What does Drift and Noise on Long Pages conclude, in numbers, about “Noise and Mistaken Exits: Why a Longer Page Does Not Mean More Persuasion”?

Answer: Chapter 2 carries the main argument: with a drift of 0.30 and noise of 1.0 the probability of a mistaken exit is 0.3543, and with a clutter coefficient of 0.06 the best page has 6 sections, while without clutter the ceiling disappears.

Scope of the numbers: Chapter 4 pushes a page forward section by section: deleting the feature table of an illustrative page gains 0.0081, and the best and worst orders of sections differ by 0.1764. Marked boxes separate published evidence, runnable Python calculations (20 programs), classical sources and the author's hypotheses.

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: 0.30, 0.3543, 1.0

P(leave) P(leave) E[T] P(pay), v-0.30
noise s v=+0.15 v=+0.30 v=+0.30 start 0.8
0.30.0344 0.0013 3.3249 0.0003
0.50.2315 0.0832 2.7788 0.0483
0.70.3515 0.2271 1.8190 0.1573
1.00.4256 0.3543 0.9710 0.2655
1.40.4618 0.4241 0.5063 0.3286
2.00.4813 0.4626 0.2495 0.3644
max |general formula - Bogacz ER|5.6e-17
limit as s grows: P(pay) -> Z/A0.50 (start 1.0), 0.40 (start 0.8)
Show the full program
# Chapter 2. p21: noise raises the chance of hitting the wrong barrier,
# and which barrier is "wrong" depends on the sign of the drift.
# 模型同 p11:A = 2,Z = 1(不偏),s 从 0.3 扫到 2.0。
# 起点居中时,误撞率 ER = 1 / (1 + exp(2 v (A/2) / s^2)),
# 即 Bogacz 等 2006 年给出的形式,这里与一般公式逐行互验。
# 漂移为正(页面信息总体支持购买)时,噪音推高「离开」概率;
# 漂移为负时,噪音反而推高「付款」概率,这些付款是误撞。参数为示意。
import math

A, Z = 2.0, 1.0


def p_pay(v, s, a=A, z=Z):
    if abs(v) < 1e-12:
        return z / a
    k = 2.0 * v / (s * s)
    return (1.0 - math.exp(-k * z)) / (1.0 - math.exp(-k * a))


def mean_time(v, s, a=A, z=Z):
    if abs(v) < 1e-12:
        return z * (a - z) / (s * s)
    return (a * p_pay(v, s, a, z) - z) / v


def er_bogacz(v, s, a=A):
    return 1.0 / (1.0 + math.exp(2.0 * abs(v) * (a / 2.0) / (s * s)))


print("        P(leave)  P(leave)  E[T]      P(pay), v=-0.30")
print("noise s v=+0.15   v=+0.30   v=+0.30   start 0.8")
for s in (0.3, 0.5, 0.7, 1.0, 1.4, 2.0):
    print("%.1f     %.4f    %.4f    %.4f    %.4f"
          % (s, 1.0 - p_pay(0.15, s), 1.0 - p_pay(0.30, s),
             mean_time(0.30, s), p_pay(-0.30, s, z=0.8)))
gap = max(abs((1.0 - p_pay(0.30, s)) - er_bogacz(0.30, s))
          for s in (0.3, 0.5, 0.7, 1.0, 1.4, 2.0))
print("max |general formula - Bogacz ER| = %.1e" % gap)
print("limit as s grows: P(pay) -> Z/A = %.2f (start 1.0), %.2f (start 0.8)"
      % (Z / A, 0.8 / A))

Single source and citation

Fan, Y. (2026). Fokker-Planck Drift-Diffusion on Long Sales Pages: Belief-Drift Rates and First-Passage Times to the Payment Barrier (Version 1.0.0), Chapter 2: Noise and Mistaken Exits: Why a Longer Page Does Not Mean More Persuasion. 5D Library. DOI pending

DOI pending · monograph page goes live after publication · 5DL-B-240 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026240,
  author    = {Fan, Yafu},
  title     = {Fokker-Planck Drift-Diffusion on Long Sales Pages: Belief-Drift Rates and First-Passage Times to the Payment Barrier},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

Other chapter answers in this book