Commonwealth v. Davis

NOTICE: All slip opinions and orders are subject to formal
revision and are superseded by the advance sheets and bound
volumes of the Official Reports. If you find a typographical
error or other formal error, please notify the Reporter of
Decisions, Supreme Judicial Court, John Adams Courthouse, 1
Pemberton Square, Suite 2500, Boston, MA, 02108-1750; (617) 557-
1030; SJCReporter@sjc.state.ma.us

SJC-13014

COMMONWEALTH vs. MATTHEW DAVIS.

Suffolk. February 1, 2021. - May 17, 2021.

Present: Budd, C.J., Gaziano, Lowy, Cypher, Kafker,
& Georges, JJ.

Armed Assault with Intent to Murder. Assault and Battery.
Attempt. Firearms. Electronic Surveillance. Global
Positioning System Device. Evidence, Videotape,
Photograph, Authentication, Identification, Scientific
test. Practice, Criminal, Probation, Required finding.

Indictments found and returned in the Superior Court
Department on May 16, 2016.

The cases were tried before Peter M. Lauriat, J.

After review by the Appeals Court, the Supreme Judicial
Court granted leave to obtain further appellate review.

David Rassoul Rangaviz, Committee for Public Counsel
Services, for the defendant.
Andrew S. Doherty, Assistant District Attorney, for the
Commonwealth.
Maria Gonzalez Calvet, of the District of Columbia, Daniel
W. Richards, of California, Michael A. Morales, of New York,
Radha Natarajan, Katharine Naples-Mitchell, & Kirsten V. Mayer,
for New England Innocence Project & another, amici curiae,
submitted a brief.
 2

LOWY, J. On September 15, 2015, a Black man with long hair

wearing a red or pink shirt or sweatshirt fired multiple shots

at the driver's side window of a moving blue sedan. The driver

of the sedan, who was uninjured, fled from the scene and did not

testify at trial. The sole civilian witness who testified at

trial did not witness the shooting itself but did see a Black

man with braids and a red shirt running away from the location

of the shooting.

The defendant, Matthew Davis, became a suspect after police

made an inquiry whether anyone wearing a global positioning

system (GPS) device at the relevant time was in the vicinity of

the shooting. Due to his probation on a Federal drug charge,

the defendant was wearing a GPS ankle monitor called an

"ExactuTrack 1" (ET1), manufactured by BI, Inc. (BI). Data from

the defendant's GPS device showed he was at the location where

the shooting took place very close in time to the shooting, and

his speed matched the shooter's movements, according to

surveillance footage and testimony from the civilian witness.

After a jury trial, the defendant was convicted of armed assault

with intent to murder, G. L. c. 265, § 18 (b), and related

charges.

On appeal, we consider whether the GPS evidence was

sufficiently reliable to be admitted. We conclude that the
 3

judge abused his discretion in admitting the speed evidence,

where the ET1's ability to measure speed had never been formally

tested. Because this error was prejudicial, we reverse the

defendant's convictions.

We also address the defendant's argument that the evidence

was insufficient to support his convictions and hold that it was

sufficient. Finally, we address other arguments the defendant

raised on appeal that may recur at retrial, including whether

maps of the GPS data violated the defendant's confrontation

rights, whether a cell phone video recording (video) of

surveillance footage was properly authenticated, and whether it

was proper for the prosecutor to ask the jury to identify the

defendant as the shooter based on footage that did not show the

shooter's face.1

Background. Because the defendant raises a sufficiency

challenge, we recite the facts the jury could have found, in the

light most favorable to the Commonwealth, reserving certain

details for later discussion. See Commonwealth v. Latimore, 378

Mass. 671, 676-677 (1979).

On the morning of September 15, 2015, at the corner of

Baker Avenue and Quincy Street in the Dorchester section of

1 We acknowledge the amicus brief submitted by the New
England Innocence Project and Charles Hamilton Houston Institute
for Race & Justice.
 4

Boston, a man fired multiple gunshots at a blue sedan.

Dispatchers received a 911 call reporting the shooting at 10:28

A.M. Responding police officers found an unoccupied blue sedan,

with multiple bullet holes in the front window, crashed into a

light pole. Several shell casings and bullet fragments were on

the ground.

One of the responding officers, Sergeant Thomas Carty,

canvased the area for potential witnesses as well as any cameras

that may have captured the shooting. He noticed a video camera

affixed to a residential property on Baker Avenue. A resident

of that address allowed Carty to view the surveillance video,

but the resident did not know how to download it or copy it to

another device. Carty instead used his cell phone to record a

video of the surveillance video as it played on a computer

screen.

The resulting video -- which is not very high resolution --

shows a Black man with long braids or dreadlocks in a red or

pink shirt or sweatshirt wearing a gray hat or cap. The man

runs towards an intersection raising his arm while holding a

handgun. As the man holds up the gun, a blue sedan is driven

into the frame from the opposite direction and then collides

with a light pole at the corner of the intersection. After the

crash, the driver gets out of the car and runs down the street.

A little over a minute later, a man who appears to be the driver
 5

returns to the car and gets into the driver's seat, before

getting out of the car and again jogging away, across the

street. The video is not high enough resolution and is taken

from too far away to discern any features of the shooter's face.

At approximately 10:30 A.M. on the day of the shooting, a

woman named Ilene Rock was standing on Bodwell Street near the

corner of Columbia Road -- a couple of blocks away from the

location of the shooting -- when she heard a noise that sounded

like gunshots or a car backfiring. Shortly after hearing the

noise, she saw a Black man with thin braids wearing a red shirt

run past her with his hand in his pocket.2 The man came within

five or six feet of her, but she did not get a good look at his

face because she was focused on his hands in his pocket. The

man ran down Bodwell Street and turned right onto Columbia Road

toward Quincy Street. Shortly thereafter, Rock heard sirens and

saw police.

2 Rock described the man she saw running as having "thin
braids" and a "red shirt." Thus, when referring to her
testimony, we use that terminology. The shooter's hair and
dress are less clear from the video. Thus, when describing the
shooter in the video, we describe his hair as "braids or
dreadlocks" and his clothing as a "red or pink shirt or
sweatshirt." The defendant asserts that he had dreadlocks, not
braids, around the time of the shooting. The photograph of him
taken the day after the shooting that was admitted in evidence
appears to confirm this.
 6

At a later date, police showed Rock a photographic array

that contained an image of the defendant. Rock later testified

that she "saw a few people that [she] thought looked similar [to

the man she saw running], but [she] couldn't make a positive

identification of him" because she "wasn't sure of the facial

features enough to make a selection." On three of the eight

photographs, she made the notations "maybe the person I saw,"

"This might be him 80%," and "This is possibly the man I saw

running." None of those three photographs depicted the

defendant.

As part of their investigation, police made an inquiry into

whether anyone wearing a GPS device was in the area at the time

of the shooting.3 The defendant, who was wearing a GPS ankle

monitor as part of his probation on a Federal drug case, had

been. The defendant's GPS device purported to show that at

10:25 A.M. on the day of the shooting, he had been on Columbia

Road near Brunswick Street travelling northeast at twenty-two

miles per hour. At 10:26 A.M., he was still travelling

northeast on Columbia Road, now at thirty-two miles per hour.

At 10:27 A.M., he was at the corner of Quincy Street and Baker

Avenue -- the location of the shooting -- travelling at ten

miles per hour. At 10:28, he was on the corner of Baker Avenue

3 The record does not describe with whom police made this
inquiry.
 7

and Bodwell Street travelling at eight miles per hour. At

10:29, he was on Columbia Road between Quincy and Bodwell

travelling at eleven miles per hour. At 10:30, he was on Quincy

Street travelling at twenty-two miles per hour. At 10:31, he

was on Church Street travelling at fifteen miles per hour. At

10:32, he had stopped at his home.

Approximately a week after the shooting, officers executed

a search warrant at the defendant's home. They found a red

long-sleeved crew neck sweatshirt under a pile of clothes in the

defendant's bedroom. The sweatshirt tested negative for gunshot

primer residue.

After a jury trial in October 2017, the defendant was found

guilty of armed assault with intent to murder, G. L. c. 265,

§ 18 (b), and multiple lesser charges.4 The defendant then filed

a timely notice of appeal, and the Appeals Court affirmed his

4 In addition to armed assault with intent to murder (count
one), the defendant was also found guilty of attempted assault
and battery by means of discharging a firearm, G. L. c. 265,
§ 15F (count two); carrying a firearm without a license, G. L.
c. 269, § 10 (a) (count three); possession of ammunition without
a firearm identification card, G. L. c. 269, § 10 (h) (1) (count
four); and carrying a loaded firearm without a license, G. L.
c. 269, § 10 (n) (count five). On count three, the defendant
had been indicted as an armed career criminal, G. L. c. 269,
§ 10G (b). After trial, he pleaded guilty to being an armed
career criminal on that count. Count four was then dismissed at
the request of the Commonwealth.
 8

convictions. See Commonwealth v. Davis, 97 Mass. App. Ct. 633

(2020). We granted further appellate review.

Discussion. 1. Admission of expert testimony. Before the

GPS evidence was introduced at trial, the judge conducted a voir

dire of the Commonwealth's expert, James Buck, manager of

product development at BI. Buck provided an overview of how GPS

technology works in general, as well as described the ET1 model

in particular. He stated that at any given time, there are

twenty-four active satellites circling the globe. Signals from

the satellites transmit to a receiver -- as in a GPS device --

and that data can be used to determine the device's speed and

location. The more satellites from which a device is receiving

signals at any given time, the more accurate the speed and

location data will be.

GPS technology works slightly differently for speed from

how it does for location. To triangulate a device's location,

it must receive signals from a minimum of three satellites. To

determine a device's speed and direction, on the other hand, it

must receive signals from at least four satellites.5 With

5 This is because speed measurements are based on a
different frequency and utilize the Doppler effect. In essence,
the speed data is not based merely on doing algebra to calculate
the average speed between two location points. Instead, the
device uses a different frequency to take a reading every
millisecond of a satellite's position in the sky, and then,
 9

respect to the ET1 specifically, Buck testified that it records

the wearer's location and speed once per minute and sends the

data over a cellular network to BI's headquarters in Colorado.6

Buck stated that BI had conducted formal testing of the ET1's

ability to measure location, but its ability to measure speed

had never been formally tested.

a. Gatekeeper reliability. Before offering expert

testimony such as Buck's, the proponent must establish a

sufficient foundation for a judge to determine whether the

expert's opinion satisfies gatekeeper reliability. See

Commonwealth v. Patterson, 445 Mass. 626, 639 (2005), overruled

on other grounds by Commonwealth v. Britt, 465 Mass. 87 (2013)

("Trial judges serve a gatekeeper function with respect to

expert opinion testimony based on specialized knowledge"). See

also Mass. G. Evid. §§ 104(a), 702 (2021). "If the process or

theory underlying [an] . . . expert's opinion lacks reliability,

that opinion should not reach the trier of fact." Patterson,

supra, quoting Commonwealth v. Lanigan, 419 Mass. 15, 26 (1994).

based on that, uses the Doppler effect to determine the GPS
device's velocity and direction.

6 Specifically, Buck testified that the device takes a
sample every fifteen seconds, and then selects the best sample
of the four to log. Although it is not entirely clear from the
record, it appears the "best sample" is the one during which the
device was communicating with the highest number of satellites.
 10

Until 1994, we used the Frye test -- also called the

general acceptance test -- to determine if proposed expert

testimony was sufficiently reliable to reach the jury.

Canavan's Case, 432 Mass. 304, 310 (2000). See Lanigan, 419

Mass. at 24; Frye v. United States, 293 F. 1013, 1014 (D.C. Cir.

1923). Under that test, "we required that in most circumstances

'the community of scientists involved [must] generally accept[]

the theory or process' for it to be admitted in evidence"

(citation omitted). Canavan's Case, supra. The test proved to

be useful because "if there is general acceptance of a theory or

process in the relevant scientific community, [it] is likely

reliable." Id. "However, we recognized that 'strict adherence

to the Frye test' could result in reliable evidence being kept

from the finder of fact. For example, a new theory or process

might be 'so logically reliable' that it should be admissible,

even though its novelty prevents it from having attained general

acceptance in the relevant scientific community." (Citations

omitted.) Id. In short, situations could arise where our law

of evidence lagged behind recognizing what was otherwise

reliable science.

To account for this circumstance, in Lanigan, 419 Mass. at

26, we adopted in part the United States Supreme Court's

reasoning in Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579,

593-594 (1993), which set forth five nonexclusive factors a
 11

judge should consider in determining the reliability of proposed

scientific evidence. The five nonexclusive factors are "whether

the scientific theory or process (1) has been generally accepted

in the relevant scientific community; (2) has been, or can be,

subjected to testing; (3) has been subjected to peer review and

publication; (4) has an unacceptably high known or potential

rate of error; and (5) is governed by recognized standards."

Commonwealth v. Powell, 450 Mass. 229, 238 (2007).

Although Daubert and Lanigan dealt specifically with

scientific evidence, we have since recognized their application

to scientific, technical, and other specialized knowledge. See

Commonwealth v. Pytou Heang, 458 Mass. 827, 844 (2011).

Further, the Daubert-Lanigan factors are nonexclusive, and we

have recognized the potential need to consider other factors

depending on the nature of the expert testimony. See, e.g.,

Commonwealth v. Hinds, 487 Mass. 212, 220-222 (2021) (discussing

application of Daubert-Lanigan factors to social sciences);

Canavan's Case, 432 Mass. at 314 n.5 ("Differing types of

methodology may require judges to apply differing evaluative

criteria to determine whether scientific methodology is

reliable"). Using these nonexclusive factors to determine

reliability is known as the Daubert-Lanigan test.

We have not, however, entirely abandoned the Frye test. In

Lanigan, 419 Mass. at 26, we noted that in many cases general
 12

acceptance will "be the significant, and often the only, issue."

Lanigan's progeny make clear that reliability can still be

established by general acceptance alone, without regard to the

other Daubert-Lanigan factors. See Patterson, 445 Mass. at 640-

641 (citing cases).

If a theory or methodology has been established as reliable

in our courts using one of these two standards in the past, then

a judge may take judicial notice of its reliability. See

Daubert, 509 U.S. at 592 n.11 ("theories that are so firmly

established as to have attained the status of scientific law,

such as the laws of thermodynamics, properly are subject to

judicial notice"). See also Mass. G. Evid. § 201.

In contrast, when proposed expert testimony uses a new

theory, or new methodology to apply an accepted theory, the

proponent must establish its reliability using a Daubert-Lanigan

analysis. See Patterson, 445 Mass. at 648-649. For example, in

Patterson, the Commonwealth sought to introduce fingerprint

evidence. Id. at 627. We held that the judge acted well within

her discretion in finding that using the theory underlying

fingerprint analysis was generally accepted. Id. at 641.

However, we also held that the judge erred by failing to conduct

a Daubert-Lanigan analysis to determine whether a new technique

based on that accepted theory was reliable. Id. at 648-649.

Similarly, in Commonwealth v. Camblin, 471 Mass. 639, 645 (2015)
 13

(Camblin I), S.C., 478 Mass. 469 (2017) (Camblin II), we held

that the judge erred by failing to conduct a Daubert-Lanigan

analysis to determine the reliability of a new generation of

breathalyzer devices, which used different techniques from those

previously found reliable by our courts.

Under both Daubert-Lanigan and Frye, we review a judge's

decision to admit expert testimony as satisfying gatekeeper

reliability under the abuse of discretion standard. See

Canavan's Case, 432 Mass. at 312. "While our review under this

standard is deferential and limited, it is not perfunctory. A

judge's findings must apply the correct legal standard to the

facts of the case and must be supported by an examination of the

record." Patterson, 445 Mass. at 639.7

7 The defendant also cites to cases that pertain not to the
Daubert-Lanigan reliability standard, but to the necessity of a
foundational showing of accuracy. See Commonwealth v. Torres,
453 Mass. 722, 737 (2009); Commonwealth v. Whynaught, 377 Mass.
14, 17 (1979); Commonwealth v. Podgurski, 81 Mass. App. Ct. 175,
185 (2012). Usually the line between a preliminary question of
fact upon which admissibility depends, such as gatekeeper
reliability, see Mass G. Evid. §§ 104(a), 702, and conditional
relevance, see Mass G. Evid. § 104(b), is clear. See P.C.
Giannelli, Understanding Evidence § 7.03 (5th ed. 2018)
(discussing conditional relevancy). Yet when the proposed
evidence is the product of a technical device, such as it is in
this case, the line between these evidentiary principles can
blur. "Radar evidence illustrates this point. The reliability
of evidence based on radar depends on (1) the validity of the
underlying theory (e.g., the Doppler effect), (2) the validity
of the technique applying that theory (e.g., the particular
model of radar), and (3) the proper application of the technique
on a particular occasion (e.g., use of tuning forks to calibrate
 14

With this framework in mind, we turn to the GPS evidence of

speed and location admitted in this case.

b. Speed. As will become clear, the Commonwealth failed

to lay the proper foundation to admit the ET1's speed

measurements under either the Daubert-Lanigan or the Frye test.

See generally Mass. G. Evid. §§ 104(a), 702. Thus, the judge

erred by admitting the speed evidence without the proper

foundation.

The defendant does not dispute the reliability of GPS

technology over-all, nor do we. The defendant's objection is to

the reliability of the ET1 model specifically. Nevertheless, we

briefly discuss the reliability of GPS theory and methodology to

give context to the issues raised.

First, it is clear from both the record and our case law

that GPS technology is generally accepted as reliable. The

judge found that GPS evidence has "been accepted in the field

for a substantial number of years by virtually most populations

the instrument)." Id. at § 24.04 endnote 39. In that scenario,
the first two elements would be subject to gatekeeper
reliability under Daubert-Lanigan or Frye. The third element,
in contrast, would in most instances be a matter of conditional
relevance. By inquiring into the third element, the judge would
ask not whether the device was reliable, but rather whether a
reasonable jury could find by a preponderance of the evidence
that it was functioning properly at the time it was used. In
essence, the relevance of a measuring device is conditioned on
whether a jury could find that it was functioning properly. See
Mass. G. Evid. § 104(b).
 15

in the world." In describing the history of GPS technology,

Buck testified that, as early as 1957, researchers at Johns

Hopkins University theorized that satellites could be used to

triangulate location. Our case law confirms that GPS technology

is "widely used and acknowledged as a reliable relator of time

and location data." Commonwealth v. Thissell, 457 Mass. 191,

198 (2010) (Thissell II).

Second, because GPS technology uses a different methodology

to measure speed from how it does location, that methodology,

too, would need to satisfy gatekeeper reliability. See Mass. G.

Evid. §§ 104(a), 702. See, e.g., Patterson, 445 Mass. at 628

(new methodology based on accepted underlying theory must pass

gatekeeper reliability). Had the defendant objected to the

reliability of the method used by GPS technology, in general, to

measure speed, we expect that the Commonwealth would have been

able to demonstrate that the use of GPS to measure speed is

generally accepted or meets the requirements of Daubert-Lanigan.

The defendant's objection at trial and his argument on

appeal pertain to the reliability of the ET1 device

specifically. He argues that the ET1 does not meet the

requirements of Daubert-Lanigan. Chief among the defendant's

concerns is the fact that the ET1's ability to measure speed has
 16

never been formally tested.8 Given the complete lack of formal

testing of the ET1 model for speed, there is also no known error

rate. Moreover, the defendant asserts that because the ET1 is

proprietary, it is impossible to say whether the methodology it

employs is generally accepted. The proprietary nature also

means it has not been subject to peer review.9 Finally, its

accuracy is not governed by any recognized standards.

We agree with the defendant that if a new model of a device

is objected to on reliability grounds, it must pass gatekeeper

reliability under either Daubert-Lanigan or Frye. It is not

sufficient to show merely that GPS technology is, in general,

reliable without making any showing pertaining to the

reliability of a particular model of a device. The Commonwealth

could meet that burden by showing that the new model itself

satisfies the Daubert-Lanigan factors -- for example, that it

8 Buck testified that BI had not done any formal testing to
ensure that speed data was accurate. Moreover, he had not
"figured out and formulated a way to successfully [test speed]
reliably and repeatedly."

BI's informal testing consisted of having ten employees at
any given time wear its devices and report back if there were
any "speed irregularities." But BI only conducts a "general
review of the [data] points"; BI has not compared the ET1 speed
data to any independent measurements of speed.

9 Buck testified that there is "one person in the industry"
who tests and compares different GPS devices, but the results of
his testing are not made available to any of the companies in
question, including BI.
 17

has been tested or peer reviewed. That is not the only way,

however, to show that a new model is reliable. For example, if

an older model has previously been found reliable, the proponent

need only show that the new model applies the same methodology

as that prior one.10 Given that devices generally tend to

improve, that will generally be sufficient to show that the new

device, too, is reliable. Here, the Commonwealth made neither

showing. It only showed that the GPS technology is a reliable

theory. For the speed data, it has not shown that the ET1

itself -- either through testing or through its similarity to a

generally accepted device -- reliably applies that accepted

theory. Thus, the judge abused his discretion in admitting the

ET1 speed evidence.11

Because on retrial the Commonwealth may again attempt to

lay the proper foundation for the speed evidence, we comment on

the remainder of the analysis. If the Commonwealth attempts to

show that a new model of a device is reliable by asserting that

it is similar to a prior model, the defendant may object and

10This showing must be based on facts and data, not a
conclusory statement that the devices are the same.

11In addition, because speed had never been tested, Buck
did not have any margin of error within which the speed data was
accurate. This could lead the jury to overvalue its accuracy.
For the location data, in contrast, Buck did not say that the
ET1 could pinpoint one's exact location; he said that it was
accurate within a certain number of feet.
 18

move for a Daubert-Lanigan hearing on the new device. This is

essentially what occurred in Camblin I, 471 Mass. at 642.

There, the Commonwealth sought to introduce evidence from a

particular model of breathalyzer (Alcotest) that had not

previously been reviewed by our courts. Id. at 640. The

defendant moved for discovery of the device's computer source

code, and that request was granted pursuant to a nondisclosure

agreement. Id. at 642. The defendant retained experts to

examine the source code. Id. The defendant then filed

affidavits and reports contending that the source code contained

thousands of errors and argued that the Alcotest used methods

different from previous machines that had been reviewed by our

courts. Id. at 644. On appeal, we held that because neither

statute nor existing case law offered guidance about the

reliability of the Alcotest's methodology, the judge should have

held a Daubert-Lanigan hearing. Id. at 650.12 We remanded the

12 This is but one example of how a defendant may assert
that a new model of device uses a different methodology from
previous models. We leave for another day how much a defendant
needs to show to assert a device uses a new methodology in order
to raise the issue. At a minimum, however, "a defendant must
file an appropriate pretrial motion stating the grounds for the
objections and request a hearing." Commonwealth v. Sparks, 433
Mass. 654, 659 (2001).

Further, we note that while in many scenarios it may be
sufficient for the Commonwealth to show that a device applies
the same methodology as prior versions, courts are not required
to admit evidence from a device merely because such evidence has
 19

case for that hearing, and then, in Camblin II, 478 Mass. at

469-470, held that the judge did not abuse his discretion in

finding that the Alcotest satisfied the Daubert-Lanigan

standard.13

c. Location. Next, the defendant objects to the admission

of the ET1's location data. We hold that the judge did not

abuse his discretion in admitting the location data.

previously been admitted. See Commonwealth v. Shanley, 455
Mass. 752, 763 n.15 (2010) ("we have not 'grandfathered' any
particular theories or methods for all time"). We similarly
leave for another day how much a defendant needs to show to call
into question the reliability of a generally accepted device.
See, e.g., Commonwealth v. Neal, 392 Mass. 1, 17-18 (1984)
(rejecting defendant's argument that generally accepted model of
breathalyzer was unreliable in light of recent discovery of its
susceptibility to radio frequency interference).

13Given that the issue could also arise on retrial, we
briefly comment on the difference between gatekeeper reliability
and conditional relevance in this scenario. If the defendant
objects -- as he did here -- to the reliability of the ET1 model
as a whole, then the Commonwealth bears the burden of showing
that the ET1 passes gatekeeper reliability. See Mass. G. Evid.
§§ 104(a), 702. See, e.g., Camblin I, 471 Mass. at 640
(reliability of Alcotest device). On the other hand, if the
defendant objects to whether the specific ET1 device worn by the
defendant on September 15, 2015, was functioning properly, then
the issue is likely a matter of conditional relevance, for which
the Commonwealth also bears the burden of laying the proper
foundation. See Mass G. Evid. § 104(b). See, e.g.,
Commonwealth v. Torres, 453 Mass. 722, 737 (2009) (whether
measuring device was calibrated); Commonwealth v. Neal, 392
Mass. 1, 19 (1984) (whether particular breathalyzer unit was
accurate at time test was performed); Commonwealth v. Whynaught,
377 Mass. 14, 17 (1979) (whether individual radar speedmeter was
calibrated); Commonwealth v. Podgurski, 81 Mass. App. Ct. 175,
185-186 (2012) (whether individual scale was calibrated).
 20

Unlike the speed data, the ET1's location data had been

formally tested. During voir dire, Buck described in detail the

circular error of probability test BI had used to ensure the

ET1's location data was accurate. The test -- which was

conducted at BI's headquarters in the suburbs of Boulder,

Colorado -- involved leaving the device in a stationary position

for six hours, and then recording the location data provided

from the satellites and plotting it on a scatter graph. Buck

described the area as an industrial park, with no buildings over

about three stories tall. He stated that the test showed that

ninety-eight percent of the ET1's location points are within

sixteen feet of the expected circle, and fifty percent are

within three feet.

Against this conclusion, the defendant argues that the

testing was not sufficient because it occurred in prime

conditions that did not simulate real-world accuracy, especially

in an urban environment. Buck stated that the accuracy of a GPS

device can be affected by tall buildings, which can block the

signal between the satellite and the device.14 At trial, Buck

14Specifically, Buck testified that the satellite signal
can be blocked by tall buildings, which have a "multipath
effect." Buck testified, "In other words, it's bouncing off of
a building to get to the actual receiver which then delays your
time of arrival which then increases the likelihood that you're
going to have an error in [location data]."
 21

stated that the device is tested on the rooftop in Colorado to

get a "baseline" accuracy so that it can be compared to prior or

future models. Buck also acknowledged that the ET1 had never

been tested in Boston or in any other urban environment. That

is why, he said, in the device manual BI tends to err on the

side of caution in reporting the radius within which a GPS

device can determine someone's location. In this case, the

manual indicated that the device was accurate to plus or minus

ninety feet.15 That gives a far wider margin of error than the

Colorado test, which determined that ninety-eight percent of the

time the location points are accurate within sixteen feet. At

trial, when asked about the accuracy of each data point

individually, Buck testified that they were within a margin of

error of between forty feet and one-half the size of the court

room.16

The judge did not abuse his discretion in ruling that the

defendant's objection went to the weight of the evidence, not

its admissibility. Buck acknowledged that the ET1 was likely

less accurate in an urban environment than in Colorado, where

15It is unclear from the record whether this manual
corresponded to the ET1 itself, or to a prior BI GPS model.
Buck stated that the manual was from 2008 or 2009.

16The differing degrees of accuracy for each data point
depended on how many satellites the GPS device was communicating
with at any given time. Buck went through each minute of the
defendant's GPS data and posited its accuracy.
 22

the baseline test occurred. Consequently, Buck did not allege

that BI's location points in an urban area would be within the

same three- to sixteen-foot radius in which they had been in its

baseline test in Colorado. His testimony shows, however, that

the Colorado test is still helpful as a baseline. Unlike for

speed, where there was no test of any kind to show that the

ET1's speed data reliably applied an accepted methodology, for

location data the Colorado test was sufficient to show that the

ET1's location data did reliably apply an accepted methodology.

Thus, the location data met the requirements of gatekeeper

reliability, and the judge did not abuse his discretion in

admitting it.

d. Prejudicial error. Because the objection to the speed

GPS data on Daubert-Lanigan grounds was preserved at trial, we

review for prejudicial error. "[T]he Commonwealth must show

that any error 'did not influence the jury, or had but very

slight effect.'" Commonwealth v. Dargon, 457 Mass. 387, 399

(2010), quoting Commonwealth v. Flebotte, 417 Mass. 348, 353

(1994). To this end, the Commonwealth argues that the speed

data was not prejudicial because the other evidence was

compelling and, to some extent, the speed data was "irrelevant"

because the location data was itself incriminating. We

disagree.
 23

In the Commonwealth's closing argument, the prosecutor made

more than ten explicit references to speed. When describing the

GPS evidence, the prosecutor stated that when looking "not only

[at] the locations, but also the speed, a bigger story will come

up for this particular case." The prosecutor used the speed

data to match the defendant's movements to those of both the

shooter in the video as well as the man Rock saw running. The

prosecutor stated that at the first two data points -- 10:25

A.M. and 10:26 A.M. -- the defendant's speed was twenty-two and

thirty-two miles per hour, showing he was likely in a car

travelling down Columbia Road. At 10:27 A.M. and 10:28 A.M.,

the defendant's speed slowed down to what was likely the speed

of someone running. This, the prosecutor said, was consistent

with the shooter in the video running at the time of the

shooting and immediately after. Further, the speed was

consistent with Rock's testimony of seeing a man in a red shirt

running down the street with his hands in his pockets --

presumably fleeing the shooting. After that, the defendant's

speed increases again, showing that he is likely back in a car.

The defendant's speed was not merely duplicative of his

location. It was crucial evidence used to correlate the

defendant's movements to those of the shooter in the video and

to the man Rock saw running. Without the speed, a jury would

have only been able to infer that the defendant was in the area
 24

where the shooting took place. With the speed, however, the

jury could match the defendant's movements to those of the

shooter in the video and the man Rock saw, thereby presenting a

compelling narrative that the defendant was the shooter. Thus,

we cannot say that it "did not influence the jury, or had but

very slight effect." Dargon, 457 Mass. at 399, quoting

Flebotte, 417 Mass. at 353. We hold that admitting the speed

data was prejudicial error and the defendant's convictions must

be reversed.

2. Sufficiency of the evidence. The defendant moved for a

required finding of not guilty at the close of the

Commonwealth's case, arguing the evidence was insufficient to

support a conviction. The motion was denied. He reaffirms this

argument on appeal, asserting that the admitted evidence is

insufficient to prove his identity as the shooter where the

Commonwealth's case is circumstantial and relies on "piling

inference upon inference" (quotation and citation omitted).

Commonwealth v. Ashford, 486 Mass. 450, 455 (2020). We

disagree.

In assessing the sufficiency of the evidence, we consider

"whether, after viewing the evidence in the light most favorable

to the prosecution, any rational trier of fact could have found

the essential elements of the crime beyond a reasonable doubt."

Latimore, 378 Mass. at 677, quoting Jackson v. Virginia, 443
 25

U.S. 307, 319 (1979). "Circumstantial evidence is sufficient to

find someone guilty beyond a reasonable doubt and inferences

drawn from such circumstantial evidence 'need only be reasonable

and possible; it need not be necessary or inescapable.'"

Commonwealth v. Grandison, 433 Mass. 135, 141 (2001), quoting

Commonwealth v. Lodge, 431 Mass. 461, 465 (2000). "A conviction

may not, however, be based on conjecture or on inference piled

upon inference." Commonwealth v. Jones, 477 Mass. 307, 316

(2017).

"If the evidence at trial was legally insufficient to

sustain a verdict, a new trial would violate the prohibition

against double jeopardy and would therefore be impermissible."

Commonwealth v. Bacigalupo, 455 Mass. 485, 489 (2009).

"Ordinarily, in determining the sufficiency of the evidence, we

include evidence improperly admitted." Id. at 490. Thus, we do

not exclude the improperly admitted speed evidence from our

analysis and do not comment on whether the admitted evidence

minus the speed data would have been sufficient to support a

conviction. See id.

The Commonwealth primarily relied on three pieces of

evidence to establish the defendant's guilt: the GPS data, the

video, and Rock's testimony. We analyze each in turn.

First, the GPS evidence established both the defendant's

speed and his location. It showed that he was at the
 26

intersection where the shooting took place at 10:27 A.M. on the

day of the shooting.17 The 911 call reporting the shooting was

made approximately one minute later, at 10:28:24 A.M. It is

true that "mere presence at the scene of a crime, without more,

is not sufficient to support a conviction." Commonwealth v.

Mazza, 399 Mass. 395, 399 (1987). Yet the GPS established not

only his location, but also his speed. As discussed supra, the

speed evidence helped to establish his identity as the shooter

17The defendant argues that the Commonwealth failed to show
the precise time the shooting took place, and given the
importance of the purportedly precise GPS data, such evidence
was necessary to establish the defendant's guilt. The defendant
argues that if the shooting occurred at 10:27 A.M. -- when the
defendant was at the intersection where the shooting took place
-- then his location would have been incriminating, but at any
other moment his location would have been exonerating.

At trial, the defendant compared the 911 call to the video,
arguing that one could infer that the shooting took place at
10:28:07 A.M., when the defendant was a block away. This
argument was premised on the theory that a man on his cell phone
appearing approximately one minute and thirty seconds into the
video was the 911 caller, and that when the 911 caller asked
someone "are you all right?" that was him speaking to the
victim, who had briefly returned to the car. While that was a
cogent argument, the jury were not required to accept it.
Whether the man on his cell phone in the video was the 911
caller is unclear. Moreover, one of the responding officers
testified that she "believe[d] there was more than one call that
morning." Thus, even if the man on his cell phone in the video
was in fact calling 911, that does not necessarily mean he is
the same caller whose call was played for the jury.
 27

by matching his movements to those of both the shooter as well

as the man Rock saw in a red shirt shortly after the shooting.18

Next, the video of the shooting showed that the gunman

appeared to be a Black man with long braids or dreadlocks and a

long-sleeved red or pink shirt or sweatshirt. As the prosecutor

argued in closing, the video was evidence that on the corner of

Quincy Street and Baker Avenue, the shooter fired several shots

at a moving car. Thus, the video was probative of both the

shooter's actions and his intent to kill. As discussed infra,

however, because the shooter's features were not clear from the

video, it cannot alone give rise to an inference that the

defendant was this shooter.

Finally, Rock testified that she heard what could have been

gunshots, and then saw a Black man with thin braids and a red

shirt with his hands in his pocket run down Bodwell Street

toward Columbia Road and turn right onto Columbia Road.19 The

18The defendant disputes the Commonwealth's proposed
inference from the speed data that the defendant was travelling
in a car, then got out and started running, and then reentered
the car minutes later. Absent the video or Rock's testimony,
the speed data might not be probative of guilt. Yet combined
with that evidence, the speed data was probative of the
defendant's identity as the shooter.

19Rock described the runner as having "thin braids" and did
not describe him as having a beard. The defendant points out
that he had dreadlocks, not braids, as depicted in the
photograph taken the day after the shooting. Further, he had a
light beard at the time.
 28

runner's path matched the defendant's GPS device's tracked

movements. The fact that Rock failed to identify the defendant

in a photographic array is fodder for cross-examination, but

given that Rock admitted that she did not get a good look at the

runner's face because she was focused on his hands, it does not,

in the light most favorable to the Commonwealth -- and when

combined with the other evidence showing the defendant's speed

and location -- preclude the inference that the defendant was

the man she saw running.

We conclude that while the evidence at trial was not by any

means overwhelming, it was sufficient to sustain the defendant's

convictions. See Jones, 477 Mass. at 318. Evidence of the GPS

speed and location data, the video, and Rock's observations

permit the reasonable inference that the defendant was the

shooter.

3. Other issues. We turn now to other issues raised on

appeal that may recur upon retrial.

a. Maps depicting GPS evidence. The defendant argues that

admitting maps depicting his location information violated his

confrontation rights under the Sixth Amendment to the United

States Constitution and art. 12 of the Massachusetts Declaration

of Rights because no one from the third-party company that

generated the maps testified at trial. Additionally, the

defendant argues admission of the maps violated the rule against
 29

hearsay. Because this issue was not raised at trial, we review

for a substantial risk of a miscarriage of justice. See

Commonwealth v. AdonSoto, 475 Mass. 497, 504 (2016). We

conclude it does not create such a risk.

At trial, the Commonwealth introduced maps showing the

defendant's latitude and longitude points reported from the ET1

from 10:25 A.M. to 10:32 A.M. Buck testified that the maps were

created by BI collecting the latitudes and longitudes of GPS

points over time and sending them to a third-party mapping

company. The mapping company would then produce a map

encompassing all the points.20 Finally, BI would plot the points

onto the map. Although the record is not entirely clear how the

points are plotted on the map, it appears they are generated by

a computer.21

20Specifically, Buck stated at voir dire: "What we do is
if you're looking to provide a map we can do it for [twenty-
four] hours, or [fifteen] minutes, whatever you want. What we
do is we take the specifics of the request, we gather those
latitude[s] [and] longitudes, we send them off to a mapping
company, such as Google, and then they render back the maps that
would be containing those latitude[s] [and] longitude[s] and
then we take and we put dots on the map representing the
latitude[s] [and] longitude[s] for display . . . ."

21At voir dire in response to a question asking how BI
ensures the points on the maps are accurate, Buck testified:
"Because nobody can get to them, nobody can do anything with
them. When they get into the SQL database they're encrypted so
that nobody can actually change the data within the database."
 30

"Hearsay requires a 'statement,' i.e., 'an oral or written

assertion or . . . nonverbal conduct of a person, if it is

intended by the party as an assertion.'" Commonwealth v.

Thissell, 74 Mass. App. Ct. 773, 776-777 (2009), S.C., Thissell

II, 457 Mass. 191, quoting Commonwealth v. Whitlock, 74 Mass.

App. Ct. 320, 326 (2009). See Mass. G. Evid. § 801(a). Whether

a computer record contains a statement depends on whether the

record is "computer-generated," "computer-stored," or a hybrid

of both. Thissell II, supra at 197 n.13. Computer-generated

records are created solely by the mechanical operation of a

computer and do not require human participation. Commonwealth

v. Royal, 89 Mass. App. Ct. 168, 171-172 (2016). For this

reason, they cannot be hearsay.

With the exception of the defendant's name, all of the

information included in the maps was computer-generated. The

latitude, longitude, and speed points in the text boxes were

generated by the GPS technology. The maps themselves were

rendered by a computer at the third-party mapping company. And

the dots on the map were rendered by BI's computer system.

Thus, because the maps -- with the exception of the defendant's

name -- were computer generated, they do not contain a statement

and are not hearsay. Further, because the maps were not

hearsay, they did not violate the confrontation clause. See
 31

Pytou Heang, 458 Mass. at 854, citing Commonwealth v. Hurley,

455 Mass. 53, 65 n.12 (2009).

b. Authentication of the surveillance video. The

Commonwealth introduced a cell phone video of surveillance video

that allegedly depicted the shooting. The defendant objected at

trial, arguing that the underlying surveillance video had not

been authenticated. He renews this argument on appeal.22

Because defense counsel preserved this argument at trial, we

review "to determine whether the judge abused [his] discretion

and, if so, whether the error resulted in prejudice to the

defendant." Commonwealth v. Connolly, 91 Mass. App. Ct. 580,

586 n.6 (2017). We hold that the judge did not abuse his

discretion in admitting the surveillance video.

"To satisfy the requirement of authenticating or

identifying an item of evidence, the proponent must produce

evidence sufficient to support a finding that the item is what

the proponent claims it is." Mass. G. Evid. § 901(a). See

Mass. G. Evid. § 104(b). See also Commonwealth v. Purdy, 459

Mass. 442, 448-449 (2011). Authenticating a surveillance video

is "typically . . . done through one of two means -- having an

eyewitness testify that the video is a fair and accurate

22 The defendant does not argue that Carty's cell phone
video was not authenticated. His argument both at trial and on
appeal pertains to the underlying surveillance video.
 32

representation of what he saw on the day in question, or having

someone testify about the surveillance procedures and the

methods used to store and reproduce the video material."

Connolly, 91 Mass. App. Ct. at 586.

These are not, however, the exclusive ways a video can be

authenticated. In addition, "[e]vidence may be authenticated by

circumstantial evidence alone, including its '[a]ppearance,

contents, substance, internal patterns, or other distinctive

characteristics.'" Commonwealth v. Siny Van Tran, 460 Mass.

535, 546 (2011), quoting Mass. G. Evid. § 901(b)(4). See Purdy,

459 Mass. at 448-449 ("A judge making a determination concerning

the authenticity of a communication sought to be introduced in

evidence may look to 'confirming circumstances' that would allow

a reasonable jury to conclude that this evidence is what its

proponent claims it to be"); Commonwealth v. Nardi, 452 Mass.

379, 396 (2008), quoting Commonwealth v. LaCorte, 373 Mass. 700,

704 (1977) (proof of authenticity may take form of testimony

"that circumstances exist which imply that the thing is what its

proponent represents it to be").

The defendant relies on Connolly, 91 Mass. App. Ct. at 586-

588, for the proposition that where secondary evidence is

introduced regarding the contents of a video, the underlying

video must also be authenticated. This is true. "Of course,

had the video been available at trial, the Commonwealth would
 33

have had to authenticate it before it could be admitted." Id.

at 586. A proponent is not exempt from our rules of

authentication if he or she introduces a video of a video,

instead of introducing the underlying video itself.

Here, however, the underlying video was authenticated

through plentiful circumstantial evidence indicating a jury

could find that it was what it purported to be. Carty testified

that he saw a car, with its driver's side door open, that had

crashed into a pole at the intersection of Quincy Street and

Baker Avenue. Across the street from where the car was crashed,

Carty found seven shell casings and several bullet fragments.

In addition, Carty said multiple still photographs were fair and

accurate representations of the scene of the shooting and the

crash. Among these photographs were multiple of a blue sedan --

the same color and body style as that of the car in the cell

phone video of the surveillance video -- that had crashed into a

light pole bearing signs designating Quincy Street and Baker

Avenue. At least one of the photographs depicted a sign in

front of the crashed car; the sign advertised a church and was

black and white with red lettering. In the cell phone video, the

same sign is visible in front of the car.

In addition, Carty viewed the surveillance video in the

immediate aftermath of the shooting after he personally

approached the resident to whom the surveillance system
 34

belonged. That mitigates concerns that the video could have

been manipulated. Compare Connolly, 91 Mass. App. Ct. at 588

(fact that officer did not obtain surveillance video until

month after incident and did not testify about circumstances

that led him to view video raised concern that video could have

been manipulated). Finally, Rock's testimony that she heard a

sound like a car backfiring or gunshots, and then saw a Black

man with braids, wearing a red shirt, and with his hands in his

pockets, provides further circumstantial evidence to

authenticate the video. Thus, the circumstantial evidence was

sufficient to enable a reasonable jury to find that the video

was what it purported to be. The judge did not abuse his

discretion in admitting it.23

c. Commonwealth's argument that jurors could identify the

defendant from a video. Finally, the defendant argues that it

was improper for the Commonwealth to urge the jury to identify

23The defendant argues that the video's time stamp raises
questions about its authenticity. For most of the cell phone
video of the surveillance video, the time stamp in the upper-
right corner is not visible. However, sixteen seconds into the
cell phone video, the surveillance video time stamp appears to
show "9/15/2015 11:33." The defendant argues that even if we
assume that the time stamp is off by an hour, 10:33 A.M. would
not match up with the supposed time of the shooting, according
to the defendant's GPS data points. It is true that an accurate
and more visible time stamp could have helped to authenticate
the video. See Connolly, 91 Mass. App. Ct. at 588. Yet, given
the substantial other circumstantial evidence, a time stamp was
not necessary to authenticate the video.
 35

him as the shooter based on grainy video of someone who only

matched his generic description. This issue is unpreserved, and

thus we review for a substantial risk of a miscarriage of

justice. See AdonSoto, 475 Mass. at 504.

In the Commonwealth's opening, the prosecutor twice urged

the jury to identify the defendant based on the video. The

prosecutor stated, "You are going to be able to see who the

person is on that video and you are going to be able to compare

it to the person sitting in [the defendant's] chair. I submit

to you . . . , you will be able to tell that it's Mr. Davis."24

In closing, the prosecutor did not explicitly suggest that the

jury could identify the defendant based on the video. He only

stated that the video, in conjunction with the other evidence,

showed that the defendant had committed the crimes.25

24Later, the prosecutor stated, "You are going to see the
video . . . . There was someone that looks incredibly similar
to Mr. Davis raising his hand letting off seven rounds at a car
. . . ."

25 For example, the prosecutor stated, "[T]here are numerous
pieces of evidence which will allow you to find beyond a
reasonable doubt that Mr. Davis is the person depicted in that
video."

The closest the prosecutor came in the closing to asking
the jury to identify the defendant based on the video was
commenting that the defendant "[h]appens to look like the
shooter." We discern no error with this statement because
unlike the opening, the prosecutor did not state that the jury
could identify the defendant based on the video alone. Instead,
he merely stated that the defendant's appearance was consistent
with the shooter's.
 36

The defendant and amici liken the prosecutor's comments to

a single-suspect showup identification without any of the

procedural protections required for such an identification. See

Commonwealth v. German, 483 Mass. 553, 563 (2019), quoting

Supreme Judicial Court Study Group on Eyewitness Evidence:

Report and Recommendations to the Justices 92 (July 25, 2013)

("most significant of pre-identification warnings is that 'the

offender may or may not be in the photo array or lineup, or the

person being shown in a showup'"); Commonwealth v. Forte, 469

Mass. 469, 477 (2014) ("An identification stemming from a

videotape containing only one individual is analogous to a one-

on-one identification, which is considered inherently

suggestive"). Further, the defendant and amici argue that the

general characteristics of being a Black man with long braids or

dreadlocks are not enough to support reasonable suspicion, let

alone the inference that one could identify the defendant based

on the video. See Commonwealth v. Warren, 475 Mass. 530, 535-

536 (2016) (general description of three Black males, two with

dark clothing and one with red "hoodie" not sufficient for

reasonable suspicion); Commonwealth v. Cheek, 413 Mass. 492, 496

(1992) (description of "black male with a black 3/4 length goose

[jacket]" not enough for reasonable suspicion because it "could

have fit a large number of men who reside in . . . a

predominately black neighborhood of the city"). Moreover, amici
 37

argue that the characterization of a Black man with braids is a

prime trigger for implicit bias, due to stereotypical

associations about criminality based on this description.

"The proper function of an opening is to outline in a

general way the nature of the case which the counsel expects to

be able to prove or support by evidence" (citation omitted).

Commonwealth v. Sylvia, 456 Mass. 182, 188 (2010). "The

prosecutor's expectation must be 'reasonable and grounded in

good faith.'" Id., quoting Commonwealth v. Fazio, 375 Mass.

451, 456 (1978). "Absent a showing of bad faith or prejudice

. . . the fact that certain evidence fails to materialize is not

a ground for reversal" (citation omitted). Id. "[A] claim of

improper [opening statement] by the prosecutor must be judged in

light of the entire [statement], the judge's instructions to the

jury, and the evidence actually introduced at trial" (citation

omitted). Commonwealth v. Barbosa, 477 Mass. 658, 669 (2017).

We agree with the defendant that the Commonwealth's

suggestions that the jury could identify the defendant based on

the video were unreasonable. The video is not high enough

resolution and is taken from too far away to be able to discern

any features of the shooter's face. Cf. Commonwealth v.

Vasquez, 482 Mass. 850, 861 (2019) (eyewitness unfamiliar with

suspect would likely be unable to make identification based on

poor quality video). All one can see is that the shooter is a
 38

Black man with long hair in braids or dreadlocks that extend

down to his midback.26 As amici point out, braided hairstyles

are not uncommon among Black people. Given the shooter's common

hairstyle and the inability to see any of his facial features,

it was unreasonable for the Commonwealth to ask the jury to

identify the defendant as the shooter in the video. See Sylvia,

456 Mass. at 188 (prosecutor's statements in opening must be

based on reasonable expectation of what evidence will show).

Because we reverse the defendant's convictions on other grounds,

we need not determine whether these improper statements gave

rise to a substantial risk of a miscarriage of justice.

Conclusion. The judgments against the defendant are

reversed, the verdicts are set aside, and the case is remanded

to the Superior Court for a new trial and further proceedings

consistent with this opinion.

So ordered.

26At trial, to show what the defendant looked like around
the time of the shooting, the Commonwealth introduced a
photograph of him taken the following day. The photograph
depicts the defendant's face and shows that his hair is in what
appear to be thin dreadlocks. However, because the dreadlocks
extend behind the defendant's shoulders, a viewer can only tell
that they are longer than shoulder length and not whether they
are as long as the shooter's, which extend to about midback.
Even if the length of the defendant's hair was similar to that
of the perpetrator in the video, such evidence still would have
fallen short of evidence from which the jury could have
identified the defendant as the perpetrator depicted in the
video.