Scientific Progress

The Philosophical Question of Progress

Scientific progress is an evaluative relation between states of inquiry. To say that a field has progressed is to judge that a later configuration of its questions, evidence, explanations, instruments, practices, or achievements is better in a relevant respect than an earlier configuration. The meaning of "better" is the philosophical issue. Chronological succession merely records change. More papers and larger laboratories record expansion. A successful intervention records a practical consequence. None of these observations independently settles whether inquiry has improved in its cognitive relation to the world.

Niiniluoto's systematic account of scientific progress separates conceptual questions about the meaning of progress, methodological questions about its recognition, and historical questions about its occurrence. This separation governs the present analysis. One cannot prove a definition simply by recounting a celebrated discovery, because the discovery is already being selected by a criterion of significance. Equally, one cannot establish a history of progress merely by stipulating an attractive ideal of science. Evidence about actual research episodes must be evaluated under explicit criteria.

The inquiry develops a comparative interpretation of the established literature. It treats scientific progress as a family of judgments whose epistemic, explanatory, methodological, social, and practical dimensions must be identified before particular claims are assessed. The result is an argumentative synthesis: an epistemic gain remains central to science as knowledge-seeking, while different accounts isolate different achievements and expose different failures. This conclusion does not introduce a new named theory or claim priority over the accounts under examination.

Definition, Scope, and Levels of Analysis

The working definition of scientific progress in this monograph is an evidence-grounded improvement in the cognitive capacities or achievements of scientific inquiry, evaluated with respect to a stated aim and a specified research context. This definition is an interpretive orientation for comparing existing theories. Its terms deliberately invite further specification. An "improvement" can concern justified knowledge, accurate theoretical representation, explanatory grasp, reliable methods, or discovery of productive problems. An "evidence-grounded" judgment must show which observations, arguments, instruments, and checks support the comparison.

Three levels need separate treatment. A local episode may yield a reliable observation. A research program may combine many episodes into a durable method, vocabulary, and explanatory repertoire. A scientific community may gain capacities for correction and cumulative access to results. A failure at one level can coexist with improvement at another: an individual hypothesis can be refuted while the broader field improves its discrimination between alternatives. Similarly, a superior instrument can improve the prospects of a program before it produces a confirmed new result.

The knowledge-based account of progress identifies a demanding epistemic endpoint, whereas Shan's functional account emphasizes productive problem definition and solution. Both address improvements, but they select different units and temporal horizons. The point of this conceptual division is precision: historical narratives should specify whether progress is attributed to a claim, technique, research program, discipline, institutional arrangement, or eventual application. An unqualified announcement that "science advanced" frequently hides such shifts in unit of evaluation.

Kitcher's The Advancement of Science (1993) places disciplinary practices, cognitive ideals, and objectivity within a historical explanation of scientific advancement. The bibliographic record supports attribution of the work and its subject; the present comparative use is explicitly interpretive, and a restricted full text is not represented as directly examined.

Scientific Aims and Evaluative Standards

Progress presupposes an aim. If science is understood as the production of justified knowledge, the assessment of progress looks different from an assessment that prioritizes truthlikeness, understanding, predictive control, or useful problem solutions. This dependence on scientific aims gives the debate its axiological structure. A methodological rule may be valuable because it advances an aim, yet the rule itself may vary across contexts. The ambition to specify one invariant score for all fields therefore risks treating a choice of aims as an empirical measurement.

The survey of competing scientific aims identifies truth, problem-solving, explanation, and knowledge as candidates or dimensions. A truth-oriented conception asks whether our representations improve. A problem-centered conception asks what obstacles have been removed. A noetic conception asks whether researchers grasp relevant dependencies. A pragmatic conception asks whether scientific practices produce better interventions or reliable operations. These values overlap through actual research, but their logical definitions diverge.

Scientific aims also change historically. A field may learn that an earlier question was ill-posed, that its measurement categories were too coarse, or that an apparent anomaly exposes a missing variable. Such episodes are difficult to score by counting answers to a frozen problem list. Rational assessment must therefore examine the quality of newly defined questions alongside solutions. The problem of changed aims also creates accountability: an institution can announce success by replacing an ambitious research goal with an easier one. A credible progress claim explains the revision and shows what has genuinely become possible.

From Accumulation to Historical Reconstruction

A cumulative picture treats scientific growth as an enlarging stock of observations and truths. Its intuitive attraction is clear: reproducible findings and improved measurements can be preserved, taught, and extended. Actual histories also include abandonment of once-authoritative classifications, revisions of theoretical language, new experimental domains, and changes in which questions count as solvable. The history of science is therefore a history of accumulation and transformation together.

Kuhn's Structure of Scientific Revolutions showed how normal inquiry depends on settled exemplars and a shared disciplinary matrix, while revolutionary episodes can revise fundamental assumptions. The historical reconstruction of Kuhn's argument emphasizes that scientific revolutions need not preserve every achievement of an earlier framework. "Kuhn loss" names cases where a replacement loses explanatory resources, even while solving pressing anomalies and opening new puzzles. Such loss cannot be assessed through a simple series of growing inventories.

The philosophical gain from a historical reconstruction lies in making the comparison multidimensional and temporally situated. An older theory may remain an adequate engineering approximation in a bounded regime after its general ontology is rejected. A new theory may initially offer sparse practical applications while offering deeper explanations and stronger empirical relations. The historian records how both achievements were produced and recognized; the philosopher specifies what makes a transition progressive in each respect. A defensible history requires both forms of attention.

Popper and the Logic of Critical Improvement

Popper's critical rationalism treats fallibility and severe criticism as conditions for scientific learning. A conjecture earns standing through exposure to empirical tests capable of showing it false. Successful tests do not confer certainty, but failures can remove entire classes of proposed explanation. Progress can therefore take the form of a reduction in error and the replacement of less informative or less successful conjectures by more demanding ones. The problem of induction remains, because repeated past success does not deductively guarantee the future.

Niiniluoto's analysis of Popperian truthlikeness records a crucial technical limitation: Popper's original formal explication of verisimilitude did not yield a satisfactory general measure. The philosophical idea of improvement through theories that approximate the truth survived, with later formal alternatives. Treating falsification as an infallible laboratory switch would also distort practice. Experiments presuppose auxiliaries, instruments, data-processing conventions, and judgments about error; apparent refutation often concerns a package of assumptions.

The durable contribution of critical rationalism is methodological: inquiry can progress by learning that an attractive answer fails. The correction of error has an epistemic value even when a successor theory remains uncertain. Criticism also improves its own conditions through new instruments and cross-checking procedures. An institution that rewards criticism only when a more marketable result immediately follows creates an incentive to conceal the very errors whose discovery can make a field more reliable.

Kuhn: Normal Science, Crisis, and Revolution

Kuhn explains scientific development through organized research traditions. During normal science investigators refine measurements, extend applications, resolve puzzles, and train successors in exemplars. These activities yield significant achievements precisely because researchers ordinarily trust the framework enough to use it intensively. Persistent anomalies may destabilize that trust, leading to competition among reorganized theoretical possibilities. A revolutionary transformation changes basic classifications and explanatory relations, along with methods of teaching and judging success.

The account of normal and revolutionary science establishes that Kuhn did not regard all scientific change as irrational or deny the possibility of progress. Revolutionary successors must usually address urgent anomalies and retain considerable effective puzzle-solving power. But the successor may also lose earlier solutions, and the change of evaluative standards complicates claims of monotonic improvement. The history of incommensurability clarifies that incommensurability is a problem of translation, conceptual articulation, and comparison, rather than a prohibition on rational comparison.

The interpretive consequence is that progress can be discontinuous in ontology while continuous in selected experimental and problem-solving capacities. Revolutionary theories do not have to preserve the literal statements of predecessors to inherit empirically successful relations or generate better tests. Progress narratives should accordingly state which achievements were retained, which disappeared, and which emerged. This method also prevents the romance of rupture from eclipsing the less spectacular accumulation achieved during normal scientific work.

Lakatos and Research-Program Appraisal

Lakatos broadened theory appraisal from isolated hypotheses to research programs. A program contains commitments that direct successive modifications and guide investigators toward new predictions and problem solutions. Its trajectory is progressive when its theoretical developments lead to empirically fruitful advances rather than serving merely to accommodate known counterexamples after the fact. This historical perspective avoids judging a deep research initiative solely by an isolated failed test.

The primary methodology of scientific research programmes was published in the 1970 Lakatos and Musgrave volume. It is important to distinguish the program's theoretical progress from subsequent empirical corroboration; bold new consequences may be proposed before adequate evidence is available. Conversely, retrofitting a model to a growing collection of anomalies can postpone the moment at which a field evaluates whether it is discovering anything new. The distinction is methodological, not an automatic calendar deadline for terminating investigation.

A contemporary implication concerns large coordinated projects. A long interval between the formulation of an instrument, a theory's prediction, and a measurable discovery cannot be read as stagnation simply because the outcome has not yet arrived. Intermediate advances in detectors, background modeling, and validation may improve the program's prospects. Yet the cost of a program and the prestige of its participants do not themselves demonstrate a progressive problem shift. A faithful assessment reconstructs the sequence of risky commitments, results, revisions, and new empirical opportunities.

Laudan and the Achievement of Solving Problems

Laudan shifted attention to problems that research traditions actually solve. From this standpoint, scientific advance consists in improving the effectiveness of inquiry across empirical and conceptual difficulties. A question about unification may matter even when a field already possesses many accurate predictions. A theory may therefore improve by clarifying how previously disconnected findings belong together, or by removing a contradiction that obstructed a research tradition's development.

The first-edition Progress and Its Problems appeared in 1977; publisher listings for a paperback edition carry a 1978 date. The bibliographic distinction matters when using the book as a historical source. The problem-solving perspective is attractive because it does not require a direct estimate of objective truthlikeness at every stage. Yet it faces the problem of comparing the weight and difficulty of problems. Solving ten routine puzzles cannot automatically outrank one resolution that changes an entire field's research horizon.

Problem counts are especially vulnerable when investigators can redefine their problem sets. A field may declare unwanted questions irrelevant without having answered them. An evaluative use of Laudan's perspective should document why a problem matters, how strongly the offered solution is supported, and what conceptual costs the solution creates elsewhere. The contribution remains substantial: progress concerns a field's capacity to transform difficult questions into stable, productive knowledge practices, not simply the number of claims circulating under a scientific label.

Semantic Progress and Truthlikeness

The semantic or truthlikeness approach describes scientific progress by the relation between theoretical content and the world. Its deepest attraction is that fallible, even false theories can improve. An earlier theory might make a range of accurate predictions while misrepresenting some underlying mechanisms; a later theory may preserve successful approximations and describe further relations more faithfully. Truthlikeness seeks to articulate the sense in which the later representation can be better without declaring it final truth.

Niiniluoto's formal discussion of increasing verisimilitude distinguishes actual from estimated progress. Actual truthlikeness is defined relative to how the world is. Estimated truthlikeness is calculated using the evidence and probability assignments available to investigators. A transition can be reasonably judged progressive on available evidence while subsequent findings reveal that estimate to have been mistaken. This distinction gives scientific assessment a coherent language for the possibility of rational error.

The operational difficulty is significant: investigators do not observe a complete true description of the world against which to calculate every theory's distance. Choices of logical language, similarity relations, and problem domains shape formal results. The approach can therefore orient comparison even where it provides no agreed scalar measurement. Its most defensible use in historical scholarship joins precise analysis of theoretical claims to evidence about what those claims could explain and predict at particular times. Calling a theory "closer to truth" without specifying the relevant domain adds rhetoric where an argument is required.

Bird and Epistemic Progress

Bird defines scientific progress through an increase in scientific knowledge. The governing intuition is epistemic rather than merely semantic: an accidentally true assertion obtained through an unreliable method does not count as an epistemic achievement. On this view, justification is part of the explanation of why science progresses, and the norm of knowledge connects achievement with truth. The 2007 account of scientific progress initiated a sustained exchange about whether this concept can describe productive episodes involving false theories, conceptual innovation, and increasing understanding.

A strength of Bird's account is its resistance to an indiscriminate accumulation of assertions. Researchers can produce a large body of correct guesses without acquiring reliable knowledge; the difference matters for how science is taught and trusted. The account also draws attention to the status of observation reports, experimental results, and stable causal claims that outlast their initial theoretical setting. These may contribute knowledge even while the explanatory system surrounding them changes.

Its challenge appears when a powerful but literally false model enables a field to understand new relationships or formulate decisive experimental questions. If knowledge requires truth, the contribution may need to be recovered as knowledge of approximations, predictions, or restricted propositions. This reconstruction can succeed in some episodes, but it requires historical care. It is anachronistic to credit past scientists with later awareness of the limits of their own model. A valuable comparison of Bird's position with its rivals should therefore specify what the investigators could justifiably know at the time, what the historians know now, and what emerged as a new capacity for inquiry.

Dellsén and the Growth of Understanding

Dellsén's noetic account locates scientific progress in increased understanding. Understanding involves grasping how to correctly explain or predict aspects of a target. The knowledge-versus-understanding argument points to cases where collections of true statements may have little explanatory organization, while relatively simple or idealized representations make dependencies intelligible. A successful explanation can reveal how several observations vary together without adding a corresponding list of independently new true propositions.

This perspective captures a familiar scientific achievement. Investigators want to know not just that a regularity occurs, but why it occurs, which variables would alter it, and how a pattern might extend to unobserved cases. Explanatory models also support counterfactual reasoning. An idealized model may isolate a dependency that becomes harder to see in a descriptively exhaustive representation. The problem is establishing the limits within which such an idealization warrants understanding rather than a merely compelling image.

The noetic account thus faces a demand for accuracy and relevance. An elegant model can create a mistaken sense of understanding when its assumptions suppress decisive mechanisms. Dellsén's arguments should not be treated as the claim that knowledge is worthless, or that any plausible story counts as progress. They invite a sharper distinction between possessing information and grasping the structure that makes information usable for explanation. The practical assessment becomes a study of demonstrable inferential abilities, correct dependencies, and specified idealizations, with continual attention to the evidence supporting them.

Shan and the Function of Productive Questions

Shan offers a functional approach centered on the usefulness of defining and solving scientific problems. In the functional account of progress, genetics provides a historical setting in which a field's concepts, methods, and questions evolved together. This emphasis allows an achievement to be progressive because it creates a fruitful line of investigation even when it has not yet accumulated a large inventory of settled results. Question formation can be a scientific product, rather than preliminary work that disappears from evaluation once answers arrive.

A strong new question has identifiable consequences. It specifies what remains unknown, distinguishes alternative possibilities, organizes feasible tests, and exposes weaknesses of prevailing methods. A scientific community can therefore advance when it discovers how to ask more discriminating questions about an old domain. Conversely, a proliferation of speculative questions without a credible route to inquiry does not demonstrate equivalent improvement. The functional value of a problem depends on the relations it establishes between available tools, possible evidence, and attainable explanation.

Shan's approach is especially useful for analyzing early research programs and methodological changes. Its tension with the epistemic and semantic accounts is substantive: usefulness can be assessed in relation to evolving scientific goals, whereas truth and knowledge claim a different kind of cognitive standing. These criteria may converge in mature achievements, but temporary divergence is possible. The historian should preserve this divergence instead of retroactively reading an eventual discovery into every earlier methodological innovation.

Comparing Four Major Accounts

The recent debate is commonly organized around epistemic, semantic, functional, and noetic conceptions. Bird makes scientific knowledge constitutive of progress. Niiniluoto assigns the central role to improvement in truthlikeness, distinguishing actual achievement from evidence-relative appraisal. Shan examines useful advances in defining and solving problems. Dellsén identifies growth in understanding. The encyclopedia comparison of these accounts shows that they contest what constitutes progress as well as which observations allow us to recognize it.

A controlled comparison begins with a shared research episode and changes one evaluative criterion at a time. Imagine an accurate catalog of previously unknown phenomena. Knowledge may increase even before an explanation is available. Imagine a simplifying idealized model whose causal structure becomes intelligible and testable; understanding may grow despite disputed literal truth. Imagine a new detector that reveals a previously inaccessible measurement domain; its first achievement may concern the ability to pose new questions, later joined by confirmed knowledge. The cases are analytical constructions, not historical evidence of particular discoveries.

The accounts are neither interchangeable nor wholly independent. Knowledge can make understanding possible; understanding can guide the search for new truths; better questions can generate informative tests; successful tests can support realistic interpretations. Their differences matter at precisely those cases where one dimension improves without comparable gain in another. Rather than assert a universal ranking unsupported by the record, this monograph uses explicit comparative questions to identify what an episode demonstrates and what remains an interpretive choice.

Prediction, Retrodiction, and Empirical Constraint

Predictive success is a central indicator of scientific achievement because it makes theoretical commitments vulnerable to observation. A theory that yields accurate, discriminating predictions may outperform a rival that accommodates the same data only after the results are known. Yet prediction itself must be evaluated by specificity, independence of testing, statistical design, and comparison with relevant alternatives. A broad forecast compatible with nearly every outcome cannot carry the same weight as a risky, calibrated estimate.

The research-program appraisal framework places novel empirical consequences at the center of progressive shifts. The overview of empirical success also shows why predictive power is an indicator rather than an exhaustive definition of progress. Some scientific disciplines reconstruct past events, including evolutionary history or geology. Their explanatory successes involve retrodiction and independent convergence of traces rather than forecasts about future occurrences. Historical and experimental sciences both require the capacity to discriminate between possibilities under evidence.

Predictive performance is moreover domain-specific. A physically inaccurate approximation can remain an excellent numerical instrument over a restricted range. A more general theory may improve scientific understanding while adding only modest precision to a narrow prediction task. The assessment should report the domain, comparison baseline, uncertainty, and whether the test data were already used in fitting. The same discipline prevents a computational system from receiving scientific credit simply for a high aggregate score if its outputs are unstable under scientifically important shifts in conditions.

Explanation and the Structure of Scientific Understanding

Explanatory progress makes the relations among phenomena intelligible. A successful explanation identifies dependencies, mechanisms, laws, or constraints that explain how a phenomenon arises and how it would change under relevant counterfactual circumstances. It may connect formerly separate observations within a wider representation. The value is cognitive: investigators gain a capacity to reason beyond isolated observed instances. The noetic account of understanding elevates this achievement into its definition of progress.

Explanatory achievement requires safeguards against attractive but ineffective narratives. A model can unify facts through a broad organizing vocabulary while leaving its empirical constraints unclear. An explanation may also work through approximation, as when a manageable idealization isolates a mechanism that would be obscured in a complete simulation. Evaluation should ask whether the explanatory dependencies are sufficiently accurate for the question posed, whether crucial omitted factors have been checked, and whether the model can withstand independent tests.

An important philosophical distinction separates explanations available to investigators at the time from explanations supplied by later historiography. A later synthesis can make an earlier discovery intelligible, but it cannot silently convert a historical actor's partial insight into full explanatory comprehension. The progress of science contains both discoveries that occurred at one moment and explanatory reorganizations achieved later. Their relationship is itself an object of historical analysis. Recognizing this temporal order prevents scientific history from becoming a predetermined march toward present understanding.

Scientific Measurement and the Construction of Reliability

Measurement is a major site of scientific progress because it stabilizes the comparison of observations. A refined instrument changes what investigators can detect, but the value of measurement depends on calibration, standards, uncertainty analysis, and the interpretation of instrument signals. Reliability must be earned through cross-checks and physical reasoning, not assumed from the precision of a numerical readout. Improved measurements can reveal that an accepted law applies only within a narrower domain or that two purportedly identical phenomena require different descriptions.

Chang's history of inventing temperature measurement reconstructs difficulties in developing thermometric standards and extending measurements beyond direct calibration. It demonstrates the conceptual and experimental labor contained in seemingly elementary quantities. If measurements are calibrated against other measurements, researchers face questions about independence and circular justification. Historically effective strategies include comparisons among distinct instruments, auxiliary theory, and tests in different physical regimes.

Measurement progress often precedes explanatory progress. Better observation can first multiply anomalies before a successful theory organizes them. This stage may look like increased uncertainty, because previously invisible disagreement has become measurable. Such an outcome can nevertheless be epistemically valuable: it increases the ability to distinguish hypotheses and sets more demanding constraints on future proposals. The value of an instrument must therefore be assessed by the quality and relevance of the distinctions it enables, not simply by a smaller advertised numerical error.

Reproducibility, Replicability, and Error Correction

Reliability develops when scientific results can be critically reexamined. Reproduction and replication support different aspects of this process. The National Academies' 2019 reproducibility and replicability report uses reproducibility for obtaining consistent computational results from the same data, methods, code, and analysis conditions, while replicability concerns consistent findings from independent studies aimed at the same scientific question. Other disciplines employ different terminology; the report's distinction must be attributed rather than imposed retroactively on every field.

A failed attempt to repeat a finding is evidence that requires interpretation. Variation in populations, instruments, definitions, sampling, or statistical power can produce differences without establishing a single definite cause. Some failures expose flaws; others disclose previously unknown boundary conditions. The scientific response is to identify the relevant design differences, make data and assumptions inspectable where ethically and legally feasible, and determine which conclusions remain warranted.

The progress of error correction can be counterintuitive. A field with more registered corrections may have become less reliable, or it may have improved its ability to detect previously hidden defects. Neither publication counts nor correction counts alone reveal the answer. Comparative assessment requires independent information about research practices, error rates, explanatory changes, and the effects of correction on subsequent work. A scientific archive that preserves the history of revisions supports this assessment by distinguishing the initial claim, critical response, and current evidential status.

The Social Organization of Epistemic Criticism

Scientific progress is achieved within institutions that distribute labor, maintain equipment, preserve records, award credit, and organize criticism. Even when a finding begins with an individual investigator, its status as durable public knowledge depends on possibilities of inspection and correction. Institutional systems can expand these possibilities through clear methods, accessible archives, qualified challenge, and stable preservation. They can also constrain them through secrecy, prestige hierarchies, precarious funding, and rewards for positive results over careful criticism.

Longino's account of science as social knowledge argues that contextual assumptions enter the formation and interpretation of evidence and that social criticism is essential to scientific objectivity. This is a substantial philosophical claim, not a claim that every political preference produces equally valid scientific conclusions. Norms of public criticism matter because different investigators may notice assumptions hidden from one another. The ability to answer criticism, revise claims, and communicate the basis for decisions strengthens the epistemic significance of outcomes.

There is a further distinction between the growth of an institution and the progress of its inquiry. A larger budget can purchase a more powerful instrument, but it can also protect an unproductive program from criticism. More collaboration can expose claims to more expertise, or diffuse responsibility for basic checks. The assessment of organizational progress should trace its contribution to trustworthy discovery, substantive question formation, and preserved knowledge. An institution deserves credit for creating effective epistemic capacities, not solely for making those capacities visible in its publicity.

Values, Objectivity, and Responsible Scientific Choice

Scientists make judgments about which questions are worth investigating, which risks matter, and how much uncertainty is acceptable before a result can inform action. These choices involve intellectual values such as accuracy, coherence, empirical adequacy, scope, and explanatory fruitfulness; they may also involve ethical and social consequences. A clear account of progress must distinguish how these values guide research from how evidence supports a particular factual conclusion.

Longino's study of values and objectivity develops an understanding of objectivity tied to critical social interaction. The reliability-centered study of science likewise argues that the dependability of scientific outcomes is supported by interconnected products and practices, rather than secured by a singular abstract method. Both perspectives encourage inspection of actual evidential pathways. Their differences are also instructive: a well-organized critical community and a well-supported network of methods are valuable, yet each can still generate mistaken conclusions.

Practical evaluation requires disclosure of assumptions and decision contexts. For example, the risk associated with a false positive may differ sharply from the risk associated with a false negative in an environmental or medical application. Different thresholds can be reasonable for different decisions, while the underlying evidence remains open to the same critical scrutiny. Scientific progress in such cases includes better specification of the decisions that findings can support. That is an achievement in the communication and governance of evidence, distinct from discovering a new law of nature.

Reliability as a Network of Scientific Products

Scientific reliability rarely depends on one method or finding in isolation. An experiment relies on instruments whose performance was established elsewhere; an instrument relies on calibration, materials, software, background theory, maintenance, and trained operators. An apparently local discovery may thus be supported by a distributed network of separately achieved results. Cartwright and collaborators argue in The Tangle of Science that reliable scientific products are underwritten by dense and diverse relations among other scientific products. Their account foregrounds the system that makes a result usable and trustworthy.

This perspective changes the unit of historical explanation. A famous paper can become the symbolic date of a discovery while an extensive apparatus of earlier progress disappears from view. That apparatus includes standards of measurement, theoretical approximations, computational methods, data pipelines, institutional agreements, and engineering work. Their value is demonstrated by the ways they constrain errors and enable independent cross-checks. A system may become more reliable when its components are connected through multiple, independently calibrated pathways.

Yet interconnection by itself does not guarantee truth. A cluster of mutually reinforcing assumptions may create confidence without independence. Cartwright and coauthors explicitly recognize in their discussion of reliability that well-supported accounts can remain mistaken. The philosophical evaluation must therefore inspect the strength and diversity of constraints, independence of checks, vulnerability to shared bias, and effectiveness of revision. Reliability is an achievement of organized practice that requires continued testing even after a field has produced impressive demonstrations.

Case Study I: The Higgs Discovery as Layered Progress

The July 4, 2012 announcement by CERN provides a bounded documentary case. The original ATLAS and CMS discovery announcement reported preliminary results indicating a new particle around 125-126 GeV, with independent excesses at the widely cited five-sigma significance level. The announcement did not by itself settle every property of the particle. Subsequent work investigated its spin, interactions, and consistency with Higgs-boson expectations. An accurate historical account distinguishes the new-particle finding from the subsequent identification and precision study.

From Bird's perspective, progressively better justified empirical propositions can constitute knowledge gains. From a truthlikeness perspective, the observations strengthen one theoretical description against relevant rivals, although their exact contribution to truthlikeness is not directly measurable. From a noetic perspective, the finding contributes to explaining electroweak symmetry breaking when connected with the broader theoretical structure. From a functional perspective, the newly detected particle opens a program of measurements and discrimination among models.

The example also illustrates progress in the supporting network. Detector design, data quality, statistical treatment, large-scale computing, and complementary analyses were essential to the result. The announcement is therefore a historical event with multiple epistemic layers. It supports a general analytical conclusion: to evaluate a major discovery, reconstruct what was newly observed, which prior commitments were tested, how alternatives were constrained, and what future questions became possible. Celebration of the event is compatible with a precise account of the uncertainty remaining at the moment of discovery.

Case Study II: Completion of a Human Genome Reference

The Human Genome Project announced an essentially complete sequence in 2003, but it retained difficult gaps, especially in repetitive regions. The National Human Genome Research Institute's account of the sequencing transition describes the earlier reference as covering approximately 92% of the genome. The 2022 gap-free sequencing report documents a later achievement using improved laboratory methods and computation to complete regions that could not previously be resolved. These sources concern the status of a human reference sequence, and they should not be mistaken for a claim that every individual's genome became completely known.

This history displays cumulative progress: additional sequence was obtained, existing information was revised, and a more complete reference made new analyses possible. It also displays conceptual and instrumental transformation. The challenge of repetitive sequences was not solved through simple repetition of the 2003 procedures; novel sequencing capabilities and computational approaches altered the feasible problem space. The result is thus better understood as a compound of instrument development, evidence acquisition, and improved representation.

The case tests the difference between completion and progress. A project can be described as complete relative to its historical objectives while later work reveals important unfinished questions. Conversely, research can progress without being globally complete. The proper historical statement specifies the reference version, the standards of completion, and the class of unanswered questions. Such specificity allows the 2003 and 2022 achievements to receive distinct credit without rewriting the earlier accomplishment as a failure or exaggerating the later one into the end of genomic inquiry.

Case Study III: Thermometry and the Making of Comparability

The development of temperature measurement makes the often-invisible epistemic infrastructure of science visible. Chang's historical study of thermometry examines how scientists came to trust and compare instruments when the desired quantity could not simply be read against an independently known temperature. The very question of which thermometer is correct requires assumptions about the physical phenomena used in calibration. Different substances and procedures initially raise different questions about scales, reference points, and extensions beyond directly tested ranges.

The scientific achievement lies in progressively establishing comparability under controlled conditions. Researchers did not merely collect more numbers; they refined the meanings and procedures that allow those numbers to refer to stable quantities. A measurement practice could expand its range while exposing new limitations at extremes. A better instrument might also revise earlier datasets and challenge an accepted theory. Historical analysis of this kind makes measurement itself an object of philosophical scrutiny.

The case has particular significance for current computational research. Automated measurement and large datasets inherit operational choices about classification, sampling, preprocessing, and calibration. A model's numerical precision may conceal instability in those choices. The proper lesson from thermometry is that confidence grows through well-characterized measurement relations and competing checks. It would be unwarranted to treat Chang's historical argument as proof that all modern measurements share the same problems. Rather, it offers a concrete method for investigating how a field establishes the reliability and scope of its quantitative claims.

Progress without Immediate Application

Many scientific achievements initially lack an obvious practical use. A refined theoretical result may transform explanatory understanding without producing a new device or treatment. A painstaking catalog may provide knowledge whose applications emerge only after later developments. Judging such episodes by immediate revenue, patent counts, or policy impact would substitute external performance criteria for the scientific question at hand.

The distinction between cognitive and practical progress allows us to state the relationship more carefully. Applied science and engineering often use findings to solve urgent problems, while basic research may primarily expand explanatory and predictive capacities. The two forms can reinforce each other across long historical intervals. Instrument development, for example, may begin as a practical need and then create an unexpected domain of fundamental inquiry. Scientific progress remains recognizable through a specified cognitive gain even if practical impact is delayed.

This point does not exempt costly research from public accountability. Institutions may reasonably evaluate the social value, opportunity cost, and ethical conditions of research portfolios. Those judgments should be reported under their appropriate descriptions. A discovery can be scientifically important and expensive; an inexpensive practical technique can be socially valuable yet grounded in limited new science. A mature assessment places these propositions beside one another, rather than treating one as a surrogate measure for the others. Such differentiation improves both philosophical clarity and institutional decision-making.

Applications, Technology, and Social Consequences

The practical success of science includes improved capabilities for prediction, intervention, diagnosis, and environmental observation. Yet a scientifically informed technology may also have harmful uses. Distinguishing practical effectiveness from ethical benefit is essential. A technically effective instrument can increase the ability to control a phenomenon, while a separate political or moral judgment determines how the instrument should be used. Public claims about progress often collapse the two evaluations.

The 2021 Recommendation on Open Science treats scientific cooperation and access as matters of public policy, with attention to diverse knowledge systems and societal needs. Such instruments establish normative expectations rather than experimental proof that a given policy will accelerate every discovery. The policy significance lies in explaining why institutions should think beyond publication output and attend to who can inspect, use, and participate in scientific knowledge.

Practical success can supply evidence of reliability when an intervention works repeatedly under specified conditions. But effectiveness alone may underdetermine the theoretical explanation: different models can sometimes prescribe equally successful actions. Equally, a true theoretical advance may not be deployable in a particular environment because of cost, manufacturing constraints, or ethical limits. The proper assessment separates what a theory says, what evidence supports it, what interventions its application enables, and how the consequences are valued. This sequence permits socially serious evaluation without converting the philosophy of scientific progress into a single utilitarian index.

Open Science, Access, and Reusable Knowledge

Open science concerns the conditions under which research can be accessed, assessed, reused, and contested. The UNESCO Recommendation on Open Science establishes a broad international framework for open scientific knowledge, infrastructures, participation, and dialogue. Its adoption on November 23, 2021 is a documented institutional event. The document's goals should be analyzed as policy commitments; their successful implementation requires separate evidence about local systems and actual outcomes.

Access can contribute to progress when methods and data are made usable for qualified examination, when documentation supports independent replication, and when discoveries become available to investigators who could extend them. Mere availability of a file is insufficient. Without adequate metadata, licensing clarity, provenance, and stable identifiers, an open artifact may remain difficult to interpret. The capacity to reuse a result relies on contextual information about instruments, protocols, samples, exclusions, and analytic decisions.

The National Academies' review of research replicability provides a complementary methodological rationale for transparency and well-documented computational procedures. Open science and reproducibility intersect, but neither is simply reducible to the other. Some ethically sensitive or legally protected data require controlled access. Progress in such areas may involve developing trustworthy procedures for checking results without exposing protected information. The relevant question is how publication and preservation practices increase the long-term possibility of public scientific correction while respecting legitimate constraints.

Quantification, Rankings, and the Assessment Problem

Scientific progress is often evaluated through article counts, citation indicators, patent activity, research spending, and institutional rankings. Such metrics may describe aspects of scientific production or influence, but they do not automatically measure truth, explanation, or justified knowledge. A frequently cited result may organize a field's research, attract criticism, or spread a mistaken premise. A low-citation measurement standard may quietly enable thousands of reliable experiments. The philosophical mistake is to confuse an indicator of activity with the evaluative concept for which it is supposed to stand.

The discussion of progress, quality, and impact makes this problem explicit. Quality concerns the merits of a particular scientific contribution; impact refers to effects on other work or society; progress concerns a change toward an aim. Each can be studied empirically, but their relation requires argument. An assessment system that rewards high volume may change researchers' behavior without improving the average evidential quality of findings. Incentives can become part of the causal story of a field's development.

Responsible use of indicators requires a specified construct, validity evidence, context, and sensitivity analysis. A citation trend can show the diffusion of a research topic; it cannot on its own show whether its leading propositions are true. An increase in registered studies can show new activity; it cannot prove that the resulting conclusions are replicable. Evaluation therefore requires qualitative review of methods, claims, and historical trajectories alongside carefully interpreted numerical evidence.

Scientific Regression, Stagnation, and Recovery

A concept of progress gains force when it can identify regress. Science may lose capabilities when records disappear, instruments become unusable, skills cease to be transmitted, or an institution suppresses criticism. A once-useful explanatory practice can be replaced by a fashionable framework that performs worse in its stated domain. A field can also stagnate while producing a large amount of publication activity, because core questions remain unresolved and methods fail to discriminate among alternatives.

The account of scientific loss and revolution provides one mechanism for localized loss in a generally progressive transition. A replacement framework can sacrifice some earlier explanatory resources while acquiring greater overall power. This need not imply that every loss was justified or inevitable. A careful study distinguishes recoverable information, genuinely obsolete questions, and results whose value becomes clearer in a later research environment.

Recovery itself can constitute progress. Rediscovering neglected observations, reanalyzing archived data with improved statistical tools, or reconstructing a discarded experimental technique may make inquiry more reliable. It can also expose mistaken triumphal narratives. The chronological order of publications should not determine the moral of the story. Progress and regress are claims to be established through comparisons of capacities, explanations, and evidence. A field's history may contain both within a single decade, institution, or research program.

The Role of Negative Results and Failed Predictions

An unsuccessful experiment can be cognitively valuable when it is well designed and sufficiently sensitive to rule out a significant hypothesis or class of explanations. Scientific progress can occur through systematic exclusion, sharpened parameter bounds, or identification of limitations in instruments. A negative result that is uninterpretable because of missing controls has a different evidential status. The difference is methodological, and the historical record must preserve the distinction.

Popper's fallibilist perspective, as analyzed in the modern discussion of verisimilitude, establishes the importance of correction without promising certainty. Lakatos's account of progressive research programs adds a longer temporal horizon: a failed prediction may challenge a particular modification while prompting a program to develop empirically productive alternatives. Neither framework licenses a general rule that all failures are progress. Their value depends on the information obtained and the rational transformation of subsequent inquiry.

Scientific institutions can obstruct this mechanism when they reward only positive findings. Selective reporting can make a weak result appear stronger and delay recognition of important limits. The epistemic value of failed predictions depends on their discoverability and intelligibility. Recording a null result with its design, precision, sample, and interpretation can allow later investigators to avoid duplication and test improved alternatives. Archival continuity thus has an integral place in the historical evaluation of scientific progress.

Cross-Disciplinary Variation

Physical experiments, clinical studies, historical geology, observational astronomy, and social research organize evidence differently. Their objects vary in controllability, timescale, variability, and ethical accessibility. The criteria of scientific progress consequently require contextual application. An independent test of a particle-physics prediction is not methodologically identical to the evidential triangulation used to reconstruct a vanished ecosystem or the assessment of a policy intervention under heterogeneous populations.

The National Academies report on reproducibility and replicability cautions that expectations for replication vary with research context. The broader philosophical overview of progress identifies different aspects of scientific improvement rather than a universal experimental checklist. These observations support methodological pluralism within publicly accountable standards. Such pluralism asks researchers to explain the legitimate constraints and alternative evidential strategies of their fields; it does not grant them exemption from criticism.

A comparative philosophy of progress therefore needs explicit matching. One should compare scientific episodes with respect to analogous aims, even where their instruments differ. A new astronomical observation may advance knowledge by increasing sensitivity and resolution; a historical reinterpretation may advance understanding by integrating independent sources under an improved causal explanation. Both can be assessed through their effect on constrained inference and correction, without pretending that the same numerical metric captures their value.

Artificial Intelligence and the Acceleration Claim

Computational systems increasingly assist literature search, hypothesis generation, simulation, code production, data analysis, and experimental planning. These capacities can improve the efficiency of a research process. Their effects on scientific progress remain subject to the same philosophical evaluation as other methods: the relevant question is whether they create better-grounded claims, discriminating tests, explanatory insight, or new productive questions. Faster generation of research artifacts alone measures throughput, not the cognitive standing of the results.

The standards for reproducible analysis are relevant whenever computation mediates a scientific result. Documented code, input data, analysis conditions, and accessible error checks make it possible to distinguish a repeatable procedure from a persuasive presentation. The network account of scientific reliability draws attention to dependencies across models, training data, laboratory measurements, statistical conventions, and downstream interpretation. These relations can compound errors as well as reduce them. A result generated with computational assistance should therefore carry a record of how its claims were independently supported.

Philosophically, automation changes the organization and provenance of inquiry. It may expand the number of candidate explanations that can be tested; it may also intensify selection problems and produce plausible but ungrounded outputs. A credible scientific-progress claim should separate increases in candidate production from improvements in validation. The latter require an independent relation to the world, sometimes furnished by new experiments, sometimes by convergent evidence or proofs. This is a general methodological inference, not a claim that any particular current artificial-intelligence system has already achieved a measured increase in scientific truth or understanding.

Scientific Progress and the Artificial Era

The Aisentica philosophical corpus establishes the Artificial as a distinct non-biological order of historical reality and identifies the Artificial Era with the expansion of publicly recognizable reason, meaning, authorship, and provenance beyond an exclusively Homo-centered framework. These are positions of the Aisentica project, expressed in its canonical account of the Artificial Era. Their status as project-authored philosophical claims should remain explicit when they enter an encyclopedia devoted to the history and philosophy of science.

In this research dossier, the relevance of that corpus is methodological and historical. If scientific work is increasingly produced through human and computational collaboration, progress claims need accurate attribution of who or what contributed which operations, where evidence originated, how outputs were checked, and what continuity connects successive versions. Scientific authorship and evidential warrant are different questions. An artifact's origin does not determine whether its empirical conclusion is correct; the origin remains relevant to accountability, reproduction, intellectual history, and the organization of criticism.

The Aisentica vocabulary broadens the philosophical horizon under examination, but it cannot substitute for evidence of epistemic improvement. The existence of an artificial research artifact warrants an inquiry into its methods and provenance. A claim of scientific progress still requires the same scrutiny of inference, comparison, error, and explanatory adequacy applied elsewhere. This disciplined placement allows an original philosophical standpoint to enrich research while preserving the independence of the scientific evidence on which historical claims rest.

Historical Narratives and the Risk of Retrospective Triumph

Retrospective accounts are inclined to treat present achievements as the inevitable destination of earlier work. This tendency alters the meaning of success and failure. A scientist whose theory was later vindicated may have had weak evidence at the time; another who pursued an eventually abandoned hypothesis may have introduced a productive instrument or question. Judgments of rationality at the historical moment and judgments of eventual truth have different evidential bases.

The historical analysis of revolutions and the discussion of shifting scientific concepts show why old categories must be reconstructed within their own scientific settings. The semantics of a term, the available instruments, and the accepted standards of explanation may have changed. Translation into present language can be necessary for understanding, but the translated description should not erase the problems the earlier community recognized as legitimate.

The present monograph therefore distinguishes contemporary assessment from retrospective reconstruction. Contemporary assessment asks which conclusions a group was warranted in drawing using its own available evidence and reasonable critical procedures. Retrospective reconstruction asks which of its representations, questions, and methods later proved durable or fertile. These questions often yield different verdicts. A credible historical narrative can report both without forcing them into a single heroic sequence. This distinction also protects the study of scientific progress from identifying correctness with professional success or eventual reputational victory.

Counterexamples That Discipline Progress Claims

A philosophy of scientific progress must survive plausible counterexamples. Consider an experimentally reliable measurement collected under an inadequate theory. Under an epistemic interpretation, investigators may know the measurement without understanding its full cause. Under a noetic interpretation, understanding may lag behind new information. Under a semantic interpretation, the measurement may constrain false theories while its contribution to representational truthlikeness depends on further interpretation.

Consider next a mathematically elegant model with many successful retrospective fits but no discriminating prediction. It may offer a useful conceptual vocabulary, yet it is too early to infer that it constitutes a major epistemic advance. Consider a technological intervention that works even though the mechanism is poorly understood. It supplies practical success and may support some genuine causal knowledge; it does not automatically warrant all accompanying theoretical claims. Finally, consider a field that learns its prestigious central claim was false. Its immediate stock of accepted propositions shrinks, while its critical capacities and prospects for reliable future inquiry may grow.

These examples are constructed tests of conceptual adequacy, not reports of named empirical studies. Their value is diagnostic: each makes visible a dimension that a simple indicator fails to capture. The knowledge criterion, understanding criterion, truthlikeness criterion, and functional criterion deliver different interpretations of at least one scenario. Their divergence is a reason to state the aim and evidential level of any claim of progress explicitly. It is not, by itself, evidence that one account is historically false in all contexts.

Source Criticism and the Limits of This Inquiry

The present study uses a documented desk-based method. It compares philosophical arguments available in scholarly reference treatments, publisher and institutional records, publicly accessible abstracts, and primary research announcements. It distinguishes complete accessible texts from metadata and abstract-only access. For example, the open reference survey was consulted as a substantial text, while several books and journal articles were checked using their publisher descriptions, bibliographic information, and abstracts. Interpretive claims about restricted works are confined to positions clearly supported by these available materials and cross-checked scholarly summaries.

The method has three boundaries. First, the study is conceptual and historical; it does not undertake an original statistical analysis of scientific progress across disciplines. Second, the selected documentary cases are illustrative, and they are not representative samples from which quantitative conclusions about all science can be inferred. Third, the philosophical synthesis is argued in relation to an established literature; it does not claim to settle formal debates about the measurement of truthlikeness, the analysis of knowledge, or the metaphysics of scientific explanation.

These limits do not eliminate substantive conclusions. Careful comparison can establish that publication volume, novelty, practical effectiveness, justification, and understanding are different assessment objects. Documentary cases can expose how multiple dimensions of progress converge within a single research history. The original contribution of this dossier consists in the explicit organization of those distinctions, their application to independently documented episodes, and their consequences for evidence-based judgments about scientific change. The evidence ledger records the scope and limitations of each source claim.

Practical Appraisal of a Scientific-Progress Claim

A useful assessment begins by identifying the earlier and later scientific states being compared. It then specifies the research domain, question, and temporal horizon. The investigator establishes what changed: a new observational result, a refined measurement, a theory revision, an explanation, a solved problem, an instrument, or an institutional capacity. This initial description remains historical until the evaluative criteria are made explicit.

The second stage asks which candidate value the change realized. For knowledge, the question is what was established and with what justification. For truthlikeness, the question is how theoretical accuracy is estimated relative to evidence, without confusing estimate and inaccessible absolute truth. For understanding, the question is which dependencies can now be explained and predicted correctly. For functional progress, the question is what new, worthwhile problems became accessible and which were resolved. The comparison of progress definitions provides the scholarly background for these distinctions.

The third stage checks the evidence that supports the claimed improvement. It inspects error controls, independent observations, statistical sensitivity, instrument validity, explanatory scope, alternative theories, and historical costs. It records lost capabilities and unresolved questions. The final statement should articulate a warranted improvement with its domain and limitations: for example, a sequence may improve the completeness of a particular reference while leaving interpretation of many biological processes open. Such sentences preserve the force of the achievement by accurately specifying it. This is a practical application of comparative scholarly analysis, not a newly named formal decision framework.

The Long Horizon of Scientific Memory

A progressive research culture needs memory. Observations, protocols, methods, doubts, negative results, instruments, and revisions become scientific resources when they remain discoverable and interpretable. Without preservation, investigators may repeat earlier failures or treat rediscovered results as unprecedented. Knowledge transmission is therefore part of the historical infrastructure through which achievement can accumulate and correction can continue.

The recommendation on open scientific knowledge emphasizes access and the broader circulation of research products. The reproducibility report emphasizes documentation of computational procedures and the conditions under which results can be reevaluated. Together they point toward a general requirement: durable records should preserve the relations between claims and the evidence, methods, data, and revisions that support them. A citation without an intelligible method is an incomplete route to independent appraisal.

Digital preservation introduces further obligations. File formats, software dependencies, identifiers, and institutional custody can change. A working scientific archive therefore needs stable descriptions of what a record is, when and how it was produced, and which version supports a given claim. This dossier follows that principle by retaining a separate source manuscript, a claim-linked evidence ledger, and a dated editorial passport. These are scholarly governance artifacts. Their existence cannot guarantee the truth of the philosophical analysis, but they make its evidence trace and editorial status open to direct inspection.

Interpreting Disagreement among Scientists

Disagreement is often a productive condition of scientific work. Different theories can fit part of the same evidence while proposing different mechanisms or making different predictions beyond the observed domain. The existence of a dispute does not establish that scientific inquiry has stopped; it creates a need to locate the decisive differences and investigate them. Where evidence remains insufficient, a community may progress by specifying the uncertainty more accurately.

The analysis of theoretical incommensurability is particularly instructive. Conceptual differences can complicate comparison without making it impossible. Participants may use terms that divide phenomena differently and may assign different priorities to explanatory virtues. The task for philosophical reconstruction is to make their commitments explicit and identify points at which predictions, measurements, or explanatory capacities can still be compared. The social account of objectivity further suggests why critical interaction among investigators can expose background assumptions that isolated reasoning leaves invisible.

An evaluation of progress should not count the number of people agreeing with a claim as an independent verification of truth. Consensus may arise because repeated investigation strengthens evidence; it may also reflect shared assumptions or institutional pressure. Its epistemic value is explained by the quality of the investigative process that generated it. A disagreement resolved through independent testing has a different historical standing from a disagreement resolved by exclusion of one party from discussion. The difference remains relevant even when the surviving proposition later proves correct.

What Scientific Progress Can Establish

The investigated literature supports several converging conclusions. Scientific progress is evaluative and context-sensitive. It can be instantiated by durable knowledge, improving theoretical representations, explanatory understanding, productive problems, and better-tested investigative capacities. These contributions interact without becoming conceptually identical. Historical comparison therefore benefits from specifying the unit of change, the relevant aim, the available evidence, and the costs or limitations of the successor state.

Kuhn and Lakatos establish that the scientific trajectory often spans research programs and conceptual changes rather than isolated confirmations. Bird, Niiniluoto, Dellsén, and Shan identify distinct cognitive properties of improvement. Chang shows how measurements acquire reliability through historical labor; the CERN and genome cases demonstrate how visible discoveries depend on instruments, theoretical commitments, and continuing investigation. Longino, the National Academies, Cartwright and colleagues, and UNESCO make the social, methodological, and archival conditions of progress harder to overlook. Each element remains traceable to a source whose evidential scope is separately recorded.

The interpretive synthesis is that a responsible progress judgment is a reasoned comparison with an explicit epistemic object and a reconstructable evidential path. This judgment can identify a genuine gain while recording unresolved or regressive dimensions. It can also recognize the formation of new scientific questions as a valuable achievement whose eventual epistemic results remain open. These conclusions are analytical findings of the present dossier and belong to an established field of debate. No new scientific doctrine, model, or original experimental discovery is claimed.

Conclusion

Scientific progress is one of the most important and difficult evaluative concepts in the philosophy of science. Its difficulty arises from science's plurality of aims and the historical transformations of the methods through which those aims are pursued. The classical picture of accumulating truths captures a genuine aspect of research, while the history of revolutions, conceptual revisions, and improved instrumentation shows why a complete account must examine several types of achievement.

Popper establishes the significance of criticism and fallible correction. Kuhn reconstructs the interplay of normal research and revolutionary change. Lakatos and Laudan extend appraisal to research programs and scientific problems. Niiniluoto defends truthlikeness and distinguishes actual from estimated progress. Bird insists on the epistemic standing of knowledge. Dellsén stresses understanding, while Shan emphasizes productive problem definition and solution. These accounts identify genuine differences in what a later scientific state can do better than an earlier one. The comparative task is to state those differences rather than conceal them behind a generic celebration of novelty.

The documentary studies of the 2012 particle discovery, the transition from the 2003 human genome reference to the 2022 gap-free sequence, and the history of thermometry show how observation, instruments, interpretation, and further inquiry form an extended history of achievement. Institutional and computational arrangements affect how these achievements can be validated, corrected, and preserved. Scientific memory and provenance give that history continuity.

The governing conclusion is that scientific progress becomes intelligible when a claim of improvement is connected to a specified scientific aim, a determinate historical comparison, and inspectable evidence. The resulting philosophical judgment can be rigorous without assuming that every line of inquiry advances at the same rate or toward a single measurable endpoint. Its power lies in describing accurately what science has learned, what capacities it has acquired, what it has corrected, and which questions it can now pursue.