Beat the clock on a set of questions by triaging each one, solve it for full points, gamble a quick guess, or flag and skip it.
You face a set of twelve multiple-choice questions, each with four answers, and a shared time budget of roughly half the time it would take to work them all. Solving a question is charged its full listed time and earns its points if you pick the right answer; guessing costs one second and is right one time in four; skipping costs and scores nothing. When the budget runs out, the round ends and your play is graded against an expert triage plan. The questions are small arithmetic, pattern and symbol puzzles, so no outside knowledge is needed: the skill is choosing.
Compare points per second. The easiest question pays 5 points for 10 seconds, half a point a second. The hardest pays 29 for 34 seconds, about 0.85 a second. Points rise faster than time, so in this game the hard questions are the better use of the clock, the opposite of the usual advice to grab the easy marks first. The reference plan solves in exactly that order, best value per second first, until the budget is spent.
Then it guesses on what is left. Without a penalty every guess is worth a quarter of the question's points for one second, so guess everything you cannot solve. With the penalty on, a wrong guess costs 2 points, and a blind guess on a question worth P points is worth P/4 − 1.5 on average. On the easiest question that is 1.25 − 1.5, a small loss, so skip it. On anything worth 8 points or more the guess still pays: 8 points gives +0.5, and a 29-pointer gives about +5.75.
So the penalty does not end guessing. It draws a line: skip the cheap long shots and keep gambling on the valuable ones.
Deadline is about exam strategy and expected value, where managing a time budget beats brute force.
A shared time budget where solving costs full time for full marks and guessing is nearly free is exam strategy stated as an optimisation.
Choosing which questions to solve is the knapsack problem: fill a fixed budget with the most valuable items. Deadline's reference plan uses the classic shortcut that George Dantzig proposed, taking items in order of value per unit of cost, which is quick and usually close to the best possible but not always exactly the best. That gap is why a strong player can occasionally beat the plan.
Deciding whether to guess is expected value, an idea that grew out of the problem of dividing the stakes of an unfinished game, which Blaise Pascal and Pierre de Fermat worked through in their letters of 1654. Real exams have argued over the same arithmetic for decades: the SAT penalised wrong answers until its 2016 redesign, which dropped the penalty and cut each question from five answer options to four.
That is the real structure of a timed exam and it is rarely taught. Time spent on one question is taken from every other, so the order you work in matters as much as how well you work.
Deadline is an original, free browser game made for PlayPendium: twelve knowledge-free questions per set, a clock that cannot cover them all, a penalty switch that redraws the line on guessing, and a reference plan to measure yourself against.