Isle of Arrows

Most tower-defense games ask you to defend a map someone else designed. Isle of Arrows asks you to build the map first, one randomly-drawn tile at a time — paths, towers, resources, water, terrain — fitting each piece onto a floating isle that keeps expanding while waves of enemies march toward your fortress. It's a genuinely novel combination, closer to a constraint-based puzzle game like Dorfromantik than to a conventional tower-defense entry, and reviewers across the board — even the ones who end up not recommending it — credit that concept as something no other game in the genre is doing quite this way. The disagreement that actually defines Isle of Arrows isn't about whether the idea is good. It's about whether the randomness the idea depends on is fair.

A concept that earns its praise

The presentation carries the concept a long way on its own. Reviewers consistently single out the game's art direction, tower designs and sound as genuinely beautiful, often specifically noting they'd praise it unreservedly if the rest of the game matched. The core tension — building outward while keeping a defensible, connected path intact as new tiles force awkward compromises — gives even a single run real, moment-to-moment decision-making that a fixed-map tower-defense game doesn't offer. Multiple distinct guild-based playstyles (favouring economy, terrain manipulation, or aggressive expansion) add real build variety for players willing to lean into whatever a run's draws are pushing them toward, rather than fighting for one ideal strategy every time.

Where "randomness" becomes "unfair"

That's exactly where the game's most detailed critical reviews push back. Several describe receiving cards that expand the isle's usable space without any towers to place on that new ground, or watching a run collapse because a specific tower type never appeared despite being unlocked and theoretically in the pool. More than one reviewer says the pool itself feels imbalanced: a relatively small set of common cards accounts for most of what actually gets drawn, leaving the game's own advertised synergy pieces — cards specifically designed to combo with each other — rarely seen together in the same run despite dozens of hours played. Steam's own achievement data is a useful, neutral data point here: even modest, execution-focused goals like building a straight road or covering all available ground on an isle sit at 5.5% and 2.4% completion respectively across the entire playerbase. Both sides of the fanbase point to that same number as proof of their argument — the critics call it evidence the RNG makes basic planning nearly impossible, the defenders call it evidence the game is exactly as hard as it's supposed to be, a real skill wall rather than a broken one. This review can't resolve which reading is correct, and doesn't think it should try to — both are reasonable conclusions drawn from the same honestly-reported data, and a buyer's own tolerance for high-variance systems is what will actually decide which side they land on.

A real support window that closed over three years ago

Gridpop backed the launch with a genuinely active first few months: seven separate patches between September and December 2022, plus a substantial free content update, Winter Fortress, that December. One detailed January 2023 review, itself still largely positive, flags a specific, named problem persisting despite that active patch cadence — a bonus card meant to let players skip unwanted water-tile draws had reportedly stopped working correctly for months. A further patch did arrive in July 2023, fixing several other specific bugs and adding a new guild playstyle, though nothing in its patch notes clearly confirms that particular card issue was among the fixes. Nothing in the public record has followed since — over three years without further word on a game whose core fairness debate was, by its own fans' account, still actively unresolved when support stopped.

An undo button that never arrived

One specific, narrow complaint recurs even among the game's most enthusiastic reviewers: there's no way to undo a single misplaced tile within the same turn. One of the most-upvoted positive reviews on the game's page — from a player who says they play frequently and recommend it highly — raises exactly this, describing losing runs they consider otherwise near-perfect to a single misclick that destroyed an entire planned path, with no way to correct it. That's a different category of complaint from the broader RNG-fairness debate: it's not about whether the game's randomness is fair, but about whether the game gives players the tools to execute cleanly on the decisions they do control. A single-tile undo would cost the game very little of its intended tension while removing a source of frustration multiple reviewers, on both sides of the wider RNG argument, single out independently.

Where it lands

Isle of Arrows deserves real credit for trying something tower defense hadn't tried before, and succeeding at it visually and conceptually even by the account of reviewers who don't recommend the finished product. But a randomness system this central to the experience, and this divisive among the people who've spent the most time with it, is a genuine, unresolved design tension rather than a minor complaint — and a support window that closed in mid-2023 means whatever balance work might have settled that tension isn't coming. A six reflects a beautiful, original idea that never quite made peace with its own core mechanic before development stopped.