OSTRER, J.A.D.
*693*494Convicted of purposeful murder and related firearms offenses, defendant Olajuwan Herbert principally contends his trial was irremediably tainted by a detective's reference to defendant's alleged gang membership, and by eyewitnesses' statements that they had been afraid to testify or identify themselves. The court sustained defendant's objection to the comment about gang membership, but denied his motion for a mistrial. The court held it cured any resulting prejudice by instructing the jury that there was no information in the case about gang involvement and that the jury should disregard the statement. The court overruled defense objections to the eyewitnesses' continued use of pseudonyms and references to fear of testifying, relying on its instruction to the jury that the use of pseudonyms was simply a matter of police procedure and the witnesses' desire for privacy.
We conclude a new trial is required because the court's instructions were inadequate to cure the prejudice caused by the gang *495reference. We therefore do not reach the issue of the eyewitnesses' repeated reference to pseudonyms and expressions of fear.
I.
Harold Claudio was shot to death in an alley off Thomas Street in Newark. The alley runs between a school on one side and a playground and two basketball courts on the other. The homicide occurred on June 9, 2012, at around 8:30 p.m. The principal witnesses at the trial were an investigating detective, Tyrone Crawley, and two eyewitnesses, Lizaire Arce, a cousin of the victim, and Jessica Maldonado, who happened to be in the area but knew neither the victim nor defendant.
Arce testified that the previous week, she saw Alberto Torres, a cousin of hers and the victim's, fight with defendant at the playground. Torres got the better of defendant, and Arce overheard her cousin call defendant by the nickname "Gunner"1 as he chased him from the playground. It was the first time Arce had ever seen defendant.
Arce said the next time she saw defendant, shortly before the murder, he entered the playground area with Claudio and two other men she did not know. Although it was around 8:30 p.m., Arce *694said it was bright out. She claimed she could identify defendant, although, as the parties later stipulated, she was sitting 107 feet, 11 inches from the place of the murder. Also, several teenagers were playing ball on the basketball court in front of her, and chain-link fences stood between her and defendant. She testified that she observed defendant raise his left arm while stepping closer to Claudio, then heard three gunshots, although she did not actually see a gun in defendant's hand. Arce testified *496that she saw defendant and the two others run out of the playground, across Pennsylvania Avenue, and up the block to Brunswick Avenue, where defendant entered the rear seat of a 1993 or 1994 blue Honda Accord with a silver sunroof.
Arce went to her cousin's side and saw that a bullet had struck the back of his head. Concluding he was "done," she left the area without calling 911. She spent time with a boyfriend and then a female friend but did not speak to police because, she said, she feared for herself and her family. However, she spoke to the police two days later, after her aunt, Claudio's mother, asked her to do so. Arce identified defendant from a photo array, and later identified him in court, as well.
Arce signed defendant's photo as "Jane Doe" and initialed the others she viewed "J.D." At trial, she testified she did so for "her safety." The defense objected, contending there was no evidence that defendant did anything to cause her to be fearful, and asked for an instruction to that effect. The court denied the request. Apparently referring to Arce's reasons for her two-day delay in speaking to police, as opposed to her reason for signing the photos as Jane Doe, the court said the State was entitled to explain why Arce "didn't do certain things" before the defense raised the matter on cross-examination.
Throughout the trial, the prosecutor and Detective Crawley, the State's key law enforcement witness, repeatedly referred to Arce as Jane Doe. The officer who presented the photo array to Arce also said he knew her only as Jane Doe. He testified, without an immediate objection, that he was informed she used the pseudonym to avoid identification and "retaliation."2
*497As the defense elicited, Arce testified inconsistently about the lighting conditions at the crime scene; the distance from which she observed defendant; and the presence of other persons in the area. Based on these inconsistencies, the defense challenged Arce's ability to accurately identify faces, and to observe defendant's alleged flight in a vehicle parked over a block away.
The other eyewitness, Jessica Maldonado, testified that she heard what sounded like fireworks after she parked on Thomas Street. She was on her way to a baby shower at the church across the street, at the corner of Thomas Street and Pennsylvania Avenue. A passerby told her he thought the sound was gunshots. She grabbed her three-year-old daughter out of the car and headed down the block. On the other side of the street, Maldonado saw a group of four or five men running in the opposite direction, accompanied by a man on a bicycle. For a second, she made eye contact with one man as he turned to *695look back. He was holding his pants and shirt to cover an object. She could not see it, but it had the shape of a gun.
Three days later, she selected defendant's photo from a photo array, signing the photo as "Jane Doe 2." At trial, the court overruled the defense's objection to any mention of Maldonado's use of the pseudonym, but prevented Maldonado from explaining why she used it. The State repeatedly referred to her as Jane Doe 2 thereafter. Maldonado did not make an in-court identification.
The defense elicited inconsistencies between Arce's and Maldonado's testimony. Arce alleged that a light-skinned man almost six feet tall with shoulder-length, orange-tipped dreadlocks accompanied defendant; yet, Maldonado saw no such person, testifying that all the men running from the scene had short hair and brown skin. Arce said defendant wore light-blue capri pants, a red-and-white shirt, and Air Force sneakers. Maldonado testified that defendant wore baggy, regular-length blue jeans and a white or gray tee-shirt. Arce said there were no cars parked on Thomas Street between Pennsylvania Avenue and Brunswick Avenue, which gave her an unobstructed view of defendant's flight. She did not see a *498woman with a child. Yet, Maldonado testified that it was difficult to find parking near the church, cars were parked on both sides of the street, and she was on the sidewalk with her child when the men fled past her. Arce said the four men all fled on foot. Maldonado said there was at least one man on a bicycle.
Maldonado's out-of-court identification was also challenged. She admitted that she found the identification procedure "hard"; the men in two other photos also looked "similar" to the man she had seen; and she selected defendant's photo only after the officer told her, in response to her question, that she had to pick just one photo. She testified that she ultimately picked defendant's photo because it depicted him with facial hair; however, in her 2012 statement to Detective Crawley, she said she could not tell if the man she saw had facial hair because it was getting dark and she was not wearing her glasses.
Detective Crawley testified that when defendant was taken into custody, he was wearing sneakers that fit Arce's description. However, there was no blood or other forensic evidence tying the sneakers to the homicide. Subsequent searches of defendant's residence and cellphones were fruitless. The weapon was not recovered, and video surveillance in the vicinity was unilluminating.
Crawley also admitted at trial that he had received information that someone else was responsible for the shooting. Torres, Arce's cousin, reportedly told the detective that someone named Bibble had bragged he committed the murder. But, because the State was unable to secure Torres's presence at trial, the court barred any reference to Torres's statement on hearsay grounds.
Also, before trial, a different judge, relying on State v. Goodman, 415 N.J. Super. 210, 1 A.3d 767 (App. Div. 2010), had permitted Torres to testify that Claudio's murder arose from a gang conflict. But, with Torres absent, gang references were barred.
*499Yet, despite the court's order, Crawley injected the subject of gangs twice.3 The first time, on direct examination, he was asked why he did not speak to the *696teenagers who were playing in the basketball court, though Arce had identified several of them by name. He answered it was a "high-crime, drug, gang area, and the people that live in that area are in fear of the police." The court sustained the defense's objection, denied a mistrial request, and instructed the jury to disregard the statement, which was complete "speculation and conjecture on [the detective's] part" as he had no information "as to why anyone didn't come forward, or if there were any people that could come forward." The judge did not address the statement that the crime occurred in a "gang area."
The second time, on redirect, the prosecutor asked Detective Crawley if Arce had explained how she was able to identify defendant. The court overruled a hearsay-based objection from the defense. Then, the detective stated, referring to the alleged previous altercation between defendant and Torres, "Yes, she told me, approximately a week ago, Gunner, who is a gang member ...."
The court denied another defense request for a mistrial. Instead, in a curative instruction, the judge directed the jury to disregard the detective's answer. The judge asserted there was "no information" in the case that gangs were involved - without directly addressing the detective's assertion about defendant in particular. In the same instruction, the judge decided to issue a limiting instruction regarding the eyewitnesses' use of pseudonyms, asserting both that the witnesses wanted to shield their identity and that it was merely a matter of police procedure.
The judge stated:
*500Ladies and Gentlemen, our function here -- I mean your function -- is to determine the guilt or innocence of Mr. Herbert fairly and impartially based upon the evidence.
Sometimes during the course of the trial, information comes to the attention of the jury and it has no place in this trial. In other words, it's prejudicial. As I told you from the beginning, fair and impartial means not being prejudicial; it means fair.
Now, Detective Crawley mentioned gangs a few minutes ago, unintentionally. I tell you now there's no information in this case whatsoever there's any gangs involved in this case whatsoever. Nothing whatsoever. You've heard nothing beforehand, you've heard nothing now, and that statement by Detective Crawley obviously was unintentional, number one.
Number two, I want to bring you to the -- to your attention again -- maybe I should have said this is before -- you know, the two witnesses we're [sic] giving [sic] John [sic] Doe 1 and John [sic] Doe 2; all right? There's no rationale for that other than the fact that they wanted to keep their identity private. All right?
You should not conclude, because they had Jane Doe 1, and Jane Doe 2, that carries any kind of implication whatsoever. It does not; all right?
So the gang situation, the John [sic] Doe situations are routine, -- the Jane Doe situation is routine police work repeatedly in these kind of cases, it's just done that way, and you can't consider that whatsoever. So you can't consider that, and you can't consider the gang situation.
I'm going to read you again what I read to you before. I direct that you not use this stricken testimony in your deliberations. By my striking the answer and directing that you disregard and not use this information, I'm not asking you to forget it. To the contrary, I'm asking that you remember what was stricken *697and understand that if, during your deliberations, you realize that the information is necessary to your decision, you may not use it.
Okay? Are we on the right page?
JURORS: Yes.
[ (Emphasis added).]
In cross-examination of Crawley, the defense had pointed out inconsistencies between the testimony of Arce and Maldonado, suggesting that Crawley should have conducted a more thorough investigation. On redirect, the prosecutor was permitted to elicit why those inconsistencies did not deter him from seeking an arrest warrant. For example, regarding the inconsistent clothing descriptions, Crawley was permitted to offer his opinion that Maldonado was more focused on her child, while Arce was "affixed to" defendant, because of the previous incident.
*501Defendant did not testify or call witnesses on his behalf. In summation, defense counsel responded to the eyewitnesses' use of pseudonyms, noting there was no evidence that defendant threatened or intimidated them. The defense also highlighted inconsistencies in the eyewitnesses' identifications. The State, in its summation, minimized those inconsistencies, and suggested that defendant's motive was to retaliate for the incident the week earlier, although defendant's assailant was Torres and not the victim.
The jury found defendant guilty of first-degree murder, N.J.S.A. 2C:11-3(a)(1), (2) ; second-degree unlawful possession of a handgun, N.J.S.A. 2C:39-5(b)(1) ; and second-degree possession of a firearm for an unlawful purpose, N.J.S.A. 2C:39-5(a). The court sentenced defendant to life imprisonment on the murder conviction, subject to the No Early Release Act, N.J.S.A. 2C:43-7.2, and a concurrent five-year term, subject to the Graves Act, N.J.S.A. 2C:43-6(c), on the conviction for unlawful possession of a handgun. The remaining charge was merged.
II.
In his counseled brief, defendant raises the following points for our consideration:
POINT I
DEFENDANT WAS DENIED A FAIR TRIAL AND DUE PROCESS OF LAW WHEN DETECTIVE CRAWLEY IMPROPERLY TOLD THE JURY THAT "GUNNER [DEFENDANT] IS A GANG MEMBER" AND THE TRIAL COURT DENIED DEFENDANT'S MOTION FOR A MISTRIAL.
POINT II
THE TRIAL COURT COMMITTED REVERSIBLE ERROR BY ALLOWING DETECTIVE CRAWLEY TO TESTIFY AS TO WHY HE OBTAINED AN ARREST WARRANT DESPITE THE INCONSISTENCIES BETWEEN THE TWO IDENTIFICATION WITNESSES REGARDING THE DESCRIPTIONS OF THE CLOTHING WORN BY THE SUSPECT.
POINT III
*502DEFENDANT'S SENTENCE IS MANIFESTLY EXCESSIVE AND UNDULY PUNITIVE.
In a supplemental pro se brief, defendant argues:
POINT I
DEFENDANT'S STATE AND FEDERAL CONSTITUTIONAL RIGHTS TO A FAIR TRIAL BY AN IMPARTIAL JURY AND DUE PROCESS OF LAW WERE VIOLATED WHEN THE PROSECUTOR ENGAGED IN A CONVERSATION WITH ONE OF THE JUROR[ ]S.
POINT II
*698THE TRIAL COURT ERRED IN ADMITTING DETECTIVE CRAWLEY'S TESTIMONY AS TO WHY HE OBTAINED AN ARREST WARRANT DESPITE THE INCONSISTENCIES BETWEEN THE TWO I[ ]DENTIFICATION WITNESSES REGARDING THE DESCRIPTIONS OF THE CLOTHING WORN BY THE SUSPECT (Supplemented to Point II of Primary Brief).
POINT III
THE IDENTIFICATION PROCEDURES USED BY THE POLICE WERE IMPERMISSIBLY SUGGESTIVE LEADING TO A SUBSTANTIAL LIKELIHOOD OF MISIDENTIFICATION.
POINT IV
TRIAL COURT COMMITTED REVERSIBLE ERROR BY ALLOWING THE PROSECUTION TO INTRODUCE EVIDENCE VIA IMPROPER IDENTIFICATION PROCEDURE THAT DEPRIVED DEFENDANT SUBSTANTIVE DUE PROCESS OF LAW AND RIGHT TO A FAIR TRIAL (Not Raised Below).
POINT V
THE TRIAL COURT'S CHARGE ON THE ISSUE OF IDENTIFICATION WAS FLAWED, THEREFORE VIOLATING DEFENDANT'S STATE AND FEDERAL CONSTITUTIONAL RIGHTS (Not Raised Below).
POINT VI
THE TRIAL COURT ERRED IN DENYING DEFENDANT A NEW TRIAL ON THE GROUNDS THAT THE VERDICT WAS AGAINST THE WEIGHT OF THE EVIDENCE.
POINT VII
THE TRIAL COURT'S FAILURE TO CHARGE THE JURY ON THE LESSER-INCLUDED OFFENSES OF ASSAULT DEPRIVED DEFENDANT OF DUE PROCESS AND HIS RIGHT TO A FAIR TRIAL. U.S.CONST.AMEND. VI, XIV, N.J. CONST. ART. I, ¶ 1, 10 (Not Raised Below).
III.
A.
The Court in State v. Winter, 96 N.J. 640, 646-47, 477 A.2d 323 (1984), addressed the specific issue posed here - "whether *503inadmissible evidence is of such a nature as to be susceptible of being cured by a cautionary or limiting instruction, or instead requires the more severe response of a mistrial." The Court held the decision "is one that is peculiarly within the competence of the trial judge, who has the feel of the case and is best equipped to gauge the effect of a prejudicial comment on the jury in the overall setting." Id. at 647, 477 A.2d 323. Consequently, "[a] motion for a mistrial is addressed to the sound discretion of the [trial] court; and the denial of the motion is reviewable only for an abuse of discretion." Ibid. (quoting State v. Witte, 13 N.J. 598, 611, 100 A.2d 754 (1953) ); see also State v. Yough, 208 N.J. 385, 397, 31 A.3d 271 (2011) (stating that whether a curative instruction can neutralize a prejudicial remark is within the trial court's competence); State v. Harvey, 151 N.J. 117, 205, 699 A.2d 596 (1997) (stating an appellate court must find "an abuse of discretion that results in a manifest injustice" to overturn a trial court's mistrial ruling).
The same deferential standard that applies to the mistrial-or-no-mistrial decision applies to review of the curative instruction itself. Winter, 96 N.J. at 647, 477 A.2d 323. In particular, a trial court is in the best position to assess the impact of an evidentiary ruling. See Crawn v. Campo, 136 N.J. 494, 512, 643 A.2d 600 (1994)
*699(stating that "[d]eference should be accorded to the trial court's conclusion concerning the prejudice attributable to the" trial court's rulings and "the extent to which that prejudice contributed to an unjust result").
B.
There is tension in our case law governing curative and limiting instructions. The authority is abundant that courts presume juries follow instructions. For example, in State v. Loftin, 146 N.J. 295, 390, 680 A.2d 677 (1996), the Court stated, "That the jury will follow the instructions given is presumed." The presumption is founded in part on necessity. "[T]he courts must rely upon the jurors' ability and willingness to follow the limiting instruction *504without cavil or question." State v. Manley, 54 N.J. 259, 270, 255 A.2d 193 (1969). The presumption is "[o]ne of the foundations of our jury system." State v. Burns, 192 N.J. 312, 335, 929 A.2d 1041 (2007).
Yet, some view the presumption skeptically. As Justice Jackson stated, "The naive assumption that prejudicial effects can be overcome by instructions to the jury ... all practicing lawyers know to be unmitigated fiction." Krulewitch v. United States, 336 U.S. 440, 453, 69 S.Ct. 716, 93 L.Ed. 790 (1949) (Jackson, J., concurring). Noting, if not adopting, that skeptical view, our Supreme Court has found, "There are undoubtedly situations in which notwithstanding the most exemplary charge, a juror will find it impossible to disregard such a prejudicial statement." State v. Boone, 66 N.J. 38, 48, 327 A.2d 661 (1974) (citing Krulewitch, 336 U.S. at 453, 69 S.Ct. 716 ). For example, the Court found that a limiting instruction could never cure the prejudicial effect from the admission of a defendant's prior but withdrawn guilty plea. Id. at 50, 327 A.2d 661.
The United States Supreme Court reached the same conclusion regarding the admission of a co-conspirator's confession that implicates a defendant. "[T]here are some contexts in which the risk that the jury will not, or cannot, follow instructions is so great, and the consequences of failure so vital to the defendant, that the practical and human limitations of the jury system cannot be ignored." Bruton v. United States, 391 U.S. 123, 135, 88 S.Ct. 1620, 20 L.Ed.2d 476 (1968).
Without delving into the many empirical studies on jury behavior, we believe that jury compliance with curative and limiting instructions is neither all truth nor all fiction. The answer is somewhere in between. See David A. Sklansky, Evidentiary Instructions and the Jury as Other, 65 Stan. L. Rev. 407, 423-39 (2013) (Evidentiary Instructions ) (analyzing various empirical studies). As Professor Sklansky has noted, "The reality is ... that evidentiary instructions probably do work, but imperfectly, and better under some conditions than others." Id. at 409.
*505The decision to opt for a curative or limiting instruction, instead of a mistrial or new trial, depends on at least three factors. First, a court should consider the nature of the inadmissible evidence the jury heard, and its prejudicial effect. "The adequacy of a curative instruction necessarily focuses on the capacity of the offending evidence to lead to a verdict that could not otherwise be justly reached." Winter, 96 N.J. at 647, 477 A.2d 323. Additionally, while a general charge may suffice to cure "only slightly improper" remarks, "a single curative instruction may not be sufficient to cure the prejudice resulting from cumulative errors at trial." State v. Vallejo, 198 N.J. 122, 136, 965 A.2d 1181 (2009) (quoting State v. Frost, 158 N.J. 76, 86-87, 727 A.2d 1 (1999) ).
*700Evidence that bears directly on the ultimate issue before the jury may be less suitable to curative or limiting instructions than evidence that is indirect and that requires additional logical linkages. For example, distinguishing between a co-conspirator's confession that directly implicates a defendant and a confession that only inferentially does so, the United States Supreme Court noted that "[s]pecific testimony that 'the defendant helped me commit the crime' is more vivid than inferential incrimination, and hence more difficult to thrust out of mind." Richardson v. Marsh, 481 U.S. 200, 208, 107 S.Ct. 1702, 95 L.Ed.2d 176 (1987). Consequently, "with regard to inferential incrimination[,] the judge's instruction may well be successful in dissuading the jury from entering onto the path of inference in the first place, so that there is no incrimination to forget." Ibid. Likewise, in Harvey, 151 N.J. at 205-06, 699 A.2d 596, the Court held that a curative instruction was sufficient to avoid a mistrial where the stricken testimony pertained to polygraph results of an earlier suspect, and there was substantial evidence to eliminate that suspect, rendering the prejudice to defendant "minimal." The Court emphasized that the evidence that the earlier suspect passed a polygraph did not directly prove defendant's guilt. Id. at 205, 699 A.2d 596.
Second, an instruction's timing and substance affect its likelihood of success. As for timing, our Court has held that a swift and *506firm instruction is better than a delayed one. Winter, 96 N.J. at 648, 477 A.2d 323 (noting the importance of an immediate and firm instruction to disregard an offending remark); see also Vallejo, 198 N.J. at 134-35, 965 A.2d 1181 (citing cases finding effective curative instructions). Delay may allow prejudicial evidence to become cemented into a storyline the jurors create in their minds during the course of the trial. See Evidentiary Instructions at 422 n.52; see also id. at 451 (stating "[t]he timing ... of instructions [is] likely to matter"). That is why our Supreme Court has stated - in the context of admitting evidence of other crimes under N.J.R.E. 404(b) - it is the "better practice" to give limiting instructions at the time the evidence is presented and again in the final jury charge. State v. Blakney, 189 N.J. 88, 93, 912 A.2d 140 (2006). It is thought that repeating the instruction prevents the jurors from "indelibly brand[ing] the defendant as a bad person" and blinding them from careful consideration of all the evidence in deliberations. Ibid.
As for substance, a specific and explanatory instruction is often more effective than a general, conclusory one. The "Court has consistently stressed the importance of ... specificity when trial judges provide curative instructions to alleviate potential prejudice to a defendant from inadmissible evidence that has seeped into a trial." Vallejo, 198 N.J. at 135, 965 A.2d 1181. "[B]ecause 'the inherently prejudicial nature of [404(b) ] evidence casts doubt on a jury's ability to follow even the most precise limiting instruction,' the court's instruction 'should be formulated carefully to explain precisely the permitted and prohibited purposes of the evidence ....' " State v. Fortin, 162 N.J. 517, 534, 745 A.2d 509 (2000) (quoting State v. Stevens, 115 N.J. 289, 309, 304, 558 A.2d 833 (1989) ); see also State v. Cofield, 127 N.J. 328, 341, 605 A.2d 230 (1992).
An instruction also can be more effective when it explains itself. "Because I said so" is likely to be even less effective from a judge to a jury than it is from a parent to an eight-year-old. See Evidentiary Instructions at 439 (stating, based on a review of *507empirical research, that instructions "work better when the judge gives the jury a reason to follow them"); id="p701" href="#p701" data-label="701" data-citation-index="1" class="page-label">*701id. at 452 (noting, subject to exception, that "[o]n the whole, mock jury studies do suggest that evidentiary instructions are more apt to be followed if the judge explains the reason for the underlying rule").4
Although trial judges may understandably try to avoid repeating and thereby reinforcing an offending remark, a court must describe it with enough specificity to enable the jury to follow the instruction. The instruction must be "clear enough [and] sharp enough to achieve its goal." Vallejo, 198 N.J. at 136-37, 965 A.2d 1181 (holding that an "instruction did not fulfill its purpose" where judge referred too generally to "things ... blurted out that have nothing to do with this case").
Third, a court must ultimately consider its tolerance for the risk of imperfect compliance. See Bruton, 391 U.S. at 135, 88 S.Ct. 1620 (referring to "consequences of failure so vital to" a criminal defendant). Yet, even in criminal cases involving errors of constitutional dimension, "not 'any' possibility [of an unjust result] can be enough for a rerun of the trial." Winter, 96 N.J. at 647, 477 A.2d 323 (quoting State v. Macon, 57 N.J. 325, 336, 273 A.2d 1 (1971) ). "The possibility must be real, one sufficient to raise a reasonable doubt as to whether the error led the jury to a result it otherwise might not have reached." Ibid. (quoting Macon, 57 N.J. at 336, 273 A.2d 1 ). By contrast, a non-constitutional error "shall *508be disregarded by the appellate court 'unless it is of a nature as to have been clearly capable of producing an unjust result.' " Id. at 648, 477 A.2d 323 (quoting State v. LaPorte, 62 N.J. 312, 318-19, 301 A.2d 146 (1973) ).
The United States Supreme Court has required an "overwhelming probability" that the jury cannot comply, in order to conclude a curative instruction was inadequate. Greer v. Miller, 483 U.S. 756, 766 n.8, 107 S.Ct. 3102, 97 L.Ed.2d 618 (1987) ("We normally presume that a jury will follow an instruction to disregard inadmissible evidence inadvertently presented to it, unless there is an 'overwhelming probability' that the jury will be unable to follow the court's instructions, and a strong likelihood that the effect of the evidence would be 'devastating' to the defendant." (citing Richardson, 481 U.S. at 208, 107 S.Ct. 1702 ; then citing Bruton, 391 U.S. at 136, 88 S.Ct. 1620 ) ). We note that our own Supreme Court has not expressly adopted the "overwhelming probability" standard.
C.
Applying these principles, we are constrained to conclude that the court's instructions did not cure the prejudicial impact of the detective's inadmissible statements that defendant was a gang member and the homicide occurred in a gang area. We reach this conclusion in large part because the judge's instructions missed the mark. It is one thing to assume jury compliance with a well-crafted curative or limiting instruction. It is quite another *702to assume compliance with an instruction that fails to clearly and sharply address the prejudicial aspect of the inadmissible evidence. An instruction can be curative only if the judicial medicine suits the ailment.
The detective's gang references were prejudicial. They may not be minimized as "fleeting comments" that likely escaped the jury's notice. Cf. Jackowitz v. Lang, 408 N.J. Super. 495, 505, 975 A.2d 531 (App. Div. 2009) ("Fleeting comments, even if improper, may not warrant a new trial, particularly when the verdict is fair."). Each time the detective referred to gangs, the trial came to an *509abrupt halt. The second time, when the detective called defendant a gang member, the jury gasped, according to defense counsel at sidebar. The assistant prosecutor contended the gasp followed his own loud reaction to the detective's statement. Either way, the detective's comment was not missed.
The comments filled a hole in the State's case: defendant's motive for killing Claudio. The State argued that defendant was retaliating for the beating he received a week earlier. But, that theory had a problem. The target was not the person who did the beating. Inserting the gang element provided a reason for the killing. The jury could conclude that the homicide arose out of a gang conflict involving both Claudio and Torres.
The second comment was particularly prejudicial because it directly tarred defendant as a gang member. In holding that evidence of gang membership was properly analyzed under N.J.R.E. 404(b) as evidence of other crimes and wrongs, we observed that "membership in a street gang ... is at the very least strongly suggestive of" criminal activity. Goodman, 415 N.J. Super. at 227, 1 A.3d 767. A "mere ... allegation[ ] of gang membership carries a strong taint of criminality." Ibid. (quoting United States v. Acosta, 110 F.Supp.2d 918, 931 (E.D. Wis. 2000) ). Evidence of past criminality risks conviction because the jury may conclude defendant is a bad person with a propensity to commit crimes. State v. Skinner, 218 N.J. 496, 514, 95 A.3d 236 (2014) ; State v. Rose, 206 N.J. 141, 180, 19 A.3d 985 (2011) (citing United States v. Green, 617 F.3d 233, 248-49 (3d Cir. 2010) ). "There is widespread agreement that other-crime evidence has a unique tendency to turn a jury against the defendant. 'The likelihood of prejudice is acute when the proffered evidence is proof of a defendant's uncharged misconduct.' " Stevens, 115 N.J. at 302, 558 A.2d 833 (quoting Edward J. Imwinkelried, The Need to Amend Federal Rule of Evidence 404(b) : The Threat to the Future of the Federal Rules of Evidence, 30 Vill. L. Rev. 1465, 1487 (1985) ).
The judge's instructions following the gang references missed the target. In response to Crawley's first mention of gangs, the court did not address the issue at all. Rather, the judge focused on *510the fact that Crawley was speculating about why teenagers did not, or would not, want to cooperate. So, the jury was free to use the testimony that gangs were rampant in the area for another purpose: that the crime charged was somehow related to gangs.
The judge's instruction following the second mention only partly addressed the prejudice of Crawley's comment that defendant was a gang member. The judge said, "[T]here's no information in this case whatsoever there's any gangs involved in this case." First, this statement was inaccurate, because the detective testified both that the playground was a gang area - which the court did not strike - and that defendant, a gang member, was involved. Also, the court's assertion that there was "no information in this case" about gang *703involvement did not contradict the truth of the detective's statement.
At best, the jury could understand the judge's statement to mean there was no evidence that a gang had ordered the homicide or the homicide arose out of a gang rivalry. But the judge's statement did not directly address defendant's membership, which conveyed the taint of criminality and propensity to commit crimes.5 Further, the court's direction that the jury not consider the "gang situation" suffered from vagueness.
The judge also asserted, as a fact, that Detective Crawley's statement about defendant was "unintentional." But the court's fact-finding lacked evidence in the record.6 It was also inappropriate, because it bolstered the detective's credibility in the juror's *511minds. The jury was more likely to consider the statement to be true because it was unintentional.7 *704As a result of these deficiencies, the risk of the jury's non-compliance with the court's instructions was intolerably high. The *512State's case was far from overwhelming, as it depended on the often-inconsistent testimony of two eyewitnesses. The trial judge recognized, in the course of one side-bar discussion, that this was a "very close case." And in the colloquy after Crawley's second gang reference, specifically alleging defendant was a gang member, the judge stated that if there were further errors, the State would be "bordering on a mistrial."
The question before us is not whether comprehensive and well-targeted instructions could have cured the taint of the inadmissible references to gangs and to defendant's gang membership. The instructions did not fully and clearly address the prejudicial aspects of the testimony.
Therefore, we conclude that the gang references caused substantial prejudice, which the judge's instructions did not cure. On this basis, we are constrained to reverse the conviction. Given our conclusion on this point, we need not address in this opinion whether the court erred in its instructions on the subject of the eyewitnesses' use of pseudonyms and expressions of fear.8
As for the other issues raised on appeal in the counseled and pro se briefs challenging the conviction, we find they lack sufficient merit to warrant discussion. R. 2:11-3(e)(2). We add that Crawley's explanation for the differences in the eyewitnesses' description would have been inappropriate opinion testimony; but defense counsel opened the door to the subject by challenging Crawley's reasons for obtaining an arrest warrant despite the differences.
Reversed and remanded for a new trial. We do not retain jurisdiction.