Move 1 · 18 to 6 months before
Listen the problem into existence
Conversations run long before a date is set, each answering the same two questions: what can we bring to the table, and what do we need from everyone else? Often the raw material is the previous gathering’s unfinished business, coded and turned into the next one’s problem statement. A researched framing paper is written from what surfaces, shared openly, and labeled a conversation starter rather than a finding.
Thematic analysis of prior roundsPre-calls by constituencyFraming paperKey questionsPre-gathering surveyOnline hackathon
Move 2 · 6 to 1 months before
Build the room on purpose
Streams are chosen by who is missing rather than what is topical — an invitation to those who don’t currently have a seat but should. People are then placed by declared conviction: in Seoul, every participant chose a problem in a pre-survey and found the assignment printed on their badge. Tables are composed in two modes, on purpose: mixed to cross-pollinate, uniform to go deep. Facilitators, hosts and scribes are recruited per table and trained repeatedly, months out.
Who-is-missing streamsPreference-based placementMixed vs uniform tablesFacilitator training ×4
Move 3 · Day 1
Teach people how to take part
The Istanbul forum opened not with a keynote but with ten minutes of interaction training. If you are going to ask 230 people across five languages to think together, you owe them the instructions first. Tirana spent its first five-minute slot on a question with no informational content at all — purely to get every hand on the polling tool before it mattered.
Interaction trainingLow-stakes first poll10 min frame / 20 min roomShared vocabulary
Move 4 · Days 1–3
Run to a bullseye, not a script
Every session is planned against four columns — topic, bullseye, tools, discussion questions — and the outcome is written before the questions are. The working shape is a sandwich: a poll to surface where the room actually is, a short input, table work producing one specific action, then a second poll on which action matters most. Tables submit one negotiated answer, not a scatter of opinions. Interaction runs through every session as a thread, never parked in an afternoon block, and seating is reshuffled so nobody spends three days with the same eight people.
Bullseye per sessionMax 5 live pollsGroup force-rankingTable turn after every inputReshuffled seatingNarrowing rounds 20→15→10 min
Move 5 · Every night
Capture without editorializing. Synthesize overnight.
Scribes record; they don’t moderate. A synthesis team works each evening and reports back the next morning — not a parade of table reports but one picture of where the room agrees and where it genuinely differs, which the room then corrects while it still can. Disagreement is kept as data: every table logs a consensus level, every facilitator report carries a tension line, and an open question wall runs beside the polls where anyone can post and upvote a challenge.
One scribe per tableOpen question wallConsensus level loggedOvernight synthesisSynthesis presents, room corrects
Move 6 · Day of close to 4 years after
Hand ownership out, not up
The convener doesn’t keep what the room made. A topic earns a team by attracting people willing to lead it — three committed leads at minimum, fifteen members, spanning career stage, way of working and location so no single institutional culture owns the problem. Six months on foundations, six on implications, two to four years on results, reporting twice a year, with a public commissioning so the handover is witnessed. In Seoul the vehicle was deliberately lighter: people committed to collaborative action, not to a committee.
Ground-up topic selection3 leads · 15 members6 / 6 / 24–48 monthsTwice-yearly reportsPublic commissioning
.png)
