Red Pillars Is The Hardest Game In This Bench And The Easiest One To Configure

The game, briefly

A precision platformer of the unforgiving sort, structured as roughly nine hundred single-screen challenges built from a movement set you have completely by the fourth screen. Death is instant, restart is instant, and the difficulty curve is steep enough that most players will not finish it.

It is included here as the counter-example to Splinter Arena. Both games are difficult. One delivers that difficulty through timing and spatial reasoning; the other delivers a portion of it through a physical arrangement that cannot be altered. Red Pillars demonstrates that these are separable, and that separating them costs nothing.

Input count

Three bound actions: move, jump, and grab. Peak simultaneous demand is three, which occurs constantly, and there is no state in the game requiring a fourth input. The entire movement vocabulary — wall jumps, corner grabs, the late-game momentum cancel — is expressed through timing and direction rather than additional buttons.

This matters more than the small number suggests. A three-input game with tight timing is difficult in a way almost any hand can attempt. A nine-input game with loose timing is difficult in a way many hands cannot attempt at all, and the bench records the difference.

The rebinding audit

One hundred per cent. All three actions can be reassigned, including to the same physical input if you want to break the game, which the menu allows with a warning rather than a refusal — a small courtesy that a surprising number of games get wrong.

The grab, which in the default scheme is a hold, can be switched to a toggle from the same screen. That option is not buried in an accessibility submenu; it sits next to the binding it modifies, which is where every game in the archive should put it and where roughly a fifth of them do.

Eleven timing windows under two hundred milliseconds and zero locked bindings. Difficulty and inflexibility are not the same property, and this game is the shortest proof of it in twenty-six benches.

What it gets right, in order of cheapness

• Toggle option placed beside the hold it replaces rather than in a separate menu.
• Rebinding permitted for every action, with conflicts warned about rather than blocked.
• No chord inputs and no double-taps anywhere in the scheme.
• Input display in the pause screen shows the current binding, not the default one.

Across four devices

Identical on pad and keyboard, which is what a three-input scheme with no chords produces. On the adaptive switch box it is fully playable — the first game in this bench to reach that result — because three assignable inputs is exactly what such hardware handles best.

On a flight stick it is playable and faintly absurd, which the bench records as a curiosity rather than a finding. The relevant point is that no device loses access to any part of the game.

Where it loses points

Not on inputs. The deductions are ordinary review deductions: the middle third repeats its ideas for about two hundred screens, the visual language for hazards is occasionally ambiguous against the red stone, and the final sequence is gated behind a collection requirement that adds nothing.

Eight point six is a score for a very good platformer with a flat patch, not a reward for good menus. Good menus are not an achievement; they are the floor, and the reason this review reads like praise is that most of the archive is below it.

Verdict

The best game in bench twenty-six and the least interesting to test, which is the highest compliment this publication can pay. If a three-person team can ship complete rebinding and an inline toggle option, the argument that it is expensive does not survive contact with this game.