What does Collapsing Idea-to-Delivery Latency conclude, in numbers, about “Batches and Rehearsal: the U-Shaped Cost and the Smith Predictor”?
Answer: Chapter 4 gives the U-shaped cost of batch size, with the cost per change falling from 4.19 to 1.20 once delivery overhead drops, and uses the Smith predictor to show the gains and risks of rehearsing in a model.
Scope of the numbers: Parameters are illustrative and all calculations are generated by the book's programs; it offers general, not individual, business guidance, promises no launch date or income, and every production deployment is carried out only after a person confirms it. Chapter 2 splits latency into four drags, decision, start-up, feedback and hand-off, with an illustrative median of 99.9 hours and P90 of 199.8 hours, the tail driven mainly by feedback and decision.
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: 1.20, 4.19
| A. cost per change by batch size b (illustrative units) | |
| b T | 8.0 T=1.0 wait to feedback (days) |
| 1 | 8.20 1.20 1.0 |
| 2 | 4.88 1.38 1.5 |
| 3 | 4.21 1.88 2.0 |
| 4 | 4.19 2.44 2.5 |
| 6 | 4.77 3.60 3.5 |
| 8 | 5.64 4.76 4.5 |
| 12 | 7.60 7.02 6.5 |
| 16 | 9.63 9.19 8.5 |
| B. optimum over b | 1 to 40 |
| T 8.0: best b 4, cost 4.19, sqrt(2T/h) 4.00 | |
| T 1.0: best b 1, cost 1.20, sqrt(2T/h) 1.41 | |
| C. cutting T from 8.0 to 1.0: feedback wait 2.5 | 1.0 d, |
| cost per change 4.19 | 1.20 |
| D. keeping b | 4 after T falls costs 2.44 vs 1.20 |
Show the full program
# Chapter 4. p41: the U-shaped cost of batch size, with feedback delay.
# One finished change per day. A release costs T (overhead of one
# shipment); each change waits for its batch, (b - 1) / 2 days on
# average, at a cost of delay h = 1 per change-day; a change is faulty
# with p = 0.10, and a faulty batch costs c = 2 per change in it to
# localise (more changes, harder to find). Cost per change:
# C(b) = T / b + h (b - 1) / 2 + c (1 - (1 - p) ** b)
# All values are illustrative units, not money. Wait to first outside
# feedback per change = (b - 1) / 2 + F with F = 1 day after release.
import math
H = 1.0
P = 0.10
C = 2.0
F = 1.0
def cost(b, t):
return t / b + H * (b - 1) / 2.0 + C * (1.0 - (1.0 - P) ** b)
if __name__ == "__main__":
sizes = (1, 2, 3, 4, 6, 8, 12, 16)
print("A. cost per change by batch size b (illustrative units)")
print(" b T=8.0 T=1.0 wait to feedback (days)")
for b in sizes:
print(" {:>2} {:>7.2f} {:>7.2f} {:.1f}".format(
b, cost(b, 8.0), cost(b, 1.0), (b - 1) / 2.0 + F))
print("B. optimum over b = 1 to 40")
for t in (8.0, 1.0):
best = min(range(1, 41), key=lambda b: cost(b, t))
eoq = math.sqrt(2.0 * t / H)
print(" T {:.1f}: best b {}, cost {:.2f}, sqrt(2T/h) {:.2f}"
.format(t, best, cost(best, t), eoq))
b8 = min(range(1, 41), key=lambda b: cost(b, 8.0))
b1 = min(range(1, 41), key=lambda b: cost(b, 1.0))
print("C. cutting T from 8.0 to 1.0: feedback wait {:.1f} -> {:.1f} d,"
.format((b8 - 1) / 2.0 + F, (b1 - 1) / 2.0 + F))
print(" cost per change {:.2f} -> {:.2f}".format(
cost(b8, 8.0), cost(b1, 1.0)))
print("D. keeping b = {} after T falls costs {:.2f} vs {:.2f}".format(
b8, cost(b8, 1.0), cost(b1, 1.0)))Single source and citation
Fan, Y. (2026). Collapsing Idea-to-Reality Latency: Feedback-Bandwidth Limits, Zero-Friction Toolchains and High-Velocity Execution Physics (Version 1.0.0), Chapter 4: Batches and Rehearsal: the U-Shaped Cost and the Smith Predictor. 5D Library. DOI pending
DOI pending · monograph page goes live after publication · 5DL-B-420 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026420,
author = {Fan, Yafu},
title = {Collapsing Idea-to-Reality Latency: Feedback-Bandwidth Limits, Zero-Friction Toolchains and High-Velocity Execution Physics},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}Other chapter answers in this book
- Chapter 1 · Loop Bandwidth: With the Delay Unchanged, More Gain Only Buys Oscillation
- Chapter 2 · Decomposing End-to-End Latency: Four Drags and the Slowest Tenth
- Chapter 3 · Queues: Why Full Utilisation Is Not Efficiency
- Chapter 5 · Hand-offs, Switching and Hesitation: Three Losses Along the Chain
- Chapter 6 · Toolchain Friction and Gates: Why Speed Never Skips a Gate
- Chapter 7 · Classical Sources East and West: Clumsy Speed, the Ant Hole and Maxwell on Governors
- Chapter 8 · Closing the Loop: the Latency Ledger, Field Tools and Checker Specifications