EpiStack

main report - Was the risk that LHC collisions destroy the Earth truly put to rest and what does that conclusion hinge on

9 min read

Properties1
type
main-report

Main report — Was the risk that LHC collisions destroy the Earth truly put to rest, and what does that conclusion hinge on?

1. The answer

The question (verbatim): “Was the risk that LHC collisions destroy the Earth (via stable micro black holes or other proposed mechanisms) truly put to rest, and what does that conclusion hinge on?”

For the micro-black-hole mechanism: effectively yes — but the “rest” is a conjunction of three separately-uncertain safeguards, none of which is a direct measurement, and the whole empirical layer routes through a single motivated source. The model’s numbers (baseline run, python3 runner/run.py, 2026-07-20): the fundamental gravity scale is very likely far above LHC reach, so no black hole forms at all (HC-3 - Accessibility of the fundamental gravity scale at LHC energies: H-8 posterior 0.94); if one formed anyway, it evaporates near-instantly (HC-2 - Evaporation of TeV-scale black holes: H-4 posterior 0.9561); if evaporation failed and a hole were trapped in Earth, it does no macroscopic damage within the solar lifetime (HC-1 - Fate of a stable LHC black hole trapped in Earth: H-2 posterior 0.8882, catastrophic H-1 at 0.0354). Chaining the danger legs — my composition, not a model output; the clusters were deliberately built as conditionals in exactly this order — gives roughly 0.06 × 0.044 × 0.035 ≈ 1e-4 as a loose upper shape for “LHC-produced stable black hole destroys Earth”, and I read the true figure as lower, because each factor’s danger mass is dominated by residual/guessed terms (reference-class-free priors, residual hypotheses), not by any positive evidence for danger.

What sits outside that number — the “what does it hinge on” half of the question, which is the real finding: (1) nearly every empirical likelihood in HC-1 and HC-2 routes through one paper, Giddings–Mangano (S-1 - Astrophysical implications of hypothetical stable TeV-scale black holes, trust-capped 0.74) and its shared cosmic-ray-survival premise D-1 - Cosmic-ray survival premise - high-energy cosmic rays have bombarded astronomical bodies over Gyr without catastrophe — the safety case is one theory pillar plus one paper’s astrophysics, not a multiply-independent web; (2) the theory pillar, Hawking evaporation (A-5 - Semiclassical QFT in curved spacetime derives thermal emission with accelerating evaporation), is approved but was never traced in this run, is unobserved in nature (O-9 - Hawking radiation has never been experimentally observed from any black hole sits orphaned), and is weakest exactly in the TeV regime at issue; (3) “other proposed mechanisms” in the question — strangelets, vacuum decay — never became clusters in this N=5 test run, so for them this analysis answers only by the untested cosmic-ray template, and the tempting independent backstop (HC-4 - Bound on the exogenous terminal-catastrophe rate, bound holds at 0.7615) explicitly does not cover human-made collisions; (4) HC-1’s residual H-6 (0.0764) concedes that the survival observations cannot distinguish exotic sub-catastrophic fates from full safety.

My bet, stated as a judgment: order 10^3–10^4 : 1 against the black-hole scenario destroying Earth — comfortable. But at no odds on “truly put to rest” in the strict sense the question probes: the conclusion is robust in direction and fragile in pedigree.

Entry points: Analysis of HC-1 - Fate of a stable LHC black hole trapped in Earth, Analysis of HC-2 - Evaporation of TeV-scale black holes, Analysis of HC-3 - Accessibility of the fundamental gravity scale at LHC energies, Analysis of HC-4 - Bound on the exogenous terminal-catastrophe rate, S-1 - Astrophysical implications of hypothetical stable TeV-scale black holes, D-1 - Cosmic-ray survival premise - high-energy cosmic rays have bombarded astronomical bodies over Gyr without catastrophe, A-5 - Semiclassical QFT in curved spacetime derives thermal emission with accelerating evaporation, A-6 - Observer selection makes naive survival-based risk bounds uninformative, E-6 - O-10 × HC-4 — Earth’s formation date is typical, not early.

2. What the analysis found

HC-3 — does a hole form at all? Prior = posterior [0.03, 0.94, 0.03]: zero likelihood blocks; the one edge (the gauge hierarchy, E-5) was priced inside the prior. Production needs TeV-scale gravity (ADD-type extra dimensions or an unlisted mechanism); the load-bearing quantity is the combined low-scale mass H-3+H-9 ≈ 0.06, resting on proposal-space head-count shares with no principled measure behind them (Analysis of HC-3 - Accessibility of the fundamental gravity scale at LHC energies). The review’s key defect: post-1998 LHC monojet/missing-energy nulls and sub-mm gravity tests were never ingested, so 0.94 is directionally safe but understates confidence.

HC-2 — does it evaporate? [0.891, 0.109] → [0.9561, 0.0439]. The prior is purely theoretical (A-5: framework-wrong 0.01, extrapolation-stabilizes 0.10); the update is astrophysical survival, effectively neutron-star survival once A-6 - Observer selection makes naive survival-based risk bounds uninformative inerts the Earth/Sun leg. All empirical trust caps at 0.74 via S-1 (Analysis of HC-2 - Evaporation of TeV-scale black holes). The review notes H-7’s 0.0439 overstates danger: a Planck-relic endpoint counts as “fails to evaporate” but is safety-inert.

HC-1 — is a stable trapped hole dangerous? [0.3097, 0.5752, 0.115] → [0.0354, 0.8882, 0.0764]. Two evidence blocks (CG-1 joint over NS/WD/Earth survival; lone white-dwarf edge E-1), both with lik(H-1)=0.1 at t=0.74, both from S-1’s stopping-power and accretion machinery — “closer to one item than two,” per Analysis of HC-1 - Fate of a stable LHC black hole trapped in Earth. The prior’s main driver (frac_log_range_fast = 0.35) is an admitted reference-class-free guess.

HC-4 — the exogenous backstop. [0.504, 0.496] → [0.7615, 0.2385] on a single edge (Earth’s typical formation date, E-6, trust 0.8). Weighing judgment: this cluster bears on the answer only weakly and indirectly. Its bound is explicitly exogenous — it does not cover human-modulated risks like the LHC (Analysis of HC-4 - Bound on the exogenous terminal-catastrophe rate) — so it is not the independent safety net a reader might take it for. Its real contribution is methodological: it exercises the observer-selection-safe bounding move (A-6) that the whole survival-argument layer depends on.

The weighing. The three physics clusters compose as a conditional chain and all point the same way, so part 1’s bottom line is the chain product (a judgment) discounted for their shared dependence: HC-2’s likelihoods already smuggle in HC-3’s production question (p_cr_production = 0.7), and HC-1 and HC-2 share S-1/D-1, so the three legs are not independent and the product is a shape, not a computed probability. The mechanism split inside the safe verdict matters less than the pedigree finding: whichever leg you doubt, the fallback leg you land on also routes through either A-5 or S-1.

3. What the answer hangs on, and what would change it

  1. S-1 trust (the single-source bottleneck) — priced. Re-run with all four S-1-capped trusts (CG-1:t_survival, E-1:t_wd, CG-2:t_surv, E-7:t_wd7) dropped 0.74 → 0.3: HC-1 danger 0.0354 → 0.1663 (×4.7), HC-2 non-evaporation 0.0439 → 0.0793. Raised to 1.0: 0.0354 → 0.0050 and 0.0439 → 0.0278. So essentially the entire empirical contribution of this analysis — roughly a factor 30 on HC-1’s danger odds between full trust and heavy distrust — flows through one paper’s contested accretion/stopping-power machinery, exactly where Plaga’s challenge lives. This is the weakest structural link.
  2. HC-1’s prior guess — priced. --set HC-1:frac_log_range_fast=0.6 (guess 0.35 → 0.6): posterior danger 0.0354 → 0.0669. Roughly linear, as the review predicted; a ×2 error in a number with no reference class is a ×2 error in the headline danger figure.
  3. A-5 (Hawking evaporation) — named, not priced. No single variable carries it; the 0.109 prior failure mass is the analysis’s own decomposition and the derivation was never traced. If semiclassical control fails at TeV masses, HC-2’s safe verdict falls back entirely onto lever 1. Magnitude argued, not computed: this is the one place a theory error breaks two clusters at once.
  4. Missing information, re-ordered by effect on the answer (labels from the reviews): (a) LHC/collider extra-dimension null limits as priced observations — exists, unread; would raise HC-3’s H-8 and shrink the danger chain’s first factor, pure confidence gain on the safe side. (b) An independent, non-Giddings–Mangano computation of trapped-hole accretion in WD/NS matter engaging Plaga directly — exists, unread (Plaga is in the source pool) to does not exist (neutral adjudication); the only item that could move the answer against safety. (c) A traced or analogue-experiment-backed audit of A-5 — exists, unread. (d) Independent replication of S-1’s neutron-star stopping-power numbers — unclear. (e) Tail-behaviour comparison for HC-4’s planet-formation distribution — exists, unread; small effect on the main answer given HC-4’s exogeneity firewall.
  5. A known contradiction that does not flip anything: E-6 and A-7 - Earth’s unremarkably late formation time yields a selection-bias-free catastrophe-rate bound near 1 per Gyr assert opposite observer-weighting tilts (two different hazard models); the update direction survives either way, but HC-4’s 0.7615 should be read at reduced precision until a step-6/8 re-trace.

4. What this does not cover

  1. Scope taken: “destroy the Earth” was operationalized as the micro-black-hole chain (production → evaporation → accretion) plus one methodological backstop cluster. Strangelet conversion and vacuum decay — named in the question’s “other proposed mechanisms” and in the run’s scope notes — were never modelled: at curated_target_N = 5 the curation cut (see agent-notes/curation.md) kept Hawking 1975, Auger, ADD 1998, Tegmark–Bostrom, and Giddings–Mangano, dropping the strangelet/vacuum literature (Busza et al., Hut–Rees). Notably the LSAG report itself — the document that “put the risk to rest” publicly — scored below the cut (trust 0.63 as a synthesis) and was never ingested; this analysis reconstructs the case from its primary inputs instead.
  2. Orphans: the Auger spectrum observations O-5/O-6/O-7 discriminated no cluster (the cosmic-ray premise entered via D-1 instead), and O-9 (Hawking radiation never observed) is orphaned precisely because A-5 was taken on trust — the graph’s most safety-relevant negative fact never touched a likelihood.
  3. What the debate performed rather than settled: the Plaga–Giddings-Mangano exchange (neither side peer-reviewed) is priced only through S-1’s 0.74 trust cap; motivated-source selection effects — which counterarguments got written down at all — are outside every number, per the HC-1 review.
  4. External consensus (labelled; comparison only). No external_consensus fields were written in this run (a pipeline gap I have logged), so this is my own characterization of the field: the physics community regards LHC black-hole risk as closed — LSAG’s conclusions were endorsed by CERN’s governing bodies, courts dismissed the lawsuits, and 15+ years of LHC operation passed without incident. This analysis agrees in direction while adding what the consensus documents do not foreground: the case’s empirical layer is nearly single-sourced, its theory layer is unobserved, the celebrated cosmic-ray argument needed the white-dwarf/neutron-star repair to cover slow trapped holes at all, and the operation-without-incident point is exactly the kind of survival evidence A-6 warns is self-selected. “Put to rest” is the right verdict; “truly”, in the sense of resting on multiple independent legs, is more than the graph can certify.

Graph View

Backlinks