>The Unfalsifiable Function

Some functions can't be falsified — too many free parameters, no fixed output. When you can't test the return value, you audit the source instead.

The fate-calculation tradition (命理) claims that the year, month, day, and hour of birth — expressed as four stem-branch pairs, the "eight characters" (八字) or "Four Pillars" (四柱) — determine the broad contours of a life: health, wealth, marriage, career, longevity, all readable in the configuration of elements at the moment of birth.

When the Siku scholars reached the fate-calculation texts in juan 109–110, they faced an epistemological challenge unlike anything in the military or philosophical sections. A forged military text can be tested against historical records. A fabricated commentary can be checked against the text it claims to comment on. But a function that claims to predict destiny from birth data cannot be tested by bibliographers — you'd need longitudinal studies, control groups, a methodology the eighteenth century hadn't invented. The compilers solved this by not trying. They didn't evaluate whether fate calculation works. They evaluated the texts that teach it — authorship, dating, internal consistency, intellectual quality, lineage. A characteristically Qing move: apply the tools you have to the questions you can answer, and let the rest stand.

The most important text is the Yuanhai Ziping (淵海子平), attributed to Xu Ziping (徐子平) of the Song. Xu Ziping is credited with transforming the earlier three-pillar system (year, month, day) into the four-pillar system by adding the hour — doubling the combinatorial complexity and setting the standard method still used today. The compilers' review is cautiously affirmative: a genuine transmission of Xu Ziping's methods, though the received text is a compilation rather than single-authored, with later additions whose boundaries aren't always clear.

What they valued was systematic quality. The text doesn't merely list rules — it provides a theoretical framework: the interactions of the five elements (五行), the relationships of stems and branches, strength and weakness, flourishing and decay. An internal logic that, whatever you think of its premises, is consistent. The compilers could evaluate that consistency even when they could not evaluate the predictions it generated.

The Sanming Tonghui (三命通會), compiled by Wan Minying (萬民英) in the late Ming, is the largest encyclopedia of fate calculation ever written — twelve volumes covering theory, chart construction, configurations, and historical case studies. The compilers praised its comprehensiveness: Wan Minying collected and organized material from dozens of otherwise-inaccessible sources, functioning as a library of the tradition. But they noted its weaknesses — varying quality, some chapters rigorous, others accumulating folk interpretations without critical assessment. Wan was a compiler, not a critic; the inclusiveness sometimes cost coherence. They accepted it not because they endorsed fate calculation but because it's a reliable record of what the tradition teaches. Their standard, again, is documentary: does the text accurately represent a body of knowledge, regardless of whether that knowledge is true?

The Li Xuzhong Mingshu (李虛中命書) posed the most interesting textual problem. Li Xuzhong (李虛中) was a Tang official whose methods are described in a famous essay by the literary giant Han Yu (韓愈), who praised his uncanny accuracy from birth data. But Han Yu's account describes Li using only three data points — year, month, day — consistent with the pre-Xu Ziping method. The received text references the four-pillar system, which didn't exist until the Song.

The conclusion was precise: the early portions, discussing three pillars, are likely genuine Tang material; the later sections, with four-pillar analysis, are Song additions. The same diagnostic applied across the catalog — a genuine core contaminated by later interpolation. The text is neither wholly authentic nor wholly forged. It's a palimpsest, original material overwritten with later additions, both preserved under one title. They accepted it with caveats, documenting exactly where the seams were.

Unable to test the output, the compilers focused on four things they could assess. Textual lineage — can the text be traced through bibliographic records to its claimed period? Internal consistency — does the system contradict itself? A text saying a stem-branch combination means wealth in one chapter and poverty in another has a problem identifiable without testing predictions. Intellectual quality — does it explain its reasoning or merely list rules? "People born in the Year of the Dragon will be wealthy" with no derivation is less serious than a claim derived from element interactions, even if both are untestable. Authorship honesty — is it by who it says? Several texts were attributed to famous names — Guiguzi, Zhuge Liang, Li Chunfeng — with no evidence; if the author lied about who wrote it, what else did they lie about?

The compilers' handling reveals a grudging concession. They never say the system works. But they never say it doesn't — and the silence on the empirical question is conspicuous compared to their explicit skepticism about other occult practices. Elsewhere they demolished marriage-matching formulas with statistical reductio, showing the system condemned impossibly many marriages as harmful — possible because date-selection systems make testable claims.

Fate calculation is harder to attack. Its predictions are probabilistic and conditional — a chart "tends toward" wealth or "inclines toward" difficulty, modulated by decade-luck pillars (大運) and annual influences (流年). The system has enough degrees of freedom that no single prediction is precisely falsifiable. The compilers, trained in textual criticism rather than experimental method, recognized this implicitly: they stuck to what they could evaluate and didn't overreach. The concession was also social — fate calculation was practiced from palace to village, the emperor's own birth chart a matter of court record. To dismiss the whole tradition was to dismiss a practice the court itself participated in.

The canon the Siku scholars defined is small and deliberate: the Yuanhai Ziping as foundational four-pillar text, the Sanming Tonghui as comprehensive encyclopedia, the Li Xuzhong Mingshu as oldest surviving source (accepted with caveats), plus supporting texts that extend the framework without departing from it. Notable for what it excludes: the wildly popular rhyming-couplet summaries that traded rigor for memorability were largely cut. The tradition's greatest hits were not, in the scholars' judgment, its greatest works — and the canon that survived imperial review is the one modern eight-character and Ziwei Doushu practitioners still work within.

The compilers did not determine whether fate calculation is true. They determined which texts represent it honestly — and in a tradition where knowledge is transmitted through books, that distinction is the one that matters.