Fog of war
A Socratic walk-through of the fog of war — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why do two commanders given the same information reach opposite conclusions?
Give two competent commanders the same reports — the same scout's sighting, the same intercepted message, the same silence from a flank — and they will sometimes order opposite things. One pushes forward, the other digs in. We are tempted to say one is simply better. But suppose both are able, honest, and reasoning carefully. What is it about the situation itself that lets the same evidence support contrary conclusions?
Reasoning it through
REASONING #Start with what a report actually is. Not a view of the battlefield — a sample. Someone saw part of something, from one angle, once, and described it in a few words. "A column moving north on the ridge road." Ask what that sentence rules out and the answer is: remarkably little. It is compatible with a withdrawal, a flank march, a relief of tired units, or a supply run.
That is the first ingredient, and it is not a defect of the reporting. Ambiguity is in the evidence. Once the evidence is ambiguous, something else must do the interpretive work, and the only thing available is what the commander already believed. If you arrived thinking the enemy exhausted and looking for a way out, that column is a retreat. If you arrived thinking he is massing for a blow at your left, the same column is the blow assembling. Neither is a misreading. The report does not decide between them; the prior does.
The second ingredient is time. A report describes a moment that has already passed — time to observe, to write, to travel, to be read among a hundred others. By the time it arrives, the commander is not commanding the battlefield but a picture of the battlefield as it was, and two commanders receiving the same facts at different lags are looking at different battles.
The third is that the enemy is not a landscape. He knows you are watching and can put things in front of you to be seen. Deception works best when it agrees with what the target already expects — a nasty interaction with the first ingredient, since the belief that makes you confident is exactly the belief an opponent can feed. The Allied deception before Normandy in 1944 succeeded largely because German commanders already thought the Pas de Calais the likely site; the false signals had only to confirm a belief, not overcome one.
Fourth, and least discussed: noise. Roberta Wohlstetter's study of Pearl Harbor made the point sharply — the warning signals were there in the record afterwards, embedded in a vastly larger volume of signals pointing elsewhere, and only hindsight makes the relevant ones stand out. Before the event, no report is marked as the one that matters.
Now the part that resolves the puzzle most cleanly, because it does not even require the two commanders to disagree about the facts. Suppose both think an attack on the left unlikely but possible. One holds a spare reserve and a left flank whose collapse would be fatal; the other would give up his own offensive by moving. Their costs of being wrong differ, so their thresholds for acting differ. Identical beliefs, different stakes, opposite orders — and both correct.
The analogy
THE ANALOGY #Think of two doctors reading the same shadow on the same X-ray. Neither is confused about what is on the film. But one works in a clinic where the disease is common and the treatment is mild, and the other where it is rare and the treatment is dangerous. The first orders treatment, the second orders another test. What separated them was not the image but what each brought to it and what each stood to lose.
The X-ray holds still and is not trying to fool anyone, whereas a battlefield changes while you read it and contains an opponent actively arranging what your scouts will see — so a commander's picture can be both stale and deliberately poisoned, which no radiograph is.
Clarifying the model
THE MODEL #The phrase "fog of war" is a later shorthand; Clausewitz wrote of war as the realm of uncertainty and of the friction that makes even simple things difficult, and the shorthand has since collected a mystical air he would not have wanted. Strip it back to the four ordinary mechanisms above — ambiguity, delay, deception, noise — because each has a practical countermeasure, whereas "fog" invites fatalism.
A second correction. It is easy to hear this as "the commander's beliefs bias him", with bias as a personal failing. But he has to bring a prior; evidence this ambiguous cannot be read without one, and a commander with no expectations would be paralysed rather than objective. The prior is not the flaw. The flaw is holding it so tightly that no report could overturn it, or forgetting that the enemy may have chosen what you are looking at.
And a caveat about judging afterwards. Hindsight deletes the alternatives: once we know the attack came on the left, every report consistent with it looks like a warning ignored, and the far larger number pointing right becomes invisible. Many events filed as intelligence failures were reasonable readings of a genuinely ambiguous record. Some were not — the distinction is real, but far harder to draw than the retrospective narrative suggests.
A picture of it
THE PICTURE #How to readRead top to bottom as elapsed time. The enemy generates two streams at once — one real movement and one manufactured for the scouts to find — and both arrive as a single report. The two notes are the crux: the same line of text leaves the commander's desk as an order to pursue or an order to hold, depending only on the belief it was read against. The last exchange shows the loop failing to close, since the check that would have corrected the reading cannot be answered in time.
What became clearer
WHAT CLEARED #Two commanders diverge on the same information because information is not a conclusion. Reports are ambiguous samples, they describe a past that has moved on, some were placed for you to find, and the few that matter sit inside a mass that does not. Whatever bridges that gap comes from the commander — his prior about the enemy, and his private accounting of what being wrong in each direction would cost. That is why the fog is not removed by better sensors, and why command remains a judgement about which mistake you can afford rather than a search for the facts.
Where to go next
ONWARD #- How intelligence organisations use structured techniques — red teams, competing hypotheses — to keep a prior from becoming a filter.
- Why faster reporting can, in some conditions, make the picture worse rather than better.
Key terms
TERMS #| Term | What it means |
|---|---|
| Friction | Clausewitz's term for the accumulation of small difficulties that makes the simplest action in war hard. |
| Prior | the belief a commander holds before the evidence arrives, which determines what ambiguous evidence appears to say. |
| Signal-to-noise problem | the difficulty that warning indicators are indistinguishable, in advance, from the far larger volume of irrelevant reports. |
| Deception | deliberately presenting an opponent with observable evidence, most effective when it agrees with what he already believes. |
Every term the collection defines is gathered in the glossary.