#!/usr/bin/env bash # Validate every published number against the program that produces it. # # This is the check a stranger runs. It needs no account, no key and no access to the # substrate repository: it compiles the programs in this directory, runs them against the # corpora pinned in ../corpus, and confirms that the figures printed appear in the # published study pages. It also confirms those pages carry no private reference. # # A number in a page that no program prints is not reproducible, and this harness says so. set -uo pipefail HERE="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" ROOT="$(cd "$HERE/.." && pwd)" PASS=0 ABSENT=0; FAIL=0 ok() { printf ' PASS %s\n' "$*"; PASS=$((PASS+1)); } bad() { printf ' FAIL %s\n' "$*"; FAIL=$((FAIL+1)); } have() { command -v "$1" >/dev/null 2>&1; } echo "=== 1. corpora match their pinned digests ===" for d in "$ROOT"/corpus/*/; do [ -f "$d/SHA256SUMS" ] || continue if ( cd "$d" && shasum -a 256 -c SHA256SUMS >/dev/null 2>&1 ); then ok "$(basename "$d") digests verify" else bad "$(basename "$d") DIGEST MISMATCH — the bytes are not the ones measured" fi done echo "=== 2. every program compiles and runs ===" if have xcrun || have swiftc; then SC=$(have xcrun && echo "xcrun swiftc" || echo "swiftc") # THE LAW IS COMPILED IN, NEVER COPIED. A script that consumes FusionLaw is # built multi-file against its single home — which is why the law cannot fork # again. Top-level code must be named main.swift, so the script is staged. LAWSRC="$ROOT/app/FusionCourt/Sources/FusionLaw" for p in "$HERE"/*.swift; do n=$(basename "$p" .swift) extra="" if grep -q "FusionLaw\.\|LawConstants\." "$p" 2>/dev/null; then extra="$extra $(ls "$LAWSRC"/*.swift 2>/dev/null | tr '\n' ' ')" fi # A script consuming the operating-point court — or its machine table — links that target too. if grep -q "FusionOperatingPointLaw\.\|OperatingEnvelope(\|FusionMachines\." "$p" 2>/dev/null; then extra="$extra $(ls "$ROOT/app/FusionCourt/Sources/FusionOperatingPoint"/*.swift 2>/dev/null | tr '\n' ' ')" fi # A script consuming the program's lattice slice links its home file, by mention, the # same way: the 16 vertices and every determinant have one home, never a script copy. # ONE FILE, not the FusionLattice directory: the slice is stdlib-only Int64, while the # directory also holds TendingSwarm.swift (Foundation actors), which no script uses and # which a wasm32 staging of the same rule would then have to compile. if grep -q "DensitySlice\." "$p" 2>/dev/null; then extra="$extra $ROOT/app/FusionCourt/Sources/FusionLattice/DensitySlice.swift" fi stage="$(mktemp -d)"; berr="$(mktemp)" # STAGING NAME IS A CHOICE, AND IT IS NOT THE SAME CHOICE FOR EVERY PROGRAM. # Top-level code is legal only in a file called main.swift, so that is the default. # A file declaring `@main` is the OPPOSITE case: Swift refuses that attribute in a # file that could carry top-level code, and naming it main.swift fails the build with # "'main' attribute cannot be used in a module that contains top-level code" — which # this harness would then report as the program being broken. It is not; the staging # was. Measured on wasi-sandwiched.swift, which is @main. if grep -q '^@main' "$p" 2>/dev/null; then src="$stage/$(basename "$p")" else src="$stage/main.swift"; fi cp "$p" "$src" # ONE LAW, ONE HOME — the market-shear tool. wasi-sandwiched defines no conjunct, # no pool arithmetic and no shortfall formula of its own: it compiles a VERBATIM # BYTE SLICE of extraction-exact.swift, which is exactly what its own build script # does. Building it here the same way is the point — a harness that linked a COPY # of the law would be grading a second law, and then there would be two. # # THE BOUNDARY IS CHECKED, NOT ASSERTED. Line END must close referenceFigures() and # END+2 must open the detector's own MAIN. If the detector source moves, this # REFUSES rather than slicing a law in half and reporting a green run over it. if grep -q 'CORE_SLICE_SHA256' "$p" 2>/dev/null; then CORE="$HERE/extraction-exact.swift" # THE BOUNDARY IS DERIVED FROM A NAMED MARKER, NEVER TYPED. It was the literal # 2535 here and in wasi-sandwiched-build.sh and wasi-sandwiched-one-law.sh, so # any addition to the law was a four-file change and three of the four would # have gone on quoting a stale number. Derived from the SECTION 11 marker it # tracks the law; the structural check below is unchanged and still REFUSES. # Verified both ways: the rule returns exactly 2535 on the file the constant # was written for, and follows the boundary on the file that moved. # THE SAME LOCALE FIX ALREADY LANDED IN wasi-sandwiched-build.sh AND # wasi-sandwiched-one-law.sh, and this THIRD site was missed. It is the one the # harness itself reads, so both wasi tools were still refusing a correct tree with # an EMPTY line number — the marker's dash is an EM DASH, three bytes in UTF-8, and # with LANG and LC_ALL unset awk's `.` matches ONE BYTE. Same anchored pattern as # the other two, so all three sites now agree in every locale. END=$(awk '/^\/\/ SECTION 11 /{print NR-3; exit}' "$CORE") if [ -f "$CORE" ] && [ "$(sed -n "${END}p" "$CORE" 2>/dev/null)" = "}" ] \ && sed -n "$((END+2))p" "$CORE" 2>/dev/null | grep -q '^// ====='; then sed -n "1,${END}p" "$CORE" > "$stage/core-detector.swift" { echo "let CORE_SLICE_SHA256 = \"$(shasum -a 256 "$stage/core-detector.swift" | awk '{print $1}')\"" echo "let CORE_SLICE_BYTES = \"$(wc -c < "$stage/core-detector.swift" | tr -d ' ')\"" echo "let CORE_SLICE_LINES = \"1..$END\"" echo "let CORE_SOURCE_NAME = \"extraction-exact.swift\"" echo "let CORE_SOURCE_SHA256 = \"$(shasum -a 256 "$CORE" | awk '{print $1}')\"" echo "let CORE_SOURCE_BYTES = \"$(wc -c < "$CORE" | tr -d ' ')\"" echo "let TOOL_SOURCE_SHA256 = \"$(shasum -a 256 "$p" | awk '{print $1}')\"" echo "let BUILD_UTC = \"$(date -u +%Y-%m-%dT%H:%M:%SZ)\"" } > "$stage/core-pin.swift" extra="$extra $stage/core-detector.swift $stage/core-pin.swift" else bad "$n SLICE BOUNDARY MOVED — extraction-exact.swift line $END no longer closes referenceFigures(); refusing to compile half a law" rm -f "$berr"; rm -rf "$stage"; continue fi fi out="$(cd "$ROOT/corpus/flood-lead-time" 2>/dev/null || cd "$HERE"; \ $SC -O -swift-version 5 $extra "$src" -o "/tmp/val_$n" 2>"$berr" && "/tmp/val_$n" 2>/dev/null "/tmp/out_$n.txt" else # A STALE TRANSCRIPT IS WORSE THAN NO TRANSCRIPT, and it was being left in place. # When a build failed, /tmp/out_.txt from an EARLIER run survived, so the # harness printed ONE failure and then passed every check_figure pinned to that # program against yesterday's bytes. Removing it makes those rows report SKIP — # absent, which is a third answer and not a pass. rm -f "/tmp/out_$n.txt" # DID NOT COMPILE and COMPILED BUT PRINTED NOTHING are different answers and were # printed alike, with the compiler's reason discarded to /dev/null. if [ -s "$berr" ]; then bad "$n DID NOT COMPILE — $(head -1 "$berr" | cut -c1-160)" else # A TYPE DECLARATION IS NOT A STUDY THAT PRINTED NOTHING, and grading it as one # trains a reader to step over a red row. This loop globs every *.swift in # reproduce/, so it meets both kinds. The discriminator is STRUCTURAL: 96 of the 97 # files here emit through print( / out( / emit(; a file with none of the three # carries no program at all and has nothing it could print. Measured on # shear-ledger-v1.swift, which declares 7 types and emits by no mechanism. # A declaration still owes the tree something, so it is graded on that instead: # some other file must compile against its types, or it is an orphan and this says # so. Both arms fire -- a real study with no output is still reported as one. emits=$(( $(grep -c -F 'print(' "$p") + $(grep -c -F 'out(' "$p") + $(grep -c -F 'emit(' "$p") )) decls=$(grep -cE '^(public )?(struct|enum|class|protocol|extension) ' "$p") if [ "$emits" -eq 0 ] && [ "$decls" -gt 0 ]; then users="" for ty in $(grep -oE '^(public )?(struct|enum|class|protocol) [A-Za-z0-9_]+' "$p" | awk '{print $NF}' | sort -u); do # WHOLE WORD, NOT SUBSTRING. Written as `grep -l "$ty"` this reported # shear-ledger-v1 as consumed by aso-offtarget-exact-vs-float because that # file contains "Bases" and one of the declared types is `Base` — a green row # earned by a substring, which is worse than the red row it replaced. The # type must appear delimited by a non-identifier character on both sides. u=$(grep -lE "(^|[^A-Za-z0-9_])$ty([^A-Za-z0-9_]|$)" "$HERE"/*.swift 2>/dev/null | grep -v "^$p$" | head -1) [ -n "$u" ] && users="$users $(basename "$u")" done users=$(echo "$users" | tr ' ' '\n' | sort -u | tr '\n' ' ' | sed 's/^ *//') if [ -n "$users" ]; then ok "$n declares $decls types and no program -- compiled against by $users" else bad "$n declares $decls types that NO file in reproduce/ compiles against -- an orphan declaration, not a study that printed nothing" fi else bad "$n compiled and printed nothing — every exit must print its reference figures" fi fi fi rm -f "/tmp/val_$n" "$berr"; rm -rf "$stage" done else echo " SKIP — no Swift toolchain on this host; the programs are the evidence, install Swift to run them" fi echo "=== 3. published figures appear in the program output that produces them ===" # each row:
check_figure() { local prog="$1" fig="$2" page="$3" local o="/tmp/out_$prog.txt" [ -f "$o" ] || { echo " SKIP $fig (no output from $prog)"; return; } if grep -qF -- "$fig" "$o"; then if [ -z "$page" ] || grep -qF -- "$fig" "$ROOT/$page" 2>/dev/null; then ok "$fig — printed by $prog${page:+ and present in $page}" else bad "$fig printed by $prog but ABSENT from $page" fi else bad "$fig NOT printed by $prog — the page cites a number its program does not produce" fi } check_figure lora-time-on-air "287.744" "Study-30-Sovereign-Edge-Pod.md" check_figure lora-time-on-air "1004.544" "Study-30-Sovereign-Edge-Pod.md" check_figure z8-vs-e8-lattice "240" "Study-30-Sovereign-Edge-Pod.md" check_figure unimodular-control-arms "4000" "" check_figure reentry-alumina-ledger "374.4" "Study-29-Continuous-Model-Shear.md" check_figure guadalupe-wave-ledger "180" "" check_figure rate-of-rise-common-window "885" "" check_figure flourishing-entropy-ledger "9 resolve exactly" "" check_figure flourishing-entropy-ledger "1/1" "" # --- Study 38: the loss-reserve triangle, exact against float --- check_figure reserve-triangle-exact-vs-float "STUDY38_RESERVE_TRIANGLE_EXACT_VS_FLOAT" "Study-38-Loss-Reserve-Triangle.md" check_figure reserve-triangle-exact-vs-float "SELFTEST PASS" "" check_figure reserve-triangle-exact-vs-float "control arms failed = 0" "Study-38-Loss-Reserve-Triangle.md" check_figure reserve-triangle-exact-vs-float "5,393,021" "Study-38-Loss-Reserve-Triangle.md" check_figure reserve-triangle-exact-vs-float "77,900" "Study-38-Loss-Reserve-Triangle.md" check_figure reserve-triangle-exact-vs-float "-16,662,494" "Study-38-Loss-Reserve-Triangle.md" check_figure reserve-triangle-exact-vs-float "21,770,741" "Study-38-Loss-Reserve-Triangle.md" check_figure reserve-triangle-exact-vs-float "19,041,666" "Study-38-Loss-Reserve-Triangle.md" check_figure reserve-triangle-exact-vs-float "17,138,459" "Study-38-Loss-Reserve-Triangle.md" check_figure reserve-triangle-exact-vs-float "15,618,034" "Study-38-Loss-Reserve-Triangle.md" check_figure reserve-triangle-exact-vs-float "2,390" "Study-38-Loss-Reserve-Triangle.md" check_figure reserve-triangle-exact-vs-float "f1=14043/46" "Study-38-Loss-Reserve-Triangle.md" check_figure reserve-triangle-exact-vs-float "Dorinco Rein Co" "Study-38-Loss-Reserve-Triangle.md" check_figure reserve-triangle-exact-vs-float "9,007,199,254,740,992" "Study-38-Loss-Reserve-Triangle.md" # --- Study 39: the actuarial domain, exact against float --- check_figure actuarial-domain-exact-vs-float "STUDY39_ACTUARIAL_DOMAIN_EXACT_VS_FLOAT" "Study-39-Actuarial-Domain.md" check_figure actuarial-domain-exact-vs-float "SELFTEST PASS" "" check_figure actuarial-domain-exact-vs-float "control arms failed = 0" "Study-39-Actuarial-Domain.md" check_figure actuarial-domain-exact-vs-float "16,128" "Study-39-Actuarial-Domain.md" check_figure actuarial-domain-exact-vs-float "10,878" "Study-39-Actuarial-Domain.md" check_figure actuarial-domain-exact-vs-float "2,016" "Study-39-Actuarial-Domain.md" check_figure actuarial-domain-exact-vs-float "3,990" "Study-39-Actuarial-Domain.md" check_figure actuarial-domain-exact-vs-float "6,048" "Study-39-Actuarial-Domain.md" check_figure actuarial-domain-exact-vs-float "2,835" "Study-39-Actuarial-Domain.md" check_figure actuarial-domain-exact-vs-float "1,995" "Study-39-Actuarial-Domain.md" check_figure actuarial-domain-exact-vs-float "22,973,085" "Study-39-Actuarial-Domain.md" check_figure actuarial-domain-exact-vs-float "11,561,327" "Study-39-Actuarial-Domain.md" check_figure actuarial-domain-exact-vs-float "IT 2022, from age 9, 86 steps" "Study-39-Actuarial-Domain.md" check_figure actuarial-domain-exact-vs-float "fewest agreeing significant digits" "Study-39-Actuarial-Domain.md" check_figure actuarial-domain-exact-vs-float "3,845" "Study-39-Actuarial-Domain.md" check_figure actuarial-domain-exact-vs-float "442, 547, 631" "Study-39-Actuarial-Domain.md" # --- Study 42: the exact contract, 2.7M NFIP settlements in Int128 cents --- check_figure exact-contract-nfip "STUDY42_THE_EXACT_CONTRACT" "Study-42-The-Exact-Contract.md" check_figure exact-contract-nfip "SELFTEST PASS" "" check_figure exact-contract-nfip "control arms failed = 0" "Study-42-The-Exact-Contract.md" check_figure exact-contract-nfip "Int128 String round-trip failures = 0" "Study-42-The-Exact-Contract.md" check_figure exact-contract-nfip "mesh ledger equals whole-corpus ledger = true" "Study-42-The-Exact-Contract.md" # --- Study 26: the exact discrimination court, and the exhaustive off-target screen --- check_figure mr-topology-vs-expression-exact "TOPOLOGY_EXPLAINS: 11 of 17 tumour types" "Study-26-Master-Regulator-Bonds.md" check_figure mr-topology-vs-expression-exact "EXPRESSION_ADDS : 5 of 17 tumour types" "Study-26-Master-Regulator-Bonds.md" check_figure mr-topology-vs-expression-exact "d0051d7a19daa0cb7f1a3d4f7be7433af73401bbf25fe41079f6409c066fee5a" "Study-26-Master-Regulator-Bonds.md" check_figure mr-topology-vs-expression-exact "100891344545564193334812497256" "Study-26-Master-Regulator-Bonds.md" check_figure mr-topology-vs-expression-exact "9.138e-12" "Study-26-Master-Regulator-Bonds.md" check_figure mr-topology-vs-expression-exact "4.756e-6" "Study-26-Master-Regulator-Bonds.md" check_figure mr-topology-vs-expression-exact "1.151e-1" "Study-26-Master-Regulator-Bonds.md" check_figure mr-topology-vs-expression-exact "2.685e-29" "Study-26-Master-Regulator-Bonds.md" check_figure mr-topology-vs-expression-exact "SELFTEST PASS" "" check_figure pelacarsen-offtarget-whole-transcriptome "513de7e9db6556df1895bfce4cb4d69e4816d7b45b75bee1dc8452335c2c7757" "Study-26-Master-Regulator-Bonds.md" check_figure pelacarsen-offtarget-whole-transcriptome "TGCTCCGTTGGTGCTTGTTC" "Study-26-Master-Regulator-Bonds.md" check_figure pelacarsen-offtarget-whole-transcriptome "670670" "Study-26-Master-Regulator-Bonds.md" check_figure pelacarsen-offtarget-whole-transcriptome "1467336203" "Study-26-Master-Regulator-Bonds.md" check_figure pelacarsen-offtarget-whole-transcriptome "5a320f524d73b5793518eb19b118829033713443d0f42af20a67bb31cc06cf56" "Study-26-Master-Regulator-Bonds.md" # --- Study 26 S3-COMBINATION: the eleven pairs, made re-derivable 2026-09-07 --- check_figure study26-combination-pairs-exact "STUDY26_COMBINATION_PAIRS__ELEVEN_CLEAR_ALL_THREE_CONTROLS" "Study-26-Master-Regulator-Bonds.md" check_figure study26-combination-pairs-exact "d0117523ff3b950a0741de281671c0bd977f8dc003f41972f2bfac2f50994d74" "Study-26-Master-Regulator-Bonds.md" check_figure study26-combination-pairs-exact "estradiol + AMG-208" "Study-26-Master-Regulator-Bonds.md" check_figure study26-combination-pairs-exact "olaparib + ursodeoxycholyltaurine" "Study-26-Master-Regulator-Bonds.md" check_figure study26-combination-pairs-exact "drospirenone + alpelisib" "Study-26-Master-Regulator-Bonds.md" check_figure study26-combination-pairs-exact "HMN-214 + saracatinib" "Study-26-Master-Regulator-Bonds.md" check_figure study26-combination-pairs-exact "XMD-892 + NVP-BGJ398" "Study-26-Master-Regulator-Bonds.md" # --- Rentosertib: the structural lock, the network census, the modality register --- check_figure rentosertib-structure-lock-exact "C27H30FN7O" "A-Drug-An-AI-Designed.md" check_figure rentosertib-structure-lock-exact "b5da901cbcba5535" "A-Drug-An-AI-Designed.md" check_figure rentosertib-structure-lock-exact "ZVDNXHUSIKGTSF-UHFFFAOYSA-N" "A-Drug-An-AI-Designed.md" check_figure tnik-repairs-exact "12548903" "" check_figure tnik-network-degree-exact "12017368" "" check_figure modality-register-exact "validation_arms_failed = 0" "" # --- the genome-wide CRISPR off-target map --- check_figure crispr-genome-offtarget-exact "487b4f81de2d24bd0bb11ecd1d8d42778e3a5d91b9edb33627c86dcc8df34980" "CRISPR-Genome-Off-Target-Map.md" check_figure crispr-genome-offtarget-exact "CRISPR_GENOME_OFFTARGET_EXACT__COMPLETE_ENUMERATION_IS_OBSERVER_INVARIANT" "CRISPR-Genome-Off-Target-Map.md" check_figure crispr-genome-offtarget-exact "487b4f81de2d24bd0bb11ecd1d8d42778e3a5d91b9edb33627c86dcc8df34980" "CRISPR-Genome-Off-Target-Map.md" check_figure crispr-genome-offtarget-exact "304796751" "CRISPR-Genome-Off-Target-Map.md" check_figure crispr-genome-offtarget-exact "3099750718" "CRISPR-Genome-Off-Target-Map.md" check_figure crispr-genome-offtarget-exact "L28RZ5CC6K" "CRISPR-Genome-Off-Target-Map.md" check_figure crispr-genome-offtarget-exact "D8UQ4B2T7M" "CRISPR-Genome-Off-Target-Map.md" # --- Study 37: the corpus-collapse instrument --- check_figure corpus-distinct-count-exact "1c2e05a659917cd5de0d20446483b26372c757b48cec510070644e7307b7fd36" "Study-37-Validated-Discoveries-Five-Molecules.md" check_figure corpus-distinct-count-exact "95594154026f63552bf746f535be96392bca50385b4577c72d545c3d1de67aa1" "Study-37-Validated-Discoveries-Five-Molecules.md" check_figure corpus-distinct-count-exact "74db969f653be18214608230b4aaee30498f7fcd09df9e3d7b107ddb5c8f64f7" "Study-37-Validated-Discoveries-Five-Molecules.md" check_figure corpus-distinct-count-exact "0169f514a861ece42573fbff4d4db28e27880373adb7296a304aa9ddb860c1f4" "Study-37-Validated-Discoveries-Five-Molecules.md" check_figure corpus-distinct-count-exact "CONTROL ARM 13/13 PASS" "" # --- where humans actually yield: fatigue curves vs regulator limits --- check_figure fatigue-yield-vs-regulator "FATIGUE_YIELD_VS_REGULATOR__FOUR_AGREE_ONE_CONFLICT_ONE_NOT_COMPARABLE" "Where-Humans-Actually-Yield.md" check_figure fatigue-yield-vs-regulator "d5fab6529c34126003aec8d8ea2f80cb1558ccf9feb3b4896eaef739758206a1" "Where-Humans-Actually-Yield.md" # --- the ceiling on what a genotype-only score can know: exact arcsin bracket over 10^12 --- check_figure genotype-score-ceiling-exact "GENOTYPE_SCORE_CEILING__EXACT_ARCSIN_BRACKET_OVER_1E12" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "438bfa14cf0c4ab6e013af73173d78d16d8ecae2c78a438ff034482579733359" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "arms run = 25 failed = 0" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "pins disagreeing = 0" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.573212381656" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.602142835007" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.610622725745" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.527947611149" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.544353505508" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.538042338431" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.549890300690" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.541714427173" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.530300726943" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.598988711992" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.571783146564" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.599737251748" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.550601310553" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.558775112702" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.545063908314" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.561183362582" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.629000353248" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "0.606403923961" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "725/1000" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "719/1000" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "691/1000" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "627/1000" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "278/1000" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" check_figure genotype-score-ceiling-exact "1,105" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" # THE RUN-TERMINAL CONTRACT, PRODUCER SIDE. library-admission-law.swift reads a declared # terminal by STRUCTURE; these two programs are the first that print one. The harness demands # every refusal path quote its published figures, so a figure pin cannot tell quoted from # computed — only a run that finished prints COMPLETE, and it is pinned here for both. check_figure genotype-score-ceiling-exact "RUN_TERMINAL COMPLETE" "The-Ceiling-On-What-A-Genetic-Score-Can-Know.md" # --- WHAT A GENE'S LENGTH ALREADY DECIDES --- # Complete enumeration: every base of every catchment on every chromosome, and every gene # carrying a LOEUF. These rows were dropped by a concurrent commit — the program and its page # were never absent, only the pins, so the harness checked none of its figures. check_figure nearest-gene-length-lottery-exact "RUN_TERMINAL COMPLETE" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "GENE_LENGTH_INSTRUMENT__NEAREST_GENE_LOTTERY_AND_LOEUF_DENOMINATOR" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "c11ae542d7344b95abd87ee84d065954f322d0118d2609b236fe69ae1dda4489" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "arms run = 16 failed = 0" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "pins disagreeing = 0" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "19,704" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "19,197" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "66,591" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "243,881" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "114,478" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "2,070,253,229" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "1,261,815,650" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "1,628,025,115" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "3662/1000" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "1719/1000" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "3359/1000" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "4366/1000" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "4092/1000" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "3143/1000" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "1626000" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "484000" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "1799000" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "412000" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "1686000" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "-124,600" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "-391,700" "What-A-Gene-Length-Already-Decides.md" check_figure nearest-gene-length-lottery-exact "10 of 10" "What-A-Gene-Length-Already-Decides.md" # --- MARKET SURVEILLANCE: the detector that flags the whole market ----------------- # Un-numbered family page, added 2026-09-08. Every row below was checked with grep -F on # BOTH sides before it was written here — the program's own no-argv output and the page — # because a pin on a figure the program does not print is exactly the defect this harness # exists to catch, and it has bitten this repository before. # # THESE FIVE PROGRAMS PRINT THEIR REFERENCE FIGURES ON EVERY EXIT PATH, INCLUDING THE # NO-ARGV ONE THIS HARNESS TAKES. The corpora are 1.1 GB and fetched, not committed, so a # run here measures nothing; it prints what it WOULD have measured, says so on its own # face, and exits non-zero. ABSENCE and REFUSAL are different answers, and an # uninstrumented early exit is indistinguishable from a program that never built. # # live-wire-watch is the one that could have cost this harness ten minutes on the network: # its old default probed live endpoints and watched the chain head for 600 seconds. The # no-argv path is now the ZERO-NETWORK one — self-test arms, the in-process null # population, then the figures — and --all still does what the old default did. # the phantom-mass base rate, the equities composite, the ETH conjunct set, attribution check_figure market-shear-exact "12,676,036" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "12,156,283" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "10,164,658" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "2,921,796" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "2,732,598" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "88,900" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "63,140" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "1,407 per 1,000" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "7.3 per 1,000" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "RATIO_PUBLISHED_NO_THRESHOLD" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "10,580,123" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "NOT_COMPUTABLE" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "837,472,908" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "15,393" "The-Detector-That-Flags-The-Whole-Market.md" # MAR Annex I A(f) computed in FULL, and the legitimate quoter that outscores the session check_figure af-conjunct-exact "9,589 bp" "The-Detector-That-Flags-The-Whole-Market.md" check_figure af-conjunct-exact "5,384 bp" "The-Detector-That-Flags-The-Whole-Market.md" check_figure af-conjunct-exact "7,665 bp" "The-Detector-That-Flags-The-Whole-Market.md" check_figure af-conjunct-exact "9,189 bp" "The-Detector-That-Flags-The-Whole-Market.md" check_figure af-conjunct-exact "6,825,510" "The-Detector-That-Flags-The-Whole-Market.md" check_figure af-conjunct-exact "9,716,694" "The-Detector-That-Flags-The-Whole-Market.md" check_figure af-conjunct-exact "7,032 bp" "The-Detector-That-Flags-The-Whole-Market.md" check_figure af-conjunct-exact "85.14th percentile" "The-Detector-That-Flags-The-Whole-Market.md" check_figure af-conjunct-exact "REFUTED_CONTROL_SCORES_HIGHER" "The-Detector-That-Flags-The-Whole-Market.md" check_figure af-conjunct-exact "REFUSED_SINGLE_POPULATION" "The-Detector-That-Flags-The-Whole-Market.md" # --- Market shear: the base rate on a third venue, and a fourth that cannot carry it --- # Every string below was confirmed present in BOTH the program's own no-argv output and the # page, with grep -F on each, before it was added. A pin that is only in one of the two is a # pin that passes for the wrong reason. check_figure market-shear-exact "16,165,067" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "15,952,637" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "13,582,830" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "2,405,437" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "169,275" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "57,401" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "2,948 per 1,000" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "1,061 per 100,000" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "4,636,704" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-exact "THE PREDICATE CANNOT RUN. ABSENT, NOT ZERO." "The-Detector-That-Flags-The-Whole-Market.md" # --- Market shear: which feed carries the key the predicate needs --- check_figure feed-order-identity "ORDER_IDENTITY_PRESENT" "The-Detector-That-Flags-The-Whole-Market.md" check_figure feed-order-identity "ORDER_IDENTITY_ABSENT" "The-Detector-That-Flags-The-Whole-Market.md" check_figure feed-order-identity "CANNOT_RUN_NO_KEY" "The-Detector-That-Flags-The-Whole-Market.md" check_figure feed-order-identity "30629120" "The-Detector-That-Flags-The-Whole-Market.md" check_figure feed-order-identity "13081242" "The-Detector-That-Flags-The-Whole-Market.md" check_figure feed-order-identity "reference_deep_candidate_fields_swept" "The-Detector-That-Flags-The-Whole-Market.md" check_figure feed-order-identity "trade_report_carries_order_identity" "The-Detector-That-Flags-The-Whole-Market.md" check_figure feed-order-identity "reference_deep_plu_residual_bytes" "" # --- Market shear: what the 21 NOT_KNOWN rows needed, the sub-window split, and out of sample --- check_figure extraction-exact "states REJECTED because they did not reproduce the leg 9" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "floor, unchanged 28,889,398,990,674,697,077 wei" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "Q1 14000250-14000499 31 detections 8 pools busiest 580 permille" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "Q3 14000750-14000999 18 detections 15 pools busiest 111 permille" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "NOT_KNOWN_NOT_REPRODUCIBLE_AT_THE_POOLS_OWN_FEE" "The-Detector-That-Flags-The-Whole-Market.md" # the live wire, the self-contradiction guard, and the false-positive FLOOR check_figure live-wire-watch "68,121 per 10,000,000" "The-Detector-That-Flags-The-Whole-Market.md" check_figure live-wire-watch "2,208 per 10,000,000" "The-Detector-That-Flags-The-Whole-Market.md" check_figure live-wire-watch "53.95x" "The-Detector-That-Flags-The-Whole-Market.md" check_figure live-wire-watch "IRREDUCIBLE FLOOR" "The-Detector-That-Flags-The-Whole-Market.md" check_figure live-wire-watch "100,426,957 bytes" "The-Detector-That-Flags-The-Whole-Market.md" check_figure live-wire-watch "815 HTTP requests" "The-Detector-That-Flags-The-Whole-Market.md" check_figure live-wire-watch "8,119,826" "The-Detector-That-Flags-The-Whole-Market.md" check_figure live-wire-watch "4,766 swaps" "The-Detector-That-Flags-The-Whole-Market.md" check_figure live-wire-watch "1,427 leg pairs" "The-Detector-That-Flags-The-Whole-Market.md" # ABSENCE, REFUSAL, BOT_BLOCKED and NOT_KNOWN are four different answers on every wire # verdict. Two of them are pinned so the distinction cannot quietly collapse into one. check_figure live-wire-watch "NOT_KNOWN" "The-Detector-That-Flags-The-Whole-Market.md" check_figure live-wire-watch "BOT_BLOCKED" "The-Detector-That-Flags-The-Whole-Market.md" # the independently written kernel that reproduced the set member for member check_figure market-shear-rederive "200,826 · 276,014" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-rederive "13,272 · 3,245" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-rederive "22,287" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-rederive "28 · 26" "The-Detector-That-Flags-The-Whole-Market.md" # the naive geometry the conjunct set is measured AGAINST — the discrimination, not a rate check_figure market-shear-positional "262,799" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-positional "2,433x" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-positional "101 per 1,000 flagged" "The-Detector-That-Flags-The-Whole-Market.md" check_figure market-shear-positional "4.8 per 1,000" "The-Detector-That-Flags-The-Whole-Market.md" # --- WHAT WAS TAKEN: the extraction kernel, added 2026-09-08 -------------------------- # The study measured a capability and never answered "so what". This program is that # answer: per victim, per token, in integer base units, from the pool's own arithmetic. # Every row below was checked with grep -F on BOTH sides before it was written here. # # THESE RUN WITH NO CORPUS AND NO NETWORK. The 1.1 GB Ethereum corpus is fetched, never # committed, so the no-argv path measures NOTHING and says so — it prints the published # reference figures, labels them PUBLISHED_REFERENCES_NOT_THIS_RUNS_MEASUREMENTS, and # exits 4. A gate given nothing must not exit 0, and this one does not. check_figure extraction-exact "GEOMETRY_ONLY_DETECTION_IS_NOT_PROOF_OF_INTENT" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "KEYED_PSEUDONYM_8HEX" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "CORPUS_ABSENT" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "PUBLISHED_REFERENCES_NOT_THIS_RUNS_MEASUREMENTS" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "126 brackets · 108 extractive · 104 blocks · 26 extractive actors" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "108 of 108 SET-IDENTICAL" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "47 false positives per 212,769 leg pairs" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "87 of 108 EXACT · 21 NOT_KNOWN" "The-Detector-That-Flags-The-Whole-Market.md" # the per-token integer table IS the result; the single-unit lines are DERIVED and the # page carries that label on both. Pinning both halves stops the derived one drifting free # of the measured one. check_figure extraction-exact "155,576,958,801,814,594,396,893" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "5,074,012,211,888,792,743,310" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "28,924,625,003,219,287,345,953" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "1,689,797,703,211,127,422" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "28,889,398,990,674,697,077" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "28,247,424,339,991,594,322" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "94,645,772,620" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "3,276,141,973" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "min 47 · p25 102 · MEDIAN 476 · p75 1,785 · max 9,999" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "0.170457244547709297" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "558.442133" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "0.104516232525109603" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "+491 permille" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "-500 permille" "The-Detector-That-Flags-The-Whole-Market.md" # MEASURED and PROJECTED are two answers. Both are pinned so neither can quietly become # the other — the label lives on the program's own key name, not only in the page's prose. check_figure extraction-exact "108 detections per 1,000 blocks = 13,586 s · 28 per hour" "The-Detector-That-Flags-The-Whole-Market.md" check_figure extraction-exact "686 per day · 250,390 per year" "The-Detector-That-Flags-The-Whole-Market.md" # --- THE TOOL a person can actually run on their own transaction ---------------------- # wasi-sandwiched carries NO law of its own: section 2 above compiles it against a verbatim # byte slice of extraction-exact.swift, refusing outright if that slice boundary has moved. # Given no hash it prints the reference figures and exits 4 — the same discipline. check_figure wasi-sandwiched "WAS I SANDWICHED, AND WHAT DID IT COST ME" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-sandwiched "A GATE GIVEN NOTHING MUST NOT PASS" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-sandwiched "== NOTHING WAS GIVEN ==" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-sandwiched "ZERO — no API key, no account, no registration, nobody's permission" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-sandwiched "extraction-exact.swift lines 1..2843" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-sandwiched "126 brackets · 108 extractive · 104 blocks · 26 pseudonymous actors" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-sandwiched "median 476 ten-thousandths of the output that was due — about 0.17 ETH" "The-Detector-That-Flags-The-Whole-Market.md" # The compiled-in law digest, printed by the binary and computed at build time by shasum(1) # over the slice. It is on the page so a stranger can re-derive it in one command: # sed -n '1,2535p' reproduce/extraction-exact.swift | shasum -a 256 check_figure wasi-sandwiched "6b093daa07214dd589156235cc4fcb04b21de3750d6b76f008e35baf0c9e2799" "The-Detector-That-Flags-The-Whole-Market.md" # --- THE PRE-TRADE CHECK: the same law, asked before the signature instead of after ----- # wasi-exposure carries NO law of its own either. Section 2 above compiles it against the # SAME verbatim byte slice of extraction-exact.swift that wasi-sandwiched is built from. # # NEITHER THE SLICE DIGEST NOR ITS LINE RANGE IS PINNED HERE, deliberately. Both move when # the law legitimately grows — measured 2026-09-08, the detector went from 2,688 lines to # 3,011 in one afternoon — and a pin on a moving number goes red over a change that broke # nothing. That two tools carry ONE law is a relation between two binaries, and # wasi-exposure-one-law.sh measures it by reading the digest out of both. # # Given no pool it measures nothing, says so, prints the published reference figures and # exits 4 — so every row here comes off a no-argv, no-network, no-corpus path, which is # what makes a refusal auditable. Each was checked with grep -F on BOTH sides before it # was written here; nothing that only the network path prints is pinned, because a pin # that needs the wire is a pin a stranger cannot check. check_figure wasi-exposure "BEFORE YOU SIGN: HOW MUCH OF WHAT YOU ARE DUE CAN BE TAKEN?" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-exposure "affine.earth market-shear · one pool · one trade · one shape" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-exposure "A GATE GIVEN NOTHING MUST NOT PASS" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-exposure "== NOTHING WAS GIVEN ==" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-exposure "ZERO — no API key, no account, no registration, nobody's permission" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-exposure "Ethereum blocks 14,000,000–14,000,999 · 13,586 s of one chain" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-exposure "108 inserted trades · 87 costed EXACTLY · 21 NOT_KNOWN, named" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-exposure "MEDIAN 476 ten-thousandths of the output that was due" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-exposure "MEDIAN 0.170457244547709297 ETH ≈ 558 USDC per victim" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-exposure "28,889,398,990,674,697,077 wei = 28.8894 WETH = 94,645.77 USDC" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-exposure "min 47 · p10 49 · p25 102 · MEDIAN 476 · p75 1,785 · max 9,999" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-exposure "constant-product 73 victims median 575 · concentrated 14 median 105" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-exposure "pool-relative size Kendall tau +491 permille" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-exposure "48 distinct pools, ONE carries 22 of 108 (203 permille)" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-exposure "47 false positives per 212,769 leg pairs" "The-Detector-That-Flags-The-Whole-Market.md" check_figure wasi-exposure "COULD, NOT WILL" "The-Detector-That-Flags-The-Whole-Market.md" # --- the generated-peptide novelty screen (cures family) --- check_figure protein-novelty-exact "8c50b3e877d1dac7b68244464ae679fc43ed273d9fd38e7e348b823c3e80563b" "Generated-Peptides-Against-The-Human-Proteome.md" check_figure protein-novelty-exact "bf1bc7e188e55199a2447fc20d25834db5f7f798daa818432370deb4a6b0df5e" "Generated-Peptides-Against-The-Human-Proteome.md" check_figure protein-novelty-exact "bb3691b332fb15cdd54c43bc42905478e53c4f4b01862885a7304260498cf3f7" "Generated-Peptides-Against-The-Human-Proteome.md" check_figure protein-novelty-exact "24cdbf96621e6c38fa046c7a203fcc3ea09e31fad410d9c1cb51f6d192a04204" "Generated-Peptides-Against-The-Human-Proteome.md" # Added 2026-09-07 with the three new PROTEINS library entries (the Q14258 maximum, the # residual-overlap null, the composition signature). The library admission law requires a # check_figure row in the same commit as the entry, so page and program cannot drift apart. check_figure protein-novelty-exact "LETFLAKSRPEL" "Generated-Peptides-Against-The-Human-Proteome.md" check_figure protein-novelty-exact "Q14258" "Generated-Peptides-Against-The-Human-Proteome.md" check_figure protein-novelty-exact "249852560" "Generated-Peptides-Against-The-Human-Proteome.md" # Program-side pin only, because the page carrying it is not yet in this root. The third # argument becomes Library-Of-Proteins.md the moment that page lands, and the pin is then # two-sided like the rows above it. A row pinned to a page that does not exist is a red # harness, and a red harness nobody can fix is how a harness stops being read. check_figure protein-novelty-exact "198674 ppm = 19.8674%" "" # --- the oligonucleotide off-target atlas --- check_figure oligo-offtarget-atlas-exact "OLIGO_OFFTARGET_ATLAS_EXACT__COMPLETE_ENUMERATION_IS_OBSERVER_INVARIANT" "Oligonucleotide-Off-Target-Atlas.md" check_figure oligo-offtarget-atlas-exact "321b36c694b89a45bb81668d7ea62b9c85cf0b3087e18bba586f43b230274b08" "Oligonucleotide-Off-Target-Atlas.md" check_figure oligo-offtarget-atlas-exact "TGCTCCGTTGGTGCTTGTTC" "Oligonucleotide-Off-Target-Atlas.md" # --- the landing page's own headline figures --- check_figure seasonal-and-alternative "About fifty tonnes" "SpaceX-Biosphere-Safety.md" check_figure seasonal-and-alternative "477 tonnes a season" "SpaceX-Biosphere-Safety.md" check_figure seasonal-and-alternative "2,460 tonnes a season" "SpaceX-Biosphere-Safety.md" check_figure seasonal-and-alternative "20,000 tonnes a season" "SpaceX-Biosphere-Safety.md" check_figure seasonal-and-alternative "219 tonnes every single day" "SpaceX-Biosphere-Safety.md" check_figure seasonal-and-alternative "1,594,900" "SpaceX-Biosphere-Safety.md" check_figure seasonal-and-alternative "1,825,000" "SpaceX-Biosphere-Safety.md" check_figure seasonal-and-alternative "3.240 J" "SpaceX-Biosphere-Safety.md" check_figure response-envelope-projection "0.257" "" check_figure response-envelope-projection "1.324" "" check_figure response-envelope-projection "10.767" "" check_figure atmosphere-domain-entropy "461,213,509" "SpaceX-Biosphere-Safety.md" # --- closed system + cascade --- check_figure closed-system-box-model "5.4 to 10.9 times" "SpaceX-Biosphere-Safety.md" check_figure closed-system-box-model "13.0%" "" check_figure closed-system-box-model "26.1%" "" check_figure mass-uncertainty-band "75.6%" "" check_figure mass-uncertainty-band "391 to 531" "SpaceX-Biosphere-Safety.md" check_figure mass-uncertainty-band "45 to 61 tonnes" "SpaceX-Biosphere-Safety.md" check_figure biosphere-cascade-chain "128.4%" "Study-31-Biosphere-Cascade.md" check_figure percolation-refutation "1.1 x 10^-15" "" check_figure tsat-control-arm "1000000 to 10000000" "" check_figure taxiout-floor-court "20,405 hours" "Study-32-Taxi-Out-Floor-Court.md" check_figure taxiout-floor-court "3.6 minutes" "" check_figure taxiout-floor-court "8.8 min" "Study-32-Taxi-Out-Floor-Court.md" check_figure taxiout-floor-court "9794 to 18365" "" check_figure guadalupe-wave-ledger "180 minutes" "The-Replacement-Grade.md" check_figure rate-of-rise-common-window "885 milli-ft/min" "" check_figure cost-ownership-horizon "year 5" "The-Replacement-Grade.md" check_figure cost-ownership-horizon "4,784,700" "The-Replacement-Grade.md" check_figure cost-ownership-horizon "2,160,000" "The-Replacement-Grade.md" check_figure cost-ownership-horizon "NOT BROADBAND" "" check_figure fusion-verdict-stream "FIRST MITIGATE at sample 7016" "" check_figure fusion-verdict-figure "warning lead 140 us" "" check_figure fusion-verdict-figure "140 µs" "Study-33-Fusion-Control-Verdict-Court.md" check_figure fusion-operating-court "COURT TERMINALS REACHED: 5 of 5" "" check_figure fusion-operating-court "NOT_APPLICABLE_NO_PLASMA_CURRENT" "" check_figure fusion-topology-agnostic "all three signatures byte-identical: true" "" check_figure fusion-topology-agnostic "spheromak toroidal-closure edges == 0: true" "" check_figure fusion-real-machines "REAL-MACHINE GEOMETRY GRADED: 4 WIN / 4 MISS-greenwald / 1 NOT_APPLICABLE" "" check_figure fusion-real-machines "1193661" "" check_figure fusion-determinism-digest "f49b576e073835bcab17bee10fe0eee1938774643d900b8ffe1a583b159ab3d7" "" check_figure fusion-exact-vs-float "PROOF_EXACT_VERDICT_IS_OBSERVER_INVARIANT" "" check_figure fusion-exact-vs-float "exact-refused points = 142 · float32 flips vs exact = 0 · two-pi float contradictions = 142" "" check_figure fusion-exact-vs-float "355/113 - 333/106 = 1/11978" "" # the 142 refused and the 142 contradictory are the SAME 142 points (counted 2026-10-03) check_figure fusion-exact-vs-float "same_points=142" "" check_figure fusion-affine-density-invariant "AFFINE_DENSITY_INVARIANT_IS_PI_FREE" "" check_figure fusion-affine-density-invariant "pi-ambiguous by 334 mm^2 across the bracket" "" check_figure fusion-affine-magnitude-invariance "AFFINE_INVARIANT_CARRIES_MEANING_AT_ANY_MAGNITUDE" "" check_figure fusion-affine-magnitude-invariance "float32 goes blind at 2^24 = 16777216" "" check_figure fusion-affine-magnitude-invariance "float64 goes blind at 2^53 = 9007199254740992" "" check_figure fusion-control-benchmark "VERDICT_DETERMINISTIC_10K TRUE" "Study-33-Fusion-Control-Verdict-Court.md" check_figure fusion-control-benchmark "VERDICT_RENDERED_AT_INDEX_208 TRUE" "Study-33-Fusion-Control-Verdict-Court.md" check_figure fusion-control-benchmark "VERDICT_DETERMINISTIC_10K TRUE" "" check_figure fusion-control-exact-law "5 of 5 arms hold" "Study-33-Fusion-Control-Verdict-Court.md" check_figure fusion-control-exact-law "REFUSED_OUT_OF_ENVELOPE" "Study-33-Fusion-Control-Verdict-Court.md" check_figure fusion-control-exact-law "idx=208 peak=960" "" check_figure fusion-control-verdict-court "FIVE REAL-WORLD EXPERIMENTS" "" check_figure fusion-control-verdict-court "STUDY33_FUSION_CONTROL_VERDICT_PENDING" "Study-33-Fusion-Control-Verdict-Court.md" check_figure fusion-control-verdict-court "beta_normalized=1.8" "" # --- Study 35: the safety brain that forgets (time axis) --- check_figure float-degradation-demo "8.4 s" "Study-35-The-Safety-Brain-That-Forgets.md" check_figure float-degradation-demo "99.8%" "Study-35-The-Safety-Brain-That-Forgets.md" check_figure float-degradation-demo "FLOAT_STATE_DEGRADES_OVER_TIME_INVARIANT_DOES_NOT" "Study-35-The-Safety-Brain-That-Forgets.md" check_figure drift-barrier-demo "96.5%" "Study-35-The-Safety-Brain-That-Forgets.md" check_figure drift-barrier-demo "49.6%" "Study-35-The-Safety-Brain-That-Forgets.md" check_figure drift-barrier-demo "50.3%" "Study-35-The-Safety-Brain-That-Forgets.md" check_figure drift-barrier-demo "DRIFT_BARRIER_TRAINED_MODEL_STALE_INVARIANT_FIXED" "Study-35-The-Safety-Brain-That-Forgets.md" # --- Zilganersen / Alexander disease: the exact ASO off-target screen --- check_figure zilganersen-offtarget-whole-transcriptome "ZILGANERSEN_OFFTARGET_EXACT__REAL_SEQUENCE_WITH_COMPOSITION_CONTROL" "The-Safety-Question-Made-Exact.md" check_figure zilganersen-offtarget-whole-transcriptome "edcb277ddea44820502b6446b00ed8bdcfdb08835d6785fdee7d1b370420bbaa" "The-Safety-Question-Made-Exact.md" check_figure zilganersen-offtarget-whole-transcriptome "CAGTATTACCTCTACTAGTC" "The-Safety-Question-Made-Exact.md" check_figure zilganersen-offtarget-whole-transcriptome "1467336203" "The-Safety-Question-Made-Exact.md" check_figure zilganersen-offtarget-whole-transcriptome "670670" "The-Safety-Question-Made-Exact.md" check_figure zilganersen-offtarget-whole-transcriptome "5a320f524d73b5793518eb19b118829033713443d0f42af20a67bb31cc06cf56" "The-Safety-Question-Made-Exact.md" # THE ORDER OF THE BASES — the registry-wide specificity ranking and its independent verifier. # Both programs print the published figures on EVERY exit path, refusal included, so these pins # are checked from a clean clone with no corpus present and no 1.5 GB download. The 5.36-trillion # window screen itself is NOT run here; the harness checks that the page and the program agree # on what that screen measured, which is the property a reader needs and the one that can rot. check_figure registry-specificity-ranking "REGISTRY_SPECIFICITY_RANKING__ORDER_NOT_COMPOSITION_SETS_THE_BURDEN" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "5135ebb89ca659c6cce26d749dfc1547c08c4e6fc959074b6b3ab67fc9862afb" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "5a320f524d73b5793518eb19b118829033713443d0f42af20a67bb31cc06cf56" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "e8a4711ff3d52ffae59c8c36a3dd6477307e9c4e6562ead6039302a0056c9e98" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "2d74f5c676d51d45df178f9fed7840729bd87633ae8e5a9104383f6fbd7c3fd1" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "717027798090 + 4638850669668 = 5355878467758" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "670670 transcripts, 79139 genes, 1480179158 bases" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "1467336203" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "489 in sweep A, 3162 in sweep B" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "9 pre-corpus + 8 corpus-dependent = 17" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "169 registry strands + 17 undesigned constructed 20-mers" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "18 UBIQUITOUS, 266 REFUSED, 19 NOT_KNOWN" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "BELOW-all-16 2 | ties-lowest 8 | inside 122 | ties-highest 1 |" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "ABOVE-all-16 44 | ties-all-16 9" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "0 BELOW-all-16, 23 ABOVE-all-16" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "families whose rankLo is 1 (ties favour the strand) : 19" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "families STRICTLY below all 16 controls (rank exactly 1-1): 2" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "9 of them tied ALL 16" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "registry strands in the table : 472" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "undesigned constructed 20-mers, NOT medicines: 17" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "SCREENED (target measured in few genes) : 169" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "UBIQUITOUS (perfect complement in hundreds of genes): 18" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "REFUSED (no perfect complement anywhere): 266" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "NOT_KNOWN (too short for a target to mean): 19" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "ZILGANERSEN off=324 ctrl median=787 range 141-1355 rank 3 of 17" "The-Order-Of-The-Bases.md" check_figure verify-one-family "ATAACTCTGTCCATTACCG" "The-Order-Of-The-Bases.md" check_figure verify-one-family "1468006855" "The-Order-Of-The-Bases.md" check_figure verify-one-family "991 1002 1007 1026 1051 1053 1150 1250" "The-Order-Of-The-Bases.md" check_figure verify-one-family "CTCCAACATCAAGGAAGATGGCATTTCTAG" "The-Order-Of-The-Bases.md" check_figure verify-one-family "5a320f524d73b5793518eb19b118829033713443d0f42af20a67bb31cc06cf56" "The-Order-Of-The-Bases.md" check_figure verify-one-family "ties-all-16" "The-Order-Of-The-Bases.md" check_figure verify-one-family "BELOW-all-16" "The-Order-Of-The-Bases.md" check_figure aso-offtarget-exact-vs-float "ASO_OFFTARGET_EXACT_IS_OBSERVER_INVARIANT" "The-Safety-Question-Made-Exact.md" check_figure crispr-guide-offtarget-exact-vs-float "CRISPR_OFFTARGET_EXACT_IS_OBSERVER_INVARIANT" "PM359-Prime-Editing-Certified-Before-Anyone-Is-Dosed.md" check_figure flt-nearmiss-fractal-shear "FLT_NEARMISS_FRACTAL_SHEAR__EXACT_IS_OBSERVER_INVARIANT" "Study-36-The-Language-Game-of-Fermats-Last-Theorem.md" check_figure flt-nearmiss-fractal-shear "1bba2839c16677070a986d49eb978dcd8a822c7dee30c769500d67544e861998" "Study-36-The-Language-Game-of-Fermats-Last-Theorem.md" check_figure flt-nearmiss-fractal-shear "700212234530608691501223040959" "Study-36-The-Language-Game-of-Fermats-Last-Theorem.md" check_figure valuation-crossing-ledger "2036" "The-Replacement-Grade.md" check_figure valuation-crossing-ledger "2038" "The-Replacement-Grade.md" check_figure valuation-crossing-ledger "973.7" "The-Replacement-Grade.md" check_figure pod-mesh-planetary "129:1" "The-Replacement-Grade.md" check_figure pod-mesh-planetary "35:1" "The-Replacement-Grade.md" check_figure pod-mesh-planetary "59:1" "The-Replacement-Grade.md" check_figure replacement-grade-ledger "domains declared" "" check_figure replacement-grade-ledger "49" "The-Replacement-Grade.md" check_figure replacement-grade-ledger "live PROVEN marker 13" "" check_figure replacement-grade-ledger "identity to 1/1 9" "" check_figure operator-reentry-ledger "187.3" "Impact-Study-SpaceX-Biosphere-Forcing.md" check_figure operator-reentry-ledger "0.89x" "" check_figure operator-reentry-ledger "45.8%" "Impact-Study-SpaceX-Biosphere-Forcing.md" check_figure ozone-baseline-and-state-change "CONTROL ARM PASSES" "Study-31-Biosphere-Cascade.md" check_figure radiative-baseline "3,539 mW/m2" "" check_figure radiative-baseline "43.7%" "Study-31-Biosphere-Cascade.md" check_figure radiative-baseline "1,547.3" "Study-31-Biosphere-Cascade.md" check_figure closed-system-joint-ledger "not an estimate" "Study-31-Biosphere-Cascade.md" check_figure ozone-baseline-and-state-change "-16.1%" "Study-31-Biosphere-Cascade.md" check_figure ozone-baseline-and-state-change "+8.3%" "Study-31-Biosphere-Cascade.md" check_figure ozone-baseline-and-state-change "261.7 DU" "Study-31-Biosphere-Cascade.md" check_figure ozone-baseline-and-state-change "21618" "" check_figure tsat-control-arm "REFUSED" "Study-31-Biosphere-Cascade.md" check_figure percolation-refutation "0.29" "Study-31-Biosphere-Cascade.md" check_figure biosphere-cascade-chain "15.6%" "Study-31-Biosphere-Cascade.md" check_figure biosphere-cascade-chain "+17.1% to +31.2%" "Study-31-Biosphere-Cascade.md" # --- Study 40: indefinite causal order — their experiment run in whole numbers ----------------- # The exact arm is unbounded decimal-string integers (no Int128: the build host has it and the # cells do not, and one law in two types is two laws). Every figure below is also re-derived by an # independent arbitrary-precision implementation, 75 of 75, 0 divergences. check_figure ico-causal-order-shear "ORDER_IS_AN_ARTEFACT_OF_THE_ARITHMETIC" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "7e5d40d56faaa936877a3d6ad9707637106426c4785258079be9e29ece2eb40f" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "9 / 0 of 9 settings" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "7 of 22, first at 1 - 10^-16" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "4 / 1 of 12 loops" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "9 / 1 / 1" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "18 of 18" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "1 / 6" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "5/18" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "2445399/4925449" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "333267/834917" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "16336650/166148483" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "2504751/3999701" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "13494202421495/934495065504" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "1/10000001000000021" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "49999999999999995000000000000000/4999999999999999250000000000000044999999999999999" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "10^-4 / 10^-8 / 10^-16, exact NONE" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "9999999999999999/19999999999999999" "Study-40-Indefinite-Causal-Order.md" check_figure ico-causal-order-shear "10000000000000000/19999999999999999" "Study-40-Indefinite-Causal-Order.md" # CONTROL: the precision ladder must show every width failing and the exact arm never. check_figure ico-causal-order-shear "exact integers unbounded NONE" "" check_figure ico-causal-order-shear "the double holds ONE number for both" "" # CONTROL ON THE CONTROLS. Every arm must be shown to FIRE, in both directions where it has two. # A run in which the float agreed everywhere, or the switch moved nothing, would print none of # these lines and the study's claims would be unsupported rather than merely quiet. check_figure ico-causal-order-shear "first scale at which float is fooled 10^4" "" check_figure ico-causal-order-shear "first scale at which float goes blind 10^8" "" check_figure ico-causal-order-shear "CURVATURE INVENTED" "" check_figure ico-causal-order-shear "CURVATURE ERASED" "" check_figure ico-causal-order-shear "DOUBLE SAYS NO EFFECT" "" check_figure ico-causal-order-shear "z = 1 -> switch moved 0 CONTROL PASSES" "" check_figure ico-causal-order-shear "rungs where a DEFINITE order moved anything 0" "" check_figure ico-causal-order-shear "distinct SEQUENCES across the nine cuts 1" "" # --- THE THREE LIBRARIES: one check_figure row per admitted entry, in the same commit. # The admission law's own "what must accompany an addition" requires this row and the entry # together, so a page and the program behind it cannot drift apart between commits. The sixth # PROTEINS entry — homology under substitution — has NO row here on purpose: it is HELD, its # program's output in this harness is a refusal, and a row pinned to a figure that program does # not print here would be a red harness rather than a measurement. check_figure protein-novelty-exact "bb3691b332fb15cdd54c43bc42905478e53c4f4b01862885a7304260498cf3f7" "Library-Of-Proteins.md" check_figure protein-novelty-exact "second rows 1400 residues 78694 lengths 20 to 100 labels 12" "Library-Of-Proteins.md" check_figure protein-novelty-exact "LETFLAKSRPEL" "Library-Of-Proteins.md" check_figure protein-novelty-exact "corpus K+R 1026307 of 5165782 = 198674 ppm = 19.8674%" "Library-Of-Proteins.md" check_figure protein-novelty-exact "aligned-pair match probability = 3266100739639 / 58984123166334 = 55372 ppm" "Library-Of-Proteins.md" check_figure zilganersen-offtarget-whole-transcriptome "edcb277ddea44820502b6446b00ed8bdcfdb08835d6785fdee7d1b370420bbaa" "Library-Of-Compound-Cures.md" check_figure pelacarsen-offtarget-whole-transcriptome "513de7e9db6556df1895bfce4cb4d69e4816d7b45b75bee1dc8452335c2c7757" "Library-Of-Compound-Cures.md" check_figure crispr-genome-offtarget-exact "487b4f81de2d24bd0bb11ecd1d8d42778e3a5d91b9edb33627c86dcc8df34980" "Library-Of-Compound-Cures.md" check_figure mr-topology-vs-expression-exact "TOPOLOGY_EXPLAINS: 11 of 17 tumour types" "Library-Of-Compound-Cures.md" check_figure oligo-offtarget-atlas-exact "321b36c694b89a45bb81668d7ea62b9c85cf0b3087e18bba586f43b230274b08" "Library-Of-Compound-Cures.md" check_figure z8-vs-e8-lattice "E8 : 240" "Library-Of-Material-Systems.md" check_figure z8-vs-e8-lattice "minimal vectors of L, counted : 240" "Library-Of-Material-Systems.md" check_figure z8-vs-e8-lattice "lifts of the 8 weight-3 codewords (3^3 each): 216" "Library-Of-Material-Systems.md" check_figure z8-vs-e8-lattice "27 of 81 words lift without remainder" "Library-Of-Material-Systems.md" check_figure z8-vs-e8-lattice "arms run 10, broken 0" "" check_figure z8-vs-e8-lattice "811c21506b141b7f509f96e66cea21f036192753a9f3b3504386890b7aae9541" "Library-Of-Material-Systems.md" check_figure z8-vs-e8-lattice "RUN_TERMINAL COMPLETE" "" check_figure lora-time-on-air "287.744" "Library-Of-Material-Systems.md" check_figure fusion-determinism-digest "f49b576e073835bcab17bee10fe0eee1938774643d900b8ffe1a583b159ab3d7" "Library-Of-Material-Systems.md" check_figure unimodular-control-arms "unimodular det=1: det=1 e_1 reachable in sample = true" "Library-Of-Material-Systems.md" check_figure fusion-exact-vs-float "PI_BRACKET 355/113 - 333/106 = 1/11978" "Library-Of-Material-Systems.md" # --- the registry specificity ranking (The Order Of The Bases) ------------------------------- # These programs are run by section 2 with NO argv and stdin from /dev/null, so they REFUSE and # print their published reference block on the way out. Every figure pinned below therefore comes # off a refusal path, which is exactly the property that makes a refusal auditable. check_figure registry-specificity-ranking "REGISTRY_SPECIFICITY_RANKING__ORDER_NOT_COMPOSITION_SETS_THE_BURDEN" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "5a320f524d73b5793518eb19b118829033713443d0f42af20a67bb31cc06cf56" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "5135ebb89ca659c6cce26d749dfc1547c08c4e6fc959074b6b3ab67fc9862afb" "The-Order-Of-The-Bases.md" check_figure registry-specificity-ranking "RANK 1 MEANS rankLo == rankHi == 1" "" check_figure registry-specificity-ranking "324 off-target windows at 4 mismatches" "" check_figure verify-one-family "INDEPENDENT_VERIFIER_ONE_FAMILY" "" check_figure verify-one-family "991 1002 1007 1026 1051 1053 1150 1250" "The-Order-Of-The-Bases.md" check_figure verify-one-family "ties-all-16, kRes 7, and AT kRes it is ABOVE all 16 (51 vs 0-20)" "" check_figure verify-one-family "RANK 1 MEANS burden < min(controls). A tie is not fewer places." "" # --- Study 41: the silicon cost of imposing an order (nine live cells) --- # The program needs Linux performance counters, so on this host it takes its REFUSAL path — # which prints the same reference block as a full run. That is the point: a study whose figures # vanish when the instrument is absent could not be pinned at all. check_figure silicon-shear-telemetry "ORDERING_IS_PRICED_IN_CYCLES_NOT_IN_PHYSICS" "Study-41-What-The-Ordering-Cost.md" check_figure silicon-shear-telemetry "22,846 against 4,053,186" "Study-41-What-The-Ordering-Cost.md" check_figure silicon-shear-telemetry "42-51 against 110,319-111,002" "Study-41-What-The-Ordering-Cost.md" check_figure silicon-shear-telemetry "6 of 6 rungs on 9 of 9 cells" "Study-41-What-The-Ordering-Cost.md" check_figure silicon-shear-telemetry "4 of 5 rungs" "Study-41-What-The-Ordering-Cost.md" check_figure silicon-shear-telemetry "1,499,438-1,500,676 of 3,000,000" "Study-41-What-The-Ordering-Cost.md" check_figure silicon-shear-telemetry "-45 to +938 cycles" "Study-41-What-The-Ordering-Cost.md" check_figure silicon-shear-telemetry "ABSENT is not ZERO and is not PASS" "" # --- STUDY 43 — the cross-venue ledger. Nine venue-sessions, three exchange operators, two # --- continents, two rulebooks, 2003 to 2026. market-shear-cross-venue.swift measures no # --- market: it carries the counts market-shear-exact.swift printed on each session, names the # --- file each was read out of, and RE-DERIVES every published rate from those counts in whole # --- numbers. It takes no argv, reads no file and applies no threshold, so it has exactly one # --- exit path and prints the complete ledger before taking it — including when it refuses. # --- Proven able to fail in both directions before these rows were written: move one published # --- rate by 1 and it exits 1 naming the row; break one closure identity and it does the same. PAGE43=Study-43-Almost-Every-Order-Is-Cancelled.md check_figure market-shear-cross-venue "12,676,036" "$PAGE43" check_figure market-shear-cross-venue "12,156,283" "$PAGE43" check_figure market-shear-cross-venue "2,921,796" "$PAGE43" check_figure market-shear-cross-venue "2,732,598" "$PAGE43" check_figure market-shear-cross-venue "16,165,067" "$PAGE43" check_figure market-shear-cross-venue "15,952,637" "$PAGE43" check_figure market-shear-cross-venue "53,909" "$PAGE43" check_figure market-shear-cross-venue "253,809" "$PAGE43" check_figure market-shear-cross-venue "251,097" "$PAGE43" check_figure market-shear-cross-venue "121,297" "$PAGE43" check_figure market-shear-cross-venue "120,331" "$PAGE43" check_figure market-shear-cross-venue "82,132" "$PAGE43" check_figure market-shear-cross-venue "81,648" "$PAGE43" check_figure market-shear-cross-venue "108,421" "$PAGE43" check_figure market-shear-cross-venue "108,137" "$PAGE43" check_figure market-shear-cross-venue "130,683" "$PAGE43" check_figure market-shear-cross-venue "130,451" "$PAGE43" check_figure market-shear-cross-venue "287,986" "$PAGE43" # the non-degenerate composite contrast — never the 0-against-0 cell, whose control is also 0 check_figure market-shear-cross-venue "88,900" "$PAGE43" check_figure market-shear-cross-venue "63,140" "$PAGE43" check_figure market-shear-cross-venue "7,451" "$PAGE43" check_figure market-shear-cross-venue "6,491" "$PAGE43" check_figure market-shear-cross-venue "3,157" "$PAGE43" check_figure market-shear-cross-venue "3,014" "$PAGE43" check_figure market-shear-cross-venue "1,970" "$PAGE43" # the BX/PSX contrast is CONFOUNDED and the confounder is a count, so it is pinned as one check_figure market-shear-cross-venue "4,636,704" "$PAGE43" # compute cost, stated as an envelope over a set of builds and sized on the slowest run check_figure market-shear-cross-venue "46.568 s" "$PAGE43" check_figure market-shear-cross-venue "328.043 s" "$PAGE43" check_figure market-shear-cross-venue "11.4 microseconds" "$PAGE43" check_figure market-shear-cross-venue "13.586 seconds" "$PAGE43" check_figure market-shear-cross-venue "28,734,686" "$PAGE43" check_figure market-shear-cross-venue "63,447" "$PAGE43" # the blockchain arm — already published, cited not re-measured check_figure market-shear-cross-venue "22,287" "$PAGE43" check_figure market-shear-cross-venue "4.8 per 1,000" "$PAGE43" # the three feeds that cannot carry the check, each failing a DIFFERENT way check_figure market-shear-cross-venue "13,081,242" "$PAGE43" check_figure market-shear-cross-venue "344,258" "$PAGE43" check_figure market-shear-cross-venue "772,868" "$PAGE43" check_figure market-shear-cross-venue "536,870,912" "$PAGE43" check_figure market-shear-cross-venue "68123fc57aeef19f331acd74803b4c28" "$PAGE43" check_figure market-shear-cross-venue "EMPTY AT SOURCE" "$PAGE43" # scope that must survive into the prose, and did check_figure market-shear-cross-venue "CTEST" "$PAGE43" check_figure market-shear-cross-venue "NOT MEASURED" "$PAGE43" check_figure market-shear-cross-venue "11 of 25" "$PAGE43" check_figure market-shear-cross-venue "8,407" "$PAGE43" check_figure market-shear-cross-venue "100,337" "$PAGE43" check_figure market-shear-cross-venue "136,091" "$PAGE43" check_figure market-shear-cross-venue "5,356" "$PAGE43" # no threshold anywhere, and the ledger's own verdict marker check_figure market-shear-cross-venue "RATIO_PUBLISHED_NO_THRESHOLD" "$PAGE43" check_figure market-shear-cross-venue "LEDGER_CLOSES_EXACT" "$PAGE43" # STUDY 44 — the Atlas container. Every figure is an integer count of bit patterns, # and the program greps the container sentence out of the proto rather than quoting it. check_figure atlas-container-pigeonhole-exact "ATLAS_CONTAINER_CANNOT_DISTINGUISH_ITS_OWN_ROWS" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "642d84b418825906d7bc2ec2933f2c8c50b61f3f74f11445d7c89d44ed62e631" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "037e8ca50171582db7bf63780e87cb37d8dfeb2c078573412bdd71c0d69f1ed9" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "4,294,967,296" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "16,777,214" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "4,278,190,082" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "9,000,000,000" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "9,299,252,154" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "3,099,750,718" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "4,721,809,918" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "1,065,353,216" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "7,934,646,784" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "4,607,182,418,800,017,408" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "1,000 variants in 4,278,190,082 slots forces 0 collisions" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "9,000,000,000 variants in 4,278,190,082 slots forces 4,721,809,918" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "arms: 7 run, 7 passed, 0 failed" "Study-44-The-Atlas-Container.md" check_figure atlas-container-pigeonhole-exact "Values are stored as single" "Study-44-The-Atlas-Container.md" # --------------------------------------------------------------------------------------- # STUDY 45 — which of nine billion answers a laboratory can act on. # # Arm A core needs no input and runs to COMPLETE under this harness, so its figures are # MEASURED on the grading run. Arm A genome and Arm B both need inputs this harness does not # supply — a 4.7 GB annotation and a key-gated live pull — so under it they always take a # refusal path. Each of those refusal paths prints the study's published figures, labelled # QUOTED, which is why these pins resolve at all. That is deliberate: a figure on the page # that no program prints is not reproducible, and this harness is what says so. PAGE45="Study-45-Which-Answers-A-Laboratory-Can-Act-On.md" # -- Arm A core: the genetic code, enumerated. No inputs, measured on this run. check_figure codon-consequence-exact "138 239 per 1000" "$PAGE45" check_figure codon-consequence-exact "392 680 per 1000" "$PAGE45" check_figure codon-consequence-exact "23 39 per 1000" "$PAGE45" check_figure codon-consequence-exact "184 of 576 319 per 1000" "$PAGE45" check_figure codon-consequence-exact "arms: 7 run, 7 passed, 0 failed" "" # -- Arm A genome: the coding footprint, from GENCODE v50 canonical CDS. check_figure coding-consequence-genome-exact "20,107" "$PAGE45" check_figure coding-consequence-genome-exact "197,573" "$PAGE45" check_figure coding-consequence-genome-exact "33,722,363" "$PAGE45" check_figure coding-consequence-genome-exact "101,167,089" "$PAGE45" check_figure coding-consequence-genome-exact "9,299,252,154" "$PAGE45" check_figure coding-consequence-genome-exact "1087" "$PAGE45" check_figure coding-consequence-genome-exact "103,076,262" "$PAGE45" check_figure coding-consequence-genome-exact "23,660,731 229 per 1000" "$PAGE45" check_figure coding-consequence-genome-exact "75,158,434 729 per 1000" "$PAGE45" check_figure coding-consequence-genome-exact "4,246,822 41 per 1000" "$PAGE45" check_figure coding-consequence-genome-exact "10,275 0 per 1000" "$PAGE45" check_figure coding-consequence-genome-exact "27,917,828 270 per 1000" "$PAGE45" check_figure coding-consequence-genome-exact "25b068b63a9b656aea78fdc2f2f290091ad451b5d1276b1c8bae4bf69259f111" "$PAGE45" check_figure coding-consequence-genome-exact "CODON_TABLE_SPEAKS_TO_ONLY_THE_CODING_FRACTION" "" # -- Arm B: collisions measured on the live artifact, both loci. check_figure atlas-collision-measure-exact "ATLAS_COLLISIONS_MEASURED_ON_THE_PUBLISHED_ARTIFACT" "" check_figure atlas-collision-measure-exact "arms: 13 run, 13 passed, 0 failed" "" check_figure atlas-collision-measure-exact "4,278,190,082" "$PAGE45" check_figure atlas-collision-measure-exact "4,721,809,918" "$PAGE45" check_figure atlas-collision-measure-exact "1600aa88aaa9cc3f4633b38d148984195b80d191303a6a1ec1be24e9d2fdcba0" "$PAGE45" check_figure atlas-collision-measure-exact "45054503330b28b151432e6b27170112aaded210311998e9aef263885ea1fb9a" "$PAGE45" # the published sentences, verbatim in both the program and the page check_figure atlas-collision-measure-exact "600 different values to 600 variants" "$PAGE45" check_figure atlas-collision-measure-exact "950 of every 1,000" "$PAGE45" check_figure atlas-collision-measure-exact "7 of 7" "$PAGE45" check_figure atlas-collision-measure-exact "145 of the 600 variants carry an identical pair of splice-site values" "$PAGE45" check_figure atlas-collision-measure-exact "574" "$PAGE45" check_figure atlas-collision-measure-exact "identical 167-value ATAC_ACTIVE vector" "$PAGE45" check_figure atlas-collision-measure-exact "POLYADENYLATION returns ZERO values for all 600 variants at CFTR" "" # the per-scorer rates the page tabulates check_figure atlas-collision-measure-exact "CHIP_TF_ACTIVE 1,617 930 968" "" check_figure atlas-collision-measure-exact "SPLICE_SITE_USAGE 367 950 998" "" check_figure atlas-collision-measure-exact "AVI_SCORE 1 0 0" "" # the exact near-zero band edge check_figure atlas-collision-measure-exact "2^-4 = 0.0625" "$PAGE45" # --- Study 48 — the atom already has an address (1 nm hydrogen depassivation lithography) --- P48="Study-48-The-Atom-Already-Has-An-Address.md" check_figure hdl-site-address-exact-vs-float "2,097,152 writable sites" "$P48" check_figure hdl-site-address-exact-vs-float "mis-addresses its first hydrogen site at step 8,783" "$P48" check_figure hdl-site-address-exact-vs-float "float32 accumulating mis-addresses 4,183,204 of 4,194,304 sites; float64 accumulating mis-addresses 0; float32 recomputed from the index mis-addresses 0" "$P48" check_figure hdl-site-address-exact-vs-float "the exact arm mis-addresses 0 sites in 4,194,304 steps" "$P48" check_figure hdl-site-address-exact-vs-float "one pitch added to 7.81 mm of carried travel changes nothing" "$P48" check_figure hdl-site-address-exact-vs-float "arms run 7, broken 0" "" check_figure hdl-site-address-exact-vs-float "every dimer-centre separation is the integer Δn² + 4Δm²" "$P48" check_figure hdl-site-address-exact-vs-float "the dimer bond runs across the row, not along it" "$P48" check_figure hdl-site-address-exact-vs-float "first mis-addresses at step 5,086,264" "$P48" check_figure hdl-site-address-exact-vs-float "HDL_SITE_ADDRESS__THE_LATTICE_IS_ITS_OWN_RULER" "$P48" check_figure hdl-site-address-exact-vs-float "4b299340a10e04822aac19294abdbcdb1f05b6387e0ff63e0635c37cf3e5b881" "$P48" # --- Study 49 — the phase code never needs pi (phase-only SLM wavefront synthesis) --- P49="Study-49-The-Phase-Code-Never-Needs-Pi.md" check_figure slm-phase-code-exact-vs-float "16,588,800 bits of command word" "$P49" check_figure slm-phase-code-exact-vs-float "float64 hands the device 47 different codes in 1,920 pixels, and float32 hands it 45" "$P49" check_figure slm-phase-code-exact-vs-float "float64 hands the device 16 different codes in 2,073,600 pixels, and float32 hands it 1,974" "$P49" check_figure slm-phase-code-exact-vs-float "47 of the 47 sit EXACTLY on a" "$P49" check_figure slm-phase-code-exact-vs-float "21 pixels common to both arms, 26 float64 only, 24 float32 only" "$P49" check_figure slm-phase-code-exact-vs-float "4 of 1,920 pixels take a different code depending on which end the sum started" "$P49" check_figure slm-phase-code-exact-vs-float "30.503 levels" "$P49" check_figure slm-phase-code-exact-vs-float "arms run 13, broken 0" "" check_figure slm-phase-code-exact-vs-float "drops only from 1,974 to 970" "$P49" check_figure slm-phase-code-exact-vs-float "SLM_PHASE_CODE__THE_CODE_IS_A_RATIO_OF_INTEGERS" "$P49" check_figure slm-phase-code-exact-vs-float "ec768ab33eb7d8a771afe773d267d03a8fc28dd99386d6345921a622052a071f" "$P49" echo "=== 3b. the admission law grades the three libraries in ONE run ===" # THE THREE LIBRARIES ARE GRADED TOGETHER, NEVER ONE AT A TIME. F1 — no entry filed in two # libraries — is a relation BETWEEN libraries, and a run given one library reports it NOT_KNOWN # and exits 2. Three libraries each graded alone, each reporting clean, is three runs none of # which asked the question; that is exactly how the first published state of these pages came to # carry three clean tables while the combined run refused. This harness makes the combined run # the only run, and the pages publish what THIS command prints. # # EXIT CODES: 0 = every entry admitted; 2 = nothing refused, something HELD for want of evidence # present here. Both are green. 1 (something REFUSED) and 3 (control arm failed) are not. if have xcrun || have swiftc; then SC2=$(have xcrun && echo "xcrun swiftc" || echo "swiftc") LALSTAGE="$(mktemp -d)"; cp "$HERE/library-admission-law.swift" "$LALSTAGE/main.swift" # Reuse a binary only if it is NEWER THAN THE SOURCE. Section 2 already compiled this # file once to prove it builds; this section needs it with arguments, and the law is the # largest program here. `-nt` is the whole guard: edit the law and it rebuilds. LALBUILT=0 [ -x /tmp/val_lal ] && [ /tmp/val_lal -nt "$HERE/library-admission-law.swift" ] && LALBUILT=1 if [ "$LALBUILT" -eq 1 ] || $SC2 -O -swift-version 5 "$LALSTAGE/main.swift" -o /tmp/val_lal 2>/dev/null; then /tmp/val_lal --library "$ROOT/library/proteins" \ --library "$ROOT/library/compounds" \ --library "$ROOT/library/materials" \ --reproduce "$HERE" --evidence /tmp > /tmp/out_library-admission-law-graded.txt 2>&1 LALEXIT=$? G=/tmp/out_library-admission-law-graded.txt # the control arm, as a RATCHET: every arm must pass and arms may only be added CA=$(grep -m1 '^CONTROL ARM ' "$G") cp_=$(printf '%s' "$CA" | sed -E 's#^CONTROL ARM +([0-9]+)/([0-9]+) PASS$#\1#') ct_=$(printf '%s' "$CA" | sed -E 's#^CONTROL ARM +([0-9]+)/([0-9]+) PASS$#\2#') if [ -n "$cp_" ] && [ "$cp_" = "$ct_" ] && [ "${ct_:-0}" -ge 61 ] 2>/dev/null; then ok "admission law control arm $cp_/$ct_ PASS (ratchet: at least 61 arms, all passing)" else bad "admission law control arm did not pass or has fewer than 61 arms: '$CA'" fi if [ "$LALEXIT" -eq 0 ] || [ "$LALEXIT" -eq 2 ]; then ok "the three libraries graded in ONE run, exit $LALEXIT (0 = all admitted, 2 = something HELD)" else bad "the three libraries graded in ONE run exited $LALEXIT — a refusal or a failed control arm, see $G" fi REFN=$(grep -m1 -E '^ REFUSED +[0-9]+$' "$G" | sed -E 's#^ REFUSED +##') [ "${REFN:-x}" = "0" ] && ok "0 entries refused across all three libraries" \ || bad "entries refused across the libraries: '$REFN'" GRADEDN=$(grep -m1 -E '^ entries graded +[0-9]+$' "$G" | sed -E 's#^ entries graded +##') if [ "${GRADEDN:-0}" -ge 16 ] 2>/dev/null; then ok "$GRADEDN entries graded (ratchet: the library may only grow)" else bad "only '${GRADEDN:-none}' entries graded — a checker given nothing must not report clean" fi if grep -q 'F1_NO_ENTRY_FILED_TWICE ok' "$G"; then ok "F1 — no entry filed in two libraries, over all three graded together" else bad "F1 did not clear: $(grep -m1 'F1_NO_ENTRY_FILED_TWICE' "$G")" fi for lb in PROTEINS COMPOUNDS MATERIALS; do if grep -q "^LIBRARY $lb -> ADMITTED" "$G"; then ok "LIBRARY $lb admitted by the law" else bad "LIBRARY $lb is not ADMITTED: $(grep -m1 "^LIBRARY $lb " "$G")"; fi done # NEGATIVE CONTROL. An instrument that has not been shown to refuse has measured # nothing, so the harness makes it refuse something on every run. EMPTYLIB=$(mktemp -d) /tmp/val_lal --library "$EMPTYLIB" --reproduce "$HERE" --evidence /tmp >/tmp/out_lal_empty.txt 2>&1 if [ $? -ne 0 ] && grep -q 'L1_NOT_EMPTY' /tmp/out_lal_empty.txt; then ok "negative control: the law REFUSES an empty library rather than reporting it clean" else bad "the law admitted an EMPTY library — always-green and always-red are the same defect" fi rm -rf "$EMPTYLIB" else bad "library-admission-law.swift does not compile — the libraries are ungraded, and ungraded is not a pass" fi rm -rf "$LALSTAGE"; rm -f /tmp/val_lal else echo " ABSENT no Swift toolchain — the three libraries are NOT graded this run, which is not a pass" fi echo "=== 4. the public pages carry no private reference ===" # grep -c prints 0 AND exits 1 on no match, so `|| echo 0` yields "0\n0" and breaks the # arithmetic — which is how the first version of this check reported PASS having counted # nothing. Count with a single grep -l pass and a positive control instead. BREACH=0 PAGES=$(ls "$ROOT"/*.md 2>/dev/null | wc -l | tr -d ' ') [ "$PAGES" -gt 0 ] || { bad "no pages found to scan — refusing to report a clean boundary"; BREACH=1; } # 'Sources/' alone is AMBIGUOUS: this public repo legitimately contains # clients/math-court-mcp/swift-example/Sources/main.swift. The pattern must name the # PRIVATE parents only, or it fires on our own public tree and trains readers to # ignore it — an always-red gate is as useless as an always-green one. for pat in 'cells/' 'cells/xcode/Sources/' 'LatticeRender/Sources/' '\.gaiaftcl' '/Users/' '\.swift:[0-9]' 'mortonBits' 'CapabilityRegistry'; do hits=$(grep -lE "$pat" "$ROOT"/*.md 2>/dev/null | wc -l | tr -d ' ') if [ "${hits:-0}" -gt 0 ]; then bad "private reference '$pat' in $hits page(s): $(grep -lE "$pat" "$ROOT"/*.md 2>/dev/null | xargs -n1 basename | tr '\n' ' ')" BREACH=1 fi done # POSITIVE CONTROL: the scanner must be able to find something that IS there. CTRL=$(grep -lE 'Affine' "$ROOT"/*.md 2>/dev/null | wc -l | tr -d ' ') if [ "${CTRL:-0}" -gt 0 ]; then [ "$BREACH" -eq 0 ] && ok "no private reference across $PAGES pages (scanner verified live on $CTRL)" else bad "the scanner found NOTHING at all, including its own control — it is not working" fi echo "=== 4b. every cited public artifact actually exists ===" # A page citing reproduce/foo.swift or corpus/bar that is not there is a broken # reproduction instruction — exactly the defect this repository was restructured to fix, # so the harness checks it rather than trusting that a rename kept up. MISSING=0 for cited in $(grep -ohE 'reproduce/[a-z0-9-]+\.swift|corpus/[A-Za-z0-9/_.-]+\.(tsv|md)|corpus/[A-Za-z0-9_-]+/SHA256SUMS' "$ROOT"/*.md 2>/dev/null | sort -u); do [ -e "$ROOT/$cited" ] || { bad "cited but absent: $cited"; MISSING=$((MISSING+1)); } done [ "$MISSING" -eq 0 ] && ok "every cited program and corpus path exists" echo "=== 4c. no private identifier, not just no private path ===" # The path patterns in 4 miss private IDENTIFIERS — daemon names, artifact names, internal # record fields. A boundary scan scoped to paths reported clean while three of these were # present, which is the same scoping error the substrate's float and heap gates each paid # for once. IDS=0 for pat in 'treasury-swarm' 'injector_lease_holder' 'covered_execs' 'observed_execs' \ 'com\.gaiaftcl' 'gaiaftcl-language-invariant' 'prove-fleet-byte-identity' \ 'apex-watchdog' 'prove-cell-identity' 'CertifiedUser' 'servableEntries'; do hits=$(grep -lE "$pat" "$ROOT"/*.md 2>/dev/null | wc -l | tr -d ' ') [ "${hits:-0}" -gt 0 ] && { bad "private identifier '$pat' in $hits page(s)"; IDS=1; } done [ "$IDS" -eq 0 ] && ok "no private identifier in any published page" echo "=== 4d. the library entry files carry no private reference ===" # Sections 4 and 4c scan the pages in the repository root. The library ENTRY FILES live one # level down, under library//, and were outside every one of those globs — the same # scoping error, one directory deeper. The positive control here is the standing not-advice # line, which every admitted entry must carry by clause E10: if the scanner cannot find it in # every file, either an entry has lost its line or the scanner is not working, and both are # reported rather than either being assumed. LIBFILES=$(ls "$ROOT"/library/*/*.md 2>/dev/null | wc -l | tr -d ' ') if [ "${LIBFILES:-0}" -eq 0 ]; then bad "no library entry files found to scan — refusing to report a clean boundary over nothing" else LBREACH=0 for pat in 'cells/' 'LatticeRender/Sources/' '\.gaiaftcl' '/Users/' '/home/' '/var/folders/' \ '\$HOME' 'mortonBits' 'CapabilityRegistry' 'com\.gaiaftcl' 'treasury-swarm'; do h=$(grep -lE "$pat" "$ROOT"/library/*/*.md 2>/dev/null | wc -l | tr -d ' ') if [ "${h:-0}" -gt 0 ]; then bad "private reference '$pat' in $h library entry file(s): $(grep -lE "$pat" "$ROOT"/library/*/*.md 2>/dev/null | xargs -n1 basename | tr '\n' ' ')" LBREACH=1 fi done LCTRL=$(grep -lF 'NOT_ADVICE' "$ROOT"/library/*/*.md 2>/dev/null | wc -l | tr -d ' ') if [ "${LCTRL:-0}" -eq "$LIBFILES" ]; then [ "$LBREACH" -eq 0 ] && ok "no private reference across $LIBFILES library entries (scanner verified live: NOT_ADVICE found in all $LCTRL)" else bad "the standing not-advice line is in only $LCTRL of $LIBFILES library entries — an entry lost its line, or the scanner is not working" fi fi echo "=== 4e. every library entry file is published on its library's page ===" # The gradeable artefact is library//; the artefact a stranger opens is Library-Of-*.md. # Clause L9 of the admission law gates this, and this row is the harness's own second opinion: # a count that must agree, computed a different way, from the fenced blocks themselves. PGOK=1 for pair in "proteins:Library-Of-Proteins.md" "compounds:Library-Of-Compound-Cures.md" "materials:Library-Of-Material-Systems.md"; do d="${pair%%:*}"; pg="${pair#*:}" nd=$(ls "$ROOT"/library/$d/*.md 2>/dev/null | wc -l | tr -d ' ') np=$(grep -c '^```affine-entry$' "$ROOT/$pg" 2>/dev/null | tr -d ' ') if [ "${nd:-0}" -gt 0 ] && [ "${nd:-0}" -eq "${np:-0}" ]; then ok "$pg publishes all $nd entries of library/$d" else bad "library/$d holds ${nd:-0} entries and $pg publishes ${np:-0}" PGOK=0 fi done echo "=== 5. the live surface serves ===" if have curl; then # ABSENT (unreachable / truncated) and MISS (reachable, wrong value) are different # answers and are reported as different answers. Neither counts as a pass. # # NOTE 2026-09-01: the REST path /language-invariant/games now returns a # CAPABILITIES REGISTRY and no longer carries domain_count or no_float. Those # fields live on the MCP path. The harness follows the fields, not the habit. MCP=https://affine.earth/language-invariant/mcp code=$(curl -s -o /tmp/court.json -w '%{http_code}' --max-time 90 -X POST "$MCP" \ -H 'Content-Type: application/json' \ -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"math_court","arguments":{"domain":""}}}' 2>/dev/null) size=$(wc -c < /tmp/court.json 2>/dev/null | tr -d ' ') if [ "$code" != "200" ] || [ "${size:-0}" -lt 256 ]; then echo " ABSENT live court unreachable or truncated (HTTP $code, ${size:-0} bytes) — surface not graded this run" else ok "live court answers 200 ($size bytes)" dc=$(grep -o 'domain_count[^0-9]*[0-9]*' /tmp/court.json | grep -o '[0-9]*$' | head -1) # THE CONSTANT IS THE HALF THAT GOES STALE. 48 -> 49 when biosphere joined; 49 -> 50 on # 2026-09-17 when the Coding Court's `cs` domain (roles compiler/consensus/ml_engineer) was # measured answering AFFINE_MATH_COURT on 9 of 9 cells while every page still said 49. The # fleet was right and the pages were fifteen days behind, so the pages were corrected and # this number follows them. A mismatch here means one of the two moved again: re-measure per # cell with --resolve before touching either. [ "$dc" = "50" ] && ok "court reports 50 domains (cs, the Coding Court, joined the 49)" || bad "domain_count is '$dc', pages say 50" if grep -q 'no_float' /tmp/court.json; then grep -q 'no_float[^a-z]*false' /tmp/court.json \ && bad "court reports a no_float:false domain" \ || ok "court reports no_float true" else echo " ABSENT no_float not present in the court response — not graded" fi fi else echo " SKIP — no curl" fi echo echo "=== $PASS passed · $FAIL failed ===" [ "$FAIL" -eq 0 ] && { echo "VALIDATED — every published figure reproduces from public bytes."; exit 0; } echo "NOT VALIDATED — see the failures above."; exit 1