explorational · a divergent move

Accepting
the offer

opensoptions bysuspending fromarts & design within a group risknovelty for its own sake

The reflex meets every proposal with “no, because” — and the scene dies there. Improvisation’s one law: take what the moment gives as given, and add. What does an idea become if you are not allowed to touch the brakes for three moves?

Keith Johnstone spent years watching actors kill scenes, and named the killing move: blocking — anything that refuses or cancels what a partner has just made true. Offer: “It’s my mother’s ring.” Block: “No it isn’t — we found it.” The scene has nowhere to stand. His diagnosis in Impro was blunt: blocking is fear — of where the scene might go, of losing control of it. “Those who say ‘Yes’ are rewarded by the adventures they have; those who say ‘No’ are rewarded by the safety they attain.” Skill runs the other way: “Good improvisers seem telepathic; everything looks prearranged. This is because they accept all offers made.” Accept and add — treat the offer as given, and hand back more than you took.

Deliberation has the same killing move with better manners. A proposal lands in a meeting and the fastest reflex in the room answers “no, because —” inside ten seconds. The attached reason makes it photograph as rigour; the timing gives it away, because ten seconds is enough to judge only the doorstep. The transfer from the rehearsal room: before judging a proposal, extend it three moves — each one taking everything before it as given, and adding — then see where it was going. Evaluation deferred is not evaluation abandoned: it comes back with a return date, aimed at what the idea became. The bench below enforces the discipline mechanically — a detector that flags your reflexive “but” as you type, a gift you must name before you may add, and a judgment pane that stays shut until the third move is committed.

the apparatus

A yes-and bench with a real gate.

Type a proposal you would normally kill, or draw one of twelve seeds. Extend it three times; the block-detector reads your text as you type. Every counter on this bench is structural — flags, overrides, gifts, streak — computed from your words by the rules printed at the bottom. Nothing here grades the idea; that stays your job, and the bench schedules when you do it.

The yes-and bench

three accepted moves · a lexical block-detector · judgment on a schedule

the offer — the proposal as it arrives

move 1 · accept + addlocked — give the bench an offer first

Take the offer as true. Add material to it — an owner, a mechanism, a detail the next move can stand on.

move 2 · accept + addlocked — awaits move 1

Accept everything move 1 made true. Add again — no brakes yet.

move 3 · accept + addlocked — awaits move 2

Accept the world moves 1–2 built. One more add — let the idea arrive somewhere.

acceptance streak
0 clean moves
blocks flagged / move
— · — · —
overrides logged
0
gifts named
0 / 3

    evaluation — now, and only nowlocked — opens after move 3, not before

    • Where did the idea arrive?
    • What did it become that the first sentence wasn’t?
    • Which added part would you keep even if the whole dies?
    • Now judge — and if no is right, say no to the destination, not the doorstep.

    as it arrived

    after three accepted moves

    a worked run — one seed, three moves, one honest “but”

    What is counted, exactly. The detector matches these tokens with word boundaries, case-insensitive, curly apostrophes normalised: no · but · however · actually · instead · can’t · cannot · won’t · won’t work · the problem is — plus one positional rule: a move whose first word is no / not / never / none / nothing / nobody / unfortunately is flagged as opening with a negation. “Button” never trips “but”; “notably” never trips “no”. The detector reads words, not meaning: it will flag a quoted “no” that blocks nothing, and it will pass “we could simply do what I suggested originally” — a block in a party hat. Both errors are why the override exists, and why each override is logged with its reason instead of hidden. Commit gate: at least 6 words in the add (a floor, not a grade), a named gift, and every flag rewritten away or overridden with a note. Streak: committed moves, newest backwards, containing zero flags — an overridden flag is logged, not erased, so it still breaks the streak. Seed draws: a mulberry32(seed) stream, index = ⌊u·12⌋ — uniform over the twelve, so repeats are possible; the seed is shown, so the sequence is reproducible. Committed moves are frozen; the only way back is reset. That is what makes the gate a gate.

    blocking is fear

    Blocking often converts fear into judgment.

    “No, because” is the best-dressed move in any meeting. It arrives with a reason, so it photographs as judgment; it ends development, so it functions as a brake; and it fires before the proposal has produced anything, so the reason was generated against the weakest version that will ever exist — the first sentence. Johnstone watched the same move on stage for years and read it as fear: of where the scene might go, of being changed by it. The blocker looks discerning. The scene dies anyway.

    Acceptance compounds. An accepted offer becomes material; the next move has more to take as given, and hands on more still. Three moves in, a one-line proposal has parts — an owner, an exception, a season, a mechanism — and criticism finally has something load-bearing to grip. The trade is direct: the block prices the idea at its doorstep; acceptance lets it appreciate for three moves, then judges what it actually became.

    what to try

    Three runs on the bench above.

    01

    Extend a proposal you would normally kill. Draw seeds until one annoys you — the one you would shoot down in a meeting without slowing down. That reflex is the test condition. Three full moves, gift named each time, before you are allowed a verdict.

    02

    Watch the detector catch your reflexive “but.” Type your first extension the way you would say it aloud and count the chips. Rewrite each flagged clause as an add — and when one “but” is genuinely load-bearing, override it and write the reason. Either way the log keeps the record straight.

    03

    Read what the third move built that the zeroth couldn’t have. When the judgment pane opens, read the side-by-side before writing anything. Then judge — and if no is right, say it. A used no counts as the bench working, not failing.

    deferred is not abandoned

    The unlock is the thesis.

    The judgment pane on this bench will not open early; the gate is the page’s argument in mechanism form. Three is a schedule, not a magic number — long enough for an idea to change species, short enough that the calendar stays honest and deferral cannot slide into never. When the pane opens, it asks destination questions: where did the idea arrive, what did it become, which added part survives even if the whole dies. A no delivered there refuses a developed thing, for stated reasons — it teaches the room something. A no delivered at the doorstep teaches the room only who is dangerous to bring ideas to.

    the honest bench

    What the detector can and cannot read.

    The detector is lexical and rough, and says so: a short token list matched on word boundaries, plus one rule about opening on a negation. It cannot read meaning in either direction. It will flag a quoted “no” that blocks nothing; it will pass a sweetly worded return to your own original plan, which accepts nothing at all. Both errors are why the override button exists — “but” is sometimes the hinge that carves a real exception into a rule — and why an override is logged with its reason rather than waved through. A detector that claimed to know which “but” was load-bearing would be lying; this one asks you, and writes your answer down.

    Streak, flag counts, override log, gift count: structure, all of it, computed from your text by the printed rules. Whether the extended idea is any good is your call — made later, and better-informed, which is the entire point of the schedule.

    the move ↔ the bench

    What each part of the apparatus stands for.

    The moveThe bench
    the offerthe proposal as it arrives — one sentence old, undeveloped, at its cheapest and easiest to kill.
    a blockthe brake that ends development: “no, because,” fired at the doorstep before there is anything to judge.
    the addmaterial compounding — each accepted move hands the next one more to build with.
    the named giftthe receipt of acceptance: what, exactly, you are taking as given — acceptance made checkable.
    the three-move gatejudgment scheduled, not skipped — evaluation with a return date it actually keeps.
    the destinationthe thing you finally evaluate: the idea as it became, refused or kept for stated reasons.

    how this opening fails

    The failure modes to hold in view.

    risk · novelty for its own sake

    Three enthusiastic moves down a dead end.

    Yes-and is a compounding machine, and it compounds bad premises with the same cheerful arithmetic as good ones. Three moves of momentum feel like validation; they are only mass. The deferred evaluation exists precisely to be used — acceptance is a schedule for judgment, never a verdict on the idea. If the third move arrives somewhere worthless, the bench has still done its job: your no is now specific, informed, and cheap to deliver, because you know exactly what you are refusing. The failure is skipping the pane because the momentum feels too pleasant to end.

    risk · the mirror failure

    The team that may never say no.

    Where yes-and hardens from a practice into a culture — where “don’t block” becomes a ban on the word no — evaluation is not deferred but deleted, and the group becomes a compounding machine with no brakes on any floor. The move is accept-then-judge; remove the judge and what remains is a mechanism for escalating commitment to whatever was said first. Even this page’s detector, misread as a rule about words instead of a schedule for judgment, would produce exactly that failure — which is why the third pane exists, and why it asks you to say no when no is right.

    Say yes for three moves; the no you say afterward will finally know what it is refusing.

    can you use it?

    Three questions before you go.

    RECOGNITION — “Interns run the next retrospective.” Which reply is an accept-and-add? A: “No — they lack context.” B: “Yes, and actually this suggests reviving my facilitation-rota idea.” C: “Yes — and each intern pairs with a senior who takes notes but may not speak.”

    Answer

    C. It takes the offer as given and adds material the next move can stand on. A judges the doorstep; B is a block in a party hat — it accepts nothing.

    THE NEAREST NEIGHBOR — What separates this move from a culture where no one may say no?

    Answer

    The return date. Evaluation is deferred to after move 3 and aimed at the destination; delete the judge and yes-and becomes escalating commitment.

    PRODUCTION — “Let’s let customers vote on the next feature we build.” Your reflex is already drafting the “no, because.” Write move 1 — accept + add, gift named — before looking.

    One version + the check

    Gift: customers get a real say in the roadmap. “Yes — and the ballot carries three features we have already scoped, each with a one-paragraph cost note, so the vote is informed.” Yours works if it names what it takes as given, adds something the next move can build on, and touches no brakes.