Wednesday, August 16, 2017

Can Lower Courts Identify Scrivener's Errors on Remand in Higher Courts' Opinions? A Thought on Judge Posner in Sears (Now with a Postscript on Inferences from Denials of Rehearing)

On Monday Judge Posner issued a short opinion on fees in In re Sears, Roebuck & Co. Front-Loading Washer Products Liability Litigation, a case that attained a certain notoriety at the class-certification-appeal stage for Judge Posner's generous treatment of Rule 23(b)(3)'s predominance requirement and doubling down on the propriety of that treatment after his first opinion was GVR'd in light of Comcast Corp. v. Behrend—a decision which, in light of Comcast's subsequent interment in Tyson Foods and the Court's ultimate denial of cert in Sears on further petition from Posner's disposition of the GVR, seems wise.  

However, it also seems that, as ballyhooed as Sears was as a matter of class-action doctrine, it wasn't such a great case; plaintiffs will only receive a maximum of $900,000 from the case's settlement.  (Correction: I am reliably advised that the $900,000 was awarded to a smaller or lower-damages class, which complained of a defect in the Sears washers' control units, while the larger or more seriously aggrieved moldy-washer class, whose counsel's fees were not at issue in this appeal, received at least $12 million in settlement.  Both classes were at issue in Judge Posner's prior opinions on certification.  While Posner mentions the two separate classes in his opinion, one would never know from reading it that the control-unit class, rather than both classes or the moldy-washer class, was the one that got only $900,000 in settlement.)  Nevertheless, between the multiple trips to the Seventh Circuit and Supreme Court and the rather involved certification proceedings, class counsel, quoting Posner, "claimed to have incurred $3.16 million in fees . . . [and] subsequently increased their base fee estimate to $3.25 million, having discovered additional billable time."  They also requested that their actual fees be multiplied by 1.85 to 1.9 "to account for what they claimed to be their extraordinary effort[.]"  The district court found, Posner says, "that they were entitled to a base fee of only $2,726,191, which the court multiplied by 1.75, making the total fee award $4,770,834."

Now, Posner will decide that a multiplier (which is apparently quite common) isn't called for here, for reasons which I must say are quite unclear other than his unembroidered ipse dixit that class counsel "failed to prove that a reasonable fee would exceed . . . the pre-multiplier figure sought by class counsel."  (Judge Posner does, to be fair, tersely note earlier that the district court concluded that the case "wasn't very complex," a "conclusion [which] leaves us puzzled about the court's decision nevertheless to allow a multiplier.")  However, a lack of reasoning is far from the biggest problem with this opinion.  Here are its last three sentences:
The presumption [that "fees that exceed the recovery to the class are unreasonable"] is not irrebuttable, however, and in this case the extensive time and effort that class counsel had devoted to a difficult [AS: but not very complex?] case against a powerful corporation entitled them to a fee in excess of the benefits to the class. But they failed to prove that a reasonable fee would exceed $2.7 million—the pre-multiplier figure sought by class counsel and already thrice the damages awarded the class. We therefore reverse the judgment of the district court and remand with directions to award $2.7 million—no more, no less—in fees to the class counsel.
Two things have gone awry here, the second of which is an apparent scrivener's error in what Judge Posner is ordering the district court to do on remand.  The first is the claim that "2.7 million [was] the pre-multiplier figure sought by class counsel."  It's not; they initially sought $3.16 million in pre-multiplier fees and ultimately asked for $3.25 million in pre-multiplier fees, at least if Posner's earlier statement of the facts is to be believed.  $2.7 million—actually, $2,726,191—is what the district court found they were entitled to in pre-multiplier fees.  

The second and far more consequential mistake is the "remand with directions to award $2.7 million—no more, no less—in fees to the class counsel."  The district court found that class counsel was entitled to $2,726,191 in actual pre-multiplier fees, which is undeniably "more" ($26,191 more) than $2.7 million.  The opinion nowhere suggests that that finding was erroneous, clearly or otherwise, and the whole thrust of the opinion is that class counsel should get its pre-multiplier fees, whatever they are.  Yet the district court is ordered, pretty unambiguously, I would say, to reduce those pre-multiplier fees by $26,191 to $2,700,000.  Perhaps rounded numbers can be used as shorthand for unrounded numbers elsewhere discussed, but when you "remand with directions to award [a particular round number] no more, no less," it's hard to say that the round number is shorthand for a number $26,191, or roughly 1%, greater.

If class counsel are reading this, I think a petition for panel rehearing would obviously be a very good idea, but the more interesting question is whether the district court would be bound on remand to award $2,700,000 in the event rehearing isn't sought or granted.  (See note on the significance of a hypothetical denial of rehearing.*)  I think the answer is no; I can't see a reason why the district court shouldn't be allowed to detect scrivener's errors in decretal language in the same way that it could detect them in statutes, which are legal commands that must be followed just as much as a court of appeals or Supreme Court mandate is. 

The point, however, isn't an uncontroversial one.  For example, in Johnson v. Williams, the Supreme Court held that deferential AEDPA review applied to a particular habeas claim and remanded for further proceedings, presumably under that standard, consistent with its opinion.  However, at the top of its opinion, it carelessly and inconsistently wrote that "we hold . . . that the restrictive standard of review set out in § 2254(d)(2) . . . applies, and that under that standard respondent is not entitled to habeas relief"—a point which the rest of the opinion never addressed.  The Ninth Circuit (yes, that Ninth Circuit), including Judge Reinhardt (yes, that Judge Reinhardt), though well aware that this was almost certainly a mistake, felt constrained on remand "to follow the mandate of the Supreme Court," internally inconsistent and plainly marred by sloppy drafting though it was, and deny habeas relief.  It would take a subsequent GVR in which the Court tacitly admitted it messed up for the Ninth Circuit to adjudicate the prisoner's habeas claim under the standard the Court previously told the Ninth Circuit to adjudicate it under.

My cursory research on reading opinions to contain scrivener's errors shows that district courts, in reviewing ALJ opinions (or other agency opinions), identify scrivener's errors fairly often, that courts of appeals occasionally identify scrivener's errors in district-court opinions or sentences, but that, perhaps because of the constraints of the mandate rule and lower courts' reticence to criticize their superiors, lower courts hardly ever identify mistakes in higher courts' decisions on remand, or in higher courts' decisions more broadly.  The one instance I've found of a district court identifying a mistake in one of its circuit's opinions, though not one entered in the case before it, is in this old unpublished opinion by Judge Kimba Wood, which unfortunately isn't on Google Scholar or Courtlistener's new and purportedly comprehensive database of every opinion on Pacer.  The particular mistake she caught rather puts one in mind of Justice Scalia's concurring opinion in Bock Laundry (see my previous post on mistake and scrivener's error):
Defendants understandably retreat from this interpretation. They ultimately characterize 181 East 73rd Street as holding that Standard 2 should apply to both plaintiffs and defendants, whether they are sponsors or tenants. The support for this view is in the last sentence of the text of 181 East 73rd Street. “And in this case, congressional intent and the unambiguous statutory language are in harmony—a party may recover attorneys' fees under the Abuse Relief Act only if the suit is lacking in substantial merit.” 954 F.2d at 51. The word “party” in this sentence suggests that Standard 2 applies to plaintiffs and defendants, whether they be tenants or sponsors. Taken literally, the sentence implies that the unambiguous statutory language of [15 U.S.C] § 3611(d) does not single out defendants for special treatment with respect to attorneys' fees, despite the fact that one of the two sentences in § 3611(d) states: “A defendant may recover reasonable attorneys' fees if the court determines that the cause of action filed by the plaintiff is frivolous, malicious, or lacking in substantial merit.” Because that is an absurd result that finds no support in the rest of the Second Circuit's opinion, I can only assume that the term “party” was placed in that sentence by mistake, instead of the word “defendant.”
305 E. 24th Owners Corp. v. Parman Co., No. 85 CIV. 3788 (KMW), 1992 WL 209292, at *2 (S.D.N.Y. Aug. 18, 1992) (emphasis added).  The Second Circuit would correct the mistake in its prior opinion rather more obliquely on appeal.

A postscript on rehearing

* Should the Seventh Circuit deny panel rehearing, that might seem a signal that for some unexpressed reason, the panel really thought the district court's finding on pre-multiplier fees was $26,191 off, or that three times what plaintiffs would receive was enough, 3.03 times was too much, and that class counsel would have to eat $26,191.  (Indeed, it's not completely impossible that Judge Posner was thinking the latter.)  

A lot depends, though, on the chariness with which rehearing is granted.  In Johnson, the Supreme Court denied rehearing, which the Ninth Circuit incorrectly took as a signal that the Court really meant its unexplained holding.  But after the Ninth Circuit wrote about how unexplained and inconsistent with the rest of Johnson that holding was and how compelled they were to follow it all the same, the Court quietly granted cert, vacated the Ninth Circuit's faithful application of their mandate, and remanded with directions to consider the petitioner's claim under AEDPA, thereby tacitly overruling their prior "holding."  The moral of that story seems to be that the Court doesn't grant rehearing to fix mistakes in its opinions, even if it really knows they're there.  And of course, this comes as no surprise; the Court virtually never grants rehearing.  The Court's reluctance to grant rehearing to correct mistakes, though, doesn't mean that its denials of rehearing clarify that its alleged mistakes aren't mistakes, or ratify the mistakes; quite the opposite.  It means when it denies rehearing, it isn't deciding whether its opinion is mistaken, but is deciding something else (whether, perhaps, the mistake is important enough to warrant the embarrassment of rehearing).  Since rehearing denials don't seem to tell us much about whether the Court meant what it appears to have accidentally said, a lower court is still free to determine, on my view, that it's clear from the entirety of the Court's opinion that it didn't mean to say what it did.

On the other hand, if a court usually grants rehearing when its opinions contain scrivener's errors, even if those errors are only important to the parties, the unexplained denial of rehearing probably does suggest that that court doesn't think its opinion contains a scrivener's error.  The best practice for appellate courts, however, would be to explain why they're denying a scrivener's-error-based rehearing petition so that it's clear to lower courts that the appellate court has rejected the claim of scrivener's error, rather than merely having elected not to address it for reasons besides its merits.  At the least, appellate courts should be more transparent about their rehearing standards in this regard so that reasonable inferences can be drawn from unexplained denials, instead of shaky inferences from a set of brute results on rehearing petitions.  For example, if the Seventh Circuit were to deny rehearing in Sears without explanation, a lower court would probably have to look at a large set of rehearing denials and grants to determine whether the Seventh Circuit regularly denies rehearing petitions presenting valid claims of scrivener's error, or whether a denial truly signals a merits rejection of such a claim given that all the good ones garner rehearing.

Sunday, August 13, 2017

Supreme Court 2016 Term in Review: (Apparent) Mistake, Another Textualist Possibility Unexplored in SW General

1.  Linguistic mistakes vs. linguistic accidents.

Have you ever said to someone, "I don't think that means what you think it means," or more declaratively, "that doesn't mean what you think it means"?  If so, you're familiar with the concept of linguistic mistake, as distinct from linguistic accident, or what lawyers call scrivener's or drafting error.  When you tell someone that what they said doesn't mean what they think it means, what you mean is that you think they intended to say precisely the words they said, but that, because they seem to intend to convey something other than what they said means, they are likely mistaken about the meaning of the words they said.

For example, if a wise but unpolished student writes on an exam, "the Court shouldn't of avoided the Chevron question in Esquivel-Quintana, that was a really lame move," it is possible that the student intended both to use a semicolon and write "shouldn't have" and accidentally wrote "shouldn't of" and used a comma in a hurry.  But it is just as possible and probably more likely that this student intentionally wrote "shouldn't of" because he mistakenly believes, as many people do, that "shouldn't of" is a grammatical construction that means "shouldn't have," and also quite likely that the student intentionally used a comma because he mistakenly believes that commas can link independent clauses that can only be linked by semicolons.  When Congress makes that sort of an error, it hasn't engaged in scrivener's error, properly understood.  The metaphorical scrivener's metaphorical pen didn't slip; the text Congress enacted was the text it meant to enact.  Congress was just mistaken about the meaning of the words it deliberately used.

On the other hand, if your boss (or Congress) tells you to file an appeal in no less than seven days, you won't think he meant to say what he said, or is mistaken about the meaning of what he said; you'll likely believe he meant to say "more" and accidentally said less.  You won't think he mistakenly believes that "less" means more, since no one fully conversant in English does.  When Congress makes this sort of an error, it has engaged in scrivener's error.  The scrivener's pen did slip, those members of Congress and staffers who read the text didn't catch it, and Congress voted for and enacted a text other than the text it intended to enact.

2.  Thompson's understanding of "notwithstanding" as mistake rather than alternative use.

Readers of my last post will recall that: (1) in 1998, Senator Fred Thompson introduced on the Senate floor the statute that was the subject of SW General, having "authored" that statute; (2) that the statute provided for three different ways of becoming an acting officer in subsections (a)(1) through (a)(3), and provided in subsection (b)(1) that "notwithstanding subsection (a)(1)" an acting officer with less than ninety days of prior service as first assistant to his present office couldn't continue to serve as acting officer if he were nominated to permanently fill that office; and (3) that in his remarks, Thompson said in the clearest terms that the phrase "notwithstanding subsection (a)(1)" meant that (b)(1)'s prohibition only applied to acting officers who took office under (a)(1), as if the word "notwithstanding" meant "but as to," rather than the standard meaning of "despite."  In my post, I argued that this might show that "notwithstanding" had an alternate meaning of "but as to" and was more ambiguous than it appeared.

Of course, it can't be the case that just because one Senator, even a statute's authoring Senator, thinks a word in a statute means or can mean x, it can mean x.  Were that so, no Senator could ever be mistaken about what words mean, and we certainly need some account of senatorial linguistic mistake.  

In my post, I readily conceded that if Thompson were the only person in the world with his understanding of "notwithstanding," he would be wrong, and claims that "notwithstanding" were ambiguous in virtue of his understanding would be unwarranted.  I'm probably willing to go much farther than that.  I do think it's pedantic to say that "irregardless" unambiguously means "not regardless" or is gibberish; there are simply too many people who think otherwise.  But I am not quite so unpedantic to say that "principle" is ambiguous between the meanings of "precept" and "primary" just because many people believe it's correct to say "principle objective" instead of "principal objective."  People who say "principle objective," I am inclined to claim, are mistaken about what "principle" means, not early adopters of an alternative usage.  So too may Senator Thompson have been mistaken about what "notwithstanding" means.  Indeed, while careful textualists must investigate the possibility that his understanding reflects a lesser-known but surprisingly prevalent alternative usage, I think it's likely the case that such an investigation would reveal he was mistaken (though I can't, at this time, draw the line for you between common mistakes and alternative meanings with anything approaching precision.).

3. Textualism as objectified intentionalism and the textualist understanding of mistakes.

a. theory 

Where does that leave a textualist?  With great respect to simple-minded textualists, some of whom are quite sophisticated, the simple-minded textualist will say (as the Court said) that once we've determined "notwithstanding" isn't ambiguous we take our marbles and go home.  A statute's meaning is its text's ordinary meaning—subject to the textualists' caveats about context (which in any event can't mutate ordinary meaning but only clarify it), absurdity, and screamingly obvious scrivener's error—and "notwithstanding" has only got one, so there we are.  Intentions are for the intentionalist buzzards to pick over, and what did Fred Thompson—bless his constitutionalist heart!—really know about English anyway?

With respect, as I say, to simple-minded textualists, that's not really what textualism is about.  To sophisticated textualists, textualism isn't about ordinary meaning; it's about, to quote John Manning approvingly quoting Justice Scalia, "'objectified intent'—the intent that a reasonable person would gather from the text of the law.Or to quote Judge Easterbrook approvingly quoting Justice Holmes in a proto-textualist mood, "we ask, not what this man meant, but what those words would mean in the mouth of a normal speaker of English."  That is, if a normal speaker of English wrote a given statute, what would you think he was trying to say?  

It's because textualism is a search for apparent intent on the face of the text that we textualists do much of what we do.  It's why, when a clause is ambiguous, we give it a non-redundant meaning instead of a redundant one—because we believe normal speakers don't mean to speak redundantly.  It's why, when Congress mentions one thing and not something else that's related, we read it to implicitly exclude the something else—because we imagine that Congress would have mentioned it if it didn't intend to exclude itIt's why, when a word in a list is ambiguous between a meaning that clashes with the rest of the list and a meaning that coalesces with it, we give it the latter meaning (think "pins, tacks, and nails")—because we assume Congress intends to stay on topic, rather than include wildly divergent things in the same list.  None of these moves has anything to do with semantic meaning or an ordinary meaning that hasn't been enriched by apparent intention.  The ordinary or semantic meaning of "nails" doesn't somehow contract to metal nails when it's placed next to "pins" and "tacks"; semantically, it's just as bivalent as ever.  We read "nails" to mean metal nails there because ordinary people mean or intend to convey the idea of metal nails when they talk about nails in the context of pins and tacks. 

Now, one thing we sometimes do when we are trying to determine what people appear to mean by what they say is to observe that what they appear to have meant to say isn't what they said and deduce that they're probably mistaken about what they said means.  As I said at the beginning of this post, we all have told someone that "that doesn't mean what you think it means."  If someone writes about their "principle objective," indeed if Congress writes about a "principle objective," we will not only take them to mean "principal objective" but act on that understanding, even though it is just barely possible that someone might use "principle," in its correct sense, adjectivally and mean something like "an objective concerning principles."  You could imagine, for example, a popular religious figure punningly writing that "your principle objective should be to follow the Golden Rule."  This small window of uncertainty aside, no reader of "principle objective" in a statute, however scrupulously textualist, would read it to actually mean "principle objective" because it isn't absolutely certain that Congress really intended to convey "principal objective."  Someone who writes "principle objective" appears, far more likely than not, to be mistaken about what "principle" means and to intend to convey "principal objective." 

Moreover, we don't only identify linguistic mistake when it's near-certain.  Recall, from the last post, Justice Kagan's hypothetical about someone who orders a salad, steak and fruit cup, and says that notwithstanding the order of the fruit cup, they don't want any strawberries in their meal.  This hypothetical, she reasoned, shows that "notwithstanding" means "despite (the fruit cup)," not "but as to only (the fruit cup)."  But we can imagine a similar hypothetical that would lead us to form very different intuitions about what the customer thinks "notwithstanding" means.  Suppose he ordered an iced tea, oysters and steak, and said that notwithstanding the order of oysters, he positively dislikes and wants no ice.  Does he really mean that despite the oysters, which often come on ice, he wants no ice in anything he gets, including his iced tea?  Or does he mistakenly believe "notwithstanding" means "but as to" and mean that as to the oysters he wants no ice?  (Also, by the way, note the possibility for domain restriction here, as discussed in my post on Advocate General.)

The mistake, I think, seems more probable, though we can't be certain or even overwhelmingly confident; perhaps he loathes ice in all settings and wants a cold tea, which needn't be on ice to have been "iced."  And if he ordered a soda, which usually comes out of fountains cold without adding ice, global icelessness is a perfectly possible interpretation of what he's said.  But even here, I think it more likely than not that he means only to hold the ice on his oysters, especially because he's mentioned the oysters and not the more obvious candidate for containing ice, the soda.  We can reasonably infer that the customer, more likely than not, mistakenly believes "notwithstanding" means "but as to," and should interpret his order as if he said "but as to" (unless one wants to run the risk of losing a tip by implying he doesn't know what notwithstanding means in an attempt at clarification).

As textualism is a game of identifying objectified intent, not ordinary meaning, which it will often track but which it will frequently enrich and from which it will occasionally deviate, textualism, like taking restaurant orders, is sometimes a game of identifying objectified mistake—a seeming misunderstanding on the part of Congress of what its words mean, judging by the text, and a seeming intention to convey something other than what it said, again judging by the text.  (Leave aside whether Congress factually was mistaken; that truly is an inquiry for the intentionalists.)  This is a game that textualists actually and not just theoretically play.  

b. practice

For example, in Bock Laundry, Federal Rule of Evidence 609, which Congress (not the Advisory Committee on the Federal Rules of Evidence) drafted, said that you could generally impeach witnesses with prior felonies; as to a "defendant," though, their probative value had to outweigh their prejudicial effect.  It certainly didn't seem as if Congress could have really meant to say that civil defendants, but not civil plaintiffs, received the benefit of prejudice-weighing.  That was, though, the ordinary meaning of what they said.  

Justice Scalia, acknowledging that "defendant" literally meant civil and criminal defendant alike, thought that Congress appeared to have meant something else.  Critically, he disclaimed any interest in "which meaning can be shown to have been understood by a larger handful of the Members of Congress" as a matter of legislative-historical fact.  Rather, he wanted to know "which meaning is most in accord with context and ordinary usage, and thus most likely to have been understood by the whole Congress which voted on the words of the statute (not to mention the citizens subject to it)."  That is to say, he wanted to know what, on the face of the text, Congress appeared to have most likely intended and what "the citizens" would believe Congress most likely intended—which is to say, he wanted to know what Congress objectively appeared to have mistakenly believed "defendant" to mean.

Posing that question, Justice Scalia concluded that Congress appeared to have mistakenly believed "defendant" to mean "criminal defendant" (a mistake, he noted, that is sometimes made "in normal conversation"), and on that ground interpreted "defendant" to mean "criminal defendant" rather than its literal meaning.  Scalia, admittedly, isn't quite as clear as I hope I've been that he's reasoning in terms of linguistic mistake, but my reading of his opinion is not an original or idiosyncratic one.

c. Confidence levels for mistakes vs. accidents

Before finally turning back to SW General, a word on how confident a textualist must be to make the objective-mistake move.  Textualists generally insist on certainty before identifying scrivener's error, and though some (one?) textualists think that's wrong, we're right to do so.  Scrivener's error, a doctrine about when Congress accidentally enacted the wrong text in some detail, must be clear lest textualism collapse into intentionalism; if we went around asking, as that one textualist has suggested we should, whether Congress more likely than not intended to write one thing when it said another, we'd just be asking what rules Congress intended apart from whatever it said.  We might not be purposivists—people who ask what purposes Congress had and construe statutes accordingly—but it's hard to see how we'd be textualists if we'd discard the text whenever there seemed to be a 51% chance that Congress meant to enact a different one.  

Further, because scrivener's error is a matter of legislative-historical fact, for the reason that textualists do care about actual subjective legislative intent to the minimal extent that they care whether Congress really intended to enact the bill they enacted, the search for scrivener's error on a more-likely-than-not standard would not be objective, but just as subjective and immersed in legislative history as the intentionalist interpretive enterprise generally.  Indeed, barring distinctions too numinous for all but a handful of judges to draw, it would simply be the intentionalist interpretive enterprise.

Linguistic mistakes, however, are quite another matter.  When Congress makes or seems to make a linguistic mistake, it hasn't accidentally enacted the wrong text; it's intentionally picked the words it picked because it's mistaken about what they mean and thinks they mean something else.  To read a text as linguistically mistaken isn't to discard the text, but to obey its apparent intent, which for a conceptualy rigorous textualist is what textual meaning is.  To follow a text's literal meaning when there appears on the face of the text to be a 70% chance that Congress mistakenly thought a word meant something other than what it means is to disobey objectified intent and to deviate, importantly, from what ordinary citizen-readers will take the statute to mean.

4. Mistake in SW General, and the role of actual mistakes in identifying objectified mistakes.

It is, then, a conceptual possibility that the correct textualist reading of "notwithstanding subsection (a)(1)" in SW General is "but as to subsection (a)(1)" because Congress appears, on the face of the text, to have mistakenly believed, more likely than not, that "notwithstanding" meant "but as to."  What textual indications are there, if any, that Congress "had" (that is, appears to have had, not actually had) this mistaken belief?  

I don't think there are many, nor that those there are are especially strong, but there certainly are some.  The first is the at-least minor oddity that Congress bothered to write "notwithstanding subsection (a)(1)" at all.  Conceiving of Congress, as Holmes and Easterbrook would, as a normal speaker of English, we tend to assume that Congress doesn't sprinkle cross-references into statutes unless they do some work.  What work does "notwithstanding subsection (a)(1)" do if it doesn't limit what follows to an exception to (a)(1)?  If you'll recall, (a)(1) through (a)(3) are all paths to becoming acting officers; (b)(1) conflicts with all three because it cuts acting officers' service short.  And if (b)(1) cuts (a)(2) and (a)(3) short without the help of "notwithstanding (a)(2) and (a)(3)," as the Court held, then why was it necessary to say that (b)(1) applied "notwithstanding subsection (a)(1)"?

The Court suggested that the conflict between (b)(1) and (a)(1) was especially aggravated; since (a)(1) made certain officials acting officers by operation of law and said they "shall" serve as acting officers, while (b)(1) said they "may not" continue to serve in an acting role once nominated to serve permanently, Congress would especially want to clarify that (b)(1) trumped (a)(1).  Subsections (a)(2) and (a)(3) involved, on the other hand, permissive presidential appointment to acting status, which arguably poses less of a conflict with (b)(1)All that's arguable, certainly, but does it really explain why "notwithstanding subsection (a)(1)" was needed?  If it weren't there, would anyone doubt that a provision about who "may not" serve once nominated would cut short the service of those who "shall" initially serve upon a vacancy?  Consider, on the other hand, how vital "notwithstanding subsection (a)(1)" is if it really, as its drafter said, limits (b)(1) to an (a)(1) exception.

Justice Sotomayor's dissent argues that reading "notwithstanding" to mean "despite" rather than "but as to" poses another superfluity problem; I believe that one is easily parried.  But whatever conclusions one may draw about the strength of an apparent-linguistic-mistake reading of "notwithstanding," I want to close by addressing one final point, which is the relevance of the historical fact that Senator Thompson was mistaken about notwithstanding's meaning (assuming that his use isn't an acceptable alternative use).

Textualists are not concerned with actual intentions, only "objectified" or apparent ones.  Actual intentions are for intentionalists.  However, for a linguistic mistake to be apparent, it first must be possible.  We simply don't know anyone who mistakenly thinks that "less" means "more," so when Congress says "no less than seven days" and seems to have meant "no more," we don't say that Congress seems to have been mistaken about the meaning of "no less"; we say that Congress seems to have accidentally written "no less" and think of the problem in terms of scrivener's error, for which we require certainty.  Linguistic-mistake readings of statutes are only viable when the mistake in question is conceivable, and to be conceivable, it helps if we know of some cases where it's happened.  One reason that it's so easy to assume someone thinks "principle" means "principal" is that many people do.  And one reason Scalia was willing to read "defendant" to mean "criminal defendant" in Bock Laundry is that people sometimes mean defendant that way "in normal conversation."  Indeed, many non-lawyers may mistakenly believe that the only people called "defendants" are criminal defendants.  Some of them are probably in Congress.

Senator Thompson, then, is relevant to a claim that Congress's apparent intention in writing "notwithstanding subsection (a)(1)" was to communicate the idea "but as to subsection (a)(1)," not because he shows that it was Congress's intention, but because he shows that it could have been the intention of a normal speaker using Congress's wordsOnce you read Thompson confidently asserting that "notwithstanding subsection (a)(1)" meant that (b)(1) only limited (a)(1), not (a)(2) or (a)(3), it's no longer so hard to imagine a Holmesian normal speaker of English mistakenly thinking "notwithstanding" means "but as to."  Apparently, it happens.  

* Acknowledgements to Mitchell Berman's typology of legislative intent in "The Tragedy of Justice Scalia"—one of the few truly excellent and practically useful bits of interpretive scholarship in the last few yearsfor suggesting much of this post, though I dissent from his judgment of Scalia as a tragic figure.