The Replay Notebook
Second visits, written down
Every piece here follows a full replay of a game finished at least four years earlier, with notes from both playthroughs side by side.
The Replay Notebook
Second visits, written down
Every piece here follows a full replay of a game finished at least four years earlier, with notes from both playthroughs side by side.
These pieces make comparative claims across years, which is the kind of claim most easily got wrong. What follows is the procedure that produces them, stated plainly enough that you can decide how much weight any given comparison deserves.
Three conditions, all required. The first playthrough must be at least four years old, contemporaneous written notes must exist, and the game must still run in the state it shipped in. Games that have been substantially patched since release are excluded, because then the game changed as well as the player and nothing can be isolated.
Beyond that, selection is unsystematic. I replay what I am curious about, which introduces a bias towards games I remember fondly, and that bias is worth knowing about when reading a piece that concludes a game held up.
| Condition | Requirement |
|---|---|
| Interval | Four years minimum since completion |
| Notes | Written at the time of first play |
| Game state | Original; no major post-release changes |
| Mods | None installed, either visit |
| Completion | Both playthroughs finished |
Default difficulty unless the first playthrough used something else, in which case the replay matches it and the piece says so. No guides, no walkthroughs and no community wikis during play, because consulting one converts a replay into a verification exercise.
Assist and accessibility settings are recorded rather than standardised. Where I used two driving assists in 2018 and none in 2026, that difference is data about me and is reported as such rather than being normalised away.
The first-visit notes are read once, in full, before the replay begins, and are not consulted again until the replay is finished. This is to avoid spending the second playthrough looking for what I wrote down the first time.
Second-visit notes are taken during play as usual. The comparison happens afterwards, with both sets open, and any direct quotation from the older notes in a published piece is quoted exactly, including where it is wrong.
Wall-clock time to completion, rounded to the nearest hour, including time spent stuck. Both figures appear in every piece because the difference between them is usually the most reliable measurement available, and it is nearly always a measurement of the player rather than the game.
Every piece tries to attribute a changed impression to one of three causes, and the tests are crude but stated. If a game was patched, the game changed. If contemporaneous notes show I misunderstood a system that was fully visible, the player changed. If a device that surprised me has since become a genre default, the surroundings changed.
Where two of these apply at once, which happens often, the piece says so rather than choosing the tidier explanation.
Pieces discuss structure and endings freely, because a replay of a game I finished years ago cannot be discussed otherwise. Everything covered here is at least four years old. No internal warnings are given.
Reviews of new games, previews, news, patch coverage, storefront links, purchase advice and scores. This notebook has nothing useful to say about anything released recently, by design.
One player, one memory, and one set of notes of uneven quality. My older notes were not written for publication and are sometimes too thin to settle a question. Hardware differs between visits, which affects performance impressions if not design ones. And I cannot replay a game without knowing I am going to write about it, which is a difference from the first visit that no procedure can remove.
Errors about what a game contains are corrected in place with a note. Playtimes are corrected if miscounted. Where a piece attributed a change to the wrong cause and a reader demonstrates it, the piece is amended and the amendment is described rather than quietly applied.