. What stayed in the memory was the one that made a better drama.
• Sekigahara — "Kobayakawa's betrayal decided it". The negotiations were under way before the battle, and the 30,000 on Nanguyama did nothing at all on the day. "Betrayal" is a word for compressing months into a day.
• Normandy — "rubber tanks fooled the reconnaissance". By 1944 the Luftwaffe could barely photograph England and most of the decoys were never seen by anyone. What worked was the clerical traffic of double agents. A photogenic device took the memory slot belonging to the one that worked.
• Tet — neither "America lost" nor "the war was lost by the press" stands up. The mistake both share is assuming that winning can be measured on one scale. This battle had three scorecards — loss ratios, ground, and the patience of the American electorate — and only the third had any authority.
• Dunkirk — "a miracle in which little ships lifted an army off a beach". 98,671 left from the beaches and 239,555 from the East Mole, and the preparation began on 20 May. But this correction has a second half: that story supplied the reason that held up the decision not to stop.
• Verdun — the claim that Falkenhayn planned to bleed France white. The Christmas Memorandum said to prove it has never been found, and the only source is his own memoir of 1920. The Prussian army archive burned in 1945. This correction is aimed at our own method: there is a permanent risk of mistaking a blueprint drawn afterwards for one drawn before.

The corrections come in five kinds: inflation (an event swells), omission (an event drops out of view entirely), compression (months are folded into one day’s decision), substitution (a photogenic cause takes the seat of the effective one) and units (the result is argued about while nobody has agreed what counts as winning). What they share is one thing: credit everything to a single day or a single factor and the years that built that day disappear. And the correction does not stop at the correction. After “that is not what happened”, write “then what was that shape for” — the Dunkirk case is the worked example. The rules for this are in the editorial policy.

These patterns are not boxes for classification. Use them by asking, in this order:

• What does the opponent consider rational? — space, momentum and attachment are the patterns that make his correct judgment a component of yours.
• What does his procedure assume? — if it assumes uninterrupted time, the design of time applies.
• Do his numbers and resources have room to spread? — if they do not, the design of frontage applies.
• Is there strength that can still arrive? — if there is, the design of convergence comes first.
• What line does he depend on? — if he depends on one, the design of supply applies.
• How much of his number will actually obey? — if there is a gap, the design of defection applies.
• Through what channel does he learn his own situation? — if you can reach that channel and read his reaction, the design of deception applies.
• Who finally rules on who won? — if it is not one of the people fighting, work out the design of the verdict first.
• If this is lost, what gets abandoned and what comes home? — too late to ask once you are losing. A design of what to lose can only be written while you might still win.
• Can my own command survive the shape that settles it? — if not, the design of command has to come first.
• Am I currently inside the conditions under which this pattern fails? — that is what the "fails when" line on each card is for.

More patterns do not make a better reading. Adding one is only worth it when the case does not fit any pattern already here. Which is why the next case on this site is chosen by whether it is a missing pattern, not by whether it is famous.

• About this site
• Editorial policy and sourcing
• Contact