How the courses are made

There are 23 parkour courses on this site and each one has a page quoting a medal time and a measured route. This is where those numbers come from. If a course guide says a jump is 4.1 cells and the Author time is two minutes, the method below is what makes that a measurement rather than a claim.

It is written by Muhammad Ali Akmal, who built all of it. Everything described here is in the source of the game, and the constants on this page are imported from it rather than typed out — so if a number here is wrong, the game is wrong too.

A platformer is its jump. Ours rises and falls under different gravities — the fall is 3× the rise, which is the single largest lever on how the game feels — and that makes the arc asymmetric. The textbook airtime formula is wrong for it by roughly 40%.

So the arc is not a number in a comment. It is derived from the movement config, and every course is built against the derived value. Retune the jump and the courses are re-checked against the new arc automatically; nothing silently becomes impossible because a constant was folded by hand into a level six weeks earlier.

A jump built at 100% of the character's reach is one you make by leaving the platform on the last frame at exactly full speed. That is a frame-perfect input for a move the game asks for a hundred times a run, and a course full of them is not hard, it is unfair.

Every gap is therefore designed against 80% of the computed reach. Expert courses spend that margin; the safety underneath it is not spent by anything. It also absorbs an optimism in the model, which assumes you keep full horizontal speed through the air when air control is only 0.15.

One more thing that sounds pedantic and is not: a gap is counted in empty cells between platforms, not centre to centre. Two pads three cells apart have two cells of air between them. Getting that wrong by one is how a jump becomes frame-perfect by accident.

A course is turned into a graph of every surface you can stand on, with an edge wherever the arc says one place can be reached from another. The test suite then walks it and refuses to ship a course where the goal cannot be reached from the spawn.

It checks a second thing that matters more and is easier to miss: that the goal is still reachable from every checkpoint. A resume point you cannot continue from is not a mercy, it is a soft lock — the run is alive, the clock is running, and the course is over. Hazard blocks are excluded from the graph, so a route that would have to cross lava is not counted as a route.

The same graph is solved for the cheapest route: walking edges cost distance over run speed, jump edges cost their airtime. That number is a hard physical floor. It assumes instant acceleration, perfect pathing and never looking at where you are going, so nobody will ever match it — it is not a target, it is a measurement of the course.

The Author time is that floor multiplied by a single constant representing how much slower a very good human is, and the other three medals are multiples of the Author time. That constant is the one genuinely unverified number in the system, and it is stated as such in the source: it is an assumption awaiting real play data, not a finding.

What the derivation does buy is that a medal can never be impossible, because it is anchored to a route the geometry admits; and that it cannot silently rot, because moving a single pad moves the floor and the test suite fails with the new value.

The ladder runs Tutorial → Easy → Medium → Hard → Expert → Obby. The tiers are not a difficulty number attached to a finished course — they are a statement about what the course is asking for, and the course is built to ask for it:

• Tutorial — learn the jump. Wide landings, short falls, the whole route visible.
• Easy — learn it properly. The same jump, many times, with a clock on it.
• Medium — learn to turn. Direction changes at speed, with landings still forgiving.
• Hard — one-block landings. The distances are ordinary; the margin is not.
• Expert — on the limit. Courses that spend the full design margin.
• Obby — hazards. Failure is instant rather than positional.

Each course guide prints the blocks it is built from, its footprint, its checkpoints, and the shape of the solved route — jumps, widest gap, total climb, biggest drop. All of those are read off the geometry the game actually builds, by a script, and a test fails if a course is edited and the numbers are not regenerated.

What they are not is a description of your run. The solver walks surfaces with no air control and no momentum; you will take more jumps than it does, on a different line. The figures describe the course, not the player, and the guides say so on every page rather than letting a reader assume otherwise.

There are no player counts, average completion times or satisfaction scores anywhere on this site, because there is no honest way to produce them yet. A number nobody can check is worth less than a sentence they can.

The ladder is in the order the courses are meant to be played. If you have never played this game, Neon Run is the tutorial and Harbour Steps is the gentlest thing on the site. If you want to see what the design margin looks like when it is fully spent, read Overclock.