In the
 United States Court of Appeals
               For the Seventh Circuit
                         ____________

Nos. 04-3622 & 04-3623
JOSEPH JASKOLSKI and NATIONAL
INSURANCE CRIME BUREAU,
                                            Plaintiffs-Appellees,

                                v.

RICK DANIELS, et al.,
                                    Defendants-Appellants.
                         ____________
       Appeals from the United States District Court for the
         Northern District of Indiana, Hammond Division.
             No. 2:03-CV-479—Rudy Lozano, Judge.
                         ____________
 ARGUED SEPTEMBER 14, 2005—DECIDED OCTOBER 21, 2005
                   ____________


  Before EASTERBROOK, ROVNER, and SYKES, Circuit Judges.
  EASTERBROOK, Circuit Judge. Joseph Jaskolski assisted
federal prosecutors in an investigation that led to the
indictment of Rick Daniels and three of his relatives for
insurance fraud. After the defendants (collectively
“Daniels”) were acquitted, they sued Jaskolski and his
employer, the National Insurance Crime Bureau, in state
court, charging them with the tort of malicious prosecution.
During discovery Daniels sought documents that Jaskolski
deemed to be grand jury materials protected from disclosure
by Fed. R. Crim. P. 6(e). When the state judge sided with
Daniels and ordered Jaskolski to hand over everything
2                                  Nos. 04-3622 & 04-3623

plaintiffs wanted, Jaskolski and the Bureau filed this suit
in federal court seeking an injunction. District Judge
Lozano obliged and enjoined Daniels from pursuing discov-
ery in state court; instead they must turn to District Judge
Moody, who supervised the federal grand jury and under
the injunction has exclusive authority to decide which
materials in Jaskolski’s (and the Bureau’s) files will be
released to the plaintiffs in the tort litigation.
  In this court the parties have devoted their energies to
debating whether Jaskolski played the role of “government
personnel” in the criminal prosecution—for, if he did, then
he “must not disclose a matter occurring before the grand
jury”. Fed. R. Crim. P. 6(e)(2)(B). Many persons who learn
information about a criminal investigation are free to
disclose what they know, and “[n]o obligation of secrecy may
be imposed except in accordance with Rule 6(e)(2)(B).” That
subsection covers, among others, any “person to whom
disclosure is made under Rule 6(e)(3)(A)(ii) or (iii).” Rule
6(e)(2)(B)(vii). Rule 6(e)(3)(A)(ii) in turn refers to “any
government personnel—including those of a state, state
subdivision, Indian tribe, or foreign government—that an
attorney for the government considers necessary to assist in
performing that attorney’s duty to enforce federal criminal
law”.
  During the criminal investigation, an Assistant United
States Attorney concluded that Jaskolski’s assistance was
“necessary” and informed Judge Moody that Jaskolski
would be allowed access to some grand jury materials. If
Jaskolski served the investigation as “government person-
nel” then he is forbidden to disclose what he learned,
without the federal court’s approval. One appellate decision
holds, however, that investigators who work for the Insur-
ance Crime Bureau are not “government personnel” even if
a federal prosecutor supervises their activities. See United
States v. Tager, 

(10th Cir. 1980). Other
decisions are more favorable to the idea that private
Nos. 04-3622 & 04-3623                                       3

employees detailed to assist federal prosecutors are “gov-
ernment personnel” for that prosecution. See United States
v. Lartey, 

(2d Cir. 1983); United States v.
Benjamin, 

(9th Cir. 1988). The parties (and
the United States, appearing as amicus curiae) want us to
determine the proper classification of private insurance
investigators under Rule 6(e)(3)(A)(ii).
  Single-minded attention to the meaning of “government
personnel” has led the parties (and the district judge) to
slight antecedent questions, such as what this dispute is
doing in federal court. State judges manage discovery in
state litigation, and if federal law bears on that subject then
state judges apply the federal law. Jaskolski alleged that
federal jurisdiction exists under 28 U.S.C. §1331, which
says that district courts have “original jurisdiction of all
civil actions arising under the Constitution, laws, or
treaties of the United States.” What claim arises under the
Constitution, laws, or treaties? Jaskolski does not say. An
issue depends on federal law, but it is an issue in a pending
state case.
  Section 1331 does not permit a defendant in state litiga-
tion to obtain a federal court’s resolution of each federal
point that may crop up. Only when a well-pleaded com-
plaint poses a substantial federal issue does §1331 supply
jurisdiction. See Grable & Sons Metal Products, Inc. v.
Darue Engineering & Manufacturing, 

(2005)
(citing many predecessors). Otherwise the existence of a
federal defense (or indeed any federal issue, for Rule 6(e)
does not supply a “defense” to the claim of malicious
prosecution) would allow a new federal suit to be launched.
That assuredly is not the law. See, e.g., Gully v. First
National Bank, 

(1936); Taylor v. Anderson,

, 75-76 (1914). Issues that affect discovery but
not the substantive claim flunk the well-pleaded-complaint
doctrine and so do not come within §1331. What’s more, if
the presence of a federal issue in a state case permitted a
4                                    Nos. 04-3622 & 04-3623

separate suit under §1331, it also would allow removal
under 28 U.S.C. §1441(b), and that too assuredly is not the
law. See, e.g., Holmes Group, Inc. v. Vornado Air Circula-
tion Systems, Inc., 

, 830-32 (2002); Franchise
Tax Board v. Construction Laborers Vacation Trust, 

, 9-12 (1983); Chicago v. Comcast Cable Holdings,
L.L.C., 

(7th Cir. 2004).
  Although §1331 does not supply jurisdiction, 18 U.S.C.
§3231 does. That’s the statute providing jurisdiction over
federal criminal prosecutions. Questions about the propriety
of releasing grand jury materials for use in other litigation
(such as the suit Daniels had filed) come within the federal
criminal tribunal’s ancillary jurisdiction. See, e.g., United
States v. Baggot, 

(1983); McDonnell v. United
States, 

, 1247-48 (3d Cir. 1993); American
Friends Service Committee v. Webster, 

, 71-72
(D.C. Cir. 1983); Doe v. Rosenberry, 

(2d Cir.
1958) (L. Hand, J.). See also Charles Alan Wright, 1 Federal
Practice & Procedure §109 (3d ed. 1999). So a dispute of this
kind properly may come to federal court—but only because
of the federal grand jury’s role, not (as the parties supposed)
because federal courts resolve all disagreements about the
application of federal law.
  Notice the conditional phraseology: disputes of this kind
properly may be resolved in federal court. Other statutes
limit the federal judiciary’s role in particular controversies.
One of these is the Anti-Injunction Act, 28 U.S.C. §2283: “A
court of the United States may not grant an injunction to
stay proceedings in a State court except as expressly
authorized by Act of Congress, or where necessary in aid of
its jurisdiction, or to protect or effectuate its judgments.”
Jaskolski requested, and the district court issued, an
injunction that stays proceedings in a state court. The
district court did not find that any of the statutory excep-
tions is satisfied; indeed, the court did not mention the
statute. On appeal Jaskolski gives it a nod, while Daniels
Nos. 04-3622 & 04-3623                                     5

and the United States ignore §2283. This statute does not
affect federal subject-matter jurisdiction, so Daniels has
forfeited its benefit. (Because the dispute between Daniels
and Jaskolski is private litigation, we need not consider the
possibility of abstention under Younger v. Harris, 

(1971), and its successors, a doctrine that federal courts
may need to enforce to protect state sovereignty even if the
litigants are sleepwalking. See Ohio Civil Rights Commis-
sion v. Dayton Christian Schools, Inc., 

, 625-26
(1986); International College of Surgeons v. Chicago, 

, 361 n.4 (7th Cir. 1998).)
  Likewise Daniels has forfeited the benefit of issue
preclusion (collateral estoppel). The state court already has
decided the very issue that these parties presented to the
federal judge. Under 28 U.S.C. §1738 the state decision has
the same preclusive effect in federal court that it would
have in state court. We see no reason why only federal
courts would be competent to determine whether Jaskolski
acted as “government personnel”; certainly federal courts do
not have sole authority to determine whether evidence
already in Jaskolski’s hands represents “matters occurring
before the grand jury” (Daniels contends that it is not, and
the state judge apparently agreed). At all events, §1738
means that the state law of preclusion must be followed
even when federal jurisdiction over a particular subject is
exclusive. Marrese v. American Academy of Orthopaedic
Surgeons, 

(1985). Once again, however, the
parties have ignored this subject, and as preclusion is an
affirmative defense it has been forfeited. We therefore need
not explore what preclusive effect an order compelling the
production of documents in discovery has under Indiana
law.
  A reader who expects us to turn at last to the question
whether Jaskolski acted as “government personnel” in the
investigation will be disappointed, for that issue turns out
to be non-dispositive. An affirmative answer would resolve
6                                   Nos. 04-3622 & 04-3623

the dispute in Jaskolski’s favor—but a negative answer
does not lead to victory for Daniels, so we leave the question
for another case in which the resolution matters. Recall the
language of Rule 6(e)(2)(B): “[T]he following persons must
not disclose a matter occurring before the grand jury: . . .
(vii) a person to whom disclosure is made under Rule
6(e)(3)(A)(ii)”. Disclosure was made to Jaskolski under Rule
6(e)(3)(A)(ii). Whether the disclosure was made “properly”
or “correctly” is neither here nor there. Rule 6(e)(2)(B) asks
whether disclosure has been “made under” a particular
subsection, not whether the subsection was applied cor-
rectly. This protects the prosecutor’s (and the witnesses’)
reliance interests and prevents a blunder from opening the
investigatory files.
  Daniels contends that it would be “absurd” to read Rule
6(e)(2)(B) to bypass the question whether a given person
should have received the grand jury materials now in his
possession. He invokes the doctrine that judges avoid giving
statutes absurd readings, but he misunderstands its scope.
This doctrine does not license courts to improve statutes (or
rules) substantively, so that their outcomes accord more
closely with judicial beliefs about how matters ought to be
resolved. The Supreme Court made this point recently. “It
is beyond our province to rescue Congress from its drafting
errors, and to provide for what we might think . . . is the
preferred result.” Lamie v. United States Trustee, 

, 542 (2004), quoting from United States v. Granderson,

, 68 (1994) (concurring opinion).


  In deciding how to address a subject, the
legislature—Rule 6(e) is the work of Congress rather than
the Supreme Court under the Rules Enabling Act—must
choose between a rule and a standard. Rules such as
“determine how the holder came by the information” are
easy to administer but are inevitably both too narrow in
Nos. 04-3622 & 04-3623                                      7

some situations (all rules have loopholes) and overbroad in
others. Standards such as “determine whether the prosecu-
tor acted in good faith in providing the information, or some
other error occurred” could in principle match the outcome
more closely to the legislative objective, but standards are
difficult to administer and create errors of their own. Courts
cannot ascertain intent or good faith without hearings; the
principal evidence would be oral and correspondingly
difficult to evaluate. Rules have lower administrative costs
and will be preferable unless they increase the error costs
(the sum of false positives and false negatives) by more
than the savings in administrative costs. Whether to choose
a rule or a standard is a legislative decision. Judges ought
not turn a rule into a standard; that amounts to little more
than disagreement with a legislative choice. Boosting the
level of generality by attempting to discern and enforce
legislative “purposes” or “goals” instead of the enacted
language is just a means to turn rules into standards. Cf.
Rodriguez v. United States, 

(1987). Rule
6(e)(2)(B) creates a bright line, and we must enforce it that
way.
  What Daniels labels “absurd” results are nothing but the
rough cuts inevitable with decision by rule. To observe that
error costs exist is not to justify use of a standard—first
because the choice is for political actors, and second because
we cannot be sure that Rule 6(e)(2)(B) as written produces
more costs than would a judicial attempt to assess the
prosecutor’s good faith and the assistant’s “government
personnel” status one case at a time.


  When an opinion says that courts interpret statutes to
avoid absurd results, it is not inviting judges to convert
rules into standards. Church of the Holy Trinity v. United
States, 

(1892), which invoked the norm against
absurd outcomes to make the law more in line with the
8                                  Nos. 04-3622 & 04-3623

Justices’ substantive preferences, has no modern traction.
Today the anti-absurdity canon is linguistic rather than
substantive. It deals with texts that don’t scan as written
and thus need repair work, rather than with statutes that
seem poor fits for the task at hand. In other words, the
modern decisions draw a line between poor exposition and
benighted substantive choice; the latter is left alone,
because what judges deem a “correction” or “fix” is from
another perspective a deliberate interference with the
legislative power to choose what makes for a good rule.
Admit the propriety of “fixing mistakes” and you allow a
general power to identify “mistakes,” which means a
privilege to make the real substantive decision. Even when
the statute invites modification, as the “context clause” in
some definitions does, judges are limited to considering the
linguistic context rather than trying to “improve” the
statute’s substantive effect. See Rowland v. California
Men’s Colony, Unit II Men’s Advisory Council, 

(1993).
  The only recent decision in which the anti-absurdity
canon played an important role, Green v. Bock Laundry
Machine Co., 

(1989), dealt with an incomplete
and baffling rule. Other decisions that cite the doctrine,
such as Public Citizen v. Department of Justice, 

(1989), and United States v. X-Citement Video, Inc., 

(1994), use it to avoid an unconstitutional reading.
The dearth of modern “substantive absurdity” decisions is
readily understandable. Scholars as well as judges have
recognized that a power to fix statutes substantively would
give the Judicial Branch too much leeway to prefer its views
about what makes for “good” laws over those of the Legisla-
tive Branch. See, e.g., John Manning, The Absurdity
Doctrine, 116 Harv. L. Rev. 2387 (2003); Adrian Vermeule,
Legislative History and the Limits of Judicial Competence:
The Untold Story of Holy Trinity Church, 50 Stan. L. Rev.
1833 (1998). The Supreme Court has been willing to enforce
Nos. 04-3622 & 04-3623                                    9

even statutes that seem to set traps for the unwary or
unfortunate. Dodd v. United States, 

(2005),
is a good example: The Court held that the statute of
limitations for collateral attacks on criminal convictions
may expire before the decision supporting the challenge
becomes applicable, and it rejected an argument that this
linguistically sound reading should be rejected as substan-
tively absurd.
  Another good example is United States v. Locke, 

(1985), which dealt with a statute that required certain
documents to be filed “before December 31.” Unwary
readers might read this as equivalent to “before the end of
the year,” and inevitably some did. The Court held that a
person who filed on December 31 had filed too late. The
statute was complete as written, and though the use of
“before December 31” rather than “on or before December
31” may have been a blunder there was no linguistic defect,
and hence no role for the anti-absurdity canon. The unfortu-
nate outcome for the late filers, the Court held, was just a
normal effect of any rule; no matter where the line may be
placed, someone always files one day too late. See also
Exxon Mobil Corp. v. Allapattah Services, Inc., 

(2005).
  Guidry v. Sheet Metal Workers National Pension Fund,

(1990), supplies another example. Curtis
Guidry pleaded guilty to embezzling funds from a pension
plan of which he had been a trustee. Guidry was himself
entitled to a pension from the plan, and the remaining
trustees confiscated the value of his pension in partial
fulfilment of Guidry’s obligation to repay what he stole.
Guidry filed suit, contending that this offset violated
ERISA’s anti-alienation clause. He lost in the court of
appeals, which said that an anti-alienation clause is
designed to protect pensioners from their own improvident
spending, not to enable them to retain ill-got gains. If the
enacting Congress had been asked: “Would you make an
10                                  Nos. 04-3622 & 04-3623

exception to the anti-alienation clause for persons who steal
from the pension fund?”, it likely would have answered yes.
This is the method of Riggs v. Palmer, 

, 

(1889), the famous decision holding that to avoid
(substantive) absurdity murdering heirs cannot inherit
despite the Statute of Wills. Yet the Supreme Court ruled
in Guidry’s favor. ERISA’s anti-alienation clause is coherent
as written and contains no exceptions. Whatever Congress
might have done, it had not done. To the extent that the
Justices referred to purposes at all, they conceived them
concretely rather than abstractly. What is the purpose of an
anti-alienation clause? It is to prevent appeals to the
equities case by case. In other words the Justices did not
ask, “What is the value served by this particular rule?”
Instead they asked “What is the value served by rules, in
general?” This led the Court to enforce the rule and to
rebuff efforts at reconstruction. Guidry represents today’s
interpretive norm.
  Our final example is Lamie, to which we have referred
already. Until 1994 the Bankruptcy Code provided that
courts could authorize compensation to the debtor’s attor-
ney in Chapter 7 proceedings. In 1994 the vital lan-
guage—"or to the debtor’s attorney”—vanished from 11
U.S.C. §330(a). There was some reason to think that the
deletion had been an accidental byproduct of other changes
made to this section; the phrase (or an equivalent) probably
should have been moved but not eliminated. But it was
eliminated, the statute as revised could be parsed, and the
Court therefore held that the judicial authority to award
fees to counsel in Chapter 7 cases had lapsed. The Justices
allowed that the change might be unfortunate for some
debtors, even counterproductive for the bankruptcy system
as a whole, but concluded that such observations must be
addressed to the legislature; bad consequences do not allow
creative “interpretation” to avoid 

.
“Our unwillingness to soften the import of Congress’ chosen
Nos. 04-3622 & 04-3623                                      11

words even if we believe the words lead to a harsh outcome
is longstanding. It results from ‘deference to the supremacy
of the legislature’ ”, 

quoting from 

.
  Rule 6(e)(2)(B) makes sense as written. It parses without
the assistance of a red pencil, and judges are not authorized
to add words (such as “properly”) that would change the
Rule’s substantive effect.
  A few other matters require only brief mention. Daniels
contends that an injunction was improvident because
Jaskolski did not establish irreparable injury. Yet the point
of Rule 6(e)(2)(B) is not to prevent injury to the person
making the disclosure; it is to prevent injury to innocent
persons whose names may be dragged into the mud, and
hindrance of future criminal investigations (because the
risk of disclosure will reduce some persons’ willingness to
cooperate with grand juries). These potential losses are not
quantifiable and certainly cannot be redressed by damages
paid to Jaskolski, if disclosure turns out to be unwarranted.
This is not something that can be proved by testimony in a
preliminary-injunction hearing; the district judge did not
err in dispensing with one.
   Perhaps, as Daniels contends, none of the information the
plaintiffs seek in the state litigation is “a matter occurring
before the grand jury”—though plaintiffs demanded essen-
tially everything Jaskolski had, including grand jury
transcripts, which are covered by Rule 6(e) in his hands
even if some disclosures have been made (for example, to
Daniels as a defendant in the federal prosecution). Judge
Lozano may have written the injunction too broadly to the
extent that it halts the entire discovery process in the state
litigation pending the sifting before Judge Moody. Yet
Daniels’s appellate brief does not attempt to show that any
category of information sought in the state litigation is
independent of the federal grand jury. It would only delay
12                                 Nos. 04-3622 & 04-3623

matters to send the proceeding back to Judge Lozano for a
more precise specification of the issues that must be
presented to Judge Moody. Prompt resolution by Judge
Moody of lingering disputes under federal law—principally,
separating the information Jaskolski obtained from the
prosecutors and federal investigators (protected by Rule
6(e)) from the information Jaskolski learned on his own as
an insurance investigator (not covered by Rule 6(e))—will
expedite returning control to the state court, where it
belongs.
                                               AFFIRMED

A true Copy:
      Teste:

                       ________________________________
                       Clerk of the United States Court of
                         Appeals for the Seventh Circuit




                  USCA-02-C-0072—10-21-05