The Astrolabe Consequence Curve

The Terrestrial Astrolabe is the late-game ritual that lets a player reshape the world in exchange for a permanent penalty. This page covers the astrolabe block in balance.json — the knobs that set how expensive each ritual is and how fast the reward (exotic materials) ramps up — plus the one village key that decides when the Artificer specialist who unlocks the whole system can arrive.

In plain terms

Every time the player fires the Astrolabe, they set three sliders: how much of the dug-out cavern to refill, how dense to make world discoveries, and whether to bias shallow or deep. Each ritual then does three things, permanently: the Maw chews faster, the strange-root defense degrades, and rare "exotic" materials start seeding into the ground. The values below set the price of that trade.

A "multiplier" here is a number the game multiplies the Maw's chew speed by — 1.0 means no change, 1.5 means 50 percent faster. Chew increases from rituals compound: each ritual multiplies the running stack rather than adding to it, so a ritual's price is a fixed share of the threat the player already carries. The stack never resets within a run, and it has no ceiling — a run ends when the player dies, not when the Maw stops getting faster.

This changed. The stack used to be additive (stack + increase). That made every ritual after the first progressively cheaper in relative terms — at mid sliders the first ritual cost +82 percent of current threat and the ninth cost +11 percent, so the decisions meant to be the most dangerous were the cheapest a player would ever make. The four chew_increase_* values below were rescaled at the same time, because a growth rate needs smaller numbers than an addend.

How a ritual is priced

When the player confirms a ritual, the permanent chew-rate cost is computed from the two relevant sliders (read by AstrolabeManager.compute_chew_increase), then applied multiplicatively:

chew_increase =
    chew_increase_base
  + slider_volume        × chew_increase_volume
  + discovery_position   × (discovery_max − discovery_min)  + discovery_min

new_stack = old_stack × (1.0 + chew_increase)

• slider_volume runs 0.0 (refill almost nothing) to 1.0 (refill the most).
• discovery_position is the discovery slider normalized to 0.0 (full-left, cheapest) to 1.0 (full-right, densest and most expensive).

So the cheapest possible ritual costs base + discovery_min = 0.08 + 0.22 = 0.30, i.e. it multiplies the stack by 1.30. The most expensive costs base + volume + discovery_max = 0.08 + 0.18 + 0.37 = 0.63, multiplying the stack by 1.63. A middle-of-the-road ritual sits at 1.465. (The Ravenous Maw challenge and the Harrow "ritual fury" dial multiply the increase before it is applied — see Challenge and Harrow Modifiers.)

Mid-slider is not an independent knob. Because the slider terms sum linearly, a middle setting is always exactly the average of the cheapest and most expensive ritual, no matter how the four keys below are split. Only the floor and the ceiling are tunable; the middle follows.

What the curve looks like in play

The stack (astrolabe_chew_mult) after N rituals, at three slider settings:

The stack is multiplied by the time-based pressure level (capped at 5.0) to give the "Hunger ×N" figure the player sees — see Maw Threat and Escalation.

The calibration anchor: the curve is tuned so that after 12 most-expensive rituals, the Maw eats a full 9-block glazed_vault_block column — 45,000 HP, the toughest wall in the game — in about 60 seconds, against a Maw that has fully learned that material at capped pressure. The same column at ritual 1 takes over three hours. Cheapest-slider play still buys about 15 minutes per column at ritual 12, so twelve careful rituals remain survivable where twelve greedy ones do not. tools/astrolabe_curve_check.gd asserts this target against the live JSON and prints the full table.

If you retune the four keys, re-derive from the anchor:

growth_per_ritual = (target_stack) ^ (1 / target_ritual_count)

The astrolabe block

balance.json → astrolabe. Every value here has a hardcoded fallback of the same number inside AstrolabeManager.gd; the JSON is the tuning surface and overrides the constant on load.

Four more keys live in this block but sit outside this page's scope (they tune the strange-root bounce defense and the tile-restore volume, not the consequence curve): bounce_loss_max (8.0), bounce_cd_increase_max (0.5), restore_pct_min (0.1), restore_pct_span (0.8). They are listed here only so a designer scanning the block knows every key is accounted for.

What the refill will never do

Whatever restore_pct_* is set to — even a full 100 percent refill — the restore has two hard guarantees, so raising the volume can never destroy a player's factory:

• It never fills in a standing machine. Machine footprints are excluded from the plan and re-checked as tiles are written.
• It never seals one off. If filling would leave a surviving machine in a sealed pocket with no route in, the plan gives back the shortest corridor needed to reach it.

The second guarantee costs a handful of tiles out of the restored volume, once per ritual — invisible against the thousands of tiles a real restore touches, and worth knowing only if you are reconciling an exact refill percentage.

How the exotic ramp actually plays

The exotic chance uses one formula in two places — refilled cavern tiles (rolled live at ritual time) and previously-undug ground everywhere (rolled deterministically as the player digs). Both read exotic_chance_per_use and exotic_chance_cap:

exotic_chance = clamp(exotic_chance_per_use × rituals_performed,
                      exotic_chance_per_use,   # floor after the 1st ritual
                      exotic_chance_cap)       # ceiling

At the defaults: 2 percent after ritual 1, 4 percent after ritual 2, on up, flattening at 20 percent from ritual 10 onward. Restored deep caverns get an extra depth multiplier (up to 2x) on top, so pushing the depth slider deep is the fastest route to exotics. Before the first ritual the chance is zero — nothing exotic appears until the Astrolabe has fired at least once.

Note the asymmetry this creates with the chew curve: the exotic reward flattens at ritual 10 while the hunger cost keeps compounding. Rituals past 10 buy discovery density and refilled cavern, not more exotics, at a steeply rising price. That is intentional — it is what makes the decision to stop a real one.

The discovery-density reward

The discovery slider's reward side (its cost side is the two chew_increase_discovery_* keys above) is a world-wide multiplier on how often a dug tile yields a discovery pocket. Each ritual multiplies the running density by 1.0 + slider × 0.5 — so 0.5x (full-left) to 1.5x (full-right) per ritual — compounding across rituals and clamped to a 2.5x ceiling. Full-right slider reaches that ceiling by about the third ritual; further rituals add exotics and Maw hunger, not more discovery density.

This ceiling is DISCOVERY_MULT_MAX, a code constant in AstrolabeManager.gd — not a balance.json key. It was lowered from 3.0 to 2.5 after the double-count fix below.

At the 2.5x cap, discovery coverage peaks around 50 percent of solid tiles in the richest mid layers (stone / iron) — roughly double the ~22 percent no-ritual baseline. "Bad" discoveries (Maw resonance, plague spores, hazards) are deliberately excluded from this multiplier, so a denser world is a richer one, never a more dangerous one.

Two guardrails keep density in check:

• The mushroom near-water boost is a floor, not a multiplied base. Before the fix, the density mult tripled that boost (0.15 → 0.45) and, because water-pocket density also scales with the mult, nearly every tile counted as "near water" — mushrooms alone blanketed the mid layers and coverage hit ~80 percent. Now the boost is applied as max(base × mult, boost), so it stays a local nudge.
• Raising the ceiling raises render load. More on-screen discovery hints cost frame time on weak hardware (an FPS-adaptive animation budget absorbs most of it, but higher coverage still means more static markers). Push it up only after a two-machine feel/perf check.

The gate: when the Astrolabe becomes reachable

The Artificer, the fourth Astrolabe specialist, is what unlocks the system. The Artificer only shows up once the village hits a population threshold and the other three specialists have already arrived.

Safe to change / handle with care

• chew_increase_* compounds — raise it to a power, not a sum. These multiply a stack that never resets within a run. A player who rituals ten times pays the product of ten increases. Bumping chew_increase_base from 0.08 to 0.12 looks tiny but at twelve most-expensive rituals it takes the stack from ~350 to ~560. Always test a full multi-ritual run and re-run tools/astrolabe_curve_check.gd, never a single ritual.
• There is no ceiling on the stack, by design. Do not add one to "fix" a runaway number — the run is supposed to end with the player's death. Nothing downstream needs a bound either: each Maw front consumes at most one wall column per frame regardless of how fast it chews.
• exotic_chance_cap is the whole reward economy for exotics. Exotics gate several endgame machines and recipes. Set the cap too low and a player who commits to rituals never sees enough exotic material to progress; too high and exotics stop feeling special.
• artificer_population is a soft-lock risk. It is a progression gate with no alternative path. Raise it past a reachable population and the Astrolabe, exotics, and the consequence curve all become dead content for that run.

After editing any of these, re-run the data validator and smoke test — see The Safe Change Workflow.

Related

• Weather and Storms
• Village Buildings
• Maw Threat and Escalation
• Challenge and Harrow Modifiers
• The Levers That Matter Most