The B-17 campaign — organised, planned, executed, monitored, re-adjusted

The society elects a Managing Committee — President, Vice President, General Secretary, Finance Secretary (Treasurer), Information Secretary and a bench of Executive Members — in a single, society-wide, in-person vote. Below is the official qualified field, transcribed verbatim from the Election Commissioner's scrutiny order. The Multi Founder Group slate is highlighted.

The roll holds 28,186 members. The single most important fact for strategy: 42.3% of members live outside Islamabad/Rawalpindi — in Peshawar, Karachi, Lahore and overseas — and must travel to vote in person. That absentee gap, not persuasion, is what suppresses turnout.

Election OS runs the campaign as a dependency-aware workflow of 18 steps across six phases — a guided setup that only gates what genuinely must come first and surfaces everything that can run in parallel. Below is the whole lifecycle, from an empty account to the polling-day board, mapped to the five things a campaign actually does: organise, plan, execute, monitor, re-adjust. Every step here is a real feature, driven on this data — seeding the case study exercises the same write-paths a live campaign uses, so building it end-to-end also verifies the stack. Seeing is believing.

What that looked like on this race — the numbers seeded end-to-end

Seeing is believing — the tools, on this race

Before any modelling, the panel scores every candidate — ours and the rivals — 0–10 on six attributes (name recognition, standing, organisation, incumbency, effort, likability), and writes a SWOT on the President race. This is what lets the simulator decide the contest on judgement, not a coin-flip. On this race the Multi Founder Group grades out ahead on the two attributes members weight most — organisation and campaign effort.

SWOT — the President race

These are real runs of the simulator on the President race in the competitive regime (one consolidated rival panel — the only case where the seat is in doubt). Choose a strategy:

The war-game says what matters; the decision engine turns that into what to do. Two tools compose: an AHP decision ranks the four closing-week programmes against weighted criteria, and the convergence engine folds SWOT × voter-salience into a ranked strategy board that feeds straight back into the simulator's levers.

A cross-audience poll reads the room — and states its own provenance honestly, so a poll of your own base can never masquerade as the electorate's voice. This one: 72 responses across voters, supporters and canvassers.

No projection is a single certain number, so instead of hiding the assumptions we vary the two that matter and show the win probability for every combination. Each square is the President seat's chance of winning under one pairing of our support level and the opposition's shape.

How to read it: find the ~42% support row (the plan), then scan across. It turns green the moment the opposition is split — which is why the campaign spends less on chasing more support and more on making sure at least two rival candidates stay in the race.

Strip away the percentages and the mandate is concrete: turn 3,753 valid votes into 10,695 — nearly a 3× lift — by combining persuasion (share 40% → 53%) with turnout (35% → 74%). Rejected/invalid ballots (~3% of a manual, multi-post paper vote) are modelled explicitly.

Projection at 73.6% turnout — 20,757 ballots cast at the top of the ballot (President: 20,134 valid, 623 rejected). Each post is counted separately: down-ballot roll-off thins the later races (−2 pts of participation per position), and the voters who persist skew a little more to the organised slate — so the five tallies differ, as they do on a real count sheet. Figures are projections from the war-game, not results.

Executive Members — 7 seats, plurality-at-large

Every voter marks up to 7 names; the top 7 win. This is not one fixed number per candidate — we run the count as a seeded Monte-Carlo (1,200 trials): each candidate carries an unknowable personal vote (name recognition, ballot position, who canvassed for them), modelled as variation on their strength (0–10). So each candidate gets an expected mark, an 80% range, and a probability of winning a seat — the honest output of a race not yet run. Strength today reflects the one public fact we hold about real people — organised-slate backing — so the four Multi Founder Group candidates are favoured (~93% each), and the last 3 seats are a genuine toss-up across an evenly-matched field (~30% each). This is exactly where an expert panel's ratings decide it: score stature, standing and campaign effort (AHP/SWOT) and the contested probabilities move.