How each side says we know (draft)

A design note, with its first phase built. Nothing here changes a verdict or a rating. It names the rules the ratings were already computed under, gives each its holders, and checks whether each side keeps the rule it professes.

A Protestant and a Catholic arguing over Romans 3:28, "a man is justified by faith without the deeds of the law", are seldom arguing only over Romans 3:28. Underneath is a disagreement about how anything is known from Scripture at all. The Westminster Confession says that "the infallible rule of interpretation of Scripture is the Scripture itself" (I.ix), and that the supreme judge of controversies "can be no other but the Holy Spirit speaking in the Scripture" (I.x). The Council of Trent receives, beside the written books, the traditions "received by the Apostles from the mouth of Christ himself, or from the Apostles themselves, the Holy Ghost dictating" (Session IV, in Waterworth's translation), and the Second Vatican Council says Scripture and tradition are to be accepted with the same devotion (Dei Verbum 9). The library's own evidence standard asks for neither: only the texts, and evidence anyone can check.

An argument that quietly assumes one of these answers has decided the dispute before hearing it. So the library names them.

What an epistemology is here

An epistemology is how a side holds that something is known: the rule by which it decides what counts as a reason. The library writes each one as a single value with five parts (Testimony.Logic.Epistemology):

• what it admits as grounds. Every claim in the library rests finally on one of four things: a text of Scripture; evidence anyone can check, such as a critical text, a lexicon or argued scholarship; the word of a council, a confession or a theologian; or a premise the library puts forward without a source of its own (what each position rests on);
• how an objection to a Scripture-only step must be argued. An argument moves from claims to a conclusion by steps: each step is a "because", and each claim is something the argument asserts. Some rules hold that a step resting on Scripture alone can be answered only from Scripture;
• the principles of reading it licenses, such as "a word means no more than its context needs";
• how it takes an argument from silence: never, or only when the argument says why the text would have mentioned the thing had it been so (Checking an argument for fallacies), or freely. Where a rule's source does not say, the library does not guess;
• who holds it, cited, as a claim is.

Four are encoded (TestimonyArguments.Epistemologies):

Each tradition's own authority, and no other's. The Catholic rule admits a council's, a Father's or the Church's teaching as the Catholic tradition receives it, and not Westminster's word or Calvin's. The Orthodox rule admits the Fathers and councils as the Orthodox Church receives them. Whose authority a claim rests on is read from the tradition its citation records as advancing it. That is a proxy, not a record of who wrote each work, and a Father both traditions receive is counted by the tradition that cites him.

The Catholic and Orthodox rules admit grounds alike in kind, and they still differ. Florovsky rejects Trent's model of two sources outright: for him tradition adds nothing to Scripture, but is the Church's living understanding of it, with no single office as its infallible interpreter. The library's checks of claims cannot see that difference yet; it lies in how a reading is reached, not in what it rests on.

None of the sources speaks to arguments from silence, as cited, so none of the rules does. Only the library's rule records a principle of reading, the one it cites to Silva. The others each have their own, and they will be recorded when they are cited in their own words.

The standards, given their holders

The library's credibility checks (Computed ratings) have always been run under three standards, evidence, tradition and Scripture. Each was a rule about which grounds count, written without a holder. Two turn out to say exactly what an encoded rule says (two_standards_are_rules):

• the scriptural standard is Westminster's rule;
• the evidence standard is the library's own.

The tradition standard is neither the Catholic rule nor the Orthodox one. It admits a council's, a confession's or a theologian's word from any tradition, Westminster's and Trent's alike, where each of those rules admits only its own. So a rating computed under the tradition standard is computed under every tradition's authorities at once, which is no one's rule. No rating and no verdict moved; what changed is what the page can say each standard is.

Does each side keep its own rule?

The charge of a double standard is common in debate: you ask me for Scripture and rest your own case on a confession. The library can now check it for claims. A reading is held to the rule of the side whose reading it is, and every claim it asserts is checked against that rule (ArgumentPackage.heldTo).

Which rule a reading professes is declared, reading by reading, with the reason. It is not read off the reading's citations, because a Reformed reading that cites a Catholic exegete is still a Reformed reading. In the dispute over faith alone (sola fide):

• the Reformed readings profess Westminster's rule: faith alone from Paul, from Jesus' words at Luke 7:50 and in John 6, and from Peter at Jerusalem; Paul's gospel against Trent's definition; and the Reformed answer to Chrysostom;
• Trent's two readings profess Trent's;
• the scholars' readings, Sanders and Dunn, their critics, the apocalyptic reading and Jervell, and two arguments about what Paul's word means, the word study and Romans 4, are held to the library's evidence rule: they argue as scholars, from texts and lexica;
• Aquinas, Luther and Chrysostom are held to none. None of them wrote Trent's rule, Westminster's or the Orthodox one. Aquinas has his own, and Luther's tradition has the Formula of Concord; the library cites neither as a rule. Chrysostom is a Father whom the Orthodox and the Catholic traditions both receive, and calling him either's would be a claim the library does not argue. Holding any of them to another's rule would give him a rule he did not write.

What the check found, and what changed because of it. When the check was first run, four Reformed readings failed it: faith alone from Paul, from Peter and from John 6, and the Reformed answer to Chrysostom. Each asserts that justification and sanctification are inseparable but distinct. Justification is God's pardoning a sinner and accepting him as righteous; sanctification is his making that sinner holy over a life; the claim is that the two always go together and are not the same thing. The library had cited it to the Westminster Confession (XIII.1, XI.1) and to Calvin (Institutes III.xi.6), and to nothing else. Westminster's rule admits Scripture and evidence, not a confession's or a theologian's word, so as encoded the four readings rested a claim on what their own rule refuses their rival.

The fault was in the earlier encoding, not in the readings. A Reformed reader cites the Confession as a summary of Scripture, printed with its Scripture proofs, and the library had cited the summary without the proofs. It had also cited the wrong summary. Confession XIII.1 is about sanctification alone and does not state the claim; the Larger Catechism does, at question 77: "Although sanctification be inseparably joined with justification, yet they differ". Under each clause it prints its proofs: 1 Corinthians 6:11 and 1:30 for "inseparably joined", and for the differences Romans 4:6 and 8, Ezekiel 36:27 with Hebrews 9:13–14, Romans 3:24–25, and Romans 6:6 and 14. Calvin's text for Scripture distinguishing the two is 1 Corinthians 1:30.

So the encoding was changed. The claim is now cited to the Larger Catechism, with its proofs, in both arguments that assert it, sola fide and the canonical witness. Every party held to a rule now keeps it (sola_fide_own_rules). No rating and no verdict moved, because nothing weighs a claim by what it rests on. What moved is what a reader can check: the texts are on the page.

This is what the check is for. A rule each side professes, applied to each side's own claims, found a premise the library had cited on a confession's word alone, and pointed to the wrong place. Neither showed in any verdict.

Trent's readings keep their rule, and could have failed it. Each rule admits authority only from its own tradition, so a Catholic reading resting a claim on Westminster's word would fail Trent's rule as a Reformed reading resting a claim on Trent's word fails Westminster's. Trent's reading of justification rests two claims on Catholic authority alone: that justification is not the remission of sins only but the renewal of the inward man, and that the justified grow in that renewal by good works done in grace. Trent's rule admits both, and Westminster's refuses both (trent_keeps_its_own_rule). The two rules part exactly where the two sides do.

One limit remains. The library's evidence rule admits the same grounds for claims as Westminster's, so the check could not have told the two apart: what sets Westminster's rule apart is how a Scripture-only step is answered, and steps are not yet checked.

What this does not do

Which rule is right is the question of the dispute over Scripture alone (sola scriptura), and a rule built into the solver, the program that computes verdicts, would settle that question before hearing it. So every result here stays conditional: under this rule, this reading rests on what the rule admits. No verdict or rating reads it.

It checks claims, and not yet steps. A step's citation in the library records its rating, and a step is marked as disputed by citing a reader who grants its starting points and denies its conclusion. So a Reformed step is often cited to Trent's anathema on it, its formal condemnation, which says who disputes the step and nothing about what the step rests on. Holding steps to a rule needs each step to record what licenses it, apart from who disputes it.

What comes next

1. Steps, held to a rule, once a step records what licenses it.
2. Whose authority, exactly. Each rule now admits authority from its own tradition, read from the tradition a citation records. A record of each work's own tradition would make that exact, and let a Father both traditions receive count for both.
3. Principles and silences, held to a rule. Each rule's principles of reading, cited in its own words, and the silences the fallacy checks already find. A move that reads by a principle its own rule does not license can be charged as claims are.
4. Verdicts under each rule. A hearing, the dispute weighed with only the objections a standard accepts, already weighs a dispute by one rule (Computed ratings). With the standards given their holders, a page can say "under Westminster's rule this prevails; under Trent's it is contested".
5. The other disputes, held to their rules, beginning with the one over Scripture alone, where the rule is itself the question; and rules of their own for Aquinas, Luther and Chrysostom, once each is cited.