Probability All the Way Down: When the Critic’s Judgment Becomes the Final Court

Revision note: Substantially revised July 20, 2026. The argument has been reordered according to the classical rhetorical movement, compressed where necessary, and given one consolidated Notes and Further Reading section.

At the disputed places, Lane Keister’s method moves from probable evidence through probable canons to a probable editorial conclusion. Keister does not claim that the critic is infallible or superior to Scripture. The question is functional: when the church can identify the wording of its supreme judge only after the critic has rendered a revisable verdict, which court is actually hearing the final appeal?

A textual critic does not become the final judge by calling himself one. He may insist that God alone is Lord, Scripture alone is infallible, and criticism is only a servant. He may approach his work reverently, pray for wisdom, reject unbelieving assumptions, and sincerely intend to recover rather than create the apostolic words.

All of that may be true, and in Lane Keister’s case much of it plainly is true. Keister affirms inspiration. He emphatically affirms that God preserved the original text of both Testaments by special providence through the manuscripts. He believes that readers of the King James Version and the English Standard Version possess the Word of God. He denies that any Christian doctrine has been lost through textual variation. He also rejects a simple identification of age, numerical majority, or any other single feature with originality.

The issue, therefore, is not Keister’s declared allegiance. It is the authority structure created by his method. In his essay on the canons of New Testament textual criticism, Keister says that all his canons have “probabilistic force only.” The canons may conflict. None automatically determines the result. Their relative weight must be assessed by the critic, whose judgment is essential: “It is unavoidable.”

Those admissions are realistic. Manuscripts do not arrive with inspired labels declaring their date, genealogy, or textual value. Scribal motives are not ordinarily recorded. Patristic citations can be paraphrastic. Versions can obscure their underlying Greek. Internal considerations often point in opposed directions. A human evaluator must reason through the evidence.

But if the evidence is probabilistic, the canons are probabilistic, their application is probabilistic, and their relative weight is determined by a fallible critic, then the conclusion is probabilistic as well. At the disputed locus, probability runs through the method from beginning to end. That is what “probability all the way down” means. It does not mean that every word of every modern Bible is equally doubtful. It means that wherever the method recognizes a genuine textual dispute, it possesses no non-probabilistic stopping point before the critic’s judgment identifies the reading to be printed as text.

The final court is therefore not identified by the critic’s personal attitude. It is identified by the place where the chain of appeal stops.

Keister’s Position at Its Strongest

Keister’s best response would begin by denying that probability makes the critic sovereign. God inspired the original wording. That wording possesses authority independently of the scholar. Providence preserved the relevant evidence. The critic examines that evidence with fallible but responsible tools and attempts to identify the reading that God originally gave. If a later discovery or better argument corrects the critic, the correction demonstrates that Scripture—not the critic—remains supreme.

Furthermore, textual criticism is not unusual merely because it involves judgment. Pastors exercise judgment in exegesis. Translators judge between possible equivalents. Historians assess testimony. Church courts interpret confessional and biblical language. Received Text advocates also compare editions, discuss printer’s errors, distinguish orthographic differences from substantive variants, and argue for particular readings. No human use of Scripture occurs without fallible reasoning.

Keister could also appeal to proportion. He reports more than ninety-percent agreement between the Received or Majority tradition and prominent early manuscripts. Most textual decisions are routine. Many variants are spelling differences, transpositions, or obvious copying accidents. It would be misleading to speak as though every verse were suspended over an abyss of doubt.

All of these qualifications should be granted.

The confessional objection is not that Keister uses probability, exercises judgment, or admits local uncertainty. Nor is it that a critic who changes his mind has thereby denied Scripture. The objection concerns the formal office of that judgment at precisely those places where the witnesses disagree and where the wording of Scripture is therefore in question.¹

Exegesis ordinarily asks what an acknowledged text means. Translation asks how that text should be represented in another language. Textual criticism, at a contested locus, asks a logically prior question: Which wording is the text that must then be interpreted and translated? Fallible judgment operates in all three tasks, but it does not occupy the same office in each. A minister interpreting Romans stands under the words of Romans. An eclectic editor deciding between two competing forms of Romans determines which words the minister will encounter above the apparatus as Romans.

The editor does not create their divine authority. He does, however, decide which claimant presently receives the status of text.

The Chain of Probability

Consider what must occur when reading A and reading B compete.

Stage: Documentary identification

Critical question: What does each witness actually read?

Why the answer remains probabilistic: Damage, corrections, ambiguous letters, and layered hands may require interpretation

Stage: Dating and provenance

Critical question: How old is the witness, and where did it originate?

Why the answer remains probabilistic: Paleographic ranges and geographic assignments are historical estimates

Stage: Genealogy

Critical question: Are several witnesses independent or descended from a common source?

Why the answer remains probabilistic: Contamination and lost intermediaries make relationships inferential

Stage: Patristic and versional evidence

Critical question: Which reading did an early father or translation know?

Why the answer remains probabilistic: Citation, paraphrase, retroversion, and dating can be uncertain

Stage: Transcriptional explanation

Critical question: Which reading would more plausibly generate the other?

Why the answer remains probabilistic: The critic reconstructs an unobserved act and possible scribal motive

Stage: Intrinsic probability

Critical question: Which wording would the author more likely have written?

Why the answer remains probabilistic: Style, context, and theology must be interpreted

Stage: Weighting the canons

Critical question: Which considerations control when they conflict?

Why the answer remains probabilistic: Keister supplies no binding or revealed hierarchy of weights

Stage: Editorial conclusion

Critical question: Which reading should be printed as text?

Why the answer remains probabilistic: The conclusion inherits the defeasibility of the preceding judgments

None of this proves that the conclusion is irrational. A conclusion may be exceedingly well supported. It may command broad scholarly agreement. The point is that the method does not transcend its premises. A chain composed of defeasible judgments yields a defeasible result.

Nor can the critic escape the chain by invoking the canons themselves. The canon that older witnesses tend to support older readings is a probability. So is geographical diversity. Genealogical independence must be reconstructed. The more difficult reading may be original, or it may be an accident. The reading that best explains the others depends upon an imagined history of corruption. Continuity, majority, and patristic evidence all require interpretation.

Keister is admirably explicit that no canon is a knockout. Yet that means the canons cannot constitute a court above the critic. They are considerations placed before him. He decides their competence, application, and relative force. At the end of the process, there is no further methodological appeal except to another critic, another weighting, another edition, or additional evidence. The human judgment is provisional, but it remains final for the present decision. Modern eclectic criticism thus produces a peculiar final court: its verdict governs the text currently printed, while the court remains permanently open to rehearing the case.

Material Authority and Functional Authority

The dispute becomes clearer when two kinds of authority are distinguished.

Material authority answers the question, “Why is the original reading binding?” The Christian answer is: because God spoke it. The critic did not inspire the reading, and no committee makes it divine.

Functional or epistemic authority answers a different question: “By what presently operative judgment is reading A identified rather than reading B as the wording God spoke?”

Keister locates material authority entirely in God. The controversy concerns the second question. If no received apographic form has presumptive authority, and every extant reading must be placed before the critical tribunal, then the church’s access to the materially authoritative word depends upon the critic’s identification of it.

This is why the assertion “Scripture, not the critic, is final” can be true confessionally while failing to describe the method’s operation at a disputed place. Scripture is confessed as the supreme authority in the abstract. But the particular wording that will function as Scripture is not identified until critical judgment has spoken.

Suppose reading A is printed in the ecclesiastically received text and reading B appears in several early witnesses. The critic weighs age, distribution, genealogy, scribal habits, context, and explanatory power, then places B in the text and A in the apparatus. A minister now expounds B as Scripture. A translation committee renders B. A seminary teaches B. A church member is told that A is secondary.

The editor did not make B inspired. Yet his verdict changed which extant wording receives public treatment as inspired. Calling the verdict provisional does not reduce its present functional authority. Indeed, provisionality is the problem. The church’s practical text at that locus remains answerable to the next successful critical appeal. The historical text is always the appellant; scholarly judgment is the sitting court.²

Westminster Identifies a Different Final Court

The Westminster Confession arranges authority in a carefully connected order.

Chapter 1.4 says Scripture’s authority depends not upon any man or church but wholly upon God, its truthful author. Chapter 1.5 allows evidences and the church’s testimony to move believers to reverent esteem, but locates full persuasion in the Spirit bearing witness by and with the Word. Chapter 1.8 says the Hebrew and Greek Scriptures were inspired, kept pure in all ages, and are therefore authentic, so that the church finally appeals to them. Chapter 1.10 then declares that the supreme judge can be no other than “the Holy Spirit speaking in the Scripture.”

The confessional chain is therefore:

God speaks → Scripture possesses divine authority → the Spirit witnesses by and with the Word → the inspired text is preserved and remains authentic → the church appeals finally to that Scripture → the Holy Spirit speaking there is the supreme judge.

This order does not eliminate teachers, councils, grammarians, translators, or textual scholars. Westminster later says that synods and councils determine controversies ministerially. Their decrees are received insofar as they are consonant with the Word. The subordinate court operates beneath an available superior court. That last condition is decisive. A final court must be available for appeal. An unknown wording cannot judge. A lost text cannot issue a sentence. An aggregate containing reading A, reading B, and reading C without formally distinguishing them cannot adjudicate which reading belongs to itself.

Keister may answer that the original text exists within the aggregate and criticism discovers it. Materially, that is his preservation claim. Yet until criticism performs the discovery, the aggregate offers competing claimants. The Holy Spirit speaking in Scripture cannot function as the final judge at the disputed wording until the critic tells the church which wording is Scripture.

This reverses the confessional order at the very point under dispute. Instead of the preserved text judging the human witness, human witnesses are weighed until critical judgment reconstructs the text that will thereafter judge. The critic remains subordinate to God in intention and ontology. He becomes prior to the church’s identifiable text in the order of knowledge.

That is what it means for criticism to function magisterially.

Self-Attestation Does Not Mean a Private Voice

Confessional bibliology’s appeal to a self-attesting Scripture is sometimes misunderstood as though the believer receives an inward revelation identifying every variant. That is not the claim. The Holy Spirit does not supply each Christian with a private critical apparatus. Self-attestation does not render manuscript history irrelevant, sanctify every printed reading, or make an editor incapable of error. It describes the ultimate ground upon which Scripture is received as Scripture.

Every proposed final authority faces the same logical question: by what higher authority is it validated? If it must be certified by something higher, it was never final. God can swear by no greater than himself (Heb. 6:13). His Word is truth (John 17:17), and its authority rests upon the truthfulness of the God who speaks. The Spirit’s witness does not add an external certificate to an otherwise uncertain human document. He bears witness by and with the Word.

Historical evidences may confirm, defend, and clarify. The church may recognize. Scholars may expose objections as false. But neither church nor scholar supplies Scripture’s ultimate authority. The same order must govern textual identity. The self-attestation of Scripture belongs to the inspired and preserved text, not indiscriminately to every manuscript or reading. A scribal addition does not become self-authenticating because it survives. A copying omission does not carry the Spirit’s authority. Particular witnesses must be tested ministerially.

But ministerial testing presupposes a rule. If the authentic text exists only as the result of an open-ended reconstruction, self-attestation is transferred away from the church’s concrete textual possession and attached to a perfect original whose present wording is mediated by probability. The doctrine remains verbally affirmed while the operative text is authenticated reading by reading through critical judgment. The alternative is not private mysticism. It is the public economy of inspiration, providential preservation, ecclesiastical reception, and subordinate textual examination. The church receives because God has spoken; evidence ministers to that reception; the evidence does not constitute the object received.

The Canon Shows That Fallible Recognition Can Reach Closure

The canon provides a useful analogy.

Reformed theology denies that the church created the authority of Matthew, Romans, or Revelation. The books were canonical because God inspired them. The church recognized their authority through providential means, subordinate evidences, and the Spirit’s witness. That recognition was carried out by fallible people across history.³

Yet fallible recognition does not require the canon to remain permanently open. The church does not reclassify Hebrews as noncanonical whenever a scholar proposes a new historical argument. Nor does it place 1 Clement above the line in the next edition because an ancient manuscript is discovered. Providential recognition reaches ecclesiastical closure without making the church infallible or constitutive of inspiration.

The analogy is not exact. Recognizing whole books and resolving local readings involve different kinds and scales of evidence. But the epistemological point stands: the fallibility of the recognizing instrument does not entail permanent revisability of the thing recognized. Keister’s model readily grants this concerning the canon. It is therefore insufficient simply to answer textual certainty by pointing to fallible human judgment. The question is whether God’s providential work has issued in a publicly received text, as it issued in a publicly received canon, or whether every disputed wording must remain open to critical reconstitution.

If a closed canon is compatible with fallible recognition, a stable apographic text cannot be dismissed merely because its reception also involved fallible churches, copyists, and editors.

The Critic Is Final Only at the Disputed Locus—But That Is Enough

It may seem unfair to call the critic the final court when so much of Scripture is undisputed. The qualification matters. At places where all relevant witnesses agree, criticism does not function constitutively. The words are received without a contest. At trivial copying errors, the judgment may be immediate and practically universal. At many variants, the difference has little effect on translation or exposition.

The magisterial function appears at the genuinely contested locus. There, the question is not whether Scripture is generally reliable. It is whether these words, rather than those words, belong to Scripture. If no antecedent received form governs the inquiry, the editor’s decision supplies the present answer.

One disputed clause is enough to reveal the order of authority. A court does not cease to be a court because it hears only contested cases. Indeed, contested cases are precisely where its office becomes visible. Nor does the numerical smallness of the disputed material answer the formal question. Ninety-percent agreement would be a reason for gratitude. Ninety-nine-percent agreement would be stronger still. But the question remains: at the remaining loci, who identifies the words that possess divine authority?

The assurance that no doctrine is lost also answers a different question. Doctrinal stability is not verbal identity. God inspired every canonical word, including words whose variants do not overturn a systematic-theology chapter. If the church’s rule is verbal, the authority relation must be coherent at the smallest disputed locus as well as across the broad doctrinal whole.

What Ministerial Probability Looks Like

A preservationist method does not abolish probability. It assigns probability a subordinate office. Ministerial criticism begins with the authentic Hebrew and Greek text as providentially preserved and publicly received. It distinguishes that text from every physical artifact that witnesses to it. No manuscript, printed edition, translator, or editor is inspired by transmission. Individual witnesses can omit, add, transpose, misspell, or misprint.

Evidence is then used to recognize and correct such failures, explain the history of variants, defend the received reading, and improve the accurate presentation of the text. Probability may affect the critic’s confidence about a witness, a date, a scribal event, or an editorial detail. It does not make the identity of Scripture the cumulative product of those probabilities.⁴

The burden of proof also changes. Under eclectic reconstruction, every reading is in principle exposed to fresh adjudication, and ecclesiastical reception is one item of evidence among others. Under ministerial preservation, the received text is the standard; a claim that it contains a corruption bears the burden of demonstrating how that claim accords with God’s preservation, the text’s public use, and the whole canonical witness.

This is not an argument that tradition cannot be corrected. Rome made ecclesiastical reception constitutive and insulated its traditions from Scripture. Confessional bibliology rejects that error. The church is ministerial. But a ministerial church can truly receive a divine deposit, just as fallible Israel possessed the oracles of God and fallible congregations possessed apostolic letters.

The difference is between ministerial recognition and magisterial constitution. Recognition receives, identifies, transmits, and defends what God has given. Constitution makes the church’s identifiable text epistemically dependent upon continuing scholarly reconstruction. Probability is a useful servant in the first order. In the second, it becomes the bridge across which every contested word must pass before the church may receive it.

A Final Court That Cannot Deliver Finality

Keister does not personally claim the right to dictate the church’s Bible. His proposed canons are not even offered as universally binding. The “critic” in this argument is therefore not Lane Keister as an individual. It is the role assigned to successive scholars, editors, and committees within the eclectic system.

One critic may prefer reading A. Another may give more weight to geographical diversity and choose B. A later editor may reconstruct the genealogy differently and restore A. Each sincerely seeks the original. Each submits in intention to the inspired autograph. Yet no publicly possessed apographic form stands above their competing verdicts as the court of appeal.

The result is a final court that cannot deliver finality. Its rulings govern provisionally. Its precedents may be overturned. Its evidential rules have no universally obligatory hierarchy. Its perfect constitutional document is the absent autograph, while the wording of that document at contested places is precisely what the court must reconstruct.

This is not skepticism about whether Christians have Bibles. Keister is right that they do. It is a question about the doctrine used to explain that possession. A highly reliable, doctrinally sufficient, continually improvable reconstruction is not the same confession as an authentic text kept pure in all ages and presently available for final appeal.

Keister preserves Scripture materially in the manuscript aggregate and honors it theologically as God’s Word. But his probabilistic method makes the church’s formal identification of disputed words dependent upon the critic’s verdict. The critic remains fallible, sincere, and revisable. He also remains the last operative judge before a reading becomes the text printed, translated, preached, and believed.

Probability all the way down therefore produces the church’s operative identification of the authoritative wording from the bottom up. Manuscripts supply alternatives; canons assign tendencies; critics weigh the tendencies; editions publish conclusions; churches receive the conclusions as the current best identification of the wording of the non-extant autograph. The confessional order runs the other way. God speaks; the Word authenticates itself; providence preserves it; the church receives it; criticism serves it; and every human judgment remains answerable to the Holy Spirit speaking in Scripture.

Notes and Further Reading