The Shim
Filed in place of the planned collision on the tīrtha that launders premeditated murder (ch 277 against ch 260–261).
Chapter 167. One sentence, and both halves of an impossible position are in it:
> "Following the usual practices of the world, Shiva was also miserable, though he is the one who is accomplished in many kinds of pastimes and deludes everyone."[^1]
He is miserable. He is also, by the doctrine this volume states everywhere, unattached, beyond the guṇas, "resplendent in his indifference." Both stand. Neither is qualified.
And it happens sixteen times. A sweep of Volume 2's body returns sixteen deployments of that clause, and sorting them settles what it's for: ten of the sixteen sit in front of an emotion — anger, grief, agitation, wonder, weeping, dejection. Not one sits in front of a killing, a deception, or a broken promise, and this volume has many of each.
The distribution is the giveaway. Six of the sixteen fall inside the eight chapters of the Gaṇeśa episode. Three fall in ch 194 alone, where Śiva sees Gaurī in distress. Seven are scattered across the remaining hundred and thirty.[^2]
Against that, the classical account of how minds handle contradictions. Festinger's fox wants the grapes, can't reach them, and rewrites the fact — they were sour anyway. Dissonance gets reduced by revising one of the two elements.
Nothing is revised here. The transcendence stands, stated in the very sentences where the god is weeping. The grief stands, at length, with full narrative weight and no hedging of its reality.
A third element is inserted between them: a reusable phrase whose only job is to let the pair coexist.
⚑ That is not resolution, not denial, and not compartmentalisation into separate contexts. It is a shim — a portable, deployable-on-demand connective that makes an unresolved contradiction survivable indefinitely, at a cost of about six words per use.
The fox rewrote the grapes. This text kept the grapes, kept the failure to reach them, and invented a sentence.
Shims are a distinct class of contradiction-management, and they have three properties the rewriting model doesn't.
They are countable, so strain is measurable. You cannot count a rewritten belief — the evidence is gone by construction, which is what makes dissonance research hard. You can count a shim. And where it clusters is where the contradiction is worst: six deployments in eight chapters, on the one episode where a god who cannot be moved has to bury his son. The hedge is a strain gauge, and it works even on a text whose author is a thousand years dead and cannot be interviewed.
They scale, where rewriting doesn't. Individual dissonance reduction is expensive and has to be redone for every case. A phrase can be handed to the next compiler, deployed by someone who never felt the tension, and eventually decay into filler — which is exactly what happens twice at the edges of this sample. One instance stretches to cover an act (Kṛṣṇa fighting a Śiva-devotee, ch 225) rather than a feeling; one attaches to a sage doing austerities (ch 253), who needed no cover at all.
⚑ That decay is the signature of a shim that has outlived the discomfort it was built for. Which is the strangest part: the contradiction is still fully present in the text, and by the end nobody transmitting it appears to feel it anymore.
And the wording escalates with the load. Chapter 194 has three in a row and they get stronger: following the customary practices of the world → he displayed the customary practices of the world → behaving like an ordinary person. The narrator is not stamping a form. He is adjusting the disclaimer to the size of the problem — which makes the strength of a hedge a second, finer measure of what its author found indefensible.
The vault-facing turn. This vault runs shims. `[SPECULATIVE]` · `[PLAUSIBLE — needs corroboration]` · `[PARAPHRASED]` · `[POPULAR SOURCE]` · this build's own `[n=1 OBSERVED]`. Every one of them exists to let a claim sit in ARCHIVES without either strengthening the evidence or removing the claim. That is the same move, and it is the right move — the alternative is a corpus that only holds settled things, which is not a knowledge base but a summary.
But the properties transfer. If shims are strain gauges, then the tag distribution in this vault is a map of where its evidence is thinnest, and nobody has ever read it that way. And if shims decay, then the tags will eventually be applied by pattern rather than by judgement, and a `[SPECULATIVE]` that means I hedged because that's what one does here is filler — indistinguishable from the real thing at the point of use, and only detectable by the same edge-analysis that caught ch 225 and ch 253.
- A count of this vault's own hedge-tags, by page and by domain. Cheap, grep-able, and it has never been run. If the tags cluster the way this clause clusters — heavy at a few load-bearing places, sparse everywhere else — the strain-gauge reading holds. If they're evenly spread, they're a style and this half collapses.
- An edge-check for decay. Pull the tagged claims at the extremes and ask whether the tag is doing work. A `[SPECULATIVE]` on something nobody would have believed anyway is the vault's ch 253.
- A second corpus with a countable shim. Legal boilerplate is the obvious candidate — without prejudice, for the avoidance of doubt, nothing in this clause shall — and the comparison would show whether escalation-with-load is general or Purāṇic.
- And the deflationary reading, which is real. Following the customary practices of the world may not be managing a contradiction at all. It may be straightforward līlā doctrine: the god plays, the play includes appearing to grieve, and there was never any tension for a contemporary reader — only for me, arriving with a category the text doesn't share. ⚑ That reading is live and this collision does not defeat it. What it does not explain is the distribution: a doctrine stated as doctrine would appear wherever līlā appears, and this appears where the emotional load is highest and nowhere else.