9. Safety: An Application
© 2017 Mark McBride, CC BY 4.0 https://doi.org/10.11647/OBP.0104.13
In this final chapter, I bring epistemology to the practical domain of law. As Comesaña (2005) presented a putative counterexample to the necessity of Sosa’s (1999, 2002) safety condition, so I present a putative counterexample to the necessity of the safety condition that Pardo (2010) employs in his work on knowledge and jury verdicts. My aim is not, of course, to falsify the thesis that safety is necessary for knowledge, but rather to advance discussion of the safety condition in philosophy of law (see Pardo, 2011). Towards the end, connections are drawn with my proposal in Chapter Eight.
0.1 Michael Pardo (2010: 38) recently intriguingly argues that “the goal or aim of legal proof is knowledge (or something approximating knowledge) rather than less epistemically demanding goals.”1 Pardo (52) continues, “[L]egal verdicts require more than truth and justification […] [T]ruth and justification also need to be connected in an appropriate way.” I do not want to contest (directly) this central claim of Pardo’s. My aim here is principally to show some difficulties for the account of (legal) knowledge with which Pardo (nn. 61 and 81) evidently operates, on which safety is a necessary condition. Highlighting these difficulties, however, puts pressure on Pardo’s central claim.
The plan: first, I set out and probe Pardo’s Fake Cabs case — a case that demonstrates Pardo’s commitment to safety as a necessary condition on (legal) knowledge. Second, I present a putative legal counterexample to safety as a necessary condition on (legal) knowledge. I close by presenting Pardo with a trilemma.
1. Pardo’s Fake Cabs Case
1.1 Pardo (52) presents the following case:
Fake Cabs: The plaintiff files a lawsuit against the defendant, who owns and drives the only taxicab in town, claiming she was hit by the defendant’s cab while crossing the street. She saw the cab drive away but did not see the driver. A video camera at the intersection filmed the accident, and it shows what appears to be a cab (but not the driver) hitting the plaintiff, exactly as she claimed. Now, suppose the car in the video really is the defendant’s, but also that — unknown to the jury — along with his real cab there are hundreds of other cars in the town that look identical to his cab. The jury finds for the plaintiff based on the video.
And Pardo’s (52) gloss thereon:
The verdict is true and justified (that is, there is sufficient evidence before the jury and no evidence regarding the fake cabs), and […] there is a straight-forward causal connection between the true conclusion and the evidence. But there is still something problematic about the verdict — the jury would have formed the same conclusion if it had been one of the hundreds of other identical-looking cars on the video. The fortuitous circumstances — that it just happened to be the defendant’s cab — render the relationship between the conclusion’s truth and justification accidental in a way that undermines knowledge. It also […] makes the verdict problematic.
So far, so good. But what feature does Pardo take to explain the verdict’s problematicalness?:
The verdict is problematic primarily because it is unsafe — that is, in a number of similar possible worlds the jury would have reached the same result and been in error.2 (n. 61)
Pardo thus evidently operates with an account of (legal) knowledge on which safety is a necessary condition.3
1.2 Let us put Pardo’s notion of unsafety on display alongside a cognate notion of safety. First, we can extract a general notion (pertaining to a subject S’s actual belief in a proposition p):
UNSAFETY*: In a number of similar possible worlds, S would have believed p, and been in error.
SAFETY*: It is not the case that in a number of similar possible worlds, S would have believed p, and been in error.4
For familiar reasons,5 these theses need to be refined as follows:
UNSAFETY: In a number of similar possible worlds, S would have believed p on the same basis, and been in error.
SAFETY: It is not the case that in a number of similar possible worlds, S would have believed p on the same basis, and been in error.
Second, to focus the discussion, let us put Pardo’s restricted notion of jury unsafety on display alongside a cognate notion of jury safety:
JURY UNSAFETY: In a number of similar possible worlds the jury would have reached the same result on the same basis, and been in error.
JURY SAFETY: It is not the case that in a number of similar possible worlds the jury would have reached the same result on the same basis, and been in error.6
Given certain plausible assumptions — notably, a jury forming a belief in the relevant legal proposition semantically expressed by its verdict7 — a counterexample to jury safety as a necessary condition on (legal) knowledge serves as a counterexample to safety as a necessary condition on knowledge.
1.3 We can see the intuitive appeal of Pardo’s analysis by ‘tweaking’ the facts of Fake Cabs. Consider:
Fake Cabs*: As Fake Cabs, but the video camera filming the accident is operated by Sam Spade, a private investigator, hired to investigate the defendant’s conduct, with perfect discriminative abilities in picking out the defendant’s cab.8
This tweak now makes the jury’s verdict safe: now it is not the case that in a number of similar possible worlds — say, worlds in which the defendant’s cab hits the plaintiff at a slightly different angle and/or velocity — the jury’s finding for the plaintiff would be in error. And now (modulo no other knowledge-thwarting luck in the offing) the problematicalness of the verdict vanishes and it becomes a candidate for knowledge. We can play out this tweak — namely, rendering an unsafe verdict safe — in a number of similar cases to comparable effect. So I take the point to generalize. Pardo’s explanation of the problematicalness of the verdict in Fake Cabs seems on the right lines, and safety seems a good candidate to be a necessary condition on (legal) knowledge.
2. Legal Counterexample to Safety as a Necessary Condition on Knowledge
2.1 Consider:
Insecure Mafia: The chief prosecution witness, Amoruso, an honest and reliable citizen, is ready truthfully to provide damning evidence against the guilty defendant, Baggio, a member of the Mafia — such evidence guaranteeing a guilty verdict. However, Insecure Mafia, the Mafia’s rival, has reason to believe that Baggio’s brother, Carbone, may be a member of the jury and is certain that Carbone, if on the jury and presented with damning evidence against Baggio, will successfully obstruct a guilty verdict. So as not to lose face (by dint of a rival mafioso going scot-free and being seen to do so), Insecure Mafia devises the following plan: should Carbone appear on the jury on the first day of trial, Insecure Mafia will overnight switch Baggio for Twin Baggio (an unrelated exact lookalike of Baggio) — Twin Baggio becomes the defendant — and will inform only Carbone of the switch (who is no longer motivated to obstruct the guilty verdict). Baggio has no incentive to reveal the switch, and Twin Baggio is paid handsomely by Insecure Mafia in return for his silence. Amoruso would then, unawares, untruthfully provide the same damning evidence against Twin Baggio, thereby guaranteeing an unjust guilty verdict against Twin Baggio and saving Insecure Mafia’s face — at least it would seem like a rival mafioso had been sent down. Carbone, however, is not on the jury. At the final hurdle in the tests for jury membership — Carbone having passed ninety-nine of the one hundred tests — it emerges, quite by chance, that Carbone is Baggio’s brother, and Carbone is dismissed. So the plan is not initiated, and the case proceeds normally. Amoruso provides true damning evidence against Baggio, and the jury convicts Baggio on the basis of Amoruso’s testimony.9
Plausibly, in this case, the jury knows that the defendant is guilty10 — therefore, modulo Pardo’s central claim, the goal or aim of legal proof has been realized. And yet, plausibly, in a number of similar possible worlds the jury would have reached the same result on the same basis (Amoruso’s testimony), and would have been in error. That is, in this case the jury knows that the defendant is guilty, but its verdict is not jury-safe. The jury has unsafe (legal) knowledge.
2.2 Objection (premise 1): the jury knows that Baggio is guilty (by Amoruso’s testimony). And this knowledge is safe — the jury would not easily have been in error. However, according to this objector (premise 2), the jury does not know that Baggio is the defendant; modulo the details of Insecure Mafia, the jury would easily have been in error, and if the man in the dock — the defendant — had not been Baggio, the jury would still have believed he was Baggio, and so on. Therefore, inductively (conclusion), the jury does not, after all, know that the defendant is guilty — no unsafe (legal) knowledge.11
Reply: the proponent of Insecure Mafia should deny premise 2 and affirm that the jury does (unsafely) know that Baggio is the defendant. How so? The proponent must, plausibly, say it is a perceptual-cum-testimonial analogue of the jury’s (unsafe) knowledge that the defendant is guilty in Insecure Mafia itself.12
3. Conclusion
3.1 This all suggests the following trilemma for Pardo (and how intuitively problematic he views the outcome in Insecure Mafia will have a bearing on which limb he takes and in what fashion): (1) retain his central claim that (safe) knowledge13 is the goal or aim of legal proof and deny that Insecure Mafia is a case of knowledge;14 (2) retain his central claim that (safe) knowledge is the goal or aim of legal proof and deny that Insecure Mafia is a case of unsafety;15 or (3) withdraw his central claim that (safe) knowledge is the goal or aim of legal proof. Each option has its drawbacks.16
1 As Pardo (57) notes, this claim of his takes place against the backdrop of a vast epistemological literature devoted to analyzing whether, and in what respect, knowledge is more valuable than less epistemically demanding achievements. For a comprehensive overview, see Pritchard and Turri (2007/2012). For an illuminating recent proposal, see Goldman and Olsson (2009). Finally, while Pardo (46) notes interesting issues about the relationship between systemic and case-specific epistemic considerations, Pardo’s principal focus—and I follow suit—is on specific cases.
2 Plausibly, the verdict is also (what might be called) insensitive.
3 Now that Fake Cabs has been introduced and preliminarily grappled with, note that the case from which Fake Cabs is derived — Fake Barns — originates in Goldman (1976) (with Goldman crediting Carl Ginet with the case). Fake Cabs is the first of several thought experiments in this chapter. The success of the thought experiments does not depend on their being empirically likely. (But consider the more prosaic presentation in Duff, Farmer, Marshall and Tadros (2007: 91), of a case of a defendant convicted accurately on the basis of evidence that though apparently sufficient at trial, turned out to be tainted or unreliable. Such cases that bear similarities to Fake Cabs are by no means far-fetched.) Rather, by abstracting from the untidiness of real-life cases, they serve interestingly to put pressure on concepts germane to the project of inquiry and to elicit intuitive judgments thereon. The sharpened conceptual awareness arising from consideration of these thought experiments can then serve to hone normative claims about the goal or aim of legal proof in verdicts in concrete legal systems.
In Fake Cabs the relationship between truth and justification is accidental in a knowledge-undermining way. Following the seminal paper by Gettier (1963), we can refer to such verdicts as being Gettierised. (Importantly, my thought experiment in section 2, Insecure Mafia, is not a Gettierised verdict; it is central to Insecure Mafia that knowledge is not undermined.) Objection: Pardo’s claim that legal verdicts ought to be non-Gettierised is utopian: with sufficient ingenuity we can ‘tweak’ the facts of any otherwise unproblematic verdict to Gettierise it. Reply: Pardo is claiming that legal verdicts ought to be non-Gettierised; not (contra the objection) non-Gettierisable — where non-Gettierisable verdicts are verdicts with a guarantee that the facts cannot be tweaked to Gettierise them.
4 Safety has received many different formulations — and defenses as a necessary condition on knowledge — in the epistemological literature; cf., notably, Williamson (2000: ch.5); and Sosa (2002). Pardo’s ‘in a number of similar possible worlds’ locution is regrettably imprecise. We might read ‘a number of’ as ‘some’. If we do so, Pardo’s SAFETY* would be equivalent to: In no similar possible worlds is it the case that S would have believed p, and been in error. The formulations extracted from Pardo are not ideal, but aim to stay close to Pardo’s n. 61.
5 See the famous ‘granny’ case in Nozick (1981).
6 One could construct like restricted notions for other legal fact finders.
7 I prescind from the difficult question of what is it for a jury (a group of minded individuals) to form a belief (whether it is, for example, just a matter of all — or a majority of — the individuals forming beliefs with the same content).
8 Similar tweaks are given to Fake Barns in the epistemological literature — e.g., cloud occlusion of the fake barns, binoculars trained rigidly on the real barn, etc.
9 Insecure Mafia is inspired by Comesaña’s (2005) Halloween Party case (though there are salient differences). Neta and Rohrbaugh (2004) also concoct two putative cases of unsafe knowledge, one of which involves a lottery. One may construct (more streamlined) variations on Insecure Mafia involving, for example, lotteries (e.g., where the switch is dependent on a lottery result); but my sense is that lottery cases lend an otherworldly nature to these cases. In any event, cases of this general form seem capable of being invented.
10 How does Insecure Mafia (knowledge) differ from Fake Cabs (non-knowledge)? One idea is that the threat to knowledge in Fake Cabs is actual — there really are fake cabs around the video camera — whereas the threat in Insecure Mafia is purely counterfactual — Carbone is not on the jury.
11 Consider: there is an opaque box into which I shall put either a frog, Kermit, or a donkey, Eeyore. I introduce you (an honest and reliable citizen) and a third party to Kermit but not to Eeyore. You think that I shall put Kermit in the box because I tell you, but only you, that I will. As it happens, I flip a coin to decide which animal to put in the box. The coin lands heads up, meaning, suppose, that I do in fact put Kermit in the box. Now you, pointing at the box, say to the third party, ‘The animal in this box is smaller than a bread bin.’ Does the third party know this claim? Objection (premise 1’): the third party does safely know that Kermit is smaller than a bread bin (by perception and my testimony). However (premise 2’), the third party does not know that Kermit is the animal in the box. So, inductively (conclusion’), the third party does not know that the animal in the box is smaller than a bread bin. No unsafe knowledge. And, the objector would press, Insecure Mafia is relevantly similar (though obviously disanalogous in some ways): ‘I/my’ is Insecure Mafia; Kermit is Baggio; Eeyore is Twin Baggio; ‘you’ is Amoruso; the third party is the jury; the coin flip is the jury tests; the animal in the box is the defendant; and ‘is smaller than a bread bin’ is ‘is guilty’.
12 Reply to n. 11’s objection: the proponent of Insecure Mafia should deny that n. 11’s case is relevantly similar to Insecure Mafia. Premise 2’ is true, but only because — unlike, mutatis mutandis, Insecure Mafia — the third party receives no knowledge-conferring evidence that Kermit is the animal in the box. If the third party does receive such knowledge-conferring evidence (e.g., testimony from you after my coin flip), the cases become relevantly similar, but premise 2’ becomes false.
13 By ‘(safe) knowledge’ I mean knowledge on which safety is a necessary condition.
14 In Chapter Eight, I proposed a different safety condition which could be used to argue that Insecure Mafia is a case of unsafe non-knowledge (see section 2.7 of Chapter Eight especially). (Perhaps, though, plausibly, for a world to be relevantly similar, Carbone needs to be dismissed — as in the actual world, according to Insecure Mafia. One might analogise the facts leading to Carbone’s dismissal with the element of prior decision in Halloween Party (see section 2.5 of Chapter Eight), such that that-Carbone-is-dismissed is a relevantly-safe condition. I reject the analogy: the facts leading to Carbone’s dismissal are designed to be ‘luck-infected’ in a way that the prior decision in Halloween Party is not.)
15 This option involves constructing a (plausible and motivated) modified safety condition on which the jury’s verdict in Insecure Mafia comes out safe. For an explicitly time-sensitive safety condition that could be put to such use, cf. Sainsbury (1997); Peacocke (1999: 310–28); and Williamson (2000: 124).
16 Pardo (2011) has replied to me.