Score
0
Best
-
Each job takes hours and is due at an hour. Pick the order, and pick which job to give up when they will not all fit.
Every job takes a number of hours and is due at a certain hour. You work on one at a time, back to back, in the order you tap them, and a job that finishes after its deadline is wasted. You may also leave a job out entirely. Each round shows a Par, which is the largest amount that can possibly be finished on time on that board, computed exactly rather than estimated; match it and you clear the round. Working in earliest-deadline-first order is always right, so the real decision is which job to give up when they will not all fit. From round seven the jobs are worth different amounts, and then you have to weigh value against hours instead of just counting jobs. Each round adds a job and shortens the planning clock. Three failed rounds ends the game, and your score is the number of rounds you clear.
Two decisions live in every board and only one of them is hard. The easy one is the order: among the jobs you decide to do, working the earliest deadline first is always at least as good as any other order. If two jobs both fit, doing the more urgent one first never costs you anything, so there is no cleverness to find there.
The hard one is which job to abandon, and the answer is genuinely counterintuitive. When the schedule overruns, most people drop the job whose deadline is furthest away, on the grounds that it is the least urgent. That is wrong. You should drop the job that takes the longest, because hours are what you are short of, not urgency — dropping a nine-hour job buys room for two four-hour ones, while dropping the least urgent job may free almost nothing.
This is not a rule of thumb, it is a theorem, and the measurements match it exactly. Across hundreds of boards, going in deadline order and dropping the longest job whenever the schedule overruns hits Par on every single board. Dropping the furthest deadline instead falls short on about one board in five. Stubbornly doing every job in deadline order clears essentially nothing once the slack is gone, and doing the shortest jobs first — which sounds efficient, and does squeeze in the most jobs when everything fits — clears under a tenth.
From round seven the jobs stop being worth the same, and the clean rule breaks with them. Dropping the longest job goes from hitting Par always to about four boards in ten, because now a nine-hour job worth three stars may be worth more than the two short jobs it was blocking. Trading value against hours is the harder problem, and no simple rule solves it every time — the best simple heuristic, dropping whatever gives the least value per hour, still misses one board in ten. That is where the game stops being a rule you apply and starts being a puzzle you have to look at.
One practical habit for the later rounds: build the schedule with the cheap-and-valuable jobs first and add the expensive ones only if the hours are genuinely there. It is much easier to see that a long job does not fit than to work out, after the fact, which of six scheduled jobs should come back out.