28 lines
1.1 KiB
Markdown
28 lines
1.1 KiB
Markdown
---
|
|
id: 2026-01-25T21:02:00-05:00
|
|
title: 2026-01-25 21:02:??
|
|
---
|
|
# 2026-01-25 21:02:??
|
|
|
|
[[the-failure-of-risk-management]]
|
|
|
|
If I have a general complaint about [[hubbard_2020_failure]] it's this:
|
|
Hubbard fails to recognize logical parallels between his different arguments,
|
|
so to a critical reader they appear contradictory:
|
|
|
|
In [[hubbard_2020_failure#Break It Down, Then Do the Math]]
|
|
Hubbard introduces **decomposition** as a method for reducing error in estimates,
|
|
Providing Fermi's "piano tuners in Chicago" problem as an example,
|
|
without acknowledging that _Fermi_ supplied the decomposition.
|
|
Hubbard also references [[macgregor_1994_judgemental-decomposition]]
|
|
which is a similar case.
|
|
Neither of these examples suggest that decomposition is a magic bullet,
|
|
or even that a layman's decomposition wouldn't be worse than nothing.
|
|
Despite this, Hubbard ends the section without qualifier:
|
|
"Clearly, decomposition helps estimates."
|
|
|
|
This section comes only pages after
|
|
[[hubbard_2020_failure#The Measurement Inversion]]
|
|
in which Hubbard warns against seeking detail for the sake of it.
|
|
|
|
See also [[the-failure-of-risk-management#_Exsupero Ursus_]]. |