# Site

## Pages

- [Page not found](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/404/): There is no page at this address.
- [Assessment](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/assessments/): Every piece of graded work in the course, what it is worth, and when it is due.
- [Final: Defend a number](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/assessments/defence/): One change to one constant in the specimen, defended in seven parts in under ten minutes.
- [Assignment 2: Nothing broke](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/assessments/nothing-broke/): Take one value from your Assignment 1 curve and show that moving the game there did not kill an option or hand the win to a lazy player.
- [Assignment 1: One constant, six values](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/assessments/one-constant/): Step one number in the specimen's rules evenly across six values and show what difficulty did in response.
- [Tutorial quizzes](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/assessments/tutorial-quizzes/): Twelve five-minute quizzes, one at the end of every tutorial. Each asks for the number you just measured. Best ten count, 2% each.
- [Lectures](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/lectures/): Twelve Tuesdays. One claim a week, all of them about the same small game.
- [A complaint about balance is a prediction](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/lectures/week-01/): Nobody has written it down yet. This week you write down three predictions of your own, each with a number, about a game you may not play until week ten.
- [A game that needs a screen cannot be measured](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/lectures/week-02/): If a game's rules cannot run without its renderer, nothing about it can be measured. This week the specimen runs with no window open, in 6 ms.
- [A perfect player describes nobody](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/lectures/week-03/): A perfect bot wins 12 runs out of 15 and tells you nothing about a person. This week the simulated player gets a reaction time, and the wins go away.
- [One run tells you nothing](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/lectures/week-04/): The same player on the same game ends anywhere between 29.8 s and 120 s. This week is about how many runs a claim needs before it is worth saying out loud.
- [Even steps, uneven difficulty](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/lectures/week-05/): Turn one dial in equal steps and the game does not get harder in equal steps. The first 50 ms costs 26 wins. A later 50 ms costs 0.4 s.
- [A dead option looks like a live one](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/lectures/week-06/): A choice that does nothing looks exactly like a choice that matters. This week you measure what each of the specimen's seven upgrades is worth, by taking it away.
- [A gate everyone passes asks no question](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/lectures/week-07/): A gate is a point the player must be strong enough to pass. Its size is not a design opinion. It is a multiplication, and the first version got it wrong.
- [The lazy strategy must lose](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/lectures/week-08/): If one way of playing beats every other, the rest of the design is decoration. For a while the best way to play the specimen was to run away.
- [A metric that cannot fail is noise](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/lectures/week-09/): Nine different builds, 34 kills each. A metric that gives every subject the same score is not measuring them, and it will not tell you so.
- [Playing finds what simulation never asked](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/lectures/week-10/): This is the week you finally play the game. Ten minutes with your hands finds things that nine weeks of simulated runs did not.
- [Every boss is easy for somebody](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/lectures/week-11/): Difficulty is not one number. The same first boss lets 50 runs in 51 past, or 10 in 51, depending on who is holding the controls.
- [A change is finished when it can be checked](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/lectures/week-12/): The semester ends with one change to one constant, and everything you have learned to say about why it is the right one.
- [People](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/people/): Who teaches the course, and how to reach them.
- [Tamsin Okoro](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/people/tamsin-okoro/): Nine years as a systems designer on action and survival games, most of them spent in meetings about whether a boss was too hard.
- [Wei Lindqvist](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/people/wei-lindqvist/): Runs the Thursday tutorials, and wrote the course's reference simulated player, so knows every way in which it is wrong.
- [Policies and support](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/policies/): The rules of the course, most of which are rules about numbers, and where to go when something is in the way.
- [Three predictions](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/sessions/01-three-predictions/): Read the specimen's rules, then turn three opinions about its balance into predictions that a measurement could prove wrong.
- [One headless run](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/sessions/02-one-headless-run/): Get the specimen running on your own machine with no window open, and time it.
- [A player with reaction time](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/sessions/03-a-player-with-reaction-time/): Write two simulated players for the specimen, give one of them a reaction time, and compare their win rates over the same seeds.
- [How many runs](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/sessions/04-how-many-runs/): Watch how far your median still moves as runs are added, and decide how much movement you are willing to live with.
- [Step one number](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/sessions/05-step-one-number/): Pick one constant in the specimen's rules, step it evenly across six values, and find out whether difficulty followed.
- [Take one away](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/sessions/06-take-one-away/): Ban each of the specimen's seven upgrade cards in turn and measure what your simulated player loses.
- [Size a gate](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/sessions/07-size-a-gate/): Work out from measured damage output how much health a boss needs in order to stop the player it is meant to stop, then check the arithmetic by simulation.
- [Break the game](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/sessions/08-break-the-game/): Write the laziest, strangest and most cowardly simulated players you can, and find out whether any of them beats playing properly.
- [Find the ceiling](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/sessions/09-find-the-ceiling/): Test your own instrument for saturation by comparing how much your metric moves between builds with how much it moves between sets of seeds.
- [Play it](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/sessions/10-play-it/): Ten minutes with the specimen in your hands, a list of what your simulation never told you, and one item turned into a check.
- [Three kinds of player](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/sessions/11-three-kinds-of-player/): Build three simulated players of different skill and find out which of them the specimen is actually made for.
- [Defence rehearsal](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-Anson0028/sessions/12-defence-rehearsal/): A dry run of the final defence in front of three classmates whose job is to find the missing step.