Solynth Network · Explanatory note
A first draft written to be attacked, and the ladder it climbs when the attack arrives.
Most people meet the phrase first as a fallacy. Someone argues against a distorted version of another person's position rather than the position itself, swapping the real claim for a weaker or sillier one that is easier to knock down, and then knocking it down. The name comes from a dummy stuffed with straw. It looks like an opponent, but it cannot fight back. Say that the speed limit outside a school should be lowered and get told that you apparently want to ban cars altogether, and you have watched one being built. The victory is hollow, because the original claim survives untouched while the audience is left with the impression that it has been dealt with.
The second meaning inverts the first, and it is the one that matters in a working meeting. Here a straw man is a deliberately rough first proposal, put into circulation precisely so that people will attack it. It is a simple draft intended to generate discussion of its own disadvantages and to prompt better proposals in its place, usually written by one or two people before a larger piece of work begins. The flimsiness is the same. What changes is that it is announced in advance and offered as a service. In the fallacy, the weakness is concealed from your opponent. In the proposal, the weakness is the invitation.
A third term sometimes joins the confusion. Steelmanning, meaning to state an opponent's argument in its strongest and most charitable form before answering it, is the honourable opposite of the fallacy. It is a modern coinage that borrowed the metaphor, and it is not part of the history that follows, even though the word steel appears in both.
The clearest documented history runs through the American defence establishment in the middle of the 1970s. In 1975 the High Order Language Working Group was formed to deal with the sprawl of programming languages then in use across the US Department of Defense and the UK Ministry of Defence, with the aim of finding or creating one language suitable for both. The process began with a straw man proposal and ended, after many iterations, in the language that was named Ada.
What makes the episode useful is that the working group named each successive draft after the material it was made of, so that the hardness of the document was written on its face. The requirements were circulated widely for comment across the military and civil communities, and each round of criticism produced a more resistant version. The Steelman document says as much in its own introduction, describing the progression from Strawman through Woodenman, Tinman and Ironman to the version that finally stood.
| Document | Issued | What it was |
|---|---|---|
| Strawman | Apr 1975 | First rough draft, circulated to be knocked apart. Not widely published. |
| Woodenman | Aug 1975 | Rebuilt after the first round of comment. Holds a shape, still splinters. |
| Tinman | Jan 1976 | Requirements properly stated. Dents under objection rather than breaking. |
| Ironman | Jan 1977 | A serious specification, revised again in July of the same year. |
| Steelman | Jun 1978 | The version that stood, and the basis on which Ada was designed. |
Later documents in the same programme extended the metaphor into a second family of names, including Pebbleman in 1978 and Stoneman in 1980 for the associated support environment. One frequently repeated claim should be treated with more caution: that the expression was already in use in the United States Air Force before 1975. That rests on a single secondary citation and is worth verifying independently before it appears in anything formal. The ladder itself, which is the part people remember, is solidly on the record.
The practical case for the technique rests on an asymmetry in how people think. Ask a room to design something from nothing and you get silence, circling, and eventually a decision that goes to whoever spoke with the most confidence. Put a concrete but obviously imperfect document in front of the same room and the objections arrive within minutes, which is exactly what was wanted. A straw man converts a hard generative problem into an easy evaluative one, and it does so by spending one person's afternoon rather than eight people's morning.
It also brings disagreement forward. Arguments that would otherwise surface at the point of delivery, when changing course is expensive and reputations are attached, arrive instead at the point where the only thing at stake is a document nobody has committed to. That timing is most of the value. A dispute about scope in week one is a conversation. The same dispute in month six is a project.
Nobody is offended by a scaffold. The point of the label is to say, in advance, that this is a scaffold.
Calling a document a straw man is a formal declaration that the author is not attached to it and that attacking it is the intended use rather than a discourtesy. It protects both sides of the table. The author is spared the cost of being wrong, because being wrong was the plan. The critics are spared the awkwardness of appearing to undermine a colleague's work, because they have been asked to do exactly that. When the convention holds, a room can be brutal about a document and entirely civil about each other, and the two things stop being confused.
That is the theory, and the familiar response of how dare you propose such nonsense shows how often it fails in practice. It is worth being clear about what is usually going wrong, because it is rarely the document. Sometimes the reaction is honest confusion, where someone has read a draft as a decision already taken and is objecting to being presented with a fait accompli. Sometimes it is territorial, where the offence is caused less by the content than by the fact that somebody else drew the line first, and the outrage is a bid to take the pen back. And occasionally the offence is legitimate, because a proposal has been labelled a straw man while quietly remaining the author's settled preference, and everyone in the room can smell it. That last case is the one that does lasting damage, because after it has happened once the label stops meaning anything and the convention is spent.
What holds the practice together is that ridicule has a price. The room should require that anyone rejecting the draft says what ought to replace it, since a straw man that attracts only complaints has moved nothing forward and the author has taken the risk of going first for nothing. It helps to say the version number out loud, which is precisely what the Ada working group was doing with its naming. To call something the straw man is a public admission that several more versions are expected. The person who put the stake in the ground has absorbed the awkwardness of being first on behalf of everyone else, and the correct response to that is to move the stake, not to complain that it was planted.
The technique has one real weakness that deserves stating alongside its merits. Even a document everyone agrees is provisional will shape the range of options that get considered, because people tend to argue within the frame they are handed rather than replacing it. A badly framed straw man can therefore narrow a discussion while appearing to open one, and the group can end up optimising a bad idea very thoroughly. The defence is to write the first version loosely, to state explicitly that alternative structures are welcome and not merely alternative details, and to be suspicious of a review round in which every objection is about wording. When nobody attacks the shape of the thing, the straw man has probably done its job too well.