Disorder Tolerance of Suspended Nanofilament Arrays: Why the Quarter-Pitch Acceptance Rule Fails as Filament Number Grows

1. Introduction

Spacing irregularity has worried everyone who has built a suspended nanofilament array, and with good reason, since an irregular array can open a gap wide enough for a hull member to pass between two strands untouched. The acceptance rule for such arrays, first codified in Deterrence Era, Year 21 (Tavares, 21 DE) and still applied in the fabrication audits of the Secretariat of the UN Planetary Defense Council (PDC) in Deterrence Era, Year 59 (DE 59; Thackeray, 33 DE), holds that an array is sound when the relative scatter of its spacings stays below one quarter of the pitch, whatever the number of filaments. In this paper, however, we show that the rule is wrong for large arrays. Passage is decided by the single widest gap, and the widest of many gaps grows with their number.

Two terms need fixing before we go on. The relative irregularity, written c, is the standard deviation of the spacings divided by their mean, so it is a pure number that does not depend on the pitch. A gap is critical when it is wide enough for the thinnest structural member of the hull section to clear both neighbouring strands.

The problem is now several decades old, and its parts are well understood separately, since comb statistics were characterised by DE 44 (Verhoeven, 44 DE) and extreme-value results for the largest gap in a random point sequence have been available since DE 52 (Dunmore, 52 DE), yet nobody joined the two in an acceptance criterion that a fabrication audit could apply to a finished array.

Operation Guzheng is the standing demonstration that such arrays work. Wang Miao's nanowire filaments, strung across the Gatun Locks of the Panama Canal, cut the ETO ship Judgment Day into slices and allowed the Trisolaran transmission archive to be recovered. We do not quote the geometry of that installation. Our representative Guzheng-type array is this paper's own design (N = 1,000 filaments, critical gap g_c = 2.2 pitches), and our question is how much spacing disorder an array of that kind can carry before a transiting hull passes through its gaps intact. Where we compare the installation with our tolerance in Section 4, we do so on the illustrative assumption that it resembled our representative array.

2. System Description

The representative array holds N = 1,000 parallel filaments in one plane between two anchor combs, with nominal pitch a. A hull section advances normal to the plane. If it meets no filament, it passes intact, and it can do so only through a gap of at least g_c times the pitch. We set g_c = 2.2, a design constant fixed by the thinnest structural member of our reference section, and we vary it in no calculation below. Spacings between neighbouring filaments are independent and lognormal with mean a and relative irregularity c. The lognormal form keeps spacings positive and reproduces the right-skewed distribution that drawn combs show (Verhoeven, 44 DE). Correlated pitch error, such as a slow creep of the comb along its length, is outside the model and we return to it under Failure Modes.

Bench arrays tested the model at reduced scale. We fabricated 360 arrays at three nominal irregularities (0.20, 0.28 and 0.36) and two sizes (100 and 400 filaments), 60 arrays in each of the six cells. Allocation followed the blocking we recommended for fabrication studies (Barragán Quiroga, 55 DE), with six blocks of ten arrays per cell fabricated in rotating order so that no cell was tied to a single week or a single operator. The pitch of every filament was measured optically, and the realised irregularity stayed within 0.005 of nominal in all cells. A breach was recorded when a gauge bar of width 2.2 pitches traversed the plane and passed between filaments without contact.

We also use one archived source, a pitch survey of the installation's anchor fixtures and the associated installation log, compiled at the time of the installation and released in 38 DE, held in PDC Declassified Holdings, New York (PDC Engineering Survey Office, 38 DE). It contains 276 pitch readings. We treat it as a record of fixture scatter only (the pitch at which the strands were seated), since it says nothing about the strands themselves after tensioning.

3. Analysis and Model

Let q be the probability that one gap is critical. With lognormal spacings q follows from the normal distribution function, and for N filaments the chance that at least one of the N - 1 gaps is critical is B = 1 - (1 - q)^N-1. That is, B is the breach probability, and everything below derives from it. The tolerance c* is the largest irregularity at which B stays at or below 0.01.

Table 1 gives c* for five array sizes (50 to 5,000 filaments), together with the breach probability that an array sitting exactly on the quarter-pitch limit would carry. The tolerance falls with size, but slowly. Relative to an array of 100 filaments, c* is 14% lower at 1,000 filaments and 20% lower at 5,000. A slow fall is what extreme-value theory leads us to expect, since for lognormal spacings the logarithm of the largest of N gaps grows roughly as the square root of the logarithm of N. Slow does not mean harmless. A rule fixed at 0.25 carries a breach probability of 0.043 at 100 filaments, which many audits would accept, but 0.357 at 1,000 and 0.890 at 5,000.

At the 1,000-filament tolerance of 0.191 the chance that any single gap is critical is about 1 in 100,000. No inspection of individual gaps would flag such an event as likely. Yet 999 gaps together breach one array in a hundred, and this is why a criterion built on mean or root-mean-square scatter cannot see the problem. We checked the closed form by direct simulation, a single run with a fixed seed and 20,000 arrays per condition. At c = 0.191 and N = 1,000 the simulated breach fraction was 0.0106 against 0.0102 analytically. At c = 0.25 and N = 1,000 it was 0.362 against 0.357, and at c = 0.30 and N = 400 it was 0.604 against 0.602.

The breach limit matters as well. At a looser breach limit of 0.05 the quarter-pitch rule is nearly right at 100 filaments, where c* is 0.254, but it still overshoots at 1,000 filaments, where c* is 0.211. The rule was calibrated on small arrays and carries that origin with it.

4. Validation Against Field Data

The bench arrays carry the main test, and the model enters it with no fitted parameter. Gap width g_c is fixed by the gauge bar, the spacings are lognormal with the nominal irregularity of each cell, and N is counted. Table 2 sets the observed breach fractions against the predictions.

In total 118 of 360 arrays were breached, and the model predicts 121, so the observed total falls three short.

Every predicted fraction lies inside the exact 95% interval of the observed one. The deviance of the six cells against their predictions is 5.3 on six degrees of freedom (p = 0.50), well below the 12.6 that marks the 5% level, and the cells give no sign that the model fails at either size.

Size dependence is the point at issue, and irregularity could confound it, so the primary test is stratified by c. The Cochran–Mantel–Haenszel test over the three irregularities gives χ² = 35.2 on one degree of freedom (p = 3 × 10^-9, uncorrected for continuity), with a common odds ratio for breach at 400 against 100 filaments of 6.3 (95% CI 3.3 to 12.3). All three strata point the same way, and their odds ratios (infinite at c = 0.20 with no breach at 100 filaments, then 2.7 and 21.7) spread as the model predicts (4.0, 5.0 and 17.2). A rule that sets a limit on c and ignores N predicts no such difference.

The pooled counts are descriptive only, since each size carries the same three irregularities with 60 arrays each. At 100 filaments 38 of 180 arrays were breached (21%), and at 400 filaments 80 of 180 (44%). Fisher's exact test on the pooled table gives p = 3.5 × 10^-6.

A field point of modest precision comes from the archived survey. The 276 pitch readings give a relative scatter of about 0.05, with a plausible range of 0.04 to 0.05, since reading error in the log can only add to the scatter. On the illustrative assumption that the installation resembled our representative array (N = 1,000, g_c = 2.2), the tolerance of 0.191 is between 3.8 and 4.8 times that scatter. The installation was not close to its limit, which is consistent with its having worked, and the survey tells us nothing about where the limit lies.

5. Failure Modes

The plausible alternative explanation for the size effect is strand loss. Larger arrays hold more filaments that can snap during tensioning, and a snapped strand merges two neighbouring gaps into one, so loss alone could produce more breaches at 400 filaments than at 100. We think it unlikely to be the main cause, and the data bear on it directly.

Of the 118 breached arrays, only 5 (2 at 100 filaments and 3 at 400) had a broken or missing strand at the critical gap. Dropping those 5 arrays from both numerator and denominator leaves 36 of 178 breached at 100 filaments (20%) against 77 of 177 at 400 (44%), and the contrast stays at p = 2.7 × 10^-6 by Fisher's exact test. The breach counts also match a model that contains no loss term at all. A loss effect large enough to produce the size difference would push the counts above that prediction, and the observed total sits three below it.

Model form is a further risk, and it points one way. A spacing distribution with a heavier right tail than the lognormal would raise q at every c and lower every tolerance in Table 1. We have not bounded the effect, and the true tolerance for full-scale arrays is likely somewhat below our figures.

Correlated error also matters. Creep of a comb along its length shifts neighbouring gaps together (Wetherell and Idowu, 41 DE), and we have no measurement of how it combines with independent scatter. The archived survey does not settle this, because an installation log records pitch at fixed stations and not the drift between them.

The reach of the bench arrays is short, since they stop at 400 filaments. The analytic form was confirmed by simulation up to 1,000, but the step from 400 to 1,000 and beyond rests on the model alone, and that step is the one that matters for an installation of the Guzheng type.

6. Conclusion

Tolerance to spacing disorder falls with the number of filaments, and we hold this as likely for full-scale arrays, though the evidence may fall short of establishing it. The model is extreme-value arithmetic confirmed at bench scale, so the main uncertainty lies in the extrapolation and not in the principle, and heavier spacing tails may lower the true tolerance of a full-scale array further. Strand loss, the best rival explanation, accounts for few of the bench breaches and probably does not remove the size effect.

We treat the Luo-Ji / Ye framework and the Lurker framework as compatible, with cosmic sociology serving as mechanism inside a Lurker-shaped universe, and nothing in these measurements asks for more than the laws of the laboratory.

Every array accepted at 1,000 filaments or more under the quarter-pitch rule was accepted against a breach probability that the rule never stated, and those audit records should be reopened. Going forward, acceptance limits should be written as a function of N, with no more than 0.18 as the working figure at 1,000 filaments until heavier-tailed spacing models are bounded. Fabrication studies should measure the pitch of every filament, report the largest measured gap beside the scatter, and test at least one array size within a factor of three of the intended installation.