False Precision¶

The measurement-communication fallacy of expressing a quantity with more significant figures or tighter bounds than the underlying evidence supports — a mismatch between the form of the claim and its warrant, read by audiences as unearned precision of knowledge.

Core Idea¶

False precision is the measurement-communication fallacy of expressing a quantity with more significant figures, tighter bounds, or finer resolution than the underlying measurement, estimate, or model actually supports — reporting "the population is 1,247,683" when the true uncertainty is plus or minus a hundred thousand, or "the project will complete on 14 March" when the schedule is accurate only to roughly a quarter. The mechanism is a mismatch between the form of the claim and its epistemic warrant: the extra digits are not fabricated in the sense of being deliberately invented, but they encode a precision that the evidence does not underwrite, and the conventions of numeric literacy lead audiences to read precision of representation as evidence of precision of knowledge. The pattern propagates in three steps — a raw signal with real but bounded precision (a measurement, a forecast, a model output); a presentation choice that carries more significant figures than the signal supports; and a cognitive uptake in which downstream readers treat the reported precision as genuine and update, plan, and decide as if it were. Laboratory significant-figures discipline exists precisely to interrupt this propagation: the precision of a calculated result cannot exceed the precision of its least precise input. False precision is distinct from outright fabrication (the numeric value may be accurate in expectation), from overconfidence in the probability-calibration sense (which concerns credences, not point-estimate resolution), and from the underlying measurement error itself (which is the physical or sampling fact); it is specifically the communicative failure of not disclosing or honoring that measurement error in the presentation.

Structural Signature¶

Sig role-phrases:

the underlying signal — a measurement, estimate, or model output with real but bounded precision
the least precise input — the coarsest contributor, which caps the precision any derived figure can legitimately carry
the presentation choice — reporting more significant figures, tighter bounds, or finer resolution than the signal supports
the form-content mismatch — the defect: the representation advertises a knowledge the evidence does not back, though the central value may be right
the digits-as-rigor convention — the numeric-literacy habit by which audiences read precision of representation as precision of knowledge
the cognitive uptake — downstream readers treat the reported resolution as genuine and plan, update, and decide on it
the propagation rule — the governing inequality: a result can be no more precise than its coarsest input, so surplus digits are spurious
the corrective — fix the presentation, not the measurement or the credence: round, interval, or range until form matches warrant

What It Is Not¶

Not fabrication. Nothing is invented but the precision. The reported value may be accurate in expectation — "1,247,683" can sit near the true count — and the defect is only the surplus digits, which advertise a resolution the evidence does not back. The corrective is to strip the digits, not to dispute the figure.
Not the underlying measurement error itself. Measurement error is the physical or sampling fact that the quantity is known only to within some band; false precision is the communicative failure to disclose or honor that band in the presentation. One can have large measurement error and report it honestly with no false precision, or small error reported over-precisely.
Not overconfidence in the calibration sense. Calibration concerns the probability attached to a claim — whether a stated 90 % credence is borne out. False precision concerns the resolution of a point estimate: how many significant figures it carries, not how confident the speaker is in a binary proposition. The two failures live on different axes.
Not a claim that the number is wrong. Pointing out false precision neither asserts nor implies the central value is mistaken; it says the form overstates the warrant. An estimate can be exactly right and still be reported with spurious precision, and the fix leaves the estimate intact at its true resolution.
Not vindicated by more decimal places signaling more rigor. The numeric-literacy convention that finer digits mean better knowledge is precisely the reflex the concept indicts. Extra resolution is warranted only when the inputs supply it; absent that, additional digits are not rigor but its counterfeit.

Scope of Application¶

False precision lives across the quantitative-argument and measurement-communication subfields of logic and reasoning — wherever numbers are reported to a reader who reads precision of representation as precision of knowledge. Its in-domain reach is broad because the propagation rule (a result no more precise than its coarsest input) is measurement bookkeeping that applies to any reported figure; the loose, audience-free version — a process "more precise than its inputs" — belongs to map_and_territory / measurement-uncertainty, not here.

Scientific reporting — significant-figures discipline proper, and the over-precise digital readout from an instrument with known calibration limits.
Forecasting and risk modeling — a "2.34 %" GDP forecast carrying digits its two-point standard error cannot back; risk scores reported to decimals from crude inputs.
Schedule and budget estimation — software deadlines pinned to the day for work whose historical variance runs in months, and cost estimates to the dollar from inputs uncertain by tens of percent.
Algorithmic scoring — credit, recidivism, and ranking outputs reported to two decimals when the model's calibration is good only to tens of points.
Historical and demographic data — to-the-day dates for roughly-known events and seven-figure populations for pre-census societies, where the resolution of the figure outruns the record.

Clarity¶

Naming false precision installs a habit the conventions of numeric literacy actively suppress: separating what value? from to what precision?. A reader trained to treat extra digits as a token of rigor reads a single number and absorbs both its magnitude and an implied confidence in one gulp; the label pries those apart and licenses the question "how was this measured, and to what resolution?" before any number is believed. What becomes legible is a form–content mismatch — a defect not in the reported value, which may be right in expectation, but in the representation, which advertises a knowledge the evidence does not back. That reframing is what lets an analyst reject the spurious digits while keeping the estimate, instead of having to dispute the underlying figure.

The distinction earns its keep by sorting false precision out of a neighborhood of look-alikes it is routinely merged with. It is not fabrication — nothing is invented but the precision. It is not overconfidence in the calibration sense, which concerns the probability attached to a claim, not the resolution of a point estimate. It is not measurement error, the physical or sampling fact itself; it is the communicative failure to disclose or honor that error in the presentation. And it is the mirror image of vague over-hedging, which understates precision the report actually has. Holding these apart tells a practitioner precisely where the corrective belongs — not in re-measuring, not in re-deriving the credence, but in the act of reporting: significant-figures discipline, uncertainty intervals, honest ranges, governed by the rule that a result can be no more precise than its least precise input. The sharp question the concept hands downstream users is therefore not "is this number correct?" but "does its precision survive propagation from the inputs?" — a check that disarms the extra digits before they are carried, falsely weighted, into the next decision.

Manages Complexity¶

Suspect numeric claims arrive from every quarter and in every guise — a market sized to seven significant figures from three interviews, a GDP forecast of "2.34 %" from a model whose standard error is two points, a project deadline pinned to the day for work whose variance runs in months, a recidivism score to two decimals from a crude classification, a battle dated to the day in a pre-census era — and a reader could try to vet each by reconstructing how it was produced and what its inputs could bear, a fresh forensic exercise per number. Naming false precision collapses that endless audit to a single comparison run on one quantity: the resolution of the report against the resolution its inputs can support. The diversity of subject matter — demography, economics, scheduling, scoring, history — drops out, because none of it changes the test; the only thing the reader tracks is whether the digits, bounds, or decimals advertised survive propagation from the least precise input. The governing rule is a single inequality — a result can be no more precise than its coarsest input — so the verdict reads off directly: if the reported precision exceeds what the inputs license, the extra digits are spurious however accurate the central value; if it does not, the figure may be taken at face. That branch is what makes the corrective unambiguous and identical across cases — not re-measure, not re-derive the credence, but round, interval, or range the presentation until form matches warrant — and it is what lets the analysis proceed unburdened: once the surplus precision is named and stripped, the estimate can be carried forward at its true resolution without the false weight the extra digits would otherwise have lent it downstream. The reader stops adjudicating each number's pedigree and instead runs one resolution check, reading "trust to here, no further" off the comparison of two precisions.

Abstract Reasoning¶

False precision licenses a set of moves for auditing numeric claims, all running off a single resolution comparison — the resolution of the report against the resolution its inputs can support — governed by the propagation rule that a result can be no more precise than its coarsest input.

Diagnostic — separate value from precision, then check resolution against the inputs. The signature move pries apart two things numeric literacy fuses, what value? from to what precision?, and asks of any reported number how it was measured and to what resolution before believing it. From the gap between a figure's advertised resolution and its warrant the analyst diagnoses a form–content mismatch: a market sized to seven significant figures from three interviews, a forecast of "2.34 %" from a model whose standard error is two points, a deadline pinned to the day for work whose variance runs in months. The diagnostic locates the defect not in the central value, which may be right in expectation, but in the representation, which advertises a knowledge the evidence does not back — and it identifies the load-bearing comparison by tracing to the least precise input, since that is what caps the precision the report can legitimately carry.

Interventionist — fix the presentation, not the measurement or the credence. The corrective follows fixed from the diagnosis and is unambiguous across every case: not re-measure, not re-derive the probability, but round, interval, or range the presentation until form matches warrant. Restate "USD 12,847,300,000" as "roughly USD 10–15 billion," report a forecast with its standard error, give a schedule as a range honoring its variance — and the prediction is that the spurious digits are disarmed before they propagate, the estimate surviving at its true resolution without the false weight the extra figures lent it. The intervention is precisely placed: because the failure is communicative rather than physical or epistemic-in-the-credence-sense, the move belongs in the act of reporting — significant-figures discipline, uncertainty intervals, honest ranges — and each is a claim that the rounded form now matches the warrant, checkable against the propagation rule.

Boundary-drawing — separate false precision from fabrication, overconfidence, measurement error, and over-hedging. The construct's central discipline is to keep apart the value from its precision, and the move includes rejecting the spurious digits while keeping the estimate rather than disputing the underlying figure. The construct also bounds itself against a neighborhood of look-alikes: it is not fabrication (nothing is invented but the precision), not overconfidence in the calibration sense (which concerns the probability attached to a claim, not the resolution of a point estimate), not measurement error (the physical or sampling fact itself), but specifically the communicative failure to disclose or honor that error in the presentation — and it is the mirror image of vague over-hedging, which understates precision the report actually has. Holding these apart fixes where the corrective belongs and prevents misrouting a presentation defect into a re-measurement or a re-calibration it does not need.

Predictive — precision propagates no further than the coarsest input, and audiences over-trust over-precise figures unless told. The propagation rule is predictive: the precision of a calculated quantity cannot exceed the precision of its least precise input, so the analyst forecasts how many significant figures a result may carry through a calculation before computing it, and predicts which reported digits are spurious by comparing them to the input that bounds them. The construct also predicts the cognitive uptake: downstream readers systematically read precision of representation as precision of knowledge, so over-precise inputs will be over-trusted and carried, falsely weighted, into the next decision unless the surplus precision is named and stripped first. The single resolution check thereby yields a forward verdict — if the reported precision exceeds what the inputs license the extra digits are spurious however accurate the central value, if it does not the figure may be taken at face — letting the reader read "trust to here, no further" off the comparison of two precisions rather than reconstructing each number's pedigree.

Knowledge Transfer¶

Within quantitative argumentation and measurement communication false precision transfers as mechanism, because what travels is a single resolution check governed by the propagation rule, and only the kind of number changes. Wherever a figure is reported, the same test — does the advertised precision survive propagation from the least precise input? — and the same corrective — round, interval, or range the presentation until form matches warrant — apply unchanged. In scientific reporting it is significant-figures discipline proper (a calculated result no more precise than its coarsest input); in forecasting and risk modeling it is the "2.34 %" forecast carrying digits its two-point standard error cannot back; in schedule and budget estimation it is the to-the-day deadline on work whose variance runs in months; in algorithmic scoring it is the two-decimal recidivism or credit score from a crude classification; in historical and demographic data it is the to-the-day date for a roughly-known event or the seven-figure population of a pre-census society. Across all of these the vocabulary (form-content mismatch, resolution, propagation) and the diagnostic ("how was this measured, and to what resolution?") carry intact, and the corrective stays in the act of reporting rather than in re-measuring. The construct also stays cleanly distinct from its in-domain look-alikes — fabrication, calibration-overconfidence, raw measurement error, over-hedging — in every one of these subfields, which is part of what makes the single check portable.

The honest characterization here is partly (C) instrument / measure and partly (B) shared abstract mechanism, and it is worth separating the two. The propagation rule itself — that a derived quantity's precision cannot exceed its coarsest input's — is not a causal mechanism but a piece of measurement bookkeeping, and it transfers literally wherever quantities are combined and reported, regardless of substrate: any field that propagates uncertainty through a calculation honors the same inequality, and that is instrument-reach, not metaphor. What does not travel without an audience is "false precision" as a fallacy, because it presupposes a representational system (numerals, decimal notation, significant figures) and a reader who reads precision-of-representation as precision-of-knowledge. Remove the reporter and the audience and there is no false precision of a temperature or a population — only the underlying measurement error, which is a different thing. So the cross-domain phenomenon the fallacy points at — the resolution of a representation overstating the resolution of its referent — is a (B) shared mechanism whose general form is already carried by measurement_uncertainty_and_observational_noise (the bounded fidelity of any measurement), epistemic_humility (matching expressed confidence to warrant), and the map_and_territory / representation-fidelity family (a model's resolution should not exceed the territory's). The home-bound cargo is everything specifically communicative and numeric: the digits-as-rigor convention, the significant-figures remedy, the cognitive uptake by which a reader over-trusts an over-precise figure. The boundary worth marking is therefore not mechanism-versus-metaphor but instrument-reach versus over-reading: the propagation rule and the resolution check apply wherever numbers carry uncertainty, but stretching "false precision" to non-representational situations — a process that is "more precise than its inputs," a system that "over-resolves" — drops the audience that the fallacy is defined around and should instead be stated as map_and_territory or measurement-uncertainty (see Structural Core vs. Domain Accent).

Examples¶

Canonical¶

Significant-figures propagation is the textbook instance. Suppose a rectangular plot is measured as 12.3 m by 4.5 m. The naive area is 12.3 × 4.5 = 55.35 m². But the second measurement carries only two significant figures — it is known to the nearest tenth of a metre, roughly ±0.05 m — so the product cannot be more precise than that coarsest input. Reporting "55.35 m²" advertises resolution to hundredths of a square metre that the tenth-of-a-metre measurements cannot back. The disciplined report rounds to two significant figures: 55 m². The central value is not disputed; only the surplus digits, which encode a knowledge the evidence does not underwrite, are stripped.

Mapped back: The two length readings are the underlying signal, each with real but bounded precision; 4.5 m is the least precise input, capping the result at two significant figures. Writing "55.35" is the presentation choice that outruns the signal — the form–content mismatch, since the area may be right in expectation while the digits overstate warrant. The propagation rule (no more precise than the coarsest input) delivers the corrective: round the presentation to 55 m², not re-measure the plot.

Applied / In Practice¶

Election polling routinely manufactures false precision. A survey of about 1,000 respondents carries a margin of sampling error near ±3 percentage points at 95% confidence, yet results are habitually reported to a tenth of a point — "Candidate A 47.3%, Candidate B 46.9%." Readers and headline writers then treat the 0.4-point gap as a real lead, though it sits far inside the error band and would not survive resampling. The honest presentation rounds to whole points and reports the margin of error, so the two candidates read as a statistical tie rather than a decided race.

Mapped back: The poll's point estimates are the underlying signal; the ±3-point sampling error is what caps their resolution, so a reported tenth is spurious under the propagation rule. The tenths digit exploits the digits-as-rigor convention — finer numbers read as firmer knowledge — driving the cognitive uptake by which a within-noise 0.4-point difference is planned and reported on as a genuine lead. The corrective lives in the reporting act: round to whole points and disclose the interval.

Structural Tensions¶

T1: Accurate value versus unwarranted form (the defect that hides inside correct numbers). False precision's defining subtlety is that the number can be right. "1,247,683" may sit squarely near the true count; the defect is in the surplus digits, not the central value. This is what makes the corrective so cleanly targeted — strip the digits, keep the estimate, no re-measurement — but it is also what makes the fallacy hard to catch: nothing about the figure is wrong in the way an error is wrong, so a reader checking only "is this correct?" passes it through. The very feature that lets the concept quarantine the presentation defect from the value is the feature that lets that defect travel undetected inside numbers that survive every accuracy check. Diagnostic: Is the objection that the central value is mistaken, or that its form advertises a resolution the inputs cannot back while the value itself may be fine?

T2: The digits-as-rigor convention as communicative shorthand versus its counterfeiting (the signal that betrays). Numeric literacy trains readers to take more significant figures as a token of firmer knowledge, and across honest reporting that convention is efficient — it lets a resolution be communicated in the number's own form without a separate uncertainty statement. False precision is parasitic on exactly this efficiency: it counterfeits rigor by supplying the digits the convention reads as warrant, with no warrant behind them. The concept therefore indicts a convention it cannot simply abolish, because the same shorthand does real communicative work when honored. The fix is not to stop reading digits as resolution but to make the digits match the warrant — which leaves the counterfeiting channel permanently open wherever discipline lapses. Diagnostic: Are these digits carrying resolution the inputs supply, or are they borrowing the convention's credit to advertise a knowledge that was never earned?

T3: Rounding toward honesty versus rounding into over-hedging (the corrective's own opposite failure). The remedy — round, interval, or range the presentation until form matches warrant — has a mirror-image failure the concept names explicitly: vague over-hedging, which understates the precision the report actually has. Strip too few digits and the counterfeit rigor survives; strip too many, or widen the band past the evidence, and genuine resolution the inputs do support is thrown away, misleading downstream users in the opposite direction. "Round until form matches warrant" is not a floor but a target with error on both sides, and locating it requires the same input-resolution judgment that diagnosing the original defect did. The corrective is not a safe default toward coarseness; it is a calibration that can overshoot. Diagnostic: Does the rounded form still carry every digit the least precise input licenses, or has the correction discarded real resolution the evidence supports?

T4: Resolution of a point estimate versus confidence in a claim (two axes that wear the same face). False precision lives on the resolution axis — how many significant figures a point estimate carries — while calibration-overconfidence lives on the credence axis — whether a stated 90% is borne out. The two present almost identically as "claiming to know more than you do," and are routinely merged, yet they take different corrections: false precision is fixed in the act of reporting by rounding, calibration-overconfidence by re-deriving the probability. The tension is that a single over-sure numeric claim can carry both defects at once, and naming only one misroutes the fix — round a figure whose real problem is an overstated credence and the over-confidence survives; re-calibrate a credence whose real problem is spurious digits and the digits remain. Diagnostic: Is the overstatement in the resolution of a point estimate (round it) or in the probability attached to a proposition (re-calibrate it) — or both, needing both fixes?

T5: Autonomy versus reduction (a numeric fallacy or an instance of representation-fidelity). False precision splits cleanly in two along the transfer axis. Its propagation rule — a derived quantity's precision cannot exceed its coarsest input's — is substrate-free measurement bookkeeping that applies literally wherever numbers are combined, an instrument that reaches every quantitative field. But "false precision" as a fallacy presupposes a representational system (numerals, significant figures) and a reader who reads precision-of-representation as precision-of-knowledge; remove the audience and there is no fallacy, only measurement error. So the portable structure — a representation's resolution overstating its referent's — already belongs to map_and_territory, measurement_uncertainty_and_observational_noise, and epistemic_humility, while the named fallacy stays bound to numeric communication with an over-reading audience. The tension is between an instrument-reach that is universal and a fallacy that is audience-bound. Diagnostic: Resolve toward map_and_territory / measurement-uncertainty when the situation is a representation over-resolving its referent with no reader in the loop; toward named false precision when a numeric report and an audience reading digits as knowledge are both present.

Structural–Framed Character¶

False precision sits at the framed pole of the structural–framed spectrum as a fallacy, though it is the most instructive framed-pole case in this cluster because one component of it is genuinely instrument-portable — a split the criteria expose rather than blur. On evaluative weight the fallacy scores framed: to charge "false precision" is to render a verdict that a report is defective, not to name a neutral mechanism the way measurement_uncertainty names the bounded fidelity of any reading. On human-practice-bound it is framed in the decisive sense: the fallacy presupposes a representational system (numerals, decimal notation, significant figures) and a reader who reads precision-of-representation as precision-of-knowledge, and it dissolves the instant that practice is removed — take away the reporter and the over-reading audience and there is no false precision of a temperature or a population, only the underlying measurement error, which is a different and observer-free thing. Institutional origin is framed: the digits-as-rigor convention, the significant-figures discipline offered as remedy, and the whole apparatus of numeric-literacy reporting are artifacts of a communicative tradition, not facts of nature. On vocab-travels and import-vs-recognize the fallacy is bimodal but stays framed at its core: stretching "false precision" to a non-representational situation — a process "more precise than its inputs," a system that "over-resolves" — drops the defining audience and becomes import-by-analogy.

The portable structural skeletons here are genuinely two, and the entry earns the plural by demonstrating each does distinct work. The first and primary is representation-resolution overstating referent-resolution — a map whose fineness exceeds the territory's — which is exactly what the fallacy instantiates from its umbrella parents map_and_territory, measurement_uncertainty_and_observational_noise, and epistemic_humility; the cross-domain reach of that lesson belongs to those neutral primes, while the fallacy's home-bound cargo (the digits-as-rigor convention, the significant-figures remedy, the reader's over-uptake) stays put. The second, distinct and load-bearing, is the propagation rule — a derived quantity can be no more precise than its coarsest input — which is not a causal mechanism but substrate-free measurement bookkeeping that transfers literally, as an instrument, to any field that combines uncertain quantities, with no audience required. That second skeleton is why the honest boundary for this entry is not mechanism-versus-metaphor but instrument-reach versus over-reading: the propagation inequality travels universally as bookkeeping, yet it too is a general tool the fallacy applies rather than something "false precision" uniquely owns. Its character: a normatively charged, audience-constituted numeric-communication fallacy whose only substrate-spanning content is a representation-fidelity skeleton borrowed from its parents plus a piece of measurement bookkeeping that is universal precisely because it is not the fallacy at all.

Structural Core vs. Domain Accent¶

This section decides why false precision is a domain-specific abstraction and not a prime — and here the skeleton is genuinely doubled, so being exact about the two portable parts matters as much as marking what stays home.

What is skeletal (could lift toward a cross-domain prime). Strip the numeric communication and two distinct portable structures survive, and the entry earns the plural. The first and primary is the resolution of a representation overstating the resolution of its referent — a map whose fineness exceeds the territory's, an expressed confidence outrunning its warrant. That is genuinely substrate-portable and recurs wherever a model or reading claims more fidelity than its source supports, which is exactly why it is carried by the parents the entry instantiates — map_and_territory (a representation should not out-resolve what it represents), measurement_uncertainty_and_observational_noise (the bounded fidelity of any reading), and epistemic_humility (match expressed confidence to warrant). The second, distinct and load-bearing, is the propagation rule — a derived quantity can be no more precise than its coarsest input. That is not a causal mechanism but substrate-free measurement bookkeeping that transfers literally wherever uncertain quantities are combined. Both are the core the entry shares, not what makes it distinctive.

What is domain-bound. Everything that makes this specifically false precision the fallacy is measurement-communication furniture and none of it survives extraction. It presupposes a representational system — numerals, decimal notation, significant figures — and a reader who reads precision-of-representation as precision-of-knowledge; the defect is a communicative one, not a physical or credential one. The worked apparatus is discipline-internal: the digits-as-rigor convention, the significant-figures remedy, the cognitive-uptake step by which an over-precise figure is over-trusted, and the empirical cases (the 55.35 m² plot, the tenth-of-a-point poll, the seven-figure pre-census population). The decisive test: remove the reporter and the over-reading audience and there is no false precision of a temperature or a population — only the underlying measurement error, which is an observer-free and different thing. The audience is the part that stays home; strip it and the fallacy dissolves, leaving behind only the two portable skeletons, neither of which is the fallacy itself.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose cross-domain transfer is recognition of the same mechanism, not analogy — and here the honest boundary is not mechanism-versus-metaphor but instrument-reach versus over-reading. Within quantitative argumentation and measurement communication false precision moves intact as a single resolution check — scientific reporting's significant figures, forecasting's over-precise "2.34 %", scheduling's to-the-day deadline, algorithmic scoring's two-decimal risk output, history's to-the-day date — because every arena supplies a reported number and a reader, and the corrective always lives in the act of reporting. The propagation rule itself even travels literally, as bookkeeping, to any field that combines uncertain quantities, but that reach belongs to the rule as a universal instrument, not to the fallacy. Beyond a reporting-and-reading situation the fallacy travels only by over-reading: calling a process "more precise than its inputs" or a system that "over-resolves" drops the defining audience and becomes analogy. And when the bare structural lesson is needed cross-domain — a representation's resolution should not outstrip its referent's — it is already carried, in more general and evaluatively neutral form, by map_and_territory, measurement_uncertainty_and_observational_noise, and epistemic_humility. The cross-domain reach belongs to those parents (and, for the bookkeeping, to the propagation rule as an instrument); "false precision," as a named fallacy, carries the numeric-communication audience that should stay home.

Relationships to Other Abstractions¶

Current abstraction False Precision Domain-specific

Parents (2) — more general patterns this builds on

False Precision is a kind of Representation Prime

False Precision is the defective numeric-representation species whose displayed resolution exceeds the fidelity its target and source warrant.
False Precision is part of Measurement Uncertainty and Observational Noise Prime

False Precision contains measurement uncertainty as the evidence-bounded resolution that the displayed digits or bounds fail to honor.

Hierarchy paths (3) — routes to 3 parentless roots

False Precision → Representation → Abstraction

Show alternative paths (2)

Not to Be Confused With¶

Fabrication (fudged or invented data). The manufacture of a numeric value with no measurement behind it — a figure conjured to deceive. False precision keeps a value that is real and may be right in expectation; only the surplus digits are unearned, advertising a resolution the evidence does not back. Tell: is the central value itself invented (fabrication), or is the value sound and only its reported resolution inflated (false precision)?
Overconfidence / overprecision (calibration sense). The metacognitive bias of attaching too high a probability to a claim — a stated 90% credence that is not borne out. This lives on the credence axis; false precision lives on the resolution axis, how many significant figures a point estimate carries rather than how sure the speaker is of a proposition. The two can co-occur in one over-sure numeric claim yet take different corrections. Tell: is the overstatement in the probability attached to a claim (overconfidence, fixed by re-calibrating) or in the number of digits a point estimate reports (false precision, fixed by rounding)?
Measurement error / observational noise. The physical or sampling fact that a quantity is knowable only to within some band — an observer-free property of the instrument and the world. False precision is the communicative failure to disclose or honor that band in the presentation; one can have large measurement error yet report it honestly, or small error reported over-precisely. Tell: is the band itself the issue (measurement error, present with no reporter), or is a known band being hidden behind an over-fine report (false precision, which needs a reporter and a reader)?
Over-hedging / spurious vagueness. The mirror-image defect: understating the precision a report actually has — widening a range past what the inputs support, discarding resolution the evidence licenses. False precision advertises more resolution than warranted; over-hedging advertises less, misleading downstream users in the opposite direction. Tell: does the form carry more digits than the coarsest input supports (false precision) or fewer (over-hedging)? "Round until form matches warrant" is a target with error on both sides.
The significant-figures propagation rule. The bookkeeping inequality — a derived quantity is no more precise than its coarsest input — that the entry uses as its test. This is a substrate-free instrument that travels literally wherever uncertain quantities are combined, with no audience required; false precision is the fallacy that arises only when a numeric report meets a reader who reads digits as knowledge. Tell: strip the reporter and the over-reading audience and the propagation rule still holds as arithmetic, but there is no false precision — the rule is the tool, the fallacy is its violation before an audience.
The representation-fidelity parents (map_and_territory, measurement_uncertainty_and_observational_noise, epistemic_humility). The broad, evaluatively neutral primes — a representation should not out-resolve its referent, any reading has bounded fidelity, expressed confidence should match warrant — that false precision instantiates rather than duplicates. Tell: strip the numerals, the digits-as-rigor convention, and the over-reading reader and what remains — a representation finer than its source supports — is already these parents at work, not a fallacy. (Treated fully in the sections above.)

Neighborhood in Abstraction Space¶

False Precision sits in a moderately populated region (48^th percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Social Perception & Self-Referential Bias (23 abstractions)

Nearest neighbors

Asymmetric-Amplitude Error Correction — 0.86
Implied Reader — 0.85
Headline-Body Mismatch — 0.84
Channel Richness — 0.84
Mandela Effect — 0.84

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