dissenting.
¶15 From seemingly innocuous statements, the seeds of significant change are sometimes sown. So it is with this line from the majority opinion: “With repetitive, specific admonitions and a limited distribution, there was no reasonable likelihood that the jurors had been exposed to the prejudicial news reports.” Maj. op. ¶ 13. With this application of a subtly retooled test for handling midtrial publicity, the majority effectively abandons our precedent in Harper v. People, 817 P.2d 77 (Colo. 1991). It does so by creating a de facto presumption that jurors will successfully follow instructions to avoid media exposure. Going forward, when trial courts have told jurors to avoid media coverage of a case, they need not canvas the jurors to ask about inadvertent exposure to such coverage. They need not inquire about this less-controllable threat to a fair trial, no matter how prejudicial the report at issue may be. Instead, they may rely blindly on instructing the jury to avoid the media. By allowing this, the majority reduces Harper to little more than an apparition, whose outline remains faintly present but whose essence eludes grasp.
¶16 The majority relegates our settled law to this insubstantial fate without being asked to do so and without so much as a passing nod to stare decisis. The People did not invite us to alter our precedent. And neither party shared its thoughts with us about the application of stare decisis to the case before us.
¶17 Because the majority offers no compelling rationale for departing from our holding in Harper, and because I believe that under Harper’s standai'd the trial court abused its discretion in refusing to poll the jury, I respectfully dissent.
I. Background
¶18 The facts merit elaboration. In this vehicular homicide case, Jacobson struck a taxi and killed two children’s librarians who were visiting Denver for a library conference. This high-speed, alcohol-fueled crash occurred on Peña Boulevard, near Denver International Airport, at 10 a.m.
¶19 The ease generated significant media attention. Prospective jurors were questioned about pretrial publicity during voir dire, and, more than a year after the incident, a number of them remembered hearing about it on television. The trial court was thorough in admonishing the jury to avoid media exposure, giving a detailed warning during voir dire and then providing daily reminders.
¶20 Three days into the prosecution’s case, defense counsel alerted the trial court that a local reporter was “going to run a full spread this evening on the news.” The prosecution also expressed concern about the story, dispatching its public information officer to ask if the reporter would hold the story until after trial. Before releasing the jury for the day, the court emphasized its daily admonishment to the jury to avoid exposure to news about the case, calling it “vitally important.”
¶21 Channel Four News aired the story that evening, broadcasting inflammatory facts and allegations about Jacobson that the trial court had ruled could not come into evidence at trial. The story referred to Jacobson’s past traffic incidents resulting in injury, traffic violations, and convictions— including one DUI. It also reported Jacobson was facing at least one more trial. Further, it relayed allegations that Jacobson was driving with a suspended license when this accident occurred, and had been doing so during at least one earlier crash. An interview with a victim of that earlier crash added an emotional flare: the victim implored that Jacobson should be locked up forever.
¶22 The next day, defense counsel asked the court to poll the jurors about whether they had seen media coverage about the case the previous evening. He argued that, even if jurors were following the admonition to avoid media, “in this day of media access, [it is] *553highly likely that inadvertently some juror could have been [exposed to the news story].”
¶23 The trial court had watched the story and recognized its contents were highly prejudicial, but it declined to poll the jury. It reasoned: (1) polling the jury about a news story would be “like dangling a cookie under a 2-year old’s nose,” (2) in anticipation of the news story, the afternoon before the story aired it had issued “a longer and stronger” admonition to avoid media exposure, and (3) it had noted that “several of the jurors got a very serious look on their face and nodded their heads” during that extended admonition.
II. Analysis
¶24 I begin with an overview of Harper and its well-established protocol for handling midtrial media issues. Next, I examine the majority’s decision abandoning that protocol. I then explain why the doctrine of stare decisis makes that decision a mistake and how, in departing from Harper, we are also departing from the nationwide majority rule—a rule that, ironically, Colorado helped forge more than a quarter of a century ago.
¶25 I finish by applying Harper to the facts in the case before us. Although the trial court is to be commended for proactively admonishing the jury, I conclude that on the facts of this ease, it erred by refusing to poll the jury. Consequently, I believe that faithful application of our precedent dictates that Jacobson’s conviction should be reversed.
A. Harper’s Standard for Polling a Jury About Midtrial Media
¶26 In Harper, we adopted a three-step process for evaluating the possibility of prejudice resulting from midtrial media reports when the defendant raises the issue during trial: first, “determine whether the coverage has a potential for unfair prejudice,” second, “canvass the jury to find out if they have learned of the potentially prejudicial publicity,” and third, question any exposed jurors individually to determine how much they know about the publicity and whether it affects their ability to decide the case fairly. 817 P.2d at 83 (quoting United States v. Gaggi, 811 F.2d 47, 51 (2d Cir. 1987)).
¶27 We expressly rejected our earlier decision in People v. Holmes, 191 Colo. 477, 553 P.2d 786 (1976), in which we had held that a trial court was not required to poll the jury regarding a newspaper article about the case released during trial. Harper, 817 P.2d at 82-83. In Holmes, we reasoned the trial court had admonished the jurors to ignore newspaper articles about the trial and the defendant presented no evidence that the jurors had seen the article. See Harper, 817 P.2d at 82. We had, in Holmes, “adhered to the presumption ... that the jury had followed the court’s instruction.” Harper, 817 P.2d at 82.
¶28 On the way to the three-step test in Harper, we found blind reliance on instructions inadequate. We concluded the reasoning in Holmes was flawed because it did not account for either (1) the possibility that jurors followed the court’s instructions but inadvertently encountered material about the case, or (2) the obstacles to obtaining evidence that jurors have been exposed to extraneous material, given that ethical rules (now Colorado Rule of Professional Conduct 3.5) prohibit lawyers from communicating with jurors. Id.
¶29 We therefore held:
A presumption that jurors follow court instructions not to permit themselves to be exposed to media reports, therefore, does not adequately take into account either the likelihood that a juror could acquire information without violating the court’s instructions or the difficulty of discovering whether jurors were exposed to such a report. We conclude that to resolve the issue of prejudice on the basis of such a faulty presumption fails to provide adequate protection for the defendant’s constitutionally based right to a fair trial.
Id. at 83.
¶30 We noted, “In implementing the first step, the trial court should focus principally upon whether the content of the media report is inherently prejudicial.” Id at 84. Other relevant factors, we said, include (1) whether the report contained inadmissible information, (2) whether it related to matters at issue at trial, (3) the timing of the publication, (4) the likelihood that the jury was *554exposed to it, and (5) the likely effectiveness of a trial court’s instruction to avoid such material. Id.
¶31 In seeking to safeguard a defendant’s right to a fair trial, we also issued a step-one imperative: “Doubt about the existence of prejudice should be resolved by proceeding to step two and polling the jurors as a group.” Id,
¶32 In applying that standard to the facts in Harper, we acknowledged that the trial court had instructed jurors not to read media accounts of the trial, but ultimately concluded that “[i]n light of the article’s highly prejudicial contents, some direct inquiry of the jury was necessary.” Id. at 86.
B. The Majority Dismantles Harper
¶33 The majority picks some words and phrases from Harper and reassembles them into something new. The result resembles Harper’s protocol but maintains none of its force. Defendants reaching for Harper’s step two will do so largely in vain, because the majority’s new step-one hurdle is almost insurmountable.
¶34 Critically, the majority recasts Harper’s first step in the ostensible process of simply reciting it. As Harper phrased it, the first step is “to determine whether the coverage has a potential for unfair px-ejudice.” Id. at 83 (emphasis added) (quoting Gaggi, 811 F.2d at 51). But when the majority paraphrases that same stép, it says this: “Fust, the trial court should determine whether the coverage is actually prejudicial and contains inadmissible evidence.” Maj. op. ¶ 10 (emphasis added) (citing Harper, 817 P.2d at 83). This actual-prejudice requirement is entirely the majority’s creation; it can’t be found among the pages of Harper—except where Harper described the old rule that it ultimately rejected. See Harper, 817 P.2d at 80 (“Our earlier eases required that a defendant demonstrate that actual prejudice resulted from juror exposure to extraneous information or influences in order to be awarded a new trial.”).
¶35 The majority also makes seemingly modest, but ultimately far-reaching, changes to the Harper step-one factors. Harper listed six factors: one primai'y factor—the content’s inherent prejudice, four “[ojther relevant factors,” and one final consideration—the “likely effectiveness of any instruction [to avoid media],” the last of which was the only of the six factors the Harper court took pains to diminish. Harper, 817 P.2d at 84 (“The existence of admonitions, alone, does not sufficiently neutralize news reports in the community where the trial is being held that may reasonably be believed to have come to the attention of the jurors.”). The majority minimizes what had been the most significant factor (inherent prejudice) and aggrandizes that which had been the least significant (admonitions). Maj. op. ¶¶ 11-12.
¶36 Tellingly, the majority omits altogether Harper’s admonition that “[d]oubt about the existence of prejudice [at step one] should be resolved by proceeding to step two and polling the jurors as a group,” 817 P.2d at 84. Thus, the majority significantly raises the bar to get to Harper’s second step.
¶37 In sum, while a remnant of Harper remains, the practical effect -is that we’ve returned to the pre-Harper rule: to poll a jury that has been adequately admonished, a defendant must show that a juror was actually exposed to the prejudicial story. Practically speaking, then, the defendant must know a juror saw the story before he can ask whether the juror saw the story.
C. Stare Decisis
¶38 “Under the doctrine of stare decisis courts are very reluctant to undo settled law.” Creacy v. Indus. Comm’n, 148 Colo. 429, 366 P.2d 384, 386 (1961). Stare decisis “promotes uniformity, certainty, and stability of the law.” People v. Porter, 2015 CO 34, ¶ 23, 348 P.3d 922, 927 (quoting People v. LaRosa, 2013 CO 2, ¶ 28, 293 P.3d 567, 574). Thus, this court departs from a rale established in a prior case only where “sound reasons exist for doing so.” People v. Kutlak, 2016 CO 1, ¶ 18, 364 P.3d 199, 205 (quoting People v. Blehm, 983 P.2d 779, 789 (Colo. 1999)).
¶39 Here, the majority reshuffles Harper’s priorities without offering any explanation at all. Harper provided two reasons for focusing on potentially prejudicial content and elimi*555nating unqualified reliance on instructions to avoid the media; the majority addresses neither.
¶40 First, jurors may not be able to control their exposure to media about the case. In almost any case, a juror dutifully following the court’s instructions to avoid media coverage could still inadvertently come across a story about the case. Harper, 817 P.2d at 82. News clips are frequently disseminated on online platforms not intended solely for news, such as Facebook and Twitter. See State v. Gathercole, 877 N.W.2d 421, 431 (Iowa 2016) (“Midtrial publicity that ... is shared widely on social media more likely reaches the jury than publicity disseminated only through one channel, method, or medium.”). Commercial breaks in non-news TV programs may advertise and preview the nightly news. Friends, family, or nearby strangers on a bus or at a restaurant may discuss a local news story within earshot of a juror. Moreover, given how news agencies frequently style headlines to grab attention, even a juror exposed only to a headline may be tainted.
¶41 Second, requiring defendants to show actual prejudice—that a juror was actually exposed to the harmful media—is unworkable because “ethical rules prohibit counsel from communicating with jurors or their family members during the trial.” Harper, 817 P.2d at 82. Thus, defendants are left to rely on jurors to volunteer during trial that they failed at following the court’s instructions, which seems a risky bet. See State v. Bey, 112 N.J. 45, 548 A.2d 846, 866 (1988) (“If, however, a trial court chooses to rely exclusively on its cautionary instructions, such incidents would likely go unmentioned, unless we can be certain of empanelling jurors who would volunteer their infractions to the court and counsel.”).
¶42 These reasons articulated in Harper resonate even more today. When Harper was decided in 1991, the court worried about inadvertent exposure to a short article under a subheading in a column called “News Shorts” in an interior page of a local newspaper. Harper, 817 P.2d at 79. Since then, the proliferation of the internet and technology has revolutionized how the world receives information.
¶43 For example, the majority classifies the prejudicial reports here as having “very limited distribution” because “they were available only during one channel’s evening newscast and on that channel’s website.” Maj. op. ¶ 13. But in this day and age, to watch “one channel’s evening newscast,” one need not sit in the living room during a particular time with a television tuned to a particular channel—television programs can be watched in any place at any time on an array of devices. And a link to one channel’s website can spread virally through myriad virtual avenues to millions of viewers in a matter of minutes. In 1991, a person had to seek out news actively, maybe by opening a newspaper or turning a television to a news channel. Today, a smartphone (if set to vibrate) can literally shake a person to notify him or her of a news story. Therefore, the Harper court’s concern with inadvertent exposure. to inflammatory, news is even more compelling today.
¶44 The majority ultimately offers one affirmative reason for dismantling Harper; it suggests that Harper could be harmful because polling the jury about “specific news reports” may encourage jurors to seek them out. Id This echoes the trial court’s concern that polling about the news story “would be like dangling a cookie under a 2-year old’s nose.”
¶45 But there is no need to poll about a specific news report. A trial court could merely pose a general question to jurors, i.e., “Has anyone seen or heard anything about this trial—even by accident—on the internet or the TV, in a newspaper, or from other people?” Such a general inquiry, whether posed routinely after recesses or in response to a particular news report, is sufficiently clear and unintimidating that jurors should volunteer inadvertent news exposure, but vague enough not to alert jurors to any reports that may exist. See Harper, 817 P.2d at 86 (“Framing this inquiry in general terms would minimize any implication that an article containing information adverse to the defendant had been published.”). Further, any concern that polling may tempt jurors to seek out prejudicial news—in deliberate violation of the court’s instructions—is under*556mined by the majority’s animating premise: jurors follow instructions to avoid media.
D. American Courts Disfavor the Majority’s Approach
¶46 When it reverts to an actual-prejudice standard for polling the jury, the majority shifts Colorado out of the mainstream of American jurisprudence on midtrial publicity. Under the American Bar Association Standards for Criminal Justice, polling is required upon motion when midtrial media “raises seiious questions of possible prejudice.” ABA Standards for Criminal Justice: Fair Trial and Free Press § 8-3.6(e) (3d ed. 1992). This, or its analog, is the majority approach in America. Among American jurisdictions that have established a standard for handling midtrial media, authorities show, and my independent research confirms, that a clear majority require polling of jurors upon some showing of potential for prejudice. See Fern L. Kletter, Annotation, Interrogation or Poll of Jurors, During Criminal Trial, as to Whether They Were Exposed to Media Publicity Pertaining to Alleged Crime or Trial, 55 A.L.R.6th 157 (2010) (categorizing approaches of thirty-six jurisdictions); State v. Holly, 145 N.M. 513, 201 P.3d 844, 849 (2009) (“The ABA approach has been adopted in some form by the majority of federal circuits and in several states.”).
¶47 All the more reason not to retreat from Harper. Instead, I would apply the standard as articulated in Harper to evaluate whether the trial court abused its discretion here.
E. Harper Requires Reversal
¶48 In my view, the trial court abused its discretion in refusing to poll the jury. A trial court abuses its discretion where its ruling is manifestly arbitrary, unreasonable, or unfair, or is based on an erroneous view of the law, People v. Elmarr, 2015 CO 53, ¶ 20, 351 P.3d 431, 438, or where the court fails to exercise discretion, People v. Darlington, 105 P.3d 230, 232 (Colo. 2005).
¶49 Here, the trial court’s process was inconsistent with Harper, and thus was either based on an erroneous view of the governing law or constituted a failure to exercise discretion under it. The court did not weigh the factors ai*ticulated in Harper in deciding whether to poll the jury. It relied almost exclusively on its belief that the jurors followed its instruction to avoid media—a rationale that Harper foreclosed. (To be fair, no one brought Harper to the trial court’s attention, but its decision contravenes Harper all the same.)
¶50 A consideration of the factors under step one of the Harper test demonstrates the potential prejudice as a result of the report.
¶51 Most importantly, the content of the report was inherently prejudicial, inadmissible, and closely related to the matters at trial. Jacobson was on trial for vehicular homicide and DUI, and the report suggested she had a history of drunk, reckless driving and defying traffic laws. The trial court said that the information “was absolutely not anything that’s going to come up in this trial and should not.”
¶52 The timing was prejudicial; the report occurred near the close of the prosecution’s case. Indeed, the trial court noted “the timing was unfortunate.”
¶53 As for the likelihood the jury was exposed to the report, we have little to go on. The report was on one local television channel’s nightly news and on its website, but we don’t know how widely it was shared on the internet, if at all, and the trial court made no findings about the likelihood that jurors were inadvertently exposed to it.
¶54 The trial court found its instruction adequate, noting it had given a “longer and stronger” admonition the afternoon before the report and that “several of the jurors got a very serious look on their face and nodded their heads” during that admonition. See Harper, 817 P.2d at 84 (“The court also should consider the likely effectiveness of any instruction [to avoid media] in light of the nature and manner of dissemination of the news reports.”). But Harper was clear that this factor alone does not justify declining to poll the jury. Id. (“The existence of admonitions, alone, does not sufficiently neutralize [media] that may reasonably be believed to have come to the attention of the jurors.”).
*557¶55 Because the circumstances in this ease are almost indistinguishable from the circumstances in Harper, the result should be the same. There, as here, the report contained prejudicial, inadmissible information about the defendant’s prior conduct that directly-related to the issues before the jury. See id. at 85. And where the jurors in Harper could have inadvertently seen the brief fourth-page article in the local newspaper, id. at 79, 85-86, it seems even more likely that the jurors here could have inadvertently seen the news story on the television or the internet. Although the trial court here thought its admonitions were effective, that is not enough to tip the scales away from the need to poll the jury. At a minimum, doubt existed as to the potential for prejudice from the news story, and that doubt should have been resolved in favor of ensuring a fair trial. See id. at 84. Thus, the trial court here, like the one in Harper, abused its discretion in refusing to poll the jurors to determine exposure.
¶56 Because Harper—as it was written— should control, I would affirm the judgment of the court of appeals, reverse Jacobson’s convictions, and remand the case for a new trial.
I am authorized to state that JUSTICE MÁRQUEZ and JUSTICE GABRIEL join in this dissent.