Case Information
*1 Before W OOD , Chief Judge , and B AUER and K ANNE , Circuit Judges .
W OOD , Chief Judge . Eric Curtis led a crew that robbed fi ve cell-phone stores located in suburban Chicago. He was ar- rested following the last of the heists and eventually stood trial on ten criminal charges: four counts for robbery, four counts for aiding in the brandishing of a rearm in relation to a crime of violence, a count for conspiracy, and a count for being a felon in possession of a rearm. A jury convicted him *2 on all counts save two: one for robbery and one for aiding in the brandishing of a rearm. Each acqui al was on a charge related to a robbery of a store in Joliet.
Curtis raises two issues on appeal. First, he argues that the district court should have excluded evidence of his cell-site location information (“CSLI”), which he alleges was obtained in violation of the Fourth Amendment. Second, he complains that the district court prohibited him from cross-examining witnesses about a potential source of bias, and thereby vio- lated the Sixth Amendment’s Confrontation Clause. Neither of these alleged errors is enough to disturb the judgment against him, which we a ffi rm.
I
“CSLI is location information generated by cellular phone providers that indicates which cell tower a particular phone was communicating with when a communication was made.” Orin S. Kerr, The E ect of Legislation on Fourth Amendment Pro- tection , 115 M ICH . L. R EV . 1117, 1128 (2017). It is capable of “pinpoint[ing] a phone’s location within 50 meters.” Carpenter v. United States , 138 S. Ct. 2206, 2219 (2018). Because cell phones are in constant communication with the nearest cell site—often a ffi xed to a cell tower—they can collect CSLI as frequently as several times a minute. at 2211–12. In this case, the government obtained historical CSLI for Curtis’s cell phone for a span of 314 days. The data placed Curtis in the vicinity of four of the ve stores at the time each was robbed. There was no CSLI evidence for the Joliet robbery.
The government relied on the procedures set forth in the Stored Communications Act (SCA), 18 U.S.C. § 2703, to obtain Curtis’s CSLI. The type of data it sought is considered to be *3 non-content information for SCA purposes. See 18 U.S.C. § 2703(c). That part of the SCA authorizes courts to order cell- phone providers to disclose non-content information if the government “o ers speci c and articulable facts showing that there are reasonable grounds to believe that … the records or other information sought are relevant and material to an on- going criminal investigation.” 18 U.S.C. § 2703(c)(1)(B), (d) . Curtis did not dispute the government’s compliance with the SCA, but he took the position that SCA compliance was not enough and moved to suppress the evidence. He argued that collecting CSLI without a search warrant violates the Fourth Amendment because there is a reasonable expectation of pri- vacy in CSLI. The district court denied the motion, ruling that a cell-phone user voluntarily discloses CSLI to his phone pro- vider, and that the Fourth Amendment does not protect vol- untarily disclosed information. See Smith v. Maryland , 442 U.S. 735, 743–44 (1979); United States v. Miller , 425 U.S. 435, 442–44 (1976). Curtis appeals that ruling.
The Supreme Court resolved Curtis’s Fourth Amendment argument in Carpenter v. United States , 138 S. Ct. 2206 (2018). There it decided that a person in Curtis’s position, for whom data was collected for a substantial time, maintains a legiti- mate expectation of privacy for Fourth Amendment purposes in the records of his physical movements disclosed by CSLI. See id. at 2217. It declined to say whether there was “a limited period for which the Government may obtain an individual’s historical CSLI free from Fourth Amendment scrutiny,” de- ciding only that accessing seven days’ or more worth of infor- mation was enough. Id. at 2217 n.3. In Carpenter , as here, the prosecutors had obtained court orders under the SCA, and those court orders purported to authorize the collection of the target’s cell phone records. at 2212. The Court said that *4 SCA compliance did not ma tt er, because the showing re- quired by the SCA “falls well short of the probable cause re- quired for a warrant.” Id. at 2221. The Court also rejected the applicability of the “third-party doctrine,” which (when it ap- plies) allows the collection of business records collected by a third party in the ordinary course of operations. at 2217. It remanded the case for further proceedings.
Our case stands in the same position as the Carpenter re- mand. The Court has resolved the question whether an SCA order obviates the need for the warrant, but it has not spoken to what should happen next. We must decide whether this conceded error automatically results in relief for Curtis, for whom records covering 314 days were collected. We conclude that it does not. A di erent part of Fourth Amendment juris- prudence is, in our view, dispositive: evidence obtained in good-faith reliance on a statute later declared unconstitu- tional need not be excluded. Illinois v. Krull , 480 U.S. 340, 349– 50 (1987); see also United States v. Pembrook , 876 F.3d 812, 823 (6th Cir. 2017), vacated on other grounds by Johnson v. United States , 138 S. Ct. 2676 (2018) (applying the good-faith excep- tion to CSLI obtained under the SCA); United States v. Graham , 796 F.3d 332, 363 (4th Cir. 2015), reversed on other grounds by United States v. Graham , 824 F.3d 421 (4th Cir. 2016) ( en banc ) (same); United States v. Davis , 785 F.3d 498, 511, 518 n.20 (11th Cir. 2015) (same).
Curtis’s proposed path around Krull is ambitious. He does not argue that o ffi cers obtained his CSLI in bad faith. Far from it: his motion to suppress seemingly concedes that there would have been probable cause to seek a search warrant. It is Krull itself that he a empts to push out of the picture. He *5 argues that Krull applies only to statutes authorizing admin- istrative searches. His logic proceeds in three steps. First, he urges, the good-faith exception to the exclusionary rule can- not be applied so as to insulate statutes from constitutional challenge. To do so would “destroy[] all incentive on the part of individual criminal defendants to litigate the violation of their Fourth Amendment rights.” Krull , 480 U.S. at 369 (O’Connor, J., dissenting). Second, he suggests that the Krull majority could sidestep that concern because the target of an administrative search necessarily knows that a search is im- pending. A forewarned target still has reason to “bring an ac- tion seeking a declaration that the statute is unconstitutional and an injunction barring its implementation” notwithstand- ing the good-faith exception. at 354 (majority opinion). Third, he points out that the target of an SCA order issued under section 2703(d) has no knowledge of the order until the CSLI has been collected and used in a criminal proceeding. At that late hour, a defendant has no incentive to challenge the statute because the good-faith exception permits admission of the fruits of an unconstitutional search.
Experience has shown that the good-faith exception has not had the chilling e ect that Curtis fears. Curtis, like many others, has challenged section 2703(d) of the SCA on Fourth Amendment grounds notwithstanding the risk that the ex- ception may apply. See , e.g. , Carpenter , supra ; United States v. Graham , 824 F.3d 421, 425 (4th Cir. 2016) ( en banc ); United States v. Daniels , 803 F.3d 335, 351–52 (7th Cir. 2015); Davis , 785 F.3d at 511; In re U.S. for Historical Cell Site Data , 724 F.3d 600, 608 (5th Cir. 2013). This is just what the Krull majority predicted: defendants will still “contest the validity of statutes [even] if they are unable to bene t directly by the subsequent exclu- sion of evidence … .” Krull , 480 U.S. at 353.
The exclusionary rule is designed primarily to deter un- constitutional conduct. Id. at 349. Nothing substantiates the fear that when passing laws such as the SCA “legislators are inclined to subvert their oaths and the Fourth Amendment.” Id. at 351. Even if there were a need to deter legislators, “there is nothing to indicate that applying the exclusionary rule to evidence seized pursuant to the statute prior to the declara- tion of its invalidity will act as a signi fi cant, additional deter- rent.” at 352. We conclude, therefore, that even though it is now established that the Fourth Amendment requires a war- rant for the type of cell-phone data present here, exclusion of that information was not required because it was collected in good faith.
II
The second issue Curtis raises is whether he should have been allowed to ask witnesses about what happened follow- ing the fi rst of the fi ve robberies. After that robbery, police went to the home of Ryan Rogers, Curtis’s cousin, whom po- lice suspected of organizing the fi rst robbery. A confrontation ensued, and o ffi cers shot and killed Ryan. (We use his rst name because Eric Rogers also played a role in these events.) Curtis was a witness. Several hours later, after Curtis had re- turned home, o ffi cers raided his house. At that time, Curtis asked for the name of the o ffi cer who had killed Ryan and de- clared his intent to le a complaint. Eric Rogers, Curtis’s co- conspirator, was with Curtis while these events transpired.
On the eve of trial, Curtis asked the court if he would be permi ed to cross-examine Eric, who had become a govern- ment witness, about the shooting and the ensuing interaction between Curtis and law enforcement. Cross-examination, Curtis maintained, would reveal Eric’s motivation to testify *7 against him because Eric believed that, given Curtis’s conten- tious relationship with law enforcement, the authorities would look with particular favor on any witness who cooper- ated against Curtis. The district court denied Curtis’s request without prejudice, ruling that the theory was a tt enuated and would lead the trial on a “goose chase.”
The next day, Curtis fi led a wri tt en motion seeking per- mission to cross-examine all coconspirators serving as gov- ernment witnesses about whether they knew that Curtis had accused the police of unjusti fi ably shooting Ryan. In addition to the theory articulated the day before, Curtis o ff ered two more reasons why cross-examination would reveal bias. First, he contended that asking about Curtis’s threat to fi le a com- plaint would show that law enforcement had reason to nudge each witness to turn against Curtis. Second, he argued that cross-examination would show that witnesses feared the po- lice because they knew that Ryan had been killed unjusti fi ably during the investigation of the robberies at issue. The district judge remained unmoved, explaining from the bench that Curtis’s theory was too convoluted and that there already was an abundance of evidence that the government’s witnesses were biased. The court also found that bringing up a poten- tially unjusti ed shooting would be excessively prejudicial. Curtis appeals that ruling. He has pared down his argument at this stage, contending only that the district court should have permi ed the questions to show that the witnesses be- lieved they would bene t from testifying against Curtis.
The Sixth Amendment’s Confrontation Clause guarantees the right to e ective cross-examination. United States v. Mar- tin , 618 F.3d 705, 727 (7th Cir. 2010). Among other things, it *8 entitles a defendant to cross-examine a witness about each po- tential source of bias. Id. But the Confrontation Clause is not a license to ask a witness literally anything. On the contrary, trial judges may narrow the scope of questioning for reasons such as “harassment, prejudice, confusion of the issues, the witness’ safety, or interrogation that is repetitive or only mar- ginally relevant.” Delaware v. Van Arsdall , 475 U.S. 673, 679 (1986). If denying Curtis’s desired line of cross-examination was constitutionally improper, we must vacate the convic- tions unless we can say beyond a reasonable doubt that after assuming the full damaging e ect of the potential cross-ex- amination the error was harmless. at 684.
Curtis made the district court’s job harder than necessary. He buried his strongest argument—the one raised on ap- peal—between two theories that were speculative, tenuous, and required a prejudicial account of the shooting. Even if we assume, however, that Curtis’s argument was properly raised, we are convinced beyond a reasonable doubt that any error, if there was one, was harmless.
The centerpiece of the case against Curtis was the CSLI data placing him near four of the robberies, coupled with call logs exhibiting that Curtis was in communication with admit- ted participants during each offense. Though the government induced several of Curtis’s coconspirators to testify against him, we are convinced that their testimony added little to the jury’s evaluation of the evidence. Curtis’s attorney effectively brought out the significant evidence of each witness’s bias. One by one, the witnesses admitted to having or expecting a cooperation agreement through which they stood to receive a sizable sentence reduction in their own criminal case—one *9 witness hoped for a reduction of as much as 54 years—in ex- change for testifying. Some of the witnesses also admitted to having lied initially to law enforcement about the extent of their and Curtis’s involvement, minimizing the former and overstating the latter. The verdict captures the effect of the im- peachment. Recall that Curtis was acquitted on the two charges related to the robbery that took place in Joliet and was convicted on the rest. Joliet was the sole robbery for which the government did not have CSLI evidence or evidence of what calls Curtis placed during the robbery. Only witness testi- mony tied Curtis to the Joliet robbery. The latter, it appears, was not enough by itself to satisfy the jury.
Finally, Curtis’s a orney managed to put Ryan’s death and Curtis’s threatened complaint against the police before the jury notwithstanding the district court’s adverse ruling. When an FBI agent took the witness stand, defense counsel asked him about both. Curtis testi ed in his own defense and told the jury that he had witnessed police kill his cousin and that he and his family were pursuing a complaint against the police. In his closing argument, defense counsel contended that there was an implicit conspiracy to frame Curtis. He asked the jury to recall that “Eric Curtis saw his cousin killed by the police and he stood up to the police.” And Curtis re- minded them that he had said to the o ffi cers, “I want to le a complaint. I saw they killed him for nothing.” Because of this, defense counsel argued to the jury that “[Curtis] has got a problem with law enforcement.” Continuing, defense counsel told the jury that “[Curtis] is the guy who is standing up to law enforcement. Everybody knew that. And the idea that they didn’t is nonsense.” While counsel did not say that the prosecutors directed any witness to testify against Curtis, he *10 did suggest that they made witnesses comfortable about turn- ing on Curtis. Because Curtis made his desired arguments and impeached his coconspirators, any error in denying the cross-examination was harmless.
III
Not all constitutional injuries have a remedy. In this case, good faith renders the Fourth Amendment violation non-re- dressable, and any Sixth Amendment violations were harm- less. We therefore A FFIRM the judgment.