Writing the review now — one flag on the input data at the end.
Bottom Line: 9 Kings is the rare Early Access roguelike that already knows exactly what it is — a tight, vicious, twenty-minute engine for building combo machines that shatter their own arithmetic. It thins out for veterans, and the balance still wobbles between kings, but the core loop is one of the sharpest in the genre right now.
The Gameplay Loop
The loop is metronomic: draft, place, defend, loot, repeat. Each cycle takes maybe ninety seconds. A full run runs twenty to thirty minutes. That compression is the game's most underrated design decision, because it changes the emotional math of failure. When a run dies at wave twelve to a card you couldn't answer, you don't feel robbed of an hour. You feel curious. You start another one. This is where the "one more run" pull that Steam reviewers keep citing actually comes from — not from addictive tricks, but from a failure cost low enough to make experimentation rational.
The draft-and-place rhythm has real tension in it. You're solving two problems simultaneously with one resource. A card is both a what and a where, and the correct answer to one is often the wrong answer to the other. That warlock wants to sit behind the line. The tile behind the line is the only spot left that extends your territory toward the citadel upgrade you need. Pick one. Most deck-builders let you sidestep this by making card selection the entire decision; 9 Kings refuses, and the game is substantially richer for it.
Where the Depth Actually Lives
The looting system is doing the heaviest lifting. In a conventional roguelike deck-builder, your card pool is a fixed set gated by rarity rolls. Here it's adversarial — the pool expands based on who you beat, which means your build's trajectory is partly a consequence of your own combat performance. Kill the time king early and your run becomes a different run. This creates something most games in the genre never achieve: a mid-run identity crisis that's productive rather than frustrating.
And the ceiling is genuinely high. When players say the game breaks its own math, they're being precise, not hyperbolic. Certain overlapping aura stacks produce multipliers that turn wave forty into a formality. Discovering one of these is the payoff the entire structure is built to deliver, and Sad Socket has resisted the obvious temptation to patch the fun out. The multipliers are the point. The game wants you to cheat it.
The Cracks
Now the honest part.
The balance between kings is not there yet. This is the most consistent criticism in the review corpus and it holds up. Some kings enter the run with a coherent engine already half-assembled; others need three specific loots to become functional. A run's outcome can be substantially determined before you've made a meaningful decision — early card luck deciding runs is a real complaint, and in a game this short, a bad opening draft doesn't create an interesting comeback story. It creates a twenty-minute run you already know you lost at minute four.
The content thins for veterans. That 87% recent review score against a 92% all-time score isn't a collapse, but it's a signal. Players who've been at this since Next Fest have mapped the synergy space, and the discovery engine — which is what makes the first thirty hours extraordinary — has less fuel at hour eighty. Endless Mode is the intended answer. It's a partial one. Unbounded difficulty scaling tests execution, not creativity, and this is a game whose soul is creativity.
Onboarding is terse. The interface explains what cards do. It doesn't explain what cards mean — that adjacency compounds, that positioning is a multiplier and not a modifier. Strong players figure this out in run three. Weaker ones bounce off thinking it's a shallow tower defense with cards. That's a teaching failure, not a design one, and it's fixable.
None of this is disqualifying. All of it is exactly what Early Access is for. But the game is being reviewed by 9,644 people as if it's finished, and it isn't.