Skip to content

Tensions & Lessons

Confidence

The observations are sourced; the diagnoses and proposed fixes are mine.

Tension 1 — the progression system is safest when unused

Boosters and weapons are the only progression in the game. Characters leave on a script, without warning, taking equipped gear with them permanently. The community's first tip is therefore to withhold investment. Meanwhile no encounter actually requires the boost, because difficulty is authored per board rather than tuned against player power.

So the system is punitive if you use it and inconsequential if you don't. Two mutually exclusive fixes exist, and the game picked neither:

  • Make departures legible — telegraph who is short-term, or return their gear to a shared stash on exit. Investment becomes safe and the equip screen becomes a real decision.
  • Make departures matter — keep the loss, but signal it, and let losing a kitted-out friend be the emotional beat the story keeps reaching for. Losing George's toolkit should hurt in a game about people who cannot take anything with them.

The second is the one this game's theme is asking for. It went unbuilt because nothing tells the player a departure is coming.

Tension 2 — determinism promises full disclosure, and the UI doesn't deliver

Removing the to-hit roll is the smartest decision in the design: it converts every board into something solvable and makes free restarts coherent. But a solvable board is only fair if every rule is visible, and Highwater hides several — attack-range tiles that read as forbidden, per-character weapon restrictions with no stated reason, an undocumented animation skip, booster magnitudes that aren't surfaced. A hidden rule in a deterministic system is worse than a hidden roll in a random one, because the player has been invited to plan exactly and then punished for information they could not have.

Tension 3 — traversal has no second channel

Half the runtime is boat travel with no decisions in it, carried entirely by the radio and the view. When the radio isn't landing, or the player is backtracking, the design has no fallback. Dredge faces the identical problem and solves it by making every minute at sea spend the master resource. Highwater has no economy at all, by choice — which keeps the tone unhurried but leaves its longest activity structurally empty.

The cheap fix isn't a resource system; it's a second reason to look. Even one interactable per leg — a salvage buoy, a radio-signal trace, a wreck worth a booster — would convert dead water into a small loop without disturbing the tone.

Tension 4 — the world is the protagonist, and the people are the cast of a road movie

Demagog's method builds world first, characters second, across three titles. That is why the newspapers, radio and item copy are excellent and the party is thin — and why critics praised the setting and disliked the dialogue almost unanimously. A road movie, though, is the one genre that cannot run on setting alone: it is a format about who is in the car. The systems knew this before the writing did — the party churn, the unique-verb design and the final civilian-protection objective are all doing character work the dialogue doesn't back up.

Tension 5 — authored content is a fixed budget

With no generator, no build layer and no randomness, content is 30 hand-made boards. That caps the game at 6–8 hours, makes replay value near-zero (chapter select exists for collectible mop-up, not for new play), and means every hour of runtime costs a full encounter's authoring. The team shipped this while making a second game simultaneously with ~20 people. Judged against that constraint, 30 bespoke puzzles is a strong result — the design's flaws are mostly the flaws of that budget being spent on breadth rather than depth.

Transferable lessons

  1. Removing randomness is a content decision, not a difficulty decision. No-miss combat converts fights from tuned encounters into authored puzzles. Budget accordingly: you now need a novel gimmick per fight and you cannot lean on numbers to escalate.
  2. If you remove the dice, publish the rules. Determinism raises the disclosure standard rather than lowering it.
  3. Never let a narrative event silently destroy a player's investment. Either protect it or dramatise it — the silent middle is the worst of both.
  4. Differentiate by irreplaceable verb, not by stat block. One unique interaction per character carries a whole cast and makes their loss legible in fiction and tactics at once; it is cheap in tuning and expensive only in animation.
  5. Downtime needs a channel, and one channel is fragile. Audio can carry traversal, but only if the player never has a reason to stop listening.
  6. Ending on a non-lethal objective is a strong closing statement — win by keeping the enemy's civilians alive, and the theme lands without a line of dialogue.

Equipment & Boosters · Combat · Boat & Navigation · Narrative & Theme · Design Lenses