Karst¶

Diagnose a landscape where acidic groundwater has dissolved soluble bedrock into a self-reinforcing hidden conduit network, so the surface no longer maps the subsurface drainage and porous-media terrain rules are suspended.

Core Idea¶

Karst is a class of landscape produced when slightly acidic groundwater — rainwater charged with carbonic acid from soil CO₂ — slowly dissolves soluble bedrock, typically limestone, dolomite, or gypsum, along joints and fractures over geological timescales. The dissolution is selective and self-reinforcing: flow concentrates in the widest openings, which dissolve fastest and capture more flow, progressively enlarging a network of conduits and caves while the surrounding matrix remains relatively intact. The result is a landscape that has lost its coherent surface drainage: water that would otherwise flow as streams disappears into sinkholes and swallet holes, travels through subsurface conduits at velocities orders of magnitude faster than porous-medium groundwater flow, and re-emerges at discrete karst springs often kilometers from the input point.

The diagnostic landform assemblage — sinkholes (dolines), poljes, blind valleys, disappearing streams, caves, and karst springs — is the surface expression of a subsurface void network largely invisible from above. That topographic decoupling from drainage is what makes karst terrain qualitatively different in hydrology, engineering, and ecology: contaminants introduced at a sinkhole can appear at a drinking-water spring within hours rather than months, because conduit flow bypasses the filtration and retardation that porous-media aquifers provide; foundations sink when subsurface voids collapse without warning; endemic cave fauna evolve in isolation in a habitat that has no surface-hydrological parallel. The underlying generative pattern is substrate-selective dissolution producing a hidden interior conduit network that periodically reveals itself through surface collapse.

Structural Signature¶

Sig role-phrases:

the soluble substrate — bedrock susceptible to chemical dissolution (limestone, dolomite, gypsum, salt)
the reactive medium — slightly acidic groundwater, rainwater charged with carbonic acid from soil CO₂, moving along joints and fractures
the geological-time process — slow dissolution acting over the timescales needed to enlarge openings into a network
the self-reinforcing channelisation — flow concentrating in the widest openings, which dissolve fastest and capture more flow, a positive feedback that grows a conduit network while the matrix stays intact
the hidden conduit network — the subsurface void-and-cave system largely invisible from above, through which water routes at velocities orders of magnitude faster than porous-medium flow
the surface-interior decoupling — drainage gone underground, so surface topography no longer maps the actual flow paths (the load-bearing diagnostic fact)
the diagnostic landform assemblage — the surface expression: sinkholes/dolines, poljes, blind valleys, disappearing streams, caves, karst springs
the collapse event — periodic sudden sinkhole failure by which the enlarging interior betrays itself at the surface, without warning

What It Is Not¶

Not terrain whose surface maps its drainage. The defining fact of karst is that drainage has gone underground into a conduit network decoupled from topography, so a catchment delineated from ridgelines may be wrong (water can cross a topographic divide through a conduit) and the operative question becomes "where do the conduits route water, and how fast?" — answerable by dye tracing, not by reading the land.
Not a porous-media aquifer. Karst hydrology is fast, channelised conduit flow that bypasses the filtration and retardation porous media provide, so a contaminant can reach a drinking-water spring in hours rather than months, kilometres away, with no attenuation. Darcy aquifer models and distance-and-porosity travel-time estimates are inadmissible; conduit-flow models replace them.
Not layered accumulation. Karst is the removal of substrate by dissolution, producing voids — the inverse of depositional layering. A landscape diagnosed as karst is read for a hidden conduit network, not for a readable stratigraphic record.
Not bare leakage. Conduit flow shares the unintended-pathway character of escape-and-leakage, but the karst diagnosis adds the generative dissolution process and the specific landform family (sinkholes, poljes, disappearing streams, karst springs) that leakage alone does not carry. The construct is a substrate-process-landform compound, not just an unintended flow path.
Not a metaphor for hidden organizational, security, or data structure. "Karst vulnerabilities," the "karst" organisation, and "data karst" borrow the slogan — hidden interior channels exposed by collapse — while leaving behind the soluble bedrock and carbonic-acid dissolution that make karst coherent. Tellingly, the actionable diagnosis in each comes from a different catalogued prime (latent-failure analysis, informal-network mapping, selection-bias reasoning), so the karst label adds evocation but no cross-domain intervention; off soluble-rock terrain it is slogan, not mechanism.

Scope of Application¶

Karst lives across the earth-science fields that confront soluble-rock terrain — geomorphology, hydrogeology, and the engineering and ecology disciplines that inherit its hidden-conduit consequences; its reach is bounded to dissolution-susceptible substrate, and the "karst vulnerabilities"/"data karst"/"karst organisation" readings are slogan carried by latent-failure and escape_and_leakage patterns, not habitats.

Geomorphology — the native habitat: classifying limestone plateaux, sinkholes/dolines, poljes, blind valleys, karst springs, and cave systems (the Yucatán cenotes, the Dinaric Alps, Mammoth Cave) and reading the landform assemblage as the surface expression of a void network.
Hydrogeology and water-supply engineering — karst aquifers behave unlike porous media (fast channelised conduit flow, acute contamination vulnerability), demanding dye tracing, conduit mapping, and treatment-train design for rapidly varying turbidity rather than Darcy models.
Foundation and geotechnical engineering — siting on karst terrain requires void detection ahead of construction because subsurface cavities collapse into sinkholes without surface warning.
Cave and groundwater ecology — endemic, often eyeless cave fauna evolve in isolation in a habitat with no surface-hydrological parallel, a biological consequence of the hydrological decoupling.

Clarity¶

The single word karst bundles a substrate condition (soluble bedrock), a process (dissolution by carbonic-acid groundwater), and a landform family (sinkholes, conduits, disappearing streams, karst springs) into one diagnosis, and the payoff of that bundling is a warning that the surface cannot be trusted as a map of the subsurface. In ordinary terrain a hydrologist reads drainage from topography — water runs downhill, catchment boundaries follow ridgelines, travel times scale with distance and porosity. Naming a landscape karst suspends that whole inference: the construct says drainage has gone underground into a conduit network largely invisible from above, so the operative question shifts from "where does the surface slope?" to "where do the conduits actually route water, and how fast?" — a question answerable only by dye tracing and conduit mapping, not by reading the land.

That reframing is what makes the label consequential rather than merely descriptive, because the decoupling has the same root cause across several fields that would otherwise treat their problems as unrelated. The fast, unfiltered conduit flow that lets a contaminant reach a drinking-water spring in hours instead of months; the foundation that fails when a hidden void collapses without surface warning; the cave fauna that evolve in a habitat with no surface-hydrological parallel — all follow from the one fact the diagnosis fixes, that selective dissolution has built a hidden interior that the surface periodically betrays through collapse. So "karst" does not just classify a landscape; it tells the hydrogeologist, the foundation engineer, and the ecologist that the methods calibrated on porous-media terrain do not apply, and which substitute methods do.

Manages Complexity¶

A carbonate landscape generates problems across hydrology, civil engineering, and ecology that, taken separately, look like three unrelated literatures: drinking-water springs that turn turbid and bacterial within hours of a distant rainstorm; building foundations that fail when the ground opens without warning; cave-dwelling species found nowhere else, evolving in apparent isolation. Each field, working alone, would build its own catalogue of anomalies — travel times that defy distance-and-porosity rules, collapses that defy surface inspection, habitats that defy the surface-drainage map. The single diagnosis karst compresses all three to one root fact. It bundles a substrate condition (soluble bedrock), a process (carbonic-acid dissolution along joints, concentrating self-reinforcingly in the widest openings), and a landform family (sinkholes, conduits, disappearing streams, karst springs) into one named state, and the load-bearing consequence of that bundle is a single proposition: selective dissolution has built a hidden interior conduit network, so the surface no longer maps the subsurface drainage.

That one proposition is what collapses the apparent sprawl. The fast unfiltered contaminant arrival, the unannounced foundation collapse, and the surface-less aquatic habitat are not three problems but three faces of the one decoupling, so the analyst stops tracking them independently and reads them all off the conduit network's existence. The diagnosis also resets, in a single stroke, which inferential machinery is even admissible: it suspends the ordinary terrain rules — drainage follows topography, catchments follow ridgelines, travel time scales with distance and porosity — that quietly underwrite porous-media hydrology, foundation siting, and habitat mapping, and replaces the question "where does the surface slope?" with "where do the conduits route water, and how fast?" Because the standard methods are now known to fail for one shared reason, the substitute toolkit is prescribed by the same diagnosis: dye tracing and conduit mapping instead of topographic catchment analysis, void detection instead of routine foundation survey, conduit-flow models instead of Darcy aquifer models. So what the hydrogeologist, the engineer, and the ecologist must each track shrinks to one structural question — has dissolution built a hidden interior here, and where does it run — from which the hazard profile, the failed default methods, and the correct substitute methods all follow. A scatter of qualitatively strange, field-specific anomalies collapses to a single substrate-process-landform diagnosis and the surface-betrays-interior fact it carries — the move from cataloguing each anomaly to reading every one off the same hidden network.

Abstract Reasoning¶

Karst licenses reasoning moves that all flow from one diagnosis — that selective dissolution has built a hidden conduit network so the surface no longer maps the subsurface drainage — letting the hydrogeologist, foundation engineer, and ecologist suspend the inferences that ordinary terrain underwrites, predict the hazards the decoupling produces, and substitute the methods calibrated for a hidden interior.

Diagnostic — recognize karst from substrate, process, and landform, then read the interior as hidden. The defining inference identifies a landscape as karst from a triple — soluble bedrock (limestone, dolomite, gypsum), dissolution by carbonic-acid groundwater, and the diagnostic landform assemblage (sinkholes/dolines, poljes, blind valleys, disappearing streams, caves, karst springs) — and from that diagnosis infers the load-bearing fact: a subsurface void network largely invisible from above. A second diagnostic reads a surface anomaly back to the conduit: a spring that turns turbid and bacterial within hours of a distant rainstorm is inferred to be fed by an open conduit bypassing filtration, not by porous-media flow; a foundation that fails when the ground opens without warning is inferred to sit over a void that collapsed; an endemic cave species is inferred to occupy a habitat with no surface-hydrological parallel. Each surface symptom is traced to the same hidden interior rather than diagnosed on its own terms. The sharp question the diagnosis substitutes is not "where does the surface slope?" but "where do the conduits actually route water, and how fast?"

Boundary-drawing — which default inferences are suspended. The signature move of the karst diagnosis is negative: it suspends the ordinary terrain rules that quietly underwrite porous-media hydrology, foundation siting, and habitat mapping. In ordinary terrain water runs downhill, catchment boundaries follow ridgelines, and travel time scales with distance and porosity — naming a landscape karst declares all three inadmissible, because drainage has gone underground into a conduit network decoupled from topography. So the construct draws a hard boundary between karst and non-karst reasoning: a catchment delineated from surface topography may be wrong (water can cross a topographic divide through a conduit), a travel-time estimate from distance-and-porosity may be off by orders of magnitude, and a habitat map from surface drainage may miss the subsurface entirely. It also bounds karst against its inverse and its neighbors: karst is removal of substrate producing voids, the inverse of depositional layered accumulation, and while its conduit hydrology shares the unintended-pathway character of leakage, the karst diagnosis adds the generative dissolution process and the specific landform family that bare leakage does not carry.

Interventionist / method-prescription — the failed defaults dictate the substitutes. Because the diagnosis fixes why the standard methods fail (one shared cause: the hidden conduit network), it prescribes the substitute toolkit by the same stroke. For hydrology, topographic catchment analysis is replaced by dye tracing and conduit mapping to find where water actually routes and how fast; Darcy aquifer models are replaced by conduit-flow models. For engineering, routine foundation survey is replaced by void detection before siting. For water supply, treatment calibrated on slow filtered porous-media flow is replaced by source protection at swallets and treatment-train design for rapidly varying turbidity, because the conduit delivers unfiltered surface water on a storm's timescale. The interventionist invariant: the lever is to treat the subsurface as a fast, unfiltered, topography-decoupled conduit network and to deploy the methods that can see it, not to refine the surface-based methods the diagnosis has already ruled out.

Predictive / order-of-events. The diagnosis carries a package of forecasts off the single fact of the conduit network. It predicts contaminant behavior: a pollutant introduced at a sinkhole can appear at a drinking-water spring in hours rather than months, kilometres away, because conduit flow bypasses the filtration and retardation porous media provide — so the construct forecasts fast, unattenuated, long-distance transport from a discrete input to a discrete output. It predicts collapse: subsurface voids enlarge by self-reinforcing dissolution (flow concentrates in the widest openings, which dissolve fastest and capture more flow), so the interior grows toward instability and the surface periodically betrays it through sudden sinkhole collapse without surface warning — an order-of-events the construct makes anticipable even though the trigger point is not. And it predicts ecological isolation: a habitat with no surface-hydrological parallel produces endemic cave fauna evolving in isolation, a biological consequence forecast directly from the hydrological decoupling.

Knowledge Transfer¶

Within earth science the diagnosis transfers as mechanism across every dissolution-susceptible substrate: limestone, dolomite, gypsum, salt domes, and even basalt lava-tube karst-analogs all run the same substrate-process-landform logic, and the full apparatus — the conduit-network inference, the suspension of the porous-media terrain rules, the prescribed substitute toolkit (dye tracing, conduit mapping, void detection, conduit-flow models), and the package of forecasts (fast unfiltered long-distance contaminant transport, unannounced collapse, endemic cave fauna) — carries intact across them and across hydrogeology and foundation engineering. The substrate chemistry changes (carbonic-acid dissolution of carbonate versus dissolution of evaporites) but the surface-betrays-interior fact and the methods that follow from it need no translation, because each genuinely has soluble rock, a reactive medium, and a self-reinforcing conduit network. Across the geosciences this is mechanism recurring, and the vocabulary (sinkhole, doline, swallet, conduit, karst spring) travels without loss.

Beyond the geosphere the transfer is analogy, and the boundary is the substrate-dissolution dynamic that has no off-rock referent. The familiar metaphorical extensions each carry only the slogan and not the mechanism: "karst-like vulnerability" in security (hidden privilege-escalation and lateral-movement channels surfacing as incidents), the "karst" organisation (formal surface structure concealing informal channels through which work really flows, with periodic crises exposing the gap), and "data karst" (records concealing channelised holes — selection bias, unobserved subpopulations — with surprising downstream effects) all borrow the picture of hidden interior channels exposed by sudden collapse while leaving behind the soluble bedrock, the carbonic-acid dissolution, and the specific landform family that make karst a coherent structural pattern. Tellingly, in each metaphor the actionable diagnosis comes from a different prime that is already catalogued — latent-failure analysis, informal-network and principal-agent mapping, selection-bias and unobserved-heterogeneity reasoning — so the karst label adds evocation but no substantive cross-domain intervention. The genuinely portable residue is the broader pattern karst instantiates — substrate-conditioned latent structure: a substrate that responds to a particular process by accumulating hidden interior structure the surface representation systematically underrepresents, periodically exposed by collapse — which is already served by escape_and_leakage (the unintended-pathway face, minus the dissolution process) together with latent-failure and iceberg-and-shadow patterns. The honest cross-domain move is therefore to reach for those latent-structure primes when "surface conceals a hidden interior exposed by failure" is the needed lesson, and to reserve "karst," its dissolution mechanism, and its sinkhole-and-conduit vocabulary for soluble-rock terrain, where alone the diagnosis is mechanism rather than slogan (should a future emergent candidate formalize substrate-conditioned latent structure as a substrate-independent prime, karst would be one instantiation of it, and the cross-domain metaphor would attach there) (see Structural Core vs. Domain Accent).

Examples¶

Canonical¶

The Classical Karst — the Kras (Carso) plateau straddling Slovenia and Italy — is the landscape that gave the phenomenon its name, and it is the textbook demonstration. The plateau's Cretaceous limestone has been dissolved by carbonic-acid groundwater into a maze of sinkholes and caves, and the Reka River enacts the defining decoupling: it flows on the surface until it plunges underground at the Škocjan Caves, a UNESCO-listed gorge and cave system, and vanishes. Its water travels kilometers through unmapped conduits and re-emerges near the Adriatic coast as the Timavo springs, some 35 kilometers away, having crossed beneath terrain whose surface topography gives no hint of the route. Surface streams are absent across much of the plateau precisely because drainage has gone entirely underground.

Mapped back: The plateau limestone is the soluble substrate, dissolved by carbonic-acid groundwater — the reactive medium — via the self-reinforcing channelisation that built the hidden conduit network carrying the Reka. The river sinking at Škocjan and resurging as the Timavo far away is the surface-interior decoupling made visible, and the sinkholes and disappearing stream are the diagnostic landform assemblage.

Applied / In Practice¶

The Edwards Aquifer in central Texas is karst hydrogeology managed as critical infrastructure. A cavernous limestone aquifer, it is the primary water supply for San Antonio and surrounding communities, and its recharge zone takes in surface water rapidly through sinkholes, fractures, and losing streambeds rather than through slow soil percolation. Because conduit flow bypasses filtration, the aquifer is acutely vulnerable to contamination introduced at the surface, and dye-tracing studies have documented water moving through it far faster than porous-media flow would allow. The Edwards Aquifer Authority accordingly manages the recharge zone with source protection and land-use controls rather than relying on the aquifer to filter pollutants. Its major discharge points, Comal and San Marcos Springs, host endemic species including blind salamanders adapted to the lightless conduit habitat.

Mapped back: Rapid recharge through sinkholes into the hidden conduit network is the surface-interior decoupling driving the contamination hazard the diagnosis forecasts. Dye tracing and recharge-zone protection are the prescribed substitute toolkit for a fast, unfiltered aquifer, the springs are the diagnostic landform assemblage, and the blind salamanders are the endemic cave fauna the hydrological isolation predicts.

Structural Tensions¶

T1: Efficient conduit versus runaway void (the same feedback builds the aquifer and the collapse). The self-reinforcing channelisation — flow concentrates in the widest openings, which dissolve fastest and capture more flow — is what makes a karst aquifer an efficient, fast-recharging conduit network rather than a diffuse sponge. The identical positive feedback is what grows the subsurface voids toward instability, enlarging the interior until the surface fails without warning. There is no version of the productive conduit system that is not also the enlarging cavity, because they are the same process at the same locations: every conduit that speeds water toward a spring is a void trending toward collapse. The feature that gives karst terrain its hydrological character and the feature that makes it geotechnically treacherous are one dissolution feedback, not two. Diagnostic: Is the conduit development here being read only for its drainage function, or also as the void enlargement that the same feedback is driving toward collapse?

T2: Rapid recharge versus contamination vulnerability (fast flow is the value and the hazard). Conduit flow moves water at velocities orders of magnitude above porous-media flow, and that speed is genuinely valuable — the Edwards Aquifer recharges rapidly through sinkholes and losing streams and supplies a major city. But the same fast, channelised flow bypasses the filtration and retardation that porous media provide, so a contaminant introduced at a distant sinkhole can reach a drinking-water spring in hours, unattenuated, kilometres away. The property that makes karst an efficient, quickly-replenished water source is exactly the property that makes it acutely dangerous to contaminate, and no treatment calibrated on slow filtered flow will hold. Speed of delivery and absence of natural cleansing are the same conduit fact. Diagnostic: Does the rapid recharge being relied on for supply also mean surface contaminants arrive fast and unfiltered — so protection must move upstream to the swallets rather than trusting the aquifer to filter?

T3: Binary diagnosis versus the karst–porous continuum (declaring "karst" can over-suspend). The label's power is negative and wholesale: naming a landscape karst suspends the porous-media terrain rules — topographic catchments, Darcy travel times, ridgeline divides — and prescribes the conduit toolkit instead. But real carbonate terrain is often a continuum, dually porous: mature open conduits coexist with a matrix that still drains diffusely, and many landscapes are only partially karstified. A wholesale suspension of porous-media reasoning can over-correct where matrix flow still governs part of the system, just as trusting the surface map over-corrects the other way where conduits dominate. The crisp karst/non-karst switch that makes the method choice decisive is exactly what mishandles the intermediate terrain that behaves as both at once. Diagnostic: Is this landscape conduit-dominated (suspend the porous-media rules) or a dual-porosity mix where matrix flow still governs part of the transport — and does the method acknowledge which parts are which?

T4: Anticipable in kind versus unpredictable in timing (you know collapse is coming, not when). The diagnosis makes karst hazards forecastable in character — fast unfiltered contaminant transport, endemic isolation, and sudden sinkhole collapse are all predicted from the hidden conduit network. But the same construct that makes collapse anticipable in kind leaves its trigger point unknowable: the interior enlarges silently and betrays itself only at the moment of failure, so "a collapse will occur here" carries no "when." This makes the protective response open-ended — void detection can always find more, dye tracing maps only the conduits it happens to trace, and there is no natural stopping point at which the subsurface is known to be safe. The confidence that a hidden interior exists coexists with irreducible uncertainty about where and when it will give way. Diagnostic: Is the investigation being scoped to a definite risk that can be closed out, or to a hazard known to exist but not to locate — where more surveying reduces but never eliminates the unmapped void?

T5: Autonomy versus reduction (a soluble-rock landscape or a domain instance of substrate-conditioned latent structure). Karst carries genuinely proprietary machinery — carbonic-acid dissolution of soluble bedrock, the sinkhole/doline/swallet/conduit/karst-spring vocabulary, the specific substrate-process-landform compound — and within earth science it transfers as literal mechanism across limestone, dolomite, gypsum, salt, and lava-tube analogs. But beyond the geosphere "karst vulnerabilities," the "karst organisation," and "data karst" borrow only the slogan (hidden interior channels exposed by collapse) while dropping the dissolution process and landform family, and tellingly the actionable diagnosis in each comes from a different catalogued prime — latent-failure analysis, informal-network mapping, selection-bias reasoning. What genuinely recurs is the broader pattern karst instantiates — substrate-conditioned latent structure the surface systematically underrepresents, exposed by collapse — already served by escape_and_leakage plus latent-failure and iceberg-and-shadow patterns. The tension is between a coherent geological concept and the recognition that off soluble-rock terrain it is slogan, not mechanism, its cross-domain content belonging to the latent-structure parents. Diagnostic: Resolve toward the latent-structure parents (escape-and-leakage, latent failure) when "surface conceals a hidden interior exposed by failure" is the lesson off soluble rock; toward karst when there is actual soluble bedrock, dissolving groundwater, and a conduit network.

Structural–Framed Character¶

Karst sits toward the structural end of the spectrum but stops short of the pole — mixed-structural, in the same family as isostasy and the other earth-science entries: a genuine, evaluatively neutral natural process dressed in geological vocabulary. Four of the five criteria read structural. Evaluative_weight is nil: the dissolution of soluble bedrock into a conduit network is neither good nor bad; the "hazard" and "vulnerability" language attaches to human engineering aims laid over the terrain (foundations, drinking water), not to the process itself, which praises and blames nothing. Institutional_origin is none: karst is a fact of how carbonic-acid groundwater dissolves limestone, not an artifact of any survey or agency — the Kras plateau karstified for geological ages before anyone named it, and naming picked out a thing nature already does. It is not human-practice-bound: remove every geomorphologist and the Reka still sinks at Škocjan and resurges at the Timavo, voids still enlarge and collapse, blind cave fauna still evolve in isolation — the process runs on rock, water, and time, not on a judging observer. And within its proper range cross-substrate reuse is recognition, not import: from limestone to dolomite to gypsum to salt to lava-tube analogs the same substrate-process-landform mechanism is recognized intact, carrying its full apparatus (conduit inference, dye tracing, void detection, the forecast package) without translation.

What keeps it off the structural pole — and makes it one of the most home-confined entries — is vocab_travels, which it fails decisively, together with the substrate-lock behind it. The operative vocabulary — soluble bedrock, carbonic-acid dissolution, sinkhole, doline, swallet, conduit, karst spring — is irreducibly geological and floats free of no other substrate the way "hidden interior structure the surface underrepresents" does; beyond soluble-rock terrain ("karst vulnerabilities," the "karst organisation," "data karst") the terms are pure slogan and the actionable diagnosis defaults to different primes entirely, so the transfer there is analogy, not mechanism. The portable structural skeleton is substrate-conditioned latent structure: a substrate that responds to a process by accumulating hidden interior structure the surface representation systematically underrepresents, periodically exposed by collapse. That skeleton is genuinely substrate-portable, but it is exactly what karst instantiates from its umbrella (escape_and_leakage plus the latent-failure / iceberg-and-shadow patterns), not what makes "karst" itself travel: the cross-domain reach belongs to those latent-structure parents, while the dissolution mechanism and sinkhole-and-conduit vocabulary stay home. Its character: structural in skeleton — a real, evaluatively neutral, recognized-in-nature dissolution-and-hidden-conduit process — but stated in a soluble-rock vocabulary that locks it to the geological substrate, leaving it mixed-structural rather than a free-floating prime.

Structural Core vs. Domain Accent¶

This section decides why karst is a domain-specific abstraction and not a prime, and it carries the case for its domain-specificity in one place.

What is skeletal (could lift toward a cross-domain prime). Strip the geology and a thin relational structure survives: a substrate responds to a process by accumulating hidden interior structure that the surface representation systematically underrepresents, built up by a self-reinforcing feedback and periodically exposed by sudden collapse. The portable pieces are abstract — a susceptible medium, a process that carves latent interior pathways, a positive feedback that concentrates and grows them, a decoupling of the visible surface from the real interior, and betrayal of that interior through failure events. That skeleton is genuinely substrate-portable — this is substrate-conditioned latent structure — which is exactly why the entry instantiates the catalog's escape_and_leakage (the unintended-pathway face, minus the dissolution) together with the latent-failure and iceberg-and-shadow patterns. That recurrence is mechanism, but it is the core karst shares, not what makes it distinctive.

What is domain-bound. Everything that makes the landscape karst in particular is earth-science furniture and does not survive extraction. The substrate is specifically soluble bedrock (limestone, dolomite, gypsum, salt); the process is specifically carbonic-acid groundwater dissolution along joints over geological time; the interior is a conduit-and-cave network through which water routes orders of magnitude faster than porous-media flow; and the surface expression is a specific landform assemblage — sinkholes/dolines, poljes, blind valleys, disappearing streams, karst springs. From that compound come the diagnosis's teeth: the suspension of the porous-media terrain rules (topographic catchments, Darcy travel times) and the prescribed substitute toolkit (dye tracing, conduit mapping, void detection, conduit-flow models). The decisive test: remove the soluble bedrock and the carbonic-acid dissolution — take a system that is not rock at all — and "karst" is no longer a mechanism but a slogan; tellingly, the actionable diagnosis in every off-rock invocation ("karst vulnerabilities," the "karst organisation," "data karst") defaults to a different catalogued prime (latent-failure analysis, informal-network mapping, selection-bias reasoning), so the karst label adds evocation but no cross-domain intervention.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose transfer is recognition of the same mechanism, not analogy. Karst's transfer is bimodal. Within earth science it transfers as mechanism intact across every dissolution-susceptible substrate — limestone, dolomite, gypsum, salt, lava-tube analogs, and the hydrogeology and foundation-engineering disciplines that inherit its consequences — because each genuinely has soluble rock, a reactive medium, and a self-reinforcing conduit network, so the conduit inference, the suspended terrain rules, the substitute toolkit, and the forecast package (fast unfiltered transport, unannounced collapse, endemic cave fauna) carry without translation. Beyond the geosphere it is analogy: off-rock uses borrow the hidden-interior-exposed-by-collapse picture while dropping the dissolution process and landform family, and the real diagnosis comes from other primes. And when the bare structural lesson is needed cross-domain — the surface conceals a hidden interior exposed by failure — it is already carried, in more general form, by the escape_and_leakage / latent-failure / iceberg-and-shadow parents karst instantiates. The cross-domain reach belongs to those latent-structure parents; "karst," as named, packs the carbonic-acid dissolution mechanism and the sinkhole-and-conduit vocabulary that should stay home on soluble-rock terrain. (Should a future prime formalize substrate-conditioned latent structure, karst would sit cleanly under it as one substrate, and the metaphor would attach there.)

Relationships to Other Abstractions¶

Current abstraction Karst Domain-specific

Parents (4) — more general patterns this builds on

Karst is part of, typical Subsidence Domain-specific

Karst terrains commonly contain sinkhole subsidence when dissolution voids lose support.
Karst is part of Feedback Prime

Karst conduit growth contains a reinforcing flow-dissolution feedback.
Karst is part of Flow Prime

Karst contains fast channelised groundwater flow through its conduit network.
Karst is part of Network Prime

Karst contains the hidden conduit network that replaces surface-mapped drainage.

Hierarchy paths (6) — routes to 5 parentless roots

Karst → Subsidence → Isostasy → Equilibrium → Fixed Point

Show alternative paths (5)

Not to Be Confused With¶

Porous-media (Darcy) aquifer. The contrast case karst is defined against: an aquifer where water moves slowly and diffusely through pore spaces, so travel time scales with distance and porosity and the medium filters and retards contaminants. Karst hydrology is fast, channelised conduit flow that bypasses that filtration — a contaminant reaches a spring in hours, not months. Applying Darcy models to karst is the signature error. Tell: does water move diffusely through pores at predictable velocity (porous-media), or through discrete conduits orders of magnitude faster, unfiltered (karst)?
Pseudokarst. Landforms that look like karst — caves, sinkholes, sinking streams — but form without dissolution of soluble rock: lava tubes in basalt, piping/suffosion cavities in unconsolidated sediment, talus caves. The karst diagnosis specifically requires the carbonic-acid dissolution of soluble bedrock; pseudokarst reaches the same morphology by a different process, so the conduit-hydrology inferences may or may not follow. Tell: is the void network produced by chemical dissolution of soluble rock (true karst), or by mechanical or volcanic processes that merely mimic the landforms (pseudokarst)?
Thermokarst. Collapse-and-subsidence topography — pits, hollows, irregular ground — produced by the thaw of ice-rich permafrost, which shares the "karst" name and the collapse morphology but involves melting ground ice, not bedrock dissolution or a conduit network. Same surface signature (irregular collapse depressions), entirely different mechanism and substrate. Tell: is the collapse driven by dissolving soluble rock into conduits (karst), or by thawing ground ice removing volume (thermokarst)?
Suffosion / piping sinkholes. Sinkholes formed by the mechanical washing-out of loose material into pre-existing openings (internal erosion), rather than by dissolution enlarging conduits. They can co-occur in karst (soil raveling into bedrock voids) but the piping mechanism is erosional, not chemical. Confusing them misattributes the rate and controls of collapse. Tell: is the void enlarging by dissolution of the rock itself (karst proper) or by mechanical transport of particles into an existing gap (suffosion/piping)?
The metaphorical "karst" grafts (organizational, data, security "karst"). Off-rock uses — "karst vulnerabilities," the "karst organisation," "data karst" — borrow the slogan (hidden interior channels exposed by collapse) while dropping the soluble bedrock and dissolution that make karst coherent, and the actionable diagnosis in each actually comes from a different prime (latent-failure analysis, informal-network mapping, selection-bias reasoning). These are slogan, not instances. Tell: is there actual soluble rock and dissolving groundwater (a karst instance), or is "karst" naming a hidden-structure-exposed-by-collapse resemblance whose real diagnosis belongs to another prime (a metaphor)?
Escape-and-leakage / latent-failure (the parents). The substrate-neutral umbrella karst instantiates — a substrate accumulates hidden interior structure the surface underrepresents, exposed by collapse (escape_and_leakage plus latent-failure / iceberg-and-shadow patterns). This is what carries the "surface conceals a hidden interior exposed by failure" lesson off soluble rock; karst is its dissolution-specific geological case. Tell: strip away the carbonic-acid dissolution and the sinkhole-conduit vocabulary and what remains — "hidden interior structure the surface misses, betrayed by failure" — is the latent-structure parent, treated more fully elsewhere; carry it (not "karst") beyond the geosphere.

Neighborhood in Abstraction Space¶

Karst sits in a sparse region of the domain-specific corpus (62^nd percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Geologic Landforms & Crustal Deformation (12 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-07-12