NPC combat stats carry the source server's tuning, and it is wrong where new players stand #84

New issue

Open

opened 2026-07-31 18:50:55 +00:00 by sickday · 0 comments

sickday commented

2026-07-31 18:50:55 +00:00

Owner

The NPC combat table's stats come from an era-adjacent server's own data
files, and a live sitting caught that data being wrong where it is most
visible: the level-2 goblins die at 3 hitpoints and the wiki says 5. The
converter is faithful — the source file itself says 3 — so this is source
quality, not extraction.

Measured against fifteen well-known creatures, joining the table to the
cache's own names and combat levels:

right    Chicken 3, Giant rat (3) 5, Man 7, Woman 7, Guard (21/22) 22,
         Hill Giant 35, Moss giant (42) 60, Hobgoblin (28) 29,
         Al-Kharid warrior 19, Barbarian (10+) 25, most Skeletons/Zombies
wrong    Goblin (2): 3 for 5 across all 37 ids
         Cow (2): two of four ids carry 3 for 8
         Lesser demon (82): 81 for 79
junk     a handful of hp=1 rows on real creatures — Moss giant (48) x2,
         Hobgoblin (47), Guard (1337) x11 — variant ids the source
         filed with placeholder stats

So the table is broadly right and wrong in the places new players stand. The
same caveat covers every other extracted stat (attack/strength/defence,
bonuses, attack speeds, aggression caps): they are that server's tuning
wherever it diverged from the era, and nothing here can tell which rows those
are without a second source.

The durable fix is a wiki-derived monster-stats dump as the primary source
(or a correcting overlay), with the provenance convention applied per value —
the wiki's changes sections date most rebalances, so "was this true on
2017-07-13" becomes a lookup. The drops table already came from a wiki scrape,
so the posture exists; what is missing is the stats scrape itself, which is
operator-supplied input like every other dump. The era guard stays the cache's
own combat level either way.

Until then: fights around Lumbridge end one hit early, and nothing else
observable is known to be off.

The NPC combat table's stats come from an era-adjacent server's own data files, and a live sitting caught that data being wrong where it is most visible: **the level-2 goblins die at 3 hitpoints and the wiki says 5.** The converter is faithful — the source file itself says 3 — so this is source quality, not extraction. Measured against fifteen well-known creatures, joining the table to the cache's own names and combat levels: right Chicken 3, Giant rat (3) 5, Man 7, Woman 7, Guard (21/22) 22, Hill Giant 35, Moss giant (42) 60, Hobgoblin (28) 29, Al-Kharid warrior 19, Barbarian (10+) 25, most Skeletons/Zombies wrong Goblin (2): 3 for 5 across all 37 ids Cow (2): two of four ids carry 3 for 8 Lesser demon (82): 81 for 79 junk a handful of hp=1 rows on real creatures — Moss giant (48) x2, Hobgoblin (47), Guard (1337) x11 — variant ids the source filed with placeholder stats So the table is broadly right and wrong in the places new players stand. The same caveat covers every other extracted stat (attack/strength/defence, bonuses, attack speeds, aggression caps): they are that server's tuning wherever it diverged from the era, and nothing here can tell which rows those are without a second source. The durable fix is a wiki-derived monster-stats dump as the primary source (or a correcting overlay), with the provenance convention applied per value — the wiki's changes sections date most rebalances, so "was this true on 2017-07-13" becomes a lookup. The drops table already came from a wiki scrape, so the posture exists; what is missing is the stats scrape itself, which is operator-supplied input like every other dump. The era guard stays the cache's own combat level either way. Until then: fights around Lumbridge end one hit early, and nothing else observable is known to be off.

sickday referenced this issue from a commit

2026-07-31 22:42:04 +00:00

feat: standoff combat both directions, on an ap-range pursuit

No labels

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

Revenant/Server#84

No description provided.

Rows
Columns