Attempt Logs — Hard Mode Journal

A stone table lit from above with arranged cards bearing abstract sigils, dice and coins

Data

The Attempt Curve: 4,912 Tries Plotted Against Progress

Our complete data set broken down by game and attempt band, with the point at which measurable improvement stops in each case. The flattening is remarkably consistent across genres.

Data 31 games Updated monthly

A small ship weaving through a dense lattice of glowing amber projectiles against a black void

Note

Above A Certain Density, Nobody Is Reacting To Bullets

Six hundred logged attempts comparing reaction time against pattern recall. Past roughly forty simultaneous projectiles, performance correlates with memorisation and not with reflexes at all.

Note 600 attempts 7 min

A stone fortress wall at night with braziers and an approaching army of torch lights below

Note

Siege Difficulty Is Usually An Information Problem

We audited four defence-focused strategy games for values the player needs and the interface hides. Between two and nine per game, and every one of them decided losses we had logged as errors.

Note 4 games 9 min

A close-up of a modular customisable controller with large swappable buttons on a dark desk

Essay

An Assist Option Is Not A Difficulty Setting, And The Confusion Costs Players

Input remapping, hold-to-toggle, reduced motion and colour options are accessibility features. Damage multipliers are difficulty. Filing them together has made both harder to discuss honestly.

Essay 31 games surveyed 12 min

A tiny character mid-jump between narrow platforms above spikes with ghost outlines of failed attempts

Timing

Retry Loop Times Across Every Game We Have Logged

Measured from the frame of death to the frame at which input is accepted again. The spread runs from 0.4 seconds to 38, and it predicts our enjoyment more reliably than any difficulty rating.

Timing 6 games Frame-counted

A night courtyard from above with patrolling guard silhouettes casting lantern shadows

Audit

Forty-One Cases Where Identical Inputs Produced Different Outcomes

The full inconsistency log from Lantern Watch, recorded attempt by attempt. This is the evidence behind the lowest mark we have given, and it took 163 attempts to compile.

Audit 163 attempts 41 flagged