Ninety-Four Days Inside a Card Game Economy That Quietly Doubled Its Grind

Why record this at all

Because progression economies are the one part of a live game that cannot be assessed from a single session, and because they are the part most likely to change without announcement. A balance patch to a weapon is visible immediately. A ten per cent adjustment to how quickly you accumulate a currency is invisible for weeks and then feels like your own fatigue.

So for ninety-four consecutive days one of us played the same routine in Ember Court — the same modes, the same duration, the same difficulty tier — and recorded what it produced. Same account, same time of day, no purchases of any kind.

What the log shows

For the first fifty days the daily yield was stable within expected variance. From day fifty-one it began to drift downward, and by day sixty-four it had settled at a materially lower level where it has remained for the thirty days since.

Expressed plainly: the routine that used to assemble a complete set in eighteen sessions now takes thirty-two. That is not a subtle adjustment, and it happened across a fortnight in which three patches shipped, none of which mentions acquisition, currency, rewards or drop rates.

We are not alleging deception. It is entirely possible that a change to something else — matchmaking tiers, session pacing, a reward table refactor — produced this as a side effect that nobody inside the studio has measured. That possibility is exactly why somebody outside should be measuring it.

What we ruled out

• Player skill drift. Win rate across the period stayed within two percentage points, and yield is only weakly coupled to winning in this game's structure.
• Account-level throttling. A second account started on day forty and following the same routine shows the same inflection at the same calendar date, not at the same account age.
• Seasonal event effects. No event ran during days fifty-one to sixty-four, and the two events that did occur earlier produced short spikes that returned to baseline.
• Recording error. The log was kept in duplicate by two people from day thirty onward, and the two records agree to within one per cent.

The publisher's position

We put the figures to the developer through its public community channel and received a general statement that no changes to reward rates were made in the period concerned. We accept that this is what their records show. Our records show what they show, both are published here, and the discrepancy is the finding.

What this does to the score

Very little, which surprised us. Ember Court was a 6.0 because its play was sharp and its progression was already tedious; discovering that the tedium has increased by roughly seventy-five per cent confirms an existing criticism rather than introducing a new one. The score moves to 6.1 for unrelated reasons — a genuinely good deck-editing overhaul shipped in month nine — and the economy finding is recorded in the piece rather than in the number.

The broader point is the reason we publish logs like this at all. Nobody can perceive a change of this kind from inside their own play, which means that in the absence of somebody counting, it simply happens.