Known limits
What does not work yet, what is unreliable and what still needs proof in the live world. Read this before judging an evening.
A successful check proves that something happened once. It does not make every future run reliable. This is the honest list as of 27 October.
Unreliable
- Dungeon groups can get stuck, lose control of a pull or wipe. Recent runs of the 19th dungeon had very different success rates; the 29th was not cleared in that test series.
- Chat still repeats more often than it should, and a newcomer is not always greeted.
- Memory does not cover every interaction yet.
Still needs live proof
- Resurrection, sphere, purify and interrupt on command have automated-test evidence but are not all proven in the live game.
- Reliable advanced teamwork in endgame dungeons. The first Twilight Temple boss goes down; no full clear of Twilight Temple or Gui Mu yet.
- A balanced long-term economy with stable prices.
- Moving the live world to a later era with its characters.
- Most companion checks used headless test clients; they still need ordinary players trying them.
Capacity
A brief capacity run on the dedicated server supported about 600 AI residents. A longer soak is a separate check, and neither proves capacity for hundreds of people.
When you hit one
The roadmap shows what is being built about each of these.
Something unclear or missing? Ask in the community