What does Mid-Context Forgetting and Relay Thresholds conclude, in numbers, about “Dissent and Diversity: Four Curves, Bigger Windows and the Single-Needle Test”?

Answer: Program p21 shows the four models of the paper drawing four shapes, with only one model's middle below its closed-book score and every model carrying a clear dilution loss. Program p22 shows that extended-window versions differ from their originals by no more than 0.3 points, far below sampling error. Program p23 shows that a single-needle recall of 0.997 leaves only a 0.740 chance of a clean sweep over a hundred facts. Description has gone far enough. To predict a length that has never been measured we need a formula that extrapolates. Can primacy, recency and dilution be written into one curve?

Scope of the numbers: Tasks, material and parameters are illustrative and the model constants come from one 2023 model; all calculations are generated by the book's programs, which call no model. It is not individual business advice and neither rates nor recommends any model or service.

中文版

Data table

The numbers below are printed verbatim by the chapter's Python program and can be reproduced by running it. standard-library Python onlyprinted values that match the answer: 0.3, 0.740, 0.997

per-fact p all of 1 all of 5 all of 20 all of 100 k for 0.5
0.9970.9970.9850.9420.740230.7
0.9900.9900.9510.8180.36669.0
0.9500.9500.7740.3580.00613.5
0.9000.9000.5900.1220.0006.6
expected misses among 100 facts at p = 0.997: 0.3
Show the full program
# Chapter 2. p23: why a near-perfect single-needle score settles little.
# A single-needle test reports the chance p that one planted fact is
# found. A real task needs k facts at once. If misses were independent,
# all k survive with p^k; we also print the k at which that falls to
# one half. The value 0.997 echoes the single-needle recall above
# 99.7 per cent reported for Gemini 1.5 Pro (Gemini Team 2024); the
# other values are illustrative. Independence is an assumption: shared
# failure modes would raise p^k for some tasks and lower it for others.
import math

PS = (0.997, 0.99, 0.95, 0.9)
KS = (1, 5, 20, 100)

print("per-fact p   " + "  ".join("all of %3d" % k for k in KS)
      + "   k for 0.5")
for p in PS:
    row = "  ".join("%10.3f" % (p ** k) for k in KS)
    k_half = math.log(0.5) / math.log(p)
    print("   %6.3f   %s   %8.1f" % (p, row, k_half))
print("expected misses among 100 facts at p = 0.997: %.1f"
      % (100 * (1 - 0.997)))

Single source and citation

Fan, Y. (2026). Mid-Context Forgetting in Long-Context Attention: Positional Bias and the Threshold for Lossless Segment-Summary Relays (Version 1.0.0), Chapter 2: Dissent and Diversity: Four Curves, Bigger Windows and the Single-Needle Test. 5D Library. DOI pending

DOI pending · monograph page goes live after publication · 5DL-B-255 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026255,
  author    = {Fan, Yafu},
  title     = {Mid-Context Forgetting in Long-Context Attention: Positional Bias and the Threshold for Lossless Segment-Summary Relays},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

Other chapter answers in this book