# A rejected alternative, in six lines *A reusable format for the obvious thing you did not do. Take it, fill it, use it again — once per rejected alternative, written while the reasons are still in your hands.* --- ## What this is for **This is for the alternative everybody reaches for first.** Every built thing has one: the approach a competent person would name in the first thirty seconds, that you considered properly and did not take. The reasons were clear to you at the time. They will not be clear in six weeks, and they will not be available at all to whoever inherits the thing. So the alternative gets proposed again. By a colleague, in a meeting, as a fresh idea. By a later version of you, at two in the morning, as a fresh idea. By whoever takes the work over, who has no way of knowing that anybody looked. Each time it costs the same afternoon to re-answer, and each time it is answered slightly worse, because the person answering is reconstructing rather than remembering. **The ordering is the format's real work.** Reasons written as a list get read as a heap, and a heap of small reasons is arguable — somebody knocks two of them down and the whole thing wobbles. Written most-fatal-first, the reader can stop at the top. If reason one holds, the rest are courtesy. **And the last two lines are what keep it a record rather than a defence.** A note that ends in certainty has become advocacy for a decision already taken. A note that ends by naming the thing no argument can settle stays useful to somebody who disagrees with you. --- ## The format ``` THE ALTERNATIVE the obvious one, named the way its advocate would name it FATAL the reason that ends it, and it goes first THE REST the remaining reasons, in falling order of weight UNDERNEATH the one fact those reasons are all faces of SETTLED what you are sure of, and how you know UNSETTLED what only a trial would tell you ``` An alternative above, an honest limit below, and three reasons between. ### What each line is asking **THE ALTERNATIVE.** Named the way somebody who favours it would name it, in one sentence, with its actual appeal intact. **If your statement of it already sounds weak, you have written the rebuttal into the question** and everything below is theatre. The test is whether the person most likely to propose it would read your line and say yes, that is the thing I mean. **FATAL.** The single reason that ends it, first, before any of the others. Most rejected alternatives have one — a property they simply do not have, a thing they cannot be made to do. Put it at the top so a reader can stop there. **If you cannot identify which reason is fatal, that is a finding**: it means you have several partial objections and no decisive one, and the alternative may be live after all. **THE REST.** The remaining reasons, in falling order of weight, each one short. They matter less than they look, and their real function is completeness — they stop somebody thinking the fatal reason was the only thing you considered. Two or three is usually all there are. A list of seven is a sign that the fatal one is missing and volume is standing in for it. **UNDERNEATH.** The one fact the reasons are all faces of. Three or four independent-looking objections are usually one property of the world, seen from three or four angles, and naming it is the difference between a note and an argument. **It is also the line that transfers**: the reasons are about this decision and the fact underneath is usually about a whole class of them, so this is the sentence still worth reading when the specifics have gone stale. **SETTLED.** What you are sure of, and how you know — measured, documented, tested, or read in a specification. Be narrow. This line's value comes entirely from its being shorter than you would like. **UNSETTLED.** What only a trial would tell you. Almost every rejection has one of these hiding in it: a rate, a throughput, a failure frequency, a thing people would actually do. **Naming it is what stops the note becoming a defence**, because it tells the next reader exactly where they could overturn you and what they would have to do. A rejected alternative that has no unsettled line has usually not been examined; it has been dismissed. --- ## A worked one ``` THE ALTERNATIVE Identify people in a room passively, by listening for the radios their phones already broadcast. No app, no action, nobody has to do anything. FATAL Phones invent a new random radio address and rotate it, deliberately, so they cannot be tracked passively. In one room, sixty of sixty-one addresses seen were random, and a single phone produced seven of them. A passive listener identifies nobody. It counts anonymous, rotating beacons. THE REST Getting past that needs software on every phone, which is the thing the approach was chosen to avoid. One operating system suppresses that software's broadcast whenever it is not on screen, so it is unreliable on half the room. And the radio in question tops out well below the size of a class. UNDERNEATH All three are one fact. A phone randomises because it is asserting a right not to be identified from outside. So identity can only be had when the person asserts it — which makes the choice not between two radios but between counting anonymous devices and asking people to participate. SETTLED The randomisation, the background suppression and the connection ceilings. All three are documented behaviour and the first was measured here. UNSETTLED Throughput. How many people can actually be identified per minute in a live room is the kind of number only a trial produces, and the approach that was chosen owes exactly the same test. ``` Note that the last line concedes the strongest thing the alternative's advocate could say: the chosen approach is untested on the one axis that matters most. That is the note being a record. It would have been easy and dishonest to leave it out. --- ## How to use it 1. **Write THE ALTERNATIVE before you write any reason.** Written afterwards it acquires the shape of the thing you are about to knock down, and you will not be able to see that it has. 2. **Find the fatal reason and put it first.** If ordering the reasons is difficult, that is the format telling you something — go back and check whether you rejected this on one ground or on a mood. 3. **Do UNDERNEATH last among the reasons, and expect it to take longest.** It is the only line that requires thinking rather than recall. 4. **Keep SETTLED narrow.** Everything you cannot say how you know goes in the line below it. 5. **Write it while the reasons are still in your hands** — the week you decided, not the month you were asked. This is the whole reason the format exists. --- ## Three ways it goes wrong **It becomes a defence.** Every line leans the same way, nothing is conceded, and the unsettled line is missing or contains something trivial. The tell is that a reader who preferred the alternative learns nothing from it except that you disagree. A rejection note is written for that reader and for nobody else. **The alternative is stated in its weakest form.** *We could have just used the obvious cheap thing* is not the alternative, it is a caricature of it, and a caricature invites the real version to be proposed again next month by somebody who thinks you never considered it. **Four reasons, no fact underneath.** This is the commonest outcome of a first attempt and it is not a failure — it is the point at which the format has more to give. Four independent reasons for one decision is unusual in nature. Look again; there is almost always one thing being seen four times. --- *Prayas Abhinav. Licensed [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) — copy it, change it, build on it, including commercially. Say where it came from, and if you changed it.*