Dice Forge: A Dice Engine You Build by Hand
Dice Forge is a 2017 game by Régis Bonnessée, published by Libellud, in which players compete as heroes for the favour of the gods. Its one genuinely unusual idea is that the dice are components you modify: each die has removable faces, and a large part of the game is deciding which of your own faces to destroy. It earns its reputation on that mechanic, and it is held back by the same mechanic being physically fiddly.
The forging mechanic, precisely
The part most descriptions get vague about is worth being exact on, because the exactness is the game.
Each player starts with two dice carrying basic resource faces. To upgrade, you spend gold at the Sanctuary and take one or more strictly different faces, and you must forge them immediately. Forging means using the new face as a lever to pry an old face off the die, snapping the new one into the empty slot, and setting the discarded face aside. It cannot be reattached, and faces cannot be rearranged afterwards.
That irreversibility is what gives the decision weight. A die has six faces and no more. Every upgrade is also a deletion, and the face you delete is a real cost, not an abstraction. New players consistently focus on what they are buying and not on what they are giving up, which is the single most common way a first game goes wrong.
The Sanctuary offers a limited pool: four faces per pool with three or four players, two with two players. At the low count the market is thin enough that you frequently cannot buy the face you planned for, and adapting to what is available is a larger part of a two-player game than of a four-player one.
Which face to remove
Since a die holds six faces and nothing can be reattached, the composition of the die is the actual position you are playing, and it is worth thinking about it as a shape rather than as a list of purchases.
A die loaded with one resource type produces that resource reliably and the others almost never. A balanced die produces moderate amounts of everything and excels at nothing. Neither is correct in the abstract, but they are correct at different times: early on you need small quantities of several resources to open up cheaper feats at all, while later the feats still on the table tend to demand larger amounts of one thing, which rewards concentration.
The trap is arriving at that second phase with a die you spread thin in the first. Because you cannot undo a face, a die diversified in round two is still diversified in round seven, and by then diversity is exactly the wrong shape. Sun shards deserve particular attention here, since they fund the extra action every turn as well as feats, and a die that produces none leaves that option permanently closed.
Why everyone rolling changes the maths
At the start of every player’s turn, all players roll both dice and apply the results. Not just the active player. In a two-player game the step happens twice per turn.
This is not a minor scheduling detail, it is the economic engine of the game. Your dice generate income four times per round at four players and twice per round at two. The same upgrade is therefore worth roughly twice as much at a full table, which means the correct opening differs by player count: at four players, buying production early compounds hard enough to justify delaying your first feat, while at two players the payback period is long enough that rushing upgrades can leave you behind on points with no time to catch up.
Because the game runs a fixed nine rounds (ten with three players) rather than to a target score, there is a hard horizon on that compounding. An upgrade bought in the last two rounds will roll a handful of times and rarely repays itself. The clock, not the money, is the real constraint.
The turn, and the two things you can do
A turn resolves in steps, and the choice sits in the middle one. You take either an offering or a heroic feat.
An offering is the Sanctuary purchase described above: pay gold, take faces, forge them at once.
A heroic feat means paying the listed resources, moving your pawn to that island’s portal, and taking the top card of that feat’s pile. Feats can be repeated, including more than once in the same turn, and a feat becomes unavailable only when its pile runs out.
The portal rule is the interesting one. If another player’s hero already occupies the portal you move to, that player is ousted: their pawn returns to its starting portal and they immediately receive a divine blessing, meaning a roll. So displacing an opponent costs you nothing but hands them income. It is a genuinely well-judged piece of design, because it stops portal contention from becoming purely punitive without removing the tension.
Feat cards file into three piles by how they behave: one for feats with no lasting effect, one for conditional permanent effects, and one for permanent effects that trigger on the owner’s turn. The conditional pile is left out of the introductory setup, which is the right call for a first game.
Once per turn you may spend two sun shards to buy one extra action, either an offering or a feat. The turn ends immediately afterwards. In practice this is how a strong late game is played: banking sun shards specifically to double up on feats in the final rounds is more reliable than hoping for a big roll.
Where it falls short
The faces are genuinely fiddly. Prying a face off a die with another face is not a delicate operation, and the discarded faces accumulate as loose components that are easy to lose or to mix up between players. Anyone with limited dexterity will find this a real barrier, not a quirk.
The theme is thin. Demigods, islands, divine favour: it is a coat of paint over a resource engine, and the cards do not build a narrative. That is fine if you came for the engine, but it will not carry a group that wants theme.
Luck still bites. Improving your dice narrows the range of outcomes, it does not remove them. Losing a decisive roll in round eight after playing the better game is entirely possible, and no amount of forging prevents it.
The two-player game is the weakest. A thinner Sanctuary pool and half the income mean fewer meaningful upgrade decisions. It works, but it is not where the game is at its best.
How it compares
Against Sagrada
Sagrada also puts dice at the centre, and takes the opposite approach: dice are drafted from a shared pool and placed into a grid, and their values are fixed once rolled. The challenge is spatial, fitting constrained colours and values into restricted positions.
Dice Forge asks the reverse question. Rather than solving a puzzle with the dice you are handed, you spend the game altering the distribution the dice can produce. The two reward different skills, spatial reasoning against probability and timing, which is why they sit together well in a collection rather than competing. If the drafting side appeals more, the Sagrada: Artisans expansion is the deeper end of that design.
Against Machi Koro
Machi Koro shares the idea of an engine that fires on die results, but you cannot touch the dice. You build a city of cards that activate on specific numbers, so you change what a result means rather than how likely it is.
The practical difference is agency. Machi Koro’s variance is higher and its swings are larger, which some groups enjoy and others resent. Dice Forge gives you a lever on the probability itself, which lowers the ceiling on both disasters and miracles.
Against Quacks of Quedlinburg
Quacks builds a bag rather than a die, adding ingredients that improve what you can draw. Structurally it is the closest of the three: both games let you improve a random output over time.
What Quacks has that Dice Forge does not is push-your-luck. You choose how far to keep drawing and can bust. Dice Forge spreads its tension evenly across the turn structure instead, with no single moment of nerve. Which one suits a group is mostly a question of whether they want that spike.
Teaching it
Lead with the die. Hand someone a die and a spare face and let them forge it before explaining anything else, because the mechanic sells itself in about four seconds and everything after is easier once they have felt it.
Then explain the rolling step, and make explicit that everyone rolls on everyone’s turn. This is the rule most often missed on a first teach, and missing it makes early upgrade decisions look irrational.
Use the introductory setup, which leaves out the conditional-effect feat cards. The decision space is wide enough without them, and they add most when players already understand what a round of income looks like.
Verdict
Dice Forge is worth owning if you like engine-building and want a version of it you can hold, and it is at its best at three or four players where the shared rolling step makes upgrades pay. The forging mechanic is not a gimmick: the irreversibility gives every purchase a cost, and the fixed round count keeps the whole thing from becoming a solved acceleration curve.
Skip it if your group wants theme and story, if fiddly components are a problem at your table, or if you mostly play two-player. The Rebellion expansion addresses the thin-market complaint with more faces and a larger card pool, but it needs the base game and is not the place to start.
Deja un comentario