If wishes were horses: discursive comments on attempts to prevent individuals from being unfairly burdened by their reference classes

Rambling response to Is it a Crime to Belong to a Reference Class?, holding essentially that since it's impossible to understand anything without external referents, we can do whatever we want with statistical evidence.

Notes

‘What was the aggre- gate amount of heroin (in grams) that Shonubi imported during all eight drug-smuggling trips that Shonubi is believed to have made?’

NOTER_PAGE: (2 0.45554123711340205 . 0.3704945515507125)

Judge Weinstein, could and should consider statistics compiled by the US Customs Service about the amounts of heroin imported by similar smugglers

NOTER_PAGE: (3 0.2081185567010309 . 0.36378876781223807)

in Shonubi IV, the Second Circuit expressly disavowed any reliance on a ‘direct evidence’ requirement. Instead, the Second Circuit Court of Appeals expanded and expressly reaffirmed its earlier suggestion (in Shonubi II) that a finding of the amount of heroin that Shonubi imported in his trips had to be supported by ‘specific evidence’.

NOTER_PAGE: (3 0.601159793814433 . 0.6730930427493713)

I argued that the attempted distinction between specific and non-specific evidence is almost unintelligible.

NOTER_PAGE: (4 0.398840206185567 . 0.5884325230511316)

statistical evidence should always be inadmissible, a conclusion that is wildly inconsistent with the Second Circuit’s general position

NOTER_PAGE: (4 0.43427835051546393 . 0.2673931265716681)

no meaningful way of limiting the reach of the specific evidence requirement to statistics

NOTER_PAGE: (4 0.485180412371134 . 0.19111483654652137)

‘Is it a crime to belong to a reference class?’¹

NOTER_PAGE: (5 0.14940364092906466 . 0.533061224489796)

my criticisms of the Second Circuit’s specific evidence requirement stand even if Judge Weinstein used truly rotten statistics and truly rotten statistical methods.

NOTER_PAGE: (7 0.5768989328311362 . 0.25795918367346937)

it is not always inferentially illegitimate to base inferences about the behaviour of an individ- ual on the behaviour of other people,

NOTER_PAGE: (7 0.6710608913998745 . 0.1893877551020408)

if group-to-individual inference is a vice, this vice is not obviated by avoidance of formal statistical methods

NOTER_PAGE: (7 0.7099811676082863 . 0.18775510204081633)

logic of the requirement alto- gether bars the use of statistical evidence to show individual human behaviour.

NOTER_PAGE: (7 0.7532956685499059 . 0.6489795918367347)

maintain that it is logically consistent to disallow the use of statistics about groups to draw inferences about the behaviour of individuals but allow the use of statistics about an individual to show the behaviour of the same individual.

NOTER_PAGE: (8 0.1324544883866918 . 0.17795918367346938)

dif- ference of opinion about the validity or invalidity of using statistics about groups of individuals to support or draw inferences about individuals.

NOTER_PAGE: (8 0.19334588826114252 . 0.3926530612244898)

The reference class that occasioned the birth of the Second Circuit’s specific evidence require- ment was, roughly, ‘balloon-swallowing Nigerian heroin importers arrested at JFK Airport from 1 September 1990 through 10 December 1991’.

NOTER_PAGE: (8 0.6654111738857502 . 0.15836734693877552)

suggests that the specific evidence requirement rests on a normative antipathy to inferences about individuals that rest on evidence about groups of individuals.

NOTER_PAGE: (9 0.18832391713747648 . 0.2220408163265306)

did not think that it would be fair to let the extent of the punishment inflicted on the individual Charles O. Shonubi depend even in part on the behaviour of a group of other people, which in this instance consisted of a small sub-group of (other) Nigerian nationals.

NOTER_PAGE: (9 0.32391713747645956 . 0.4285714285714286)

sensed that an extra-epistemic explanation for the specific evidence requirement just would not pan out.

NOTER_PAGE: (10 0.11550533584431891 . 0.4195918367346939)

statistical inference about an individual’s unknown behaviour can be supported only by statistics about that same individual.

NOTER_PAGE: (10 0.519774011299435 . 0.25306122448979596)

a great deal of law on the books demonstrates that group behaviour can be used as evidence in a wide variety of ways to show the behaviour of an individual.

NOTER_PAGE: (10 0.6409290646578782 . 0.27755102040816326)

if you were to tell me that I ought not consider behaviour or matters other than the behaviour or matter that is the ultimate subject of my inferential interest—I would say to you that you have effectively precluded the use of any evidence whatever.

NOTER_PAGE: (12 0.3615819209039548 . 0.43918367346938775)

to know what I am seeing when I watch X, I need to have knowledge about many things and matters other than X.

NOTER_PAGE: (12 0.610797237915882 . 0.3616326530612245)

Fascism has rejected the concept of a separate individuality, the concept of ‘a man’, and operates only with vast aggregates.

NOTER_PAGE: (13 0.24168236032642815 . 0.17877551020408164)

the question of whether it is possible to draw inferences about in- dividuals without burdening individuals with adverse inferences resulting at least in part from such individual’s inherited or unchosen attributes is both an old question and a hard one.

NOTER_PAGE: (13 0.6798493408662901 . 0.3746938775510204)

Valid inference can be unfair.

NOTER_PAGE: (14 0.21594475831763968 . 0.16163265306122448)

a choice to forfeit the making of valid inferences (e.g. by precluding the use of probative evidence) always exacts a price.

NOTER_PAGE: (14 0.35655994978028877 . 0.15183673469387757)

Court of Ap- peals never clearly stated whether it saw non-specific statistical evidence as objectionable because it thought that such evidence is socially harmful or whether the court regarded non-specific statistical evidence as objectionable only because the court thought that such evidence is bereft of inferential force.

NOTER_PAGE: (14 0.6070307595731325 . 0.14612244897959184)