Case: the deliverable list that doesn't fit

The first question for any deliverable list is whether the musts alone fit the capacity: the rule from Must, should, could that musts sit clearly below it. Answer it with arithmetic, not impressions.

Add the likely values of every line the owners marked must. Likely, not optimistic: you plan on what usually happens. On the physics list, 19 lines are marked must: curriculum 39, practice 33, tutors 23, test 23, web 26, the launch email 3 and project management 12, together 159 person-days. Capacity, the plan's own assumption, is 150 person-days between 2 November and 29 January, counting staff, freelance authors and the illustrator.

So even the musts are 9 days over. That number is the start of the conversation, not the end: before anyone reaches for the date, check whether every “must” really is one.

When the musts don't fit, the tempting conclusion is Leo's: “then it's the date after all”. Before that, run the must test on every line marked must: can we open on 1 February and reach the goal without it? Owners mark their favourite items must in good faith, so expect a few to fail.

On the physics list two do. D11 video solutions for the 40 hardest tasks (14 days): tutors still explain hard tasks in class. D22 parent progress dashboard (14 days): parents ask the tutor. Both are useful later, so they move to could, on the after-launch list.

Others look optional but pass. D16 diagnostic test: without it learners can't be placed into packages. D06 final curriculum review: it's what makes the curriculum acceptable. D23 launch email: without it the families who asked last spring never hear about physics. Judge by the goal, not by the size or the cost of the line.

Recount after the must test. Removing D11 and D22 (14 likely days each) takes the musts from 159 to 131 person-days for 17 lines, against 150 of capacity. That leaves 19 days for shoulds and coulds, and for anything that slips.

That margin is the number Leo needs: the date isn't the problem after all, scope was. The two false musts, together 28 days, were exactly what made the plan look impossible.

Keep the arithmetic visible in the decision note: 159 proposed, two items failed the test, 131 real musts, 19 days left. A sponsor who sees the sum can sign without a meeting; one who only sees “we cut two items” will ask why those two and why not others. Numbers from the list also protect the owners: Vira's and Danylo's items moved because of a test everyone can repeat, not because someone liked them less.

Each line on the list has three values. Summed, the 17 real musts run from 94 person-days (every line optimistic) to 193 (every line pessimistic), with a likely sum of 131. The temptation is to pick an extreme: 193 “to be safe”, or 94 “because the team can push”.

Both are wrong for the same reason. All lines going badly at once is unlikely, and so is all going well. Some tasks will run long, others short, and they partly cancel out. A plan on 193 would cut scope that would almost certainly fit; a plan on 94 is broken on day one. Averaging the extremes isn't a method either: it throws away the likely values the owners gave.

So plan on the likely sum and treat the range as information about uncertainty. How to protect the date against that spread, with one shared buffer instead of padding every line, is the subject of module 4.

A list can fit one limit and break another. Check each limit separately and find the one that binds, the idea from Must, should, could, now with real numbers.

Money: the 17 musts plus all five shoulds cost 503,450 UAH, under Hanna's 600,000 ceiling. Days: the same list is 131 + 26 = 157 likely person-days, 7 over the capacity of 150. So the full must + should list can't stay, however comfortable the budget looks. The ceiling would matter only for the full list including the coulds and the false musts (704,450 UAH).

That changes which shoulds to drop. Cutting by price would remove the January ads (90,000 UAH, 5 days) and keep the costly-in-time items. Cutting by days, the obvious candidates are D09 diagrams (8 days, an outside illustrator, but needing Vira's finished sets) and D10 mock exam (6 of Vira's days, the scarcest time on the project).

With 19 days left after the real musts, pick the shoulds by the days they cost and the value they bring to the goal, and keep a few days of room for slips (musts clearly below capacity, from Must, should, could).

A defensible cut keeps three shoulds. D20 waiting list (2 days): it tells you in December whether anyone wants physics, before most money is spent. D17 level report for parents (5 days): it helps parents pick a package after the test. D24 January ads (5 days): it brings families beyond those who asked last spring. Total: 131 + 2 + 5 + 5 = 143 person-days, leaving 7 days of room.

Dropped first: D09 diagrams (8 days) and D10 mock exam (6 of Vira's days, which she'd rather spend elsewhere), plus the adaptive test and the two false musts. Other combinations can be defended too; what matters is that the choice is explained by days and value, and that some room remains.

Fitting the capacity is a total: 143 person-days of work against 150 across everyone booked for the project. It doesn't yet say the work can be done by the date.

Two things can still break the calendar. One person's load: 73 of the 131 must-days sit with Vira and her authors, and Danylo holds 28. If Vira can give the launch only part of her week, her work alone may not fit before 29 January, however much room the total shows. Dependencies: demo lessons wait for the final curriculum review, the test build waits for the question bank. Work that waits can't use spare days elsewhere.

So the decision note records Vira's load as the open question, and the drop order Leo signs stays ready for January.

How it fits together: add effort by priority on likely values, test every must, check which limit binds, keep room and get a drop order signed. Module 3 turns the list into a schedule and asks whether 1 February really holds.