Bellwright: an actual sandbox, not a roadmap with WASD support Bellwright entered Early Access with broken pathfinding, rough combat and NPCs occasionally sharing one communal brain cell. It also entered Early Access as an actual game. The core loop already worked: survive, recruit villagers, build production chains, equip an army and liberate the region village by village. Since then, Donkey Crew has kept expanding and repairing that machine instead of disappearing into the Early Access fog. THE GOOD Bellwright lets you decide what kind of game you are playing today. In co-op, one player can chase quests while another hunts, gathers, plans the settlement or handles logistics. Solo, you simply change jobs whenever the mood strikes. You can: [*]Rush progression or turtle for fifty hours. [*]Build an efficient production network or live like a forest goblin. [*]Fight honourably or commit crimes against pathfinding from the nearest rock. [*]Win through melee, archery, fortifications, logistics, numbers or industrial-scale cheese. [*]Follow the campaign or spend the evening discovering why nobody is making flour. That is actual freedom. Not a dialogue wheel offering three flavours of the same outcome, but a game that lets the player decide what the evening is about. Bellwright also understands something many sandbox developers eventually forget: An overpowered strategy in a single-player game is not a public-health emergency. Weapons, armour, progression and encounters do get rebalanced, but balance is treated as a default rather than holy scripture. Difficulty settings cover raids, damage, hunger, spoilage, learning speed, durability and most other sources of suffering. There is also an official Modkit and Steam Workshop support. Some games notice that players are enjoying turrets, animals, farming or defensive engineering and immediately call the balance police. Bellwright usually fixes the machine and leaves you in the driver's seat. Want a serious medieval campaign? Fight clean and raise the difficulty. Want to build trenches, abuse terrain and establish a logistics-powered murder economy before advancing the story? Also valid. The game is not constantly leaning over your shoulder like an anxious dungeon master asking whether you are having fun correctly. The developers also kept working after the launch money arrived. Since release, Bellwright has received major settlement-management improvements, prosperity systems, animal husbandry, official modding, caravans, automated trade, fortifications, new regions and factions, planning tools, player-built roads, better army control, performance work and a frankly indecent pile of smaller fixes. Large updates are followed by hotfix trains for crashes, co-op, quests, navigation, interfaces and late-game settlements. That does not guarantee the future. It does prove that the studio has spent more than two years shipping features instead of merely announcing its intention to develop a game. Valheim found enormous success and then began developing at the speed of continental drift: months of silence followed by news that somebody had invented a new roof angle. Bellwright's cadence comes from another universe. The ownership model is refreshingly boring: [*]Buy the game. [*]Play offline. [*]Host co-op from your existing world. [*]No mandatory publisher account. [*]No microtransaction bazaar. [*]No paid gameplay-DLC staircase attached to the main menu. The AI voices are clearly disclosed placeholders. They add more than silence would, and I am not performing a toolchain exorcism over temporary voice acting in an unfinished game. THE BAD Freedom is not polish. The map is broad rather than densely authored. There is plenty of walking, repeated ambient dialogue and downtime where the simulation matters more than handcrafted discovery. The early game can also feel brutally grindy before villagers, priorities, outposts and logistics begin carrying their share of the workload. NPC automation still requires supervision. Workers ignore obvious solutions, choose scenic routes through danger or halt an entire production chain because one basket is in the wrong emotional state. Much of the game is about designing systems, watching them fail and determining whether the cause was poor planning, missing resources, bad priorities or an NPC who has silently decided to abandon material reality. If diagnosing why your settlement stopped producing food sounds like unpaid technical support, this may not be your game. THE UGLY Combat becomes good only after you personally reverse-engineer it. There is directional blocking, stamina management, spacing, equipment progression, army commands and plenty of room for preparation. The problem is that the game explains far too little. I wasted hours studying timings, testing weapons, separating bugs from mechanics and learning which encounters were simply not meant to be fought in the obvious way. There is no reliably current developer wiki, no useful public bug tracker and no definitive combat tutorial that explains the whole system without mixing current mechanics, old information, guesses and half-obsolete advice. Before that knowledge clicks, combat can resemble a medieval washing-machine accident. Group fights become chaotic. Hit detection can feel questionable. Enemies crowd together. Allies block attacks. Archery and terrain can trivialise encounters, while direct melee against several bandits can turn into a public execution within seconds. The game also does a poor job of communicating whether you lost because of your timing, equipment, stamina, positioning, army composition or some current mechanical defect. Once equipment, preparation, positioning and commands come together, battles can be satisfying. But the learning process is not elegant. It is field research conducted on your own corpse. WHO SHOULD PLAY IT? Play Bellwright if you enjoy: RimWorld, Necesse, Abiotic Factor, Cataclysm: Dark Days Ahead, Valheim, Terraria, Factorio or Mount & Blade. Not because Bellwright perfectly copies any of them, but because it occupies the same dangerous intersection of survival, logistics, construction, strategy, combat and "I will just reorganise this one production chain before bed." Its best moments come from solving problems the game did not explicitly script for you: turning a weak settlement into an industrial base, preparing a campaign, reorganising supply lines, fortifying a position or finding a completely unreasonable solution that somehow works. Study it first, or wait for 1.0, if you primarily want: [*]Kingdom Come-style tactile combat. [*]The Witcher's authored quest density. [*]Baldur's Gate 3's characters and campaign. [*]Skyrim's skill-and-magic toybox. [*]A small, polished loop that explains itself properly. Bellwright is unfinished, sprawling, grindy, occasionally stupid and unusually generous with player agency. The developers have also given UE5 enough attention to resurrect performance from the grave, although results still depend heavily on hardware, settlement size and current patch quality. It has bugs. It has cheese. It has workers who occasionally behave like medieval printers. It also has something increasingly rare: developers who keep building the game without trying to confiscate the player's fun. I would rather have this unruly machine than another polished corridor pretending to be a sandbox.