Commonwealth v. Crayton

The defendant was convicted by a Superior Court jury on two indictments of possession of child pornography, in violation of G. L. c. 272, § 29C. 1 We granted the defendant’s application for direct appellate review. In his appeal, the defendant claims that the trial judge made three errors that warrant a new trial. First, he contends that the judge erred in admitting in evidence the in-court identifications of the defendant by two eyewitnesses who had not previously participated in an out-of-court identification procedure. Second, the defendant claims that, where the defendant admitted to police that he had used library computers on the day in question but denied having used them to view child pornography, the judge erred in allowing in evidence the admission but excluding from evidence the denial. Third, he argues that the judge erred in admitting in evidence three pornographic drawings of children that were found in the defendant’s possession ten months after he allegedly viewed the child pomo- *230 graphy charged in the indictments. We establish a new standard for the admission of in-court identifications where the eyewitness had not previously participated in an out-of-court identification procedure and conclude that the in-court identifications in this case would not have been admissible under that standard. We also conclude that the judge erred in excluding from evidence the defendant’s denial and in admitting in evidence the drawings, and that the errors and the admission of the in-court identifications, considered together, resulted in unfair prejudice that requires that the convictions be vacated and a new trial ordered. 2 3 - 8

Background. We summarize the evidence at trial, reserving discussion of the evidence that pertains to the issues on appeal. On January 21, 2009, between approximately 3:30 p.m. and 4 p.m., an eighth grade student, M.S., was doing homework at a computer in the basement technology center of the Central Square branch of the Cambridge Public Library. 4 A man she described as short, white, and bald, with a “little beard” and eyeglasses was sitting at an adjacent computer to the right of her. 5 **She went to the library “[mjostly every day,” but had never seen the man before. When she looked at his computer screen, she saw an image of “a girl about ten years old, covering her chest.” She could not tell whether the girl was wearing any clothes, because she saw only a “top view” and the man was “covering] the computer screen” with the “umbrella-type” cover that was on it. 6 She “waved” at her friend, R.M., a ninth grade student, who was also in the technology center of the library, and urged him to look at the man’s computer. R.M. testified that he “just got a quick glimpse of the computer,” and could only see “a small portion” of the screen, which displayed a young child wearing no clothes. He saw only the side of the man’s face; he described the man as bald with a goatee. He went to the library every day after school, but had not seen the man before. During trial, both M.S. and R.M. *231 identified the defendant as the man that they had seen at the computer on January 21.

M.S. and R.M. walked over to Ricardo Negron, a library employee who was working at the staff desk in the technology center that afternoon, and they told him that a person was looking at children wearing no clothes on the computer. 7 Before M.S. and R.M. approached him, Negron had observed M.S. at computer no. one and a white male, “perhaps” in his “early thirties,” bald, with eyeglasses, whom he had seen before at the technology center, at computer no. two. 8 The police later showed Negron an array of photographs, but he was unable to identify anyone from the array. 9 ’ 10

Library users were required to log on to a computer by entering their library bar code, so when the two teenagers alerted Negron to what they had seen, Negron looked up the log-in information for computer no. two. While he was doing so, the man using computer no. two logged off and left the room. The log inquiry revealed that a person using the library card of an eighteen year old male, “perhaps of Asian descent,” had logged on to computer no. two at 3:08 p.m. and logged off at 3:55 p.m. 11 At some time after *232 3:55 p.m.j Negron went upstairs to speak to the library manager, Esme Green. Green went downstairs to the technology center, looked at two “video clips” saved on computer no. two, saw that they depicted an approximately twelve year old girl, “either naked or almost naked, masturbating,” and telephoned the police.

When Negron went upstairs, another library employee, Ricardo Ricard, went downstairs to staff the technology center. Having learned of the allegation, Ricard logged on to computer no. two, saw a folder on the computer with the label “W,” and looked at a video file inside the folder, which showed a nude female child. Because he was concerned that the library computers deleted all files when they were shut down for the night, Ricard transferred the folder containing the file to a universal serial bus (USB) drive, which he later gave to Green. He then disabled the computer’s “reboot” software so that the computer would retain the files that were then on it.

Ricard had not seen the man who used computer no. two on January 21, but he was aware of the man’s physical description. On January 22, when he saw a man who matched that description in the library lobby, he told Green of the man’s presence, and Green notified the police.

Detectives Brian O’Connor and Pam Clair of the Cambridge police department arrived at the library and saw the defendant at a computer with another individual. The detectives observed the defendant for approximately twenty to thirty minutes at a computer that displayed a “MySpace” profile page, “looking at MySpace.” As the defendant was leaving the library, Detective O’Connor asked to speak with him, and the defendant agreed. The defendant admitted that he had been in the library’s computer room the previous day. He said he had used one of the computers for five minutes and then switched to another computer, which he identified as computer no. two, to check his electronic mail (email). The defendant said that his e-mail address was cblizzard @yahoo.com. He also said that he did not have his own MySpace profile, but used his friend’s profile.

After this conversation, Detective O’Connor obtained the USB drive that Ricard had given to Green, seized computer no. two, and copied the folder labeled “W” onto a compact disc. After *233 obtaining a search warrant, Detective O’Connor conducted a forensic search of the hard drive of computer no. two. That search revealed twenty-seven “cookies,” which O’Connor described as “text file[s]” that store information on an Internet browser regarding a Web site that a particular user has visited on the Internet. 12 The first of these cookies, entitled “magic-Lolita(l) .txt,” was created at 3:14 p.m. on January 21; the last, entitled “www.inno-centgirls(l).txt,” was created at 3:48 p.m. that day. Detective O’Con-nor also uncovered “Yahoo searches” on computer no. two that had been conducted between 3:14 and 3:25 p.m. on January 21 using such search terms as “One hundred percent Lolita” and “Top Lolita.” Detective O’Connor also located temporary Internet files on the computer’s hard drive in which images were automatically downloaded by the Internet browser from a Web site that the user visited. In those temporary files, he found approximately 210 photographs where children were engaged in sexual acts, of which seven were printed out and admitted as exhibits at trial. These seven images were created on the computer between 3:27 and 3:50 p.m. on January 21. The detective also located six video files on the hard drive of the computer, of which two video files were located in a temporary Internet file folder and four video files were located in a folder entitled “W.” The four video files in the “W” folder, which were played for the jury, were created on the computer between 3:43 and 3:54 p.m. that day. Detective O’Connor also located a MySpace page in the temporary Internet files reflecting a log-in date and time of January 21 at 3:13 p.m. The MySpace page identifies the user as “Walter”; the e-mail address associated with the MySpace page was C-Blizzard69@MySpace.com.

Discussion. 1. In-court identifications of the defendant. Before trial, neither the police nor the prosecutor asked M.S. or R.M. to participate in an identification procedure to determine whether they could identify the man they had seen at the computer on January 21, 2009. They were never shown a photographic array or asked to view a lineup. The first time they were asked to identify the man was on April 7, 2011 — more than two years after the first and only time they had seen him — when they were asked by the prosecutor on the witness stand at trial whether they saw the man in the court room, and each identified the defendant.

The defendant moved before trial to preclude the Commonwealth from eliciting an in-court identification of the defendant *234 from any witness that had not previously made an out-of-court identification, including M.S. and R.M. The defendant contended that, under such circumstances, an in-court identification of the defendant would be inherently and unnecessarily suggestive. At a pretrial hearing on the motion, the defendant requested that a voir dire of the teenagers be conducted before any in-court identification was elicited. On the first day of trial, before either M.S. or R.M. had testified, the judge denied the motion, and also denied the request for a voir dire. The judge said that she might reconsider her ruling if the prosecutor failed to lay an adequate foundation through the eyewitnesses’ trial testimony before eliciting the in-court identifications. The judge noted that the in-court identifications could not be tainted by a suggestive pretrial identification procedure where there had been none. The judge recognized that “an in-court identification always has some suggestiveness to it,” but said that defense counsel “[could] highlight that suggestiveness” on cross-examination. The judge noted defense counsel’s objection to her ruling. Although the defendant did not renew his objection when the in-court identifications of M.S. and R.M. were elicited, where the judge noted the earlier objection, we shall treat the claim of error as preserved. See Commonwealth v. Aviles, 461 Mass. 60, 66 (2011), and cases cited.

We look first to our existing case law on the admission of eyewitness identification testimony. “Under art. 12 of the Massachusetts Declaration of Rights, an out-of-court eyewitness identification is not admissible where the defendant proves by a preponderance of the evidence, considering the totality of the circumstances, that the identification is so unnecessarily suggestive and conducive to irreparable misidentification that its admission would deprive the defendant of his right to due process.” Commonwealth v. Walker, 460 Mass. 590, 599 (2011), citing Commonwealth v. Johnson, 420 Mass. 458, 463-464 (1995), and Commonwealth v. Thornley, 406 Mass. 96, 98 (1989). In contrast with the United States Supreme Court, which has ruled under the Fourteenth Amendment to the United States Constitution that an out-of-court identification that is unnecessarily suggestive will be admissible if it is reliable under “the totality of the circumstances,” Manson v. Brathwaite, 432 U.S. 98, 110, 113 (1977), we have said that “the reliability test does little or nothing to discourage police from using suggestive identification procedures,” and that “[o]nly a rule of per se exclusion can ensure the continued protection against the danger of mistaken identification and wrongful convictions” *235 arising from suggestive identification procedures. Johnson, supra at 468, 472. See Walker, supra at 599 n.13.

In addition, where an unreliable identification arises from “especially suggestive circumstances” other than an unnecessarily suggestive identification procedure conducted by the police, we have declared that “[c]ommon law principles of fairness” dictate that the identification should not be admitted. Commonwealth v. Jones, 423 Mass. 99, 109 (1996). Our reliance on common-law principles of fairness to suppress an identification made under “especially suggestive circumstances” even where the circumstances did not result from improper police activity is also in contrast with the United States Supreme Court jurisprudence. Compare id. with Perry v. New Hampshire, 132 S. Ct. 716, 720-721 (2012).

We have applied the “unnecessarily suggestive” standard to showup identifications, where the police show a suspect to an eyewitness individually rather than as part of a lineup or photographic array. See, e.g., Commonwealth v. Phillips, 452 Mass. 617, 628-629 (2008); Commonwealth v. Martin, 447 Mass. 274, 279-281 (2006). Such “[o]ne-on-one identifications are generally disfavored because they are viewed as inherently suggestive,” Martin, supra at 279, but suggestiveness alone is not sufficient to render a showup identification inadmissible in evidence; the defendant must prove by a preponderance of the evidence that it was “unnecessarily suggestive” (emphasis in original). Commonwealth v. Figueroa, 468 Mass. 204, 217 (2014), quoting Phillips, supra at 627.

A showup identification may be unnecessarily suggestive for two reasons. First, it may be unnecessarily suggestive where there was not “good reason ... for the police to use a one-on-one identification procedure.” Martin, 447 Mass. at 279, quoting Commonwealth v. Austin, 421 Mass. 357, 361 (1995). See Commonwealth v. Meas, 467 Mass. 434, 441, cert, denied, 135 S. Ct. 150 (2014). Although “good reason” for a showup identification does not require exigent or special circumstances, see Martin, supra, quoting Austin, supra, there is generally “good reason” where the showup identification occurs within a few hours of the crime, because it is important to learn whether the police have captured the perpetrator or whether the perpetrator is still at large, and because a prompt identification is more likely to *236 be accurate when the witness’s recollection of the event is still fresh. See Figueroa, 468 Mass. at 217-218 (“good reason” existed for showup identification occurring two and one-half hours after murder); Phillips, 452 Mass. at 628-629 (good reason existed for showup identification within one hour after murder and armed robbery).

Second, “[ejven where there is ‘good reason’ for a showup identification, it may still be suppressed if the identification procedure so needlessly adds to the suggestiveness inherent in such an identification that it is ‘conducive to irreparable mistaken identification.’ ” Figueroa, 468 Mass. at 217, quoting Phillips, 452 Mass. at 628. See Commonwealth v. Leaster, 395 Mass. 96, 103 (1985) (even where showup occurs promptly after crime, “if there are special elements of unfairness, indicating a desire on the part of the police to ‘stack the deck’ against the defendant, an identification resulting from such a confrontation would be inadmissible”); Commonwealth v. Moon, 380 Mass. 751, 756-759 (1980) (identification procedure unnecessarily suggestive where police suggested name of defendant to victim and then showed him single photograph that police removed from vehicle that eyewitness thought belonged to assailant). 13

Where, as here, a prosecutor asks a witness at trial whether he or she can identify the perpetrator of the crime in the court room, and the defendant is sitting at counsel’s table, the in-court identification is comparable in its suggestiveness to a showup identification. See Commonwealth v. Carr, 464 Mass. 855, 877 (2013), quoting Commonwealth v. Bol Choeurn, 446 Mass. 510, 519-520 (2006) (“We have long recognized that ‘a degree of *237 suggestiveness inheres in any identification of a suspect who is isolated in a court room’ ”). See also Perry v. New Hampshire, 132 S. Ct. at 727 (all in-court identifications “involve some element of suggestion”). Although the defendant is not alone in the court room, even a witness who had never seen the defendant will infer that the defendant is sitting with counsel at the defense table, and can easily infer who is the defendant and who is the attorney. 14 See United States v. Archibald, 734 F.2d 938, 941, modified, 756 F.2d 223 (2d Cir. 1984) (“Any witness, especially one who has watched trials on television, can determine which of the individuals in the courtroom is the defendant...”).

In fact, in-court identifications may be more suggestive than showups. See Mandery, Due Process Considerations of In-Court Identifications, 60 Alb. L. Rev. 389, 415 (1996) (“If anything, the evidence suggests that in-court identifications merit greater protection” than pretrial identifications). At a showup that occurs within hours of a crime, the eyewitness likely knows that the police suspect the individual, but unless the police say more than they should, the eyewitness is unlikely to know how confident the police are in their suspicion. However, where the prosecutor asks the eyewitness if the person who committed the crime is in the court room, the eyewitness knows that the defendant has been charged and is being tried for that crime. The presence of the defendant in the court room is likely to be understood by the eyewitness as confirmation that the prosecutor, as a result of the criminal investigation, believes that the defendant is the person whom the eyewitness saw commit the crime. Under such circumstances, eyewitnesses may identify the defendant out of reliance on the prosecutor and in conformity with what is expected of them rather than because their memory is reliable. See id. at 417-418 (“The pressure of being asked to make an identification in the formal courtroom setting and the lack of anonymity . . . create conditions under which a witness is most likely to conform his or her recollection to expectations, either by identifying the particular person whom he or she knows the authorities desire identified, or by acting in conformity with the behavior of others they may have seen on television ...”).

Although we have adopted a “rule of per se exclusion” for unnecessarily suggestive out-of-court identifications, we have not *238 adopted such a rule for in-court identifications, despite their comparable suggestiveness. See Bol Choeurn, 446 Mass. at 519-520, quoting Commonwealth v. Napolitano, 378 Mass. 599, 604 (1979), S.C., Napolitano v. Attorney General, 432 Mass. 240 (2000) (“We have long recognized that ‘a degree of suggestiveness inheres in any identification of a suspect who is isolated in a court room’ . . . [, but that] ‘does not, in itself, render the identification impermissibly suggestive’ ”). Instead, we have excluded an in-court identification only where “it is tainted by an out-of-court confrontation . . . that is ‘so impermissibly suggestive as to give rise to a very substantial likelihood of irreparable misidentification.’ ” Carr, 464 Mass. at 877, quoting Bol Choeurn, supra at 520. In essence, we have excluded in-court identifications only where their inherent suggestiveness is magnified by the impermissible suggestiveness of an out-of-court identification. Therefore, here, where there had been no out-of-court identification to taint the in-court identification, the judge’s admission of the in-court identification conformed to our case law. We now do what a trial judge cannot do — revisit the wisdom of our case law regarding the admission of in-court identifications where the eyewitness has not earlier been asked to make an out-of-court identification.

There can be no doubt that, if the police, more than twenty-six months after the incident, had brought M.S. or R.M. to any room other than a court room on the day of trial, identified one of the persons in the room as the defendant, and asked whether the person they had seen looking at images of nude children in the library that day was in the room, we would conclude that the showup identification was unnecessarily suggestive and therefore inadmissible in evidence, especially where this had been the first identification procedure the police had attempted. The question we must confront is whether such an otherwise inadmissible showup identification becomes admissible because the showup occurred in the court room.

A first-time in-court identification differs from an out-of-court showup in three ways, so we must evaluate whether these differences justify the admission of an in-court identification that would be inadmissible if it occurred out of court. The first difference is that, with an in-court identification, the jury see the identification procedure, whereas the jury do not see a showup identification procedure unless the police videotape the procedure. “[W]hen a first-time eyewitness identification occurs in *239 court and no suggestive pretrial identification procedures were administered by the state, courts generally have concluded that the factfinder is better able to evaluate the reliability of the identification because he or she can observe the witness’s demeanor and hear the witness’s statements during the identification procedure” (emphasis in original). State v. Hickman, 355 Or. 715, 735 (2014), citing Byrd v. State, 25 A.3d 761, 766 (Del. 2011), and United States v. Domina, 784 F.2d 1361, 1368 (9th Cir. 1986), cert, denied, 479 U.S. 1038 (1987). This conclusion appears to be premised on the ability of the jury during an in-court identification to see “indications of witness certainty or hesitation during the identification process, including facial expression, voice inflection, and body language,” and to make “other observations pertinent to assessing the reliability of a person’s statements.” Hickman, supra.

We agree that a jury may be better able to assess a witness’s level of confidence during an in-court identification than through evidence of a showup, but we do not agree that this means that a jury are better able to evaluate the accuracy of an in-court identification. Social science research has shown that a witness’s level of confidence in an identification is not a reliable predictor of the accuracy of the identification, especially where the level of confidence is inflated by its suggestiveness. See Supreme Judicial Court Study Group on Eyewitness Evidence: Report and Recommendations to the Justices 19 (July 25, 2013) (SJC Study Group Report), citing State v. Lawson, 352 Or. 724, 111 (2012), and State v. Guilbert, 306 Conn. 218, 253 (2012). 15 Moreover, even if we were persuaded that there were evaluative benefits arising *240 from the jury’s ability to see the identification procedure, it would not justify admission of an inherently suggestive identification. Certainly, where there was not good reason to conduct an out-of-court showup, an identification arising from such a showup would not be admissible because the police have videotaped it.

The second difference between a first-time in-court identification and a showup is that the former occurs in court, and therefore “is subject to immediate challenge through cross-examination.” Hickman, 355 Or. at 735. Some other courts have concluded that “[wjhere a witness first identifies the defendant at trial, defense counsel may test the perceptions, memory and bias of the witness, contemporaneously exposing weaknesses and adding perspective in order to lessen the hazards of undue weight or mistake.” Id., quoting People v. Rodriguez, 134 Ill. App. 3d 582, 589 (1985), cert, denied, 475 U.S. 1089 (1986). We are not persuaded that the immediacy of cross-examination materially lessens “the hazards of undue weight or mistake” arising from a suggestive identification. Eyewitnesses are routinely subject to cross-examination regarding their showup identifications, but that does not render such identifications admissible where they are unnecessarily suggestive. Cf. Walker, 460 Mass. at 606-608. Moreover, we have previously recognized how difficult it is for a defense attorney to convince a jury that an eyewitness’s confident identification might be attributable to the suggestive influence of the circumstances surrounding the identification. See Jones, 423 Mass. at 110 (“This is not a case in which cross-examination and a judge’s jury instruction concerning eyewitness testimony can fairly protect the defendant from the unreliability of [the eyewitness’s] identification”). See also Perry v. New Hampshire, 132 S. Ct. at 737 (Sotomayor, J., dissenting), quoting Kansas v. Ventris, 556 U.S. 586, 594 n.* (2009) (“[E]yewitness identifications upend the ordinary expectation that it is ‘the province of the jury to weigh the credibility of competing witnesses.’ . . . [Jjurors find eyewitness evidence unusually powerful and their ability to assess credibility is hindered by a witness’ false confidence in the accuracy of his or her identification”). Nor is the immediacy of cross-examination likely to make the cross-examination more effective in revealing the risk of inaccuracy. In fact, such immediacy means that defense counsel has little opportunity to prepare *241 an effective cross-examination regarding the identification, because it occurred minutes earlier.

The third difference between a first-time in-court identification and a showup is that, where defense counsel has advance warning that the prosecutor intends to ask the eyewitness at trial to identify the defendant, defense counsel has the opportunity to propose alternative identification procedures that are less suggestive, “such as an in-court line-up, or having the defendant sit somewhere in the courtroom other than the defense table.” Domina, 784 F.2d at 1368-1369. See United States v. Brown, 699 F.2d 585, 594 (2d Cir. 1983) (“[W]hen a defendant is sufficiently aware in advance that identification testimony will be presented at trial and fears irreparable suggestivity, as was the case here, his remedy is to move for a line-up in order to assure that the identification witness will first view the suspect with others of like description rather than in the courtroom sitting alone at the defense table”).

We do not join those courts that have placed the burden on the defendant to avoid a suggestive in-court identification by proposing alternative, less suggestive identification procedures. See Hickman, 355 Or. at 742-743, citing Brown, 699 F.2d at 594, and Domina, 784 F.2d at 1369 (“Courts considering the admissibility of first-time in-court identifications generally have placed the burden of seeking a prophylactic remedy on the defendant”). Placing this burden on the defendant suggests that the Commonwealth is entitled to an unnecessarily suggestive in-court identification unless the defendant proposes a less suggestive alternative that the trial judge in his or her discretion adopts. See Domina, supra (“[Pjarticular methods of lessening the suggestiveness of in-court identification ... are matters within the discretion of the court”). We decline to grant the Commonwealth such an entitlement where, as here, the Commonwealth failed earlier to conduct a less suggestive out-of-court identification procedure, and the in-court identification is therefore the only identification of the defendant made by an eyewitness.

Where an eyewitness has not participated before trial in an identification procedure, we shall treat the in-court identification as an in-court showup, and shall admit it in evidence only where there is “good reason” for its admission. 16 The new rule we declare today shall apply prospectively to trials that commence *242 after issuance of this opinion, and shall apply only to in-court identifications of the defendant by eyewitnesses who were present during the commission of the crime. 17

We recognize that the “good reason” that generally justifies most out-of-court showups — i.e., “concerns for public safety; the need for efficient police investigation in the immediate aftermath of a crime; and the usefulness of prompt confirmation of the accuracy of investigatory information,” Austin, 421 Mass. at 362 — depends on the short duration of time between the crime and the showup, and will never justify an in-court showup. But there may be other grounds that constitute “good reason” for an in-court showup where there has not been a nonsuggestive out-of-court identification procedure. For instance, there may be “good reason” for the first identification procedure to be an in-court showup where the eyewitness was familiar with the defendant before the commission of the crime, such as where a victim testifies to a crime of domestic violence. Cf. Carr, 464 Mass. at 858, 874, 877 (in-court identifications not impermissibly suggestive where eyewitnesses had known defendant from neighborhood prior to murder); Commonwealth v. Cong Duc Le, 444 Mass. 431, 443 & n.9 (2005) (in-court identifications not impermissibly suggestive where witness knew defendants and identification was not issue at trial). “Good reason” might also exist where the witness is an arresting officer who was also an eyewitness to the commission of the crime, and the identification merely confirms that the defendant is the person who was arrested for the charged crime. In both of these circumstances, the in-court showup is understood by the jury as confirmation that the defendant sitting in the court room is the person whose conduct is at issue rather than as identification evidence. See People v. Rodriguez, 79 N.Y.2d 445, 449-450 & n.* (1992) (“confirmatory identification” exception to requirement of pretrial hearing on admissibility of suggestive pretrial identification applies where eyewitness and *243 defendant are “known to one another” or where defendant’s identity is not live issue at trial). And in both of these circumstances, where the witness is not identifying the defendant based solely on his or her memory of witnessing the defendant at the time of the crime, there is little risk of misidentification arising from the in-court showup despite its suggestiveness.

Although we generally place the burden on the defendant to move to suppress an identification, that makes little sense where there is no out-of-court identification of the defendant by a witness and only the prosecutor knows whether he or she intends to elicit an in-court identification from the witness. If the burden were on the defendant to move to suppress an identification in these circumstances, a defendant would need to file motions to suppress the in-court identification of witnesses whom the prosecutor might not intend to ask to make such an identification. To avoid the filing of needless motions, we place the burden on the prosecutor to move in limine to admit the in-court identification of the defendant by a witness where there has been no out-of-court identification. 18 Once the motion is filed, the defendant would continue to bear the burden of showing that the in-court identification would be unnecessarily suggestive and that there is not “good reason” for it. See Martin, 447 Mass. at 279-280, 283 n.6, citing Commonwealth v. Odware, 429 Mass. 231,235 (1999). Although we impose no restrictions on when such a motion must be filed, a prosecutor would be wise to file it in advance of trial, because, if the defendant were to prevail in suppressing the in-court identification as unnecessarily suggestive, the Commonwealth would still have time, if it chose, to conduct a less suggestive out-of-court identification procedure. 19 *244 Limiting in-court showups under the “good reason” standard need not diminish the important evidentiary role of reliable eyewitness identifications. See Walker, 460 Mass. at 604 n.16 (“eyewitness identification is ... an invaluable law enforcement tool in obtaining accurate convictions”). Reliable evidence of eyewitness identification will continue to be admissible where it arises from a nonsuggestive out-of-court identification procedure. Where a prosecutor recognizes during trial preparation that no lineup or photographic array has been shown to an eyewitness who may be able to identify the defendant, nothing bars the prosecutor from causing such an identification procedure to be conducted out-of-court before the witness takes the stand. All that is lost by barring first-time in-court showups where there is no “good reason” for such a showup is the unfair evidentiary weight of a needlessly suggestive showup identification that might be given more weight by a jury than it deserves. See id. (“eyewitness identification is the greatest source of wrongful convictions”). 20

*245 In this case, there was no “good reason” for the highly suggestive in-court identifications by M.S. and R.M., where the Commonwealth had abundant opportunity to attempt to obtain a far less suggestive out-of-court identification through a lineup or photographic array. But we cannot conclude that the judge abused her discretion in allowing the in-court identifications in evidence where their admission was in accord with the case law existing at the time of her decision, and where we only today apply the “good reason” standard to first-time in-court showups. 21 We there *246 fore reserve discussion of what to do about the admission of the suggestive in-court showups until later in this opinion, where we consider it in the context of the other claimed prejudicial errors.

2. Exclusion of the defendant’s statement of denial. When Detectives O’Connor and Clair confronted the defendant in the library on January 22 and asked to speak with him about an “incident” that had occurred the previous day, the defendant admitted that he had been in the library’s computer room and had used computer no. two to check his e-mail, but denied that he had used the library’s computers to view child pornography. The jury, however, did not learn of his denial because the judge before trial had allowed the Commonwealth’s motion in limine to exclude this denial as hearsay that was not otherwise admissible under the doctrine of verbal completeness. 22

Under the doctrine of verbal completeness, “ ‘[wjhen a party introduces a portion of a statement or writing in evidence,’ a judge has the discretion to ‘allow[ ] admission of other relevant portions of the same statement or writing which serve to “clarify the context” of the admitted portion.’ ” Commonwealth v. Aviles, 461 Mass. 60, 75 (2011), quoting Commonwealth v. Carmona, 428 Mass. 268, 272 (1998). The purpose of the doctrine is “to ensure that a party does not present ‘a fragmented and misleading version of events’ to the fact finder.” Aviles, supra, quoting Carmona, supra. “The doctrine of verbal completeness does not open the door for everything in a statement or document.” Aviles, supra, citing Kobayashi v. Orion Ventures, Inc., 42 Mass. App. *247 Ct. 492, 498 (1997). “To be admitted, ‘the additional portions of the statement must be (1) on the same subject as the admitted statement; (2) part of the same conversation as the admitted statement; and (3) necessary to the understanding of the admitted statement.’ ” Aviles, supra, quoting Commonwealth v. Eugene, 438 Mass. 343, 350-351 (2003).

Here, the defendant’s denial that he was using the library’s computers to view child pornography was on the same general subject as the other admitted statements he made to Detectives O’Connor and Clair, and was part of the same conversation, so its admissibility rested on whether its admission was necessary to a fair understanding of the admitted statements. We conclude that it was necessary. A reasonable jury would have understood from Detective O’Connor’s testimony that the “incident” he spoke to the defendant about was an allegation that the defendant had been seen viewing child pornography on a library computer. A reasonable jury might thus have expected that if the defendant had not viewed the child pornography, he would have denied it. But here, the defendant admitted that he had used library computers the previous day and, according to the detective, had specifically admitted to having used computer no. two. By excluding the defendant’s denial, the judge might have left the jury with the false impression that the defendant had not denied viewing the child pornography where an innocent person would have denied it, and therefore, there was a significant risk that a reasonable jury might have understood the other statements the defendant made to the detectives as an implied admission to having viewed the child pornography. See Commonwealth v. O’Dell, 392 Mass. 445, 447-449 (1984) (omission of defendant’s statements denying awareness that passenger had just committed a robbery “distort[ed] the meaning” of statements in which defendant admitted to driving getaway vehicle).

Under these circumstances, the defendant’s denial should have been admitted under the doctrine of verbal completeness to eliminate that risk. 23 See Commonwealth v. Watson, 377 Mass. 814, 832 (1979), S.C., 409 Mass. 110 (1991) (“If the evidence *248 used to prove the admission consists of a part of a statement, whether oral or written, by the defendant, he has the right [under the verbal completeness doctrine] to offer any other part of the same statement which tends to explain or disprove the claimed admission ... ”). The defendant’s statement that he was not using computer no. two to view child pornography contradicts the meaning that a reasonable jury could have attributed to the defendant’s statement that he used computer no. two on a day when other library users had seen someone who matched his description looking at child pornography on computer no. two (where child pornography was later found). As a matter of fairness to the defendant, his statement of denial should have been admitted in evidence, because it is an essential part of what the defendant meant to convey to the detectives, and because it contradicts the meaning that a reasonable jury might otherwise give to the defendant’s admitted statements. We therefore conclude that it was an abuse of discretion to grant the Commonwealth’s motion in limine to exclude the denial from evidence.

3. Admission of “other bad acts” evidence. Over the defendant’s objection, the judge allowed in part the Commonwealth’s motion in limine to admit in evidence three hand-drawn sketches of young girls that were found on October 29, 2009, “tucked away” with the defendant’s mail and other belongings in his cell at the Suffolk County house of correction during a “routine random cell search.” 24 All three drawings depicted very young girls who were nude or partially nude with their breasts and *249 genital areas exposed. Two of the drawings depicted girls engaged in sexual acts; one depicted a girl performing oral sex on a penis-shaped sex toy, 25 and the other depicted a girl masturbating with a sex toy.

In her final instructions to the jury, the judge said that the evidence was admitted only for the limited purpose “to show the defendant’s state of mind, his knowledge and intent,” and not to show “bad character or propensity.” 26 The defendant contends that the judge abused her discretion by admitting these drawings in evidence.

The standard for evaluating the admissibility of “other bad acts” evidence is well established. Evidence of a defendant’s prior or subsequent bad acts is inadmissible for the purpose of demonstrating the defendant’s bad character or propensity to commit the crimes charged. See Commonwealth v. Anestal, 463 Mass. 655, 665 (2012); Commonwealth v. Butler, 445 Mass. 568, 574 (2005). However, such evidence may be admissible for some other purpose, for instance, “to establish motive, opportunity, intent, preparation, plan, knowledge, identity, or pattern of operation.” Walker, 460 Mass. at 613, quoting Commonwealth v. Horton, 434 Mass. 823, 827 (2001). Even if the evidence is relevant to one of these other purposes, the evidence will not be admitted if its probative value is outweighed by the risk of unfair prejudice to the defendant. See Anestal, supra; Butler, supra, quoting Commonwealth v. Barrett, 418 Mass. 788, 794 (1994). 27

*250 As the judge correctly explained to the jury when this evidence was admitted, the questions the jury needed to decide regarding state of mind and intent were “whether the defendant knowingly and intentionally possessed visual material on January 21st, 2009, whether the defendant knew or reasonably should have known the person depicted in such visual material was under the age of eighteen, and whether the defendant had knowledge of the nature and content of such visual material.” If there were any claim that the defendant might have viewed the child pornography on January 21 by mistake or accident, without realizing their content, or any reasonable possibility that a juror might have a reasonable doubt as to whether he did, the evidence that he possessed these hand-drawn sketches in his cell more than ten months later would have been probative regarding his state of mind or intent. But the defendant’s attorney told the judge that the defendant was not claiming a defense of mistake, and the search inquiries found on computer no. two left no doubt that the person using that computer between 3:08 p.m. and 3:55 p.m. on January 21 was looking for child pornography.

The main factual issues in dispute during trial were the identity of the person using computer no. two during that timeframe, whether a person has “possession” of visual materials that he accesses on a public computer, and whether the visual materials offered in evidence depicted real children or had instead been digitally altered to look like children. The only disputed issue for which the drawings might have been probative was the issue of identity, but the jury were not permitted to consider the drawings as to this issue. Nor could the jury have been permitted to consider the drawings on the issue of identity, because “evidence of [other] bad acts is not admissible to prove identity unless there is a special mark or distinctiveness in the way the acts were committed (i.e., in the modus operandi).” Commonwealth v. Jackson, 417 Mass. 830, 836 (1994), quoting Commonwealth v. Brusgulis, 406 Mass. 501, 505 (1990). “It is not enough that there is some ‘general, although less than unique or distinct, similarity *251 between the incidents.’ ” Jackson, supra, quoting Brusgulis, supra at 507. Where there is only a general similarity, the risk is great that a jury will view the similar act as evidence of bad character or propensity rather than of identity. See United States v. Miller, 673 F.3d 688, 699-700 (7th Cir. 2012) (“Pattern evidence is propensity evidence, and it is inadmissible unless the pattern shows some meaningful specificity or other feature that suggests identity or some other fact at issue” [emphasis in original]); Brusgulis, supra at 503, 505-507 (defendant’s prior sexual assaults admitted in evidence to prove “common scheme, modus operandi, pattern of conduct and identification” were inadmissible “other bad acts” evidence, which “obviously could have an improper influence on the jury’s fact-finding function,” because similarities were “common to numerous assaults on women: a secluded site; an attempt to drag or force the victim to a more secluded area; words of threat having no unique content, spoken to obtain compliance; and abandonment of the effort because of the assailant’s concern over being discovered”). Here, where the jury were limited to consider the hand-drawn sketches only as to issues that were not in dispute, and where the drawings had only a general similarity to the child pornography found on the computer, the risk was enormous that the jury would use the drawings for the forbidden purpose of identifying the defendant as the person who viewed the child pornography on computer no. two based on his bad character and propensity to possess child pornography.

We generally “presume that a jury understand and follow limiting instructions, . . . and that the application of such instructions ordinarily renders any potentially prejudicial evidence harmless” (citations omitted). Commonwealth v. Donahue, 430 Mass. 710, 718 (2000). See Commonwealth v. Jackson, 384 Mass. 572, 579 (1981) (“We presume, as we must, that a jury understand[ ] and follow[ ] limiting instructions . . . ”). But we cannot so easily presume this to be true where the limiting instruction regarding the “bad acts” evidence effectively told the jury not to consider the evidence with respect to issues in dispute and to consider it only with respect to issues not in dispute. Faced with such an instruction, the danger is great that a jury would make the powerful natural (and forbidden) inference that the defendant’s possession of pornographic drawings of children shows that he has an interest in child pornography, so he must have been the person viewing child pornography in the library. *252 See Commonwealth v. Vasquez, 462 Mass. 827, 841-842 (2012), citing Bruton v. United States, 391 U.S. 123, 135-137 (1968) (where judge instructs jury to disregard “powerfully incriminating” evidence, “presumption that jurors could follow a judge’s limiting instruction fail[s] to inspire confidence that such an instruction could cure any prejudice or avert the risk that jurors nevertheless would consider the [evidence] against the defendant”); Greer v. Miller, 483 U.S. 756, 766 n.8 (1987), quoting Richardson v. Marsh, 481 U.S. 200, 208 (1987), and Bruton, supra at 136 (“We normally presume that a jury will follow an instruction . . . 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. . . ”).

Because the probative value of the drawings was so minimal with regard to the state of mind, knowledge, or intent of the defendant, and because the risk of unfair prejudice was so great, this is the unusual case where we conclude that it was an abuse of discretion to admit the “bad act,” even with a limiting instruction. See Anestal, 463 Mass. at 671-672 (palpable error to admit defendant’s prior act of child abuse in evidence at murder trial); Brusgulis, 406 Mass. at 507 (error to admit defendant’s prior unrelated acts of assault and attempted rape at trial for assault with intent to rape).

4. Review of the defendant’s convictions for prejudicial error. Having concluded that the in-court identifications of the defendant by M.S. and R.M. in these circumstances in the future should be suppressed as unnecessarily suggestive showup identifications without “good reason,” and that the judge erred, over objection, in excluding from evidence the defendant’s denial that he had viewed child pornography at the library, and in admitting in evidence the unfairly prejudicial hand-drawn sketches found in his cell, we turn now to the question whether the defendant’s convictions should be vacated and a new trial ordered. Where the defendant preserved his objections to each of these rulings, we review for prejudicial error. Commonwealth v. Cruz, 445 Mass. 589, 591 (2005). See Commonwealth v. Alphas, 430 Mass. 8, 23 (1999) (Greaney, J., concurring). In the unusual circumstances of this case, we include the unreliability of the in-court identifications in that calculus. Cf. Commonwealth v. Pring-Wilson, 448 Mass. 718, 736-737 (2007) (judge did not abuse discretion in applying new rule that applied prospectively to order new trial *253 where judge concluded that integrity of verdict was suspect because jury did not have benefit of relevant evidence critical to issue of whether defendant acted in self-defense).

We recognize the question to be a close one. If the in-court identifications and the bad acts evidence had not been admitted in evidence, and the defendant’s denial not been excluded, there would still be powerful evidence that the defendant was the person at computer no. two who viewed child pornography on the afternoon of January 21, 2009. The forensic examination of computer no. two leaves no doubt that the person who used that computer between 3:08 p.m. and 3:55 p.m. on January 21 searched for and viewed child pornography. The defendant is strongly tied to that computer by his admissions to Detective O’Connor on January 22 that he had used computer no. two the previous day to check his e-mail. 28 In addition, the e-mail address he gave to the detective was cblizzard@yahoo.com, which is very similar to the MySpace e-mail address used on computer no. two shortly before the child pornography was accessed from that computer.

But we do not determine whether there was prejudicial error by examining what a reasonable jury might have done if the errors had never happened. Instead, we determine whether there is a “reasonable possibility that the error[s] might have contributed to the jury’s verdict.” Alphas, 430 Mass. at 23. See Cruz, 445 Mass. at 591, quoting Commonwealth v. Flebotte, 417 Mass. 348, 353 (1994) (“[I]f we cannot find ‘with fair assurance, after pondering all that happened without stripping the erroneous action from the whole, that the judgment was not substantially swayed by the error,’ then it is prejudicial”). We cannot exclude that reasonable possibility here. The exclusion of the defendant’s denial that he had viewed child pornography at the library might have been *254 understood by the jury as an implicit admission that he had viewed it. The defendant’s hand-drawn sketches showed that he had a propensity to view child pornography. The unnecessarily suggestive in-court identifications by M.S. and R.M. were the only identifications of the defendant. Considered together, this evidence was so powerfully prejudicial that a reasonable jury might not have thought it necessary to look closely at the circumstantial evidence, and there remains the possibility that a jury who took a close look at that evidence might have concluded that it fell short of eliminating a reasonable doubt that the defendant just happened to be in the wrong place at the wrong time. In short, given the magnitude of the prejudice, we shall err in favor of the grant of a new trial.

Conclusion. For these reasons, we vacate the defendant’s judgments of conviction and remand this case to the Superior Court for a new trial.

So ordered.