Mario Flores Molina v. Merrick Garland

FOR PUBLICATION UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT MARIO RAJIB FLORES MOLINA, Nos. 19-73028 Petitioner, 20-71774 v. Agency No. A215-879-596 MERRICK B. GARLAND, Attorney General, Respondent. OPINION On Petition for Review of an Order of the Board of Immigration Appeals Argued and Submitted April 14, 2021 Pasadena, California Filed June 13, 2022 Before: Richard A. Paez and Lawrence VanDyke, Circuit Judges, and Edward R. Korman, * District Judge. Opinion by Judge Paez; Concurrence by Judge Korman; Dissent by Judge VanDyke * The Honorable Edward R. Korman, United States District Judge for the Eastern District of New York, sitting by designation. 2 FLORES MOLINA V. GARLAND SUMMARY ** Immigration The panel (1) granted Mario Rajib Flores Molina’s petition for review of the Board of Immigration Appeals’ decision affirming an immigration judge’s denial of asylum and related relief, and remanded, holding that the record compelled a finding that Flores Molina’s past experiences constituted persecution and that the Board erred in its analysis of other issues; and (2) dismissed as moot Flores Molina’s petition for review of the Board’s denial of his motion to reopen. Flores Molina was publicly marked as a terrorist and threatened with torture over social media by Nicaraguan government operatives, repeatedly verbally threatened with death by supporters of the Ortega regime, received a death threat painted on his home by masked men likely affiliated with the government, and received a second death threat— this time during a direct confrontation—after he was seriously beaten by six members of the Sandinista Youth. Flores Molina also had a near confrontation with an armed paramilitary group that located him at a hideaway. The panel explained that the threats were credible given the history and context of the Ortega regime’s killing and torture of its political opponents. The panel observed that this court has stated in various opinions that both the de novo and the substantial evidence standard of review apply to the question of whether ** This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader. FLORES MOLINA V. GARLAND 3 particular acts constitute persecution. The panel wrote that it need not address the nuances of the two standards, or which standard should apply, because the harm Flores Molina suffered rose to the level of persecution under the more deferential substantial evidence standard of review. The panel held that the record compelled the conclusion that Flores Molina’s experiences in Nicaragua constituted persecution. First, the panel wrote that this court has consistently recognized that being forced to flee from one’s home in the face of an immediate threat of severe physical violence or death is squarely encompassed within the rubric of persecution. Here, Flores Molina was forced to flee three separate times after being personally targeted with violence and threatened with death for his political views. Second, the panel wrote that this court has repeatedly held that threats may be compelling evidence of past persecution, particularly when they are specific and menacing and are accompanied by evidence of violent confrontations, near-confrontations and vandalism, as was the case here. Moreover, this court has consistently held that death threats alone can constitute persecution. The panel concluded that any reasonable adjudicator would be compelled to hold that the repeated and specific threats that Flores Molina experienced, amid violence and menacing confrontations, amount to persecution. Third, the panel wrote that that an applicant may suffer persecution based on the cumulative effect of several incidents, even if no single incident rises to the level of persecution. The panel explained that this is a fact-bound endeavor that is not reducible to a set formula, but rather requires that the relevant facts be evaluated in combination with each other to form a sufficiently negative portrait of the 4 FLORES MOLINA V. GARLAND petitioner’s experience in his or her own country that not only allows a finding of past persecution but requires it. Here, Flores Molina was repeatedly threatened and subjected to violence, in an escalating fashion, all within the welldocumented backdrop of the Ortega regime’s violent crackdown on members of the political opposition. Turning to the issue of Flores Molina’s claim that he has a well-founded fear of future persecution, the panel held that the Board erred by failing to address highly probative evidence. The panel explained that the Board cited the record selectively, relying on two news reports of the Ortega regime’s release of 100 prisoners and its intention to release more, to support its assertion that Flores Molina’s fear of future persecution was speculative, while ignoring other evidence that documented the conditions released prisoners faced, delays in releasing political prisoners, the detention and disappearance of additional activists and protesters in the interim, and gross human rights violations in Nicaragua. Moreover, the Board failed to discuss whether the repeated death threats and threats of violence Flores Molina faced were sufficient to inspire a well-founded fear of future persecution. Likewise, the Board failed to address highly probative evidence concerning the likelihood of torture. The panel remanded for the Board to consider the remaining elements of past persecution, Flores Molina’s claim for humanitarian asylum, and all of the probative evidence concerning whether Flores Molina established a well-founded fear of future persecution or clear probability of torture. Because it granted Flores Molina’s petition as to the denial of asylum and related relief, the panel dismissed as FLORES MOLINA V. GARLAND 5 moot Flores Molina’s petition as to the denial of his motion to reopen. Concurring, District Judge Korman wrote separately to address the standard of review applicable to the Board’s past persecution determination. Judge Korman wrote that although this court owes deference under the substantial evidence standard to the administrative findings of fact, whether particular facts constitute persecution for asylum purposes is a legal question reviewed de novo. Judge Korman explained that the substantial evidence standard is not a good fit for questions, like the one presented in this case, regarding the application of a legal standard to settled facts. Judge Korman agreed with the majority that the decision in this case would be the same regardless of which standard applied, but noted that he would also have concurred in a majority opinion concluding that the Board legally erred in concluding that Flores Molina’s hardships did not amount to persecution. Dissenting, Judge VanDyke would deny the petition because the record does not compel the conclusion that (1) the past harassment Molina suffered rose to the level of past persecution, or that (2) such harassment—together with the most recent country conditions evidence that was before the agency—demonstrated a well-founded fear of future persecution. Judge VanDyke would also hold that the record does not compel a contrary conclusion with respect to Flores Molina’s remaining applications for withholding of removal, humanitarian asylum, or protection under CAT, and that the Board did not abuse its discretion in denying Flores Molina’s motion to reopen. Judge VanDyke addressed the three interconnected layers of deference this court owes to immigration agency 6 FLORES MOLINA V. GARLAND decisions, which combine to form what should be one of the most deferential standards of review in our legal system. First, the scope of the court’s review is tightly circumscribed by the extraordinarily deferential standard that Congress has commanded—yielding to the agency’s determinations unless a different conclusion is compelled. Second, the court applies that extreme deference to the extraordinarily difficult and often indeterminate factual inquiries that the agency alone is charged with making. And third, as the Supreme Court recently reiterated in Garland v. Dai, 141 S. Ct. 1669 (2021), the agency enjoys extraordinary discretion in making the difficult determinations of how much credibility, weight, and persuasiveness to afford different parts of the record in reaching its factual conclusions. Judge VanDyke wrote that his view of Flores Molina’s past harms is not far from that of his colleagues— the facts present a close call, and he is sympathetic to the majority’s view that Flores Molina may have suffered past persecution. Judge VanDyke wrote that he diverges with the majority regarding its approach to the agency’s decision and the record, explaining that the majority admittedly travels a well-trodden path looking for a basis to overturn the agency instead of scouring the record as a whole looking for a way to uphold the agency if even a single reasonable factfinder could agree with its ultimate conclusion. Judge VanDyke wrote that this court’s edifice of immigration caselaw has obfuscated the correct standard of review, making the proper approach harder to see and even harder to execute. These small differences of opinion, as illustrated in this case, have been multiplied over time in many decisions, leading to the lopsided edifice that is currently improperly driving much of this court’s immigration caselaw. FLORES MOLINA V. GARLAND 7 COUNSEL Mary-Christine Sungaila (argued) and Joshua R. Ostrer, Buchalter APC, Irvine, California; Paula M. Mitchell, Attorney; Tina Kuang (argued) and Natalie Kalbakian (argued), Certified Law Students, Loyola Law School, Los Angeles, California; for Petitioners. Jeffrey R. Leist (argued), Senior Litigation Counsel; Anthony C. Payne, Assistant Director; Office of Immigration Litigation, Civil Division, United States Department of Justice, Washington, D.C.; for Respondent. OPINION PAEZ, Circuit Judge: Petitioner Mario Rajib Flores Molina (“Flores Molina”) participated in demonstrations against the ruling regime in his native Nicaragua, where he witnessed the murder of his friend and fellow protester by police and paramilitary members. Thereafter, he was publicly marked as a terrorist, threatened with torture and death by government operatives, and forced to flee his home. Flores Molina, however, was tracked down at his hideaway by armed paramilitary members, and was forced to flee for his life a second time. Flores Molina still was not safe. He was discovered, yet again, assaulted, and threatened with death by a governmentaligned group. Flores Molina ultimately fled a third time— from Nicaragua altogether—out of fear for his safety. He eventually presented himself to authorities at the United States border and sought asylum and other relief. 8 FLORES MOLINA V. GARLAND When Flores Molina sought asylum, withholding of removal and protection under the Convention Against Torture (“CAT”), an Immigration Judge (“IJ”) and the Board of Immigration Appeals (“BIA”) determined that his past experiences in Nicaragua did not rise to the level of persecution. They also determined that Flores Molina did not establish a well-founded fear of future persecution. The IJ and BIA denied all forms of relief and ordered Flores Molina’s removal to Nicaragua. Flores Molina petitions for review of the BIA’s denial of his appeal of the IJ’s decision, as well as of the BIA’s subsequent denial of his motion to reopen proceedings. Because the record compels a finding that Flores Molina’s past experiences constitute persecution and because the BIA erred in its analysis of the other issues, we grant the first petition and remand for further proceedings. Accordingly, we dismiss the second petition as moot. I. Factual and Procedural Background A. Political Context: 2018 Protests and Violent Suppression in Nicaragua The Sandinista National Liberation Front or “Frente Sandinista de Liberación Nacional” (“FSLN”) regained control of the Nicaraguan government in 2007 under Daniel Ortega. The FSLN maintains power in part through Citizen Power Councils (“CPCs”), FSLN party-based grassroots organizations that operate in neighborhoods and districts across Nicaragua. The CPCs function as intelligencegathering entities for the Nicaraguan government. They also assist the government in suppressing dissent. CPCs and police work with paramilitary groups associated with the Sandinista Party to target the homes of protesters. In recent years, they have abducted and detained protesters, and raided homes of suspected protesters across Nicaragua. FLORES MOLINA V. GARLAND 9 In April 2018, political opposition groups, university students and farmers organized protests against pension reform and government corruption. The protests developed into a wider movement in opposition to the Ortega regime, which was met with violent suppression by the FSLN, CPCs, police and paramilitary groups. Shortly after the commencement of the protest movement, the Nicaraguan parliament passed a law enabling the Ortega regime to prosecute protesters as terrorists, and to impose harsh penalties. The Guardian reported that between April and July 2018, it was estimated that over 300 protesters were killed by the police and government operatives. B. Flores Molina’s Participation in Protests and the Consequences Flores Molina is a graduate of the Autonomous National University of Nicaragua, married, and the father of two children. He has been an active member of the opposition Liberal Party since 2006. In April 2018, Flores Molina began to participate in the opposition protest movement in the city of Estelí. At the second protest he attended that month, police and paramilitary personnel shot bullets into the crowd of demonstrators. As he fled the shooting, Flores Molina learned that his friend, and fellow demonstrator, Franco Valdivia, had been shot in the head. Flores Molina stopped, turned around, and tended to Valdivia as he lay in a pool of his own blood, before Valdivia ultimately died of his injuries. Flores Molina participated in protests in Estelí throughout May 2018, as police and paramilitary members regularly shot at, wounded and killed demonstrators. As Flores Molina’s presence at the protests continued, he received escalating threats on his life from government operatives and paramilitary members. 10 FLORES MOLINA V. GARLAND Government operatives publicly circulated posts on social media identifying Flores Molina as an instigator of hate and violence and threatened to send him to “El Chipote” prison, notorious as a site for torture. Flores Molina continued to receive threats over WhatsApp and was aware of at least five public posts on Facebook that were widely circulated identifying him as a dangerous opponent of the government. The public posts galvanized Ortega supporters to locate and drive to Flores Molina’s home and verbally threaten him. Then, in June, Flores Molina found his home vandalized, with the words “Bullets to Strikers” spray painted on the walls by a group of masked individuals who arrived in an unmarked truck commonly known as the type of vehicle used by government operatives. The escalating digital and verbal threats, the death threat painted on his house, and the increasing number of killings of protesters by the Ortega regime forced Flores Molina to flee his home for safety. But five months after Flores Molina fled to a hideaway, a truck full of police officers and paramilitary members arrived at his refuge wearing ski masks, army jackets and carrying assault rifles. The paramilitary squad demanded that Flores Molina come out, climbed on the roof, and looked through the windows. Flores Molina hid in the backyard to evade detection; immediately after, he fled, for the second time, to a new hideaway. On November 20, 2018, six masked members of the proOrtega Sandinista Youth assaulted Flores Molina as he returned to his second hideaway. They struck him in the head, causing him to lose a tooth and leaving scarring on his lip. As they beat him, the attackers warned Flores Molina, “This is what happens to the ones that want to be part of the coup. And at the next encounter, we’re going to kill you.” FLORES MOLINA V. GARLAND 11 Flores Molina could not see a doctor because the hospital entrance was full of police and the paramilitary members. The United States Department of State Country Report on Human Rights Practices in Nicaragua for 2018 (“2018 State Department Report”) shows that the Ortega regime directed the Ministry of Health to deprive protesters of medical attention and instructed public hospitals and clinics not to provide medical care to wounded protesters. Flores Molina fled Nicaragua and ultimately arrived at the United States-Mexico border. He presented himself at a port of entry and requested protection. C. Administrative Proceedings Flores Molina appeared pro se and testified before an IJ on June 10, 2019, where he requested asylum, withholding of removal and protection under CAT. At the hearing, Flores Molina and the Department of Homeland Security submitted, as exhibits, news articles and country conditions reports on Nicaragua and the 2018 protest movement. The IJ found Flores Molina’s testimony consistent with the declaration he submitted in support of his application for relief, but determined that he had not shown that his past experiences constituted persecution for the purposes of asylum and withholding of removal. The IJ also held that, because Flores Molina failed to show past persecution, he had not demonstrated a well-founded fear of future persecution. The IJ denied Flores Molina’s application for asylum, withholding of removal and CAT protection. In November 2019, the BIA dismissed Flores Molina’s appeal, finding that “[t]he claimed past harm, cumulatively considered, [did] not rise to the level of past persecution” and that the “threats are not the sort of ‘extreme’ or ‘especially menacing’ threats necessary to establish past 12 FLORES MOLINA V. GARLAND persecution.” The BIA also concluded that Flores Molina lacked an objectively well-founded fear of future persecution because he was physically assaulted only once, and because the number of political activists detained in Nicaragua is small compared to the number of individuals who participated in the protest movement. As for CAT relief, the BIA determined that the past harm Flores Molina experienced did not rise to the level of torture and that the risk he would be tortured upon removal to Nicaragua was too speculative to merit relief. The BIA affirmed the IJ’s denial of asylum, withholding of removal and CAT protection. In June 2020, the BIA denied Flores Molina’s motion to reopen his removal proceedings to seek a continuance while the United States Citizenship and Immigration Services (“USCIS”) adjudicated his pending application for an immigrant visa. Flores Molina timely petitioned for review of the BIA’s denial of asylum, withholding of removal and CAT protection (No. 19-73028), and the BIA’s denial of his motion to reopen (No. 20-71774). We address both petitions. II. Standards of Review “Where the BIA conducts its own review of the evidence and law, rather than adopting the IJ’s decision, our review is limited to the BIA’s decision, except to the extent the IJ’s opinion is expressly adopted.” Rodriguez v. Holder, 683 F.3d 1164, 1169 (9th Cir. 2012) (citation and quotation marks omitted). Here, the BIA dismissed Flores Molina’s appeal, agreeing with several of the immigration judge’s findings while adding its own reasoning. Thus, we review the decisions of both the BIA and the immigration judge to FLORES MOLINA V. GARLAND 13 the extent that the BIA agreed with the immigration judge’s conclusions. Id. “We review factual findings for substantial evidence and legal questions de novo.” Guerra v. Barr, 974 F.3d 909, 911 (9th Cir. 2020). In particular, “[w]e review denials of asylum, withholding of removal, and CAT relief for substantial evidence.” Guo v. Sessions, 897 F.3d 1208, 1212 (9th Cir. 2018) (citation and quotation marks omitted). Those findings are “conclusive unless any reasonable adjudicator would be compelled to conclude to the contrary.” 8 U.S.C. § 1252 (b)(4)(B). Where the BIA does not consider all the evidence before it, either by “misstating the record [or] failing to mention highly probative or potentially dispositive evidence,” its decision is legal error and “cannot stand.” Cole v. Holder, 659 F.3d 762, 772 (9th Cir. 2011). III. Discussion We address the two petitions in turn. We first review the BIA’s denial of asylum, withholding of removal, and CAT relief to Flores Molina. 1 — 1 — Our dissenting colleague sharply criticizes our court’s immigration jurisprudence. This is not the first time he has expressed such views. See, e.g., Nababan v. Garland, 18 F.4th 1090, 1096, 1103 (9th Cir. 2021) (VanDyke, J., dissenting) (expressing “perpetual[] embarrass[ment]” by “[o]ur circuit’s immigration jurisprudence” and describing it as a “nasty habit” that we “should at least try to kick”); Reyes v. Garland, 11 F.4th 985 , 998 (9th Cir. 2021) (VanDyke, J., dissenting) (lamenting that “the Ninth Circuit’s abysmal and indefensible immigration precedents are the gifts that keep on taking”); id. at 1007 (discussing some “of our more blatant recent immigration gaffes” and describing the majority as “missing an opportunity to right our circuit’s badly listing immigration ship”); Avila-Arias v. Garland, 847 F.App’x 468, 472–73 (9th Cir. 2021) 14 FLORES MOLINA V. GARLAND A. Asylum To be statutorily eligible for asylum, Flores Molina must show that he is a refugee. 8 U.S.C. § 1158 (b)(1). A refugee is one who is “unable or unwilling to avail himself or herself of the protection of [his or her native] country because of persecution or a well-founded fear of persecution on account of race, religion, nationality, membership in a particular social group, or political opinion.” Id. § 1101(a)(42)(A). “Persecution is defined as ‘the infliction of suffering or harm . . . in a way regarded as offensive.’” Mendoza-Pablo v. Holder, 667 F.3d 1308, 1313 (9th Cir. 2012) (quoting Li v. Ashcroft, 356 F.3d 1153, 1158 (9th Cir. 2004) (en banc)). “Either past persecution or a well-founded fear of future persecution provides eligibility for a discretionary grant of asylum.” Ratnam v. INS, 154 F.3d 990, 994 (9th Cir. 1998). An individual “who establishes past persecution is presumed to have a well-founded fear of persecution.” Id. (citation omitted). (VanDyke, J., dissenting) (advocating for our court “to emulate the BIA” more and to pay special attention to deferring to the BIA when the petitioner has a criminal history); Aguilar-Osorio v. Garland, 991 F.3d 997 , 1000–01 (9th Cir. 2021) (per curiam) (declaring that the majority’s remand for the BIA to consider an issue in the first instance was “lawless” and that “[w]e make it very difficult, if not impossible, for the BIA to properly do the job Congress gave it”); Sanchez Rosales v. Barr, 980 F.3d 716, 721 (9th Cir. 2020) (VanDyke, J., dubitante) (writing separately “because [our] precedent is silly and well illustrates our court’s nasty habit of muddying immigration law”). While we note the dissent’s critique, it fails to engage with our analysis of the issues in this case. Our task is to apply existing precedent—whether or not we agree with it—as we faithfully do here. FLORES MOLINA V. GARLAND 15 The BIA determined that Flores Molina did not experience past persecution and concluded that his fear of future persecution was “too speculative” to be well-founded. 1. Past Persecution We hold that the BIA’s determination that Flores Molina did not suffer past persecution in Nicaragua is not supported by substantial evidence. 2 To show past persecution, Flores Molina “has the burden of establishing that (1) his treatment rises to the level of persecution; (2) the persecution was on account of one or more protected grounds; and (3) the persecution was committed by the government, or by forces that the government was unable or unwilling to control.” Baghdasaryan v. Holder, 592 F.3d 1018, 1023 (9th Cir. 2010) (citation omitted). Here, the BIA ended its analysis at the first element, determining that Flores Molina’s past experiences did not rise to the level of persecution. Applying our court’s binding caselaw to the record evidence, — 2 — We have previously stated that “[w]hether particular acts constitute persecution for asylum purposes is a legal question reviewed de novo.” Kaur v. Wilkinson, 986 F.3d 1216, 1221 (9th Cir. 2021) (alterations adopted) (citation omitted). We have also stated that we “review for substantial evidence the BIA’s particular determination that a petitioner’s past harm ‘does not amount to past persecution.’” Sharma v. Garland, 9 F.4th 1052 , 1060 (9th Cir. 2021) (alteration adopted) (citation omitted). We need not address whether de novo review should apply, or discuss the nuances of the two standards, because the harm Flores Molina suffered rose to the level of persecution under the more deferential “substantial evidence” standard of review. See Fon v. Garland, — F.4th —, 2022 WL 1562281 at *4 n.1 (9th Cir. May 18, 2022) (applying the “substantial evidence” standard of review and declining to address whether de novo review applies because the outcome was the same under any standard). See generally id. at *7–8 (Graber, J., concurring) (discussing the proper standard of review as applied to whether acts rise to the level of persecution); id. at *9–12 (Collins, J., concurring) (same). 16 FLORES MOLINA V. GARLAND there are three bases that compel our conclusion that Flores Molina’s past experiences constitute persecution. a. First, “[a]s we have consistently recognized, being forced to flee from one’s home in the face of an immediate threat of severe physical violence or death is squarely encompassed within the rubric of persecution, as long as the persecutors’ actions are motivated” by a protected ground. Mendoza-Pablo, 667 F.3d at 1314 (citation omitted) (finding past persecution because petitioner “fled from her home village as a result of her (eminently-reasonable) belief that her life . . . was in severe and immediate danger because Guatemalan military forces had specifically targeted the village’s inhabitants on the basis of their racial and ethnic background”); see also Knezevic v. Ashcroft, 367 F.3d 1206 , 1211–12 (9th Cir. 2004) (finding past persecution where ethnically Serbian petitioners fled their hometown to escape hostile Croatian forces because they “realized the threat of harm—and possibly death—was imminent”). Here, Flores Molina was forced to flee three separate times after being personally targeted for his political views with violence and threatened with death. 3 He was forced to — 3 — The dissent disregards our holdings in Mendoza-Pablo and Knezevic by arguing that those petitioners only suffered persecution because their homes were “completely destroyed.” Dissent 49. To be sure, various homes were burned after the petitioner’s mother fled in Mendoza-Pablo. 667 F.3d at 1311, 1314 . In Knezevic, the petitioners’ home was also destroyed after they fled and partially restored later. 367 F.3d at 1212 . While their homes were subsequently destroyed, that was not the lynchpin of our holdings in those cases. Rather, we held that the petitioners were persecuted because, like Flores Molina, they fled “in the face of an immediate threat of severe physical violence or death.” Mendoza-Pablo, 667 F.3d at 1314 . FLORES MOLINA V. GARLAND 17 flee, first, after a series of escalating threats culminated in a death threat painted on his home, second, after being discovered at his hideaway and narrowly evading detection by a group of armed paramilitary members, and third, after being assaulted and threatened with death by a gang of Sandinista Youth. The “severe and immediate danger” that he faced arose within the broader context of mass killings and violent reprisals by the Ortega regime and its affiliates against protestors like him, including the killing of his friend Valdivia that he witnessed firsthand. The repeated incidents in which Flores Molina fled were each “in the face of an immediate threat of severe physical violence or death,” and thus rise to the level of persecution. Mendoza-Pablo, 667 F.3d at 1314 . 4 b. Second, we have “repeatedly held that threats may be compelling evidence of past persecution, particularly when they are specific and menacing and are accompanied by evidence of violent confrontations, near-confrontations and vandalism.” Mashiri v. Ashcroft, 383 F.3d 1112, 1119 (9th Cir. 2004) (holding that a single written death threat was “strong evidence of persecution” particularly in light of the broader violence and escalating harm); see also Ruano v. Ashcroft, 301 F.3d 1155 , 1160–61 (9th Cir. 2002) (finding past persecution where petitioner experienced multiple death threats, “near face-to-face confrontations” with armed persecutors, and persecutors directly confronted his family); Baballah v. Ashcroft, 367 F.3d 1067, 1074 (9th Cir. 2004) (“Threats and attacks can constitute persecution even where an applicant has not been beaten or physically harmed.” — 4 — The BIA did not address the fact that Flores Molina was forced to repeatedly flee in the wake of violence and death threats. 18 FLORES MOLINA V. GARLAND (citations omitted)); Del Carmen Molina v. INS, 170 F.3d 1247, 1249 (9th Cir. 1999) (finding past persecution where the petitioner’s family had been killed by guerrilla forces and the petitioner received two threatening notes related to their killing). And we have “consistently held that death threats alone can constitute persecution.” Navas v. INS, 217 F.3d 646, 658 (9th Cir. 2000) (emphasis added) (collecting cases); see also Kaur v. Wilkinson, 986 F.3d 1216, 1227 (9th Cir. 2021) (reiterating that death threats “alone” can constitute persecution “because murder is perhaps the ultimate threat to bodily integrity” (citing Lim, 224 F.3d at 936)). Flores Molina was publicly marked as a terrorist and threatened with torture over social media by government operatives, repeatedly verbally threatened with death by supporters of the Ortega regime, received a death threat painted on his home by masked men likely affiliated with the government, and received a second death threat—this time during a direct confrontation—after he was seriously beaten by six members of the Sandinista Youth. In addition, Flores Molina had a near confrontation with an armed paramilitary group that located him at a hideaway. The threats were credible given the history and context of the Ortega regime’s killing and torture of its political opponents. Indeed, Flores Molina witnessed the killing of his friend and fellow protester when his friend was shot in the head at a demonstration. See Salazar-Paucar v. INS, 281 F.3d 1069, 1075 (9th Cir.), as amended, 290 F.3d 964 (9th Cir. 2002) (holding that “[e]vidence of harm to individuals who held the same political positions” to the petitioner supported a finding of past persecution). Such “[r]epeated death threats, especially when those threats occurred in conjunction with other forms of abuse, require a finding of past persecution.” Smolniakova v. Gonzales, 422 F.3d 1037, 1049 (9th Cir. 2005); see also Aden v. Wilkinson, 989 F.3d 1073, 1082 (9th FLORES MOLINA V. GARLAND 19 Cir. 2021) (“[W]hen the incidents have involved physical harm plus something more, such as credible death threats, we have not hesitated to conclude that the petitioner suffered persecution.” (emphasis in original)). The BIA cited two cases to support its determination that the threats Flores Molina experienced did not constitute persecution. The BIA’s reliance on those cases, however, is misplaced. The BIA cited Lim v. INS, 224 F.3d 929, 936 (9th Cir. 2000) for its conclusion that “these threats are not the sort of ‘extreme’ or ‘especially menacing’ threats necessary to establish past persecution.” But the Lim court declined to find past persecution precisely because “[n]either Lim nor his family was ever touched, robbed, imprisoned, forcibly recruited, detained, interrogated, trespassed upon, or even closely confronted.” Id. Flores Molina, by contrast, was publicly singled out on multiple occasions, his home was trespassed upon, and he was closely confronted and beaten, at the behest of a coordinated campaign by a brutal regime that credibly acted on its threats against other similarly situated political opponents. This is not a situation as the one the Lim court described where “[t]hreats themselves are sometimes hollow.” Id. Lim is not on-point. See also Ruano, 301 F.3d at 1160 (distinguishing Lim because Ruano had been “closely confronted” and pursued “by men he knew were armed”). Next, the BIA cited Gu v. Gonzales, 454 F.3d 1014, 1020 (9th Cir. 2006) for the proposition that the physical harm Flores Molina suffered at the hands of six Sandinista Youth assailants was insufficient to constitute past persecution. The BIA referred to Gu as holding that “a single instance of detention and beating, which resulted in non-serious injuries” was insufficient to establish past persecution, but ignored the surrounding circumstances that are relevant to 20 FLORES MOLINA V. GARLAND Flores Molina’s claim. Flores Molina does not contend that he experienced past persecution on the sole basis of his beating by a group of Sandinista Youth, but rather based on the sequence of escalating threats and the broader context of violence targeted at him and other political dissidents like him. The BIA’s citation to Gu does not support its analysis and does not relieve it of its obligation to consider the “totality of the circumstances” in deciding whether past persecution is shown. Guo v. Ashcroft, 361 F.3d 1194, 1203 (9th Cir. 2004); see also Aden, 989 F.3d at 1082–84 (distinguishing Gu, 545 F.3d at 1020, where Aden was physically assaulted only once but also received repeated threats within the context of “political and social turmoil” in Somalia). Further, “it is the conduct of the persecutor” that is relevant to evaluating whether past treatment rises to the level of persecution—not “the level of harm” or “subjective suffering” the petitioner experienced. Kaur, 986 F.3d at 1226 (citations omitted). Accordingly, the severity of Flores Molina’s injuries is not dispositive to whether the threats and violence constituted persecution. Id. Any reasonable adjudicator would be compelled to hold that the repeated and specific death threats that Flores Molina experienced, amid the violence and menacing confrontations to which he was subjected, amount to persecution. c. The third basis for our holding is that “[a]n applicant may suffer persecution because of the cumulative effect of several incidents,” even if no single incident rises to the level of persecution. Chand v. INS, 222 F.3d 1066, 1074 (9th Cir. 2000). “[T]he key question is whether, looking at the cumulative effect of all the incidents that [Flores Molina] suffered, the treatment he received rises to the level of FLORES MOLINA V. GARLAND 21 persecution.” Sharma v. Garland, 9 F.4th 1052 , 1061 (9th Cir. 2021) (citation omitted). Our inquiry is “a fact-bound endeavor that is not reducible to a set formula,” but rather requires that the relevant facts “be evaluated in combination with each other to form a sufficiently negative portrait of the petitioner’s experience in his or her own country that not only allows a finding of past persecution but requires it.” Id. (citations omitted). Flores Molina was repeatedly threatened, in an escalating fashion—from threats and public blacklisting on social media by government operatives, to a spray-painted death threat on the walls of his home forcing him to flee, to armed paramilitary members searching for him at his first hideaway and forcing him to flee again, to a gang of Sandinista Youth beating and threatening to kill him near his second hideaway location. See Cordon-Garcia v. INS, 204 F.3d 985, 991 (9th Cir. 2000) (“The determination that actions rise to the level of persecution is very fact-dependent, though threats of violence and death are enough.” (citations omitted)). The progression of threats and violence that Flores Molina experienced was set against the welldocumented backdrop of the Ortega regime’s violent crackdown on members of the political opposition. Where “evidence of a specific threat on [a petitioner’s] life, and here there were many, is presented in conjunction with evidence of political and social turmoil, the [petitioner] has succeeded in establishing a prima facie eligibility for asylum.” Korablina v. INS, 158 F.3d 1038, 1045 (9th Cir. 1998); see also Aden, 989 F.3d at 1083–84 (concluding that the evidence of political and social turmoil, along with the persecutors’ physical assault and continued pursuit of the applicant, compelled a finding a past persecution). 22 FLORES MOLINA V. GARLAND The two citations the BIA provided as examples of cases where the “cumulative effect of several incidents” was insufficient bear no resemblance to the threats and violence that Flores Molina experienced. In Wakkary v. Holder, we held that the petitioner’s experiences of “being beaten by youths and robbed of his sandals and pocket money in 1985 and 1990 (seventeen and twelve years, respectively, before he filed his asylum application), and being accosted by a threatening mob while his family was driving to Bible school in 1998,” did not “cumulatively amount to past persecution.” 558 F.3d 1049 , 1059–60 (9th Cir. 2009). And in Hoxha v. Ashcroft, we relied on Lim, 224 F.3d at 936 , to hold that the nonspecific and “unfulfilled” threats by various individuals with no connection to the government constituted “harassment rather than persecution.” 319 F.3d 1179, 1182 (9th Cir. 2003). Although we acknowledged that physical violence “ordinarily” establishes persecution, the one instance of violence the petitioner faced “was not connected with any particular threat,” there was “no evidence indicating that the incident was officially sponsored,” and there was “no evidence that the attackers knew who [the petitioner] was or that they showed any continuing interest in him.” Id. at 1182 , 1182 n.5. Neither of these factual scenarios are close to the sustained, repeated, specific, government-aligned, and politically motivated threats and violence to which Flores Molina was subjected. In sum, applying our caselaw to the record evidence, there are three bases that would compel any reasonable adjudicator to find that Flores Molina’s past experiences “rose to the level of persecution.” Baghdasaryan, 592 F.3d at 1023 . On remand, the BIA must address the remaining FLORES MOLINA V. GARLAND 23 elements of past persecution. 5 If Flores Molina establishes the remaining elements, “a rebuttable presumption of a wellfounded fear arises, 8 C.F.R. § 208.13 (b)(1), and the burden [then] shifts to the government to demonstrate that there has been a fundamental change in circumstances such that the applicant no longer has a well-founded fear.” Tawadrus v. Ashcroft, 364 F.3d 1099, 1103 (9th Cir. 2004) (citation and quotation marks omitted). 2. Fear of Future Persecution Because Flores Molina also challenges the BIA’s determination that he failed to demonstrate a well-founded fear of future persecution, we proceed to address that issue. The BIA was “required to evaluate all relevant evidence in the record” to determine whether Flores Molina carried his burden. Davila v. Barr, 968 F.3d 1136, 1143 (9th Cir. 2020) (citation omitted). In rejecting Flores Molina’s claim that he has a well-founded fear of future persecution, the BIA erred by failing to address highly probative evidence. The BIA cited the record selectively to support its assertion that Flores Molina’s fears of future persecution were speculative. See id. (concluding that “the BIA’s extreme selectivity in using the Country Report evidence belie[d] any attempt” to evaluate all relevant evidence). The BIA cited two May 2019 news reports about the Ortega — 5 — The BIA must determine whether the persecution Flores Molina experienced was on account of one or more protected grounds, and whether the persecution was committed by the government, or by forces that the government was unable or unwilling to control. Baghdasaryan, 592 F.3d at 1023 . The record evidence strongly suggests that Flores Molina was persecuted on account of his political opinion by individuals affiliated with Nicaragua’s ruling Ortega regime. However, that is a determination the agency must make in the first instance. 24 FLORES MOLINA V. GARLAND regime’s release of 100 prisoners and announcement of the regime’s intentions to release more. But the BIA ignored news reports that documented the conditions released prisoners faced, delays in releasing political prisoners, and that the Ortega regime detained and disappeared additional activists and protesters in the interim. The BIA also ignored the 2018 State Department Report, Amnesty International’s 2018 report on political persecution within Nicaragua, and the Inter-American Commission of Human Rights’ 2018 report on gross human rights violations in Nicaragua. Moreover, the BIA failed to discuss whether the repeated death threats and threats of violence Flores Molina faced were sufficient to “inspire a well-founded fear of future persecution.” Lim, 224 F.3d at 937 (emphasis deleted); see also id. at 936–37 (discussing cases where threats that were insufficient to establish past persecution were, nonetheless, sufficient to demonstrate a well-founded fear of future persecution); Kaiser v. Ashcroft, 390 F.3d 653, 658 (9th Cir. 2004) (“Threats on one’s life, within a context of political and social turmoil or violence, have long been held sufficient to satisfy a petitioner’s burden of showing an objective basis for fear of persecution.” (citation omitted)). All this record evidence contextualizes the Nicaraguan government’s actions and Flores Molina’s specific fears of persecution upon removal. 6 Where the BIA fails to consider highly — 6 — The BIA’s assertion that Flores Molina is not sufficiently at risk of future persecution because the “number of detained activists” in Nicaragua “is relatively small when considering that upwards of a million people participated in the aforementioned protests,” makes no sense. Assessing the threat of political persecution to Flores Molina depends on his personal circumstances, including the threats, violence and targeting he experienced prior to fleeing Nicaragua. The BIA’s calculation of the percentage of activists who have been detained as a portion of those who protested is meaningless for the purpose of assessing whether Flores Molina’s fear of future persecution is wellFLORES MOLINA V. GARLAND 25 probative record evidence, its “decision cannot stand.” Cole, 659 F.3d at 771–72. 3. Humanitarian Asylum The BIA denied Flores Molina humanitarian asylum, pursuant to 8 C.F.R. § 1208.13 (b)(1)(iii), because it determined he failed to make “a showing of past persecution.” Because the BIA erred in its finding that Flores Molina failed to show he suffered past persecution, we vacate the denial of humanitarian asylum and remand for further consideration. Flores Molina further contends that we should remand his humanitarian asylum claim to the IJ to fully develop the record related to the elements of that relief. Because the BIA failed to analyze the merits of Flores Molina’s humanitarian asylum claim, however, we remand to the BIA to assess the merits of this claim in the first instance (which the BIA may remand to the IJ in its discretion). B. Withholding of Removal The BIA’s sole basis for denying Flores Molina’s claim for withholding of removal was because it determined he had “not established eligibility for asylum.” Because the BIA erred in its denial of asylum, we remand Flores Molina’s withholding of removal claim for further consideration. founded, in light of the evidence that Flores Molina himself had been targeted. The BIA’s unreasonably deficient analysis on this point constitutes reversible error on its own. Tadevosyan v. Holder, 743 F.3d 1250 , 1252–53 (9th Cir. 2014) (“The BIA abuses its discretion when it acts arbitrarily, irrationally, or contrary to the law, and when it fails to provide a reasoned explanation for its actions.” (internal quotation marks omitted)). 26 FLORES MOLINA V. GARLAND Moreover, if the BIA determines that Flores Molina experienced past persecution on account of a protected ground, the BIA must credit Flores Molina with a rebuttable presumption of eligibility for withholding of removal. 8 C.F.R § 1208.16(b)(l)(i); Ahmed v. Keisler, 504 F.3d 1183, 1199 (9th Cir. 2007). Finally, even if the BIA determines that Flores Molina is not entitled to a presumption of eligibility for withholding of removal, it must consider all probative evidence related to Flores Molina’s fear of future persecution. See supra Discussion § I(B). C. Convention Against Torture To qualify for CAT relief, an applicant must establish that it is “more likely than not” that he would be tortured if removed to the proposed country of removal. See 8 C.F.R. § 1208.16 (c)(2). In assessing whether it is more likely than not an individual would be tortured, “all evidence relevant to the possibility of future torture shall be considered.” § 1208.16(c)(3); Maldonado v. Lynch, 786 F.3d 1155, 1162 (9th Cir. 2015) (en banc). Just as the BIA legally erred in rejecting Flores Molina’s well-founded fear of future persecution by failing to examine relevant evidence, it repeated the error in rejecting his CAT claim. See supra Discussion § I(B). The BIA failed to mention highly probative country conditions reports which reflect the Nicaraguan government’s continued jailing and mistreatment of political protesters like Flores Molina, including immediate arrests of activists at the airport as soon as they returned to the country. See Aguilar-Ramos v. Holder, 594 F.3d 701, 705 (9th Cir. 2010) (“The failure of the IJ and BIA to consider evidence of country conditions FLORES MOLINA V. GARLAND 27 [when assessing a CAT claim] constitutes reversible error.” (citations omitted)). 7 Of particular relevance to Flores Molina’s CAT claim, the 2018 State Department Report states that cases of torture are well documented and “public officials intentionally carried out acts that resulted in severe physical or mental suffering for the purposes of securing information, inflicting punishment, and psychologically deterring other citizens from reporting on the government’s actions or participating in civic actions against the government.” The report furthers states that “[m]embers of civil society and student leaders involved in the protests that began [in April 2018] were more likely than members of other groups to be subjected to such treatment.” Flores Molina is a member of the opposition Liberal Party, participated in the April 2018 protests, and his stance against the Ortega regime was repeatedly publicized online by government operatives. The BIA’s failure to consider relevant evidence of country conditions in connection with Flores Molina’s CAT claim is reversible error. Id. We therefore grant Flores Molina’s petition for review as to his CAT claim and remand for further consideration. D. Motion to Reopen Flores Molina’s second petition (No. 20-71774) seeks review of the BIA’s denial of his motion to reopen proceedings to seek a continuance while the United States Citizenship and Immigration Service adjudicates his — 7 — The BIA’s gesture to “the totality of the record,” without mentioning or discussing these country reports, does not insulate the BIA from reversal. “[W]here there is any indication that the BIA did not consider all of the evidence before it, a catchall phrase does not suffice, and the decision cannot stand.” Cole, 659 F.3d at 771–72. 28 FLORES MOLINA V. GARLAND application for an immigrant visa. Because we grant Flores Molina’s first petition (No. 19-73028), upon remand to the BIA for further proceedings, Flores Molina will no longer be subject to a final order of removal. Upon remand, Flores Molina may request that the BIA remand his case to the IJ so that he may pursue his application for an immigrant visa. Accordingly, we dismiss his second petition as moot. * * * In sum, we grant petition No. 19-73028 and hold that (1) the record evidence compels the conclusion that Flores Molina’s past experiences rise to the level of persecution, (2) the BIA legally erred by failing to consider highly probative evidence of future harm that Flores Molina may suffer in its analysis of his asylum, withholding of removal and CAT claims, and, thus, (3) the BIA’s denials of asylum, humanitarian asylum, withholding of removal, and CAT relief are remanded for further consideration. We dismiss as moot petition No. 20-71774. Petition 19-73028 GRANTED and REMANDED. Petition 20-71774 DISMISSED as moot. KORMAN, District Judge, concurring: I concur in the majority opinion. I write separately to address an issue raised by Judge VanDyke’s dissent and left unresolved in the majority opinion—specifically, the standard of review that applies to the BIA’s past persecution decision. The dissent attacks the manner in which this circuit and the majority opinion apply the substantial evidence standard provided by the Immigration and Nationality Act FLORES MOLINA V. GARLAND 29 (“INA”), which requires that “the administrative findings of fact are conclusive unless any reasonable adjudicator would be compelled to conclude to the contrary.” 8 U.S.C. § 1252 (b)(4)(B). The majority opinion opens the door to such a critique because it vacates the BIA’s past persecution finding based on reasoning that “the BIA’s determination that Flores Molina did not suffer past persecution in Nicaragua is not supported by substantial evidence.” Majority Opinion at 15. Yet there is no need to apply the substantial evidence standard to our review of the “BIA[’s] . . . analysis . . . determining that Flores Molina’s past experiences did not rise to the level of persecution.” Id. at 15. In this case, the BIA did not base its past persecution finding on a rejection of the veracity of Flores Molina’s description of his past experiences. Rather, the BIA held that Flores Molina’s “claimed past harm, cumulatively considered, does not rise to the level of past persecution.” (Emphasis added). 1 Although we owe deference under the substantial evidence standard to “the administrative findings of fact,” 8 U.S.C. § 1252 (b)(4)(B), “[w]hether particular acts constitute persecution for asylum purposes is a legal question reviewed de novo,” Kaur v. Wilkinson, 986 F.3d 1216, 1221 (9th Cir. 2021) (alterations adopted) (quoting Boer-Sedano v. Gonzales, 418 F.3d 1082, 1088 (9th Cir. 2005)); see also Pitcherskaia v. INS, 118 F.3d 641, 646 (9th — 1 — Thus, contrary to Judge VanDyke’s suggestion, this is not a case in which “we don’t know if the agency discounted Molina’s factual account of his past experiences” in denying his claims. Dissent at 41 n.3. Whether or not “the agency [will] f[i]nd [that] version of facts fully persuasive” on remand remains to be seen, id., but it strains credulity to suggest that it based the decision before us on a different set of facts than the one put forward by Flores Molina. Not to mention that such a “path may [not] reasonably be discerned” from the BIA’s opinion. Garland v. Ming Dai, 141 S. Ct. 1669 , 1679 (2021). 30 FLORES MOLINA V. GARLAND Cir. 1997) (“The meaning of ‘persecution’ . . . is a legal question reviewed de novo.”); Mendoza-Alvarez v. Holder, 714 F.3d 1161, 1163 (9th Cir. 2013) (similar). Nevertheless, there are cases that have applied substantial evidence review to such questions without substantial discussion or analysis of the issue. See, e.g., Sharma v. Garland, 9 F.4th 1052 , 1060 (9th Cir. 2021). But the substantial evidence standard is not a good fit for questions, like the one presented in this case, regarding the application of a legal standard to settled facts. No case explains how a reviewing court can ascertain whether or how a reasonable adjudicator could make the determination prescribed by 8 U.S.C. § 1252 (b)(4)(B) without making a separate determination whether the law was correctly applied by the agency to the facts it found. On its own terms, moreover, the substantial evidence standard applies only to the agency’s “findings of fact,” 8 U.S.C. § 1252 (b)(4)(B), not its application of a legal standard to those facts. Indeed, the Supreme Court held in an analogous context that “the statutory phrase ‘questions of law’” in 8 U.S.C. § 1252 (a)(2)(D), “includes the application of a legal standard to undisputed or established facts.” GuerreroLasprilla v. Barr, 140 S. Ct. 1062, 1068 (2020). See also Fon v. Garland, — F.4th —, No. 20-73166, 2022 WL 1562281 , at *9–12 (9th Cir. May 18, 2022) (Collins, J., concurring) (providing additional reasons that support applying de novo review to “the question whether [a] [p]etitioner’s harms rose to the level of persecution”); id. at *7 (Graber, J., concurring) (acknowledging that de novo review applies to such questions where “answering [it] entails primarily legal . . . work”). While I agree with the majority opinion that our decision in this case would be the same regardless of which standard FLORES MOLINA V. GARLAND 31 applies, Judge VanDyke appears to take a different view. He contends that the substantial evidence standard requires that we uphold the BIA’s past persecution decision but acknowledges that he may have reached a different conclusion under a de novo standard of review. See Dissent at 32 (“If I were the BIA-for-a-day . . . I may have been persuaded that Molina’s hardships amounted to persecution.”). I agree that, deferring to the BIA’s construction of the facts, its decision that Flores Molina’s hardships did not amount to persecution was legally erroneous. And I would have also concurred in a majority opinion that would have decided this case on that ground. VANDYKE, Circuit Judge, dissenting: We say it often: the review of agency immigration decisions must be done with an extraordinarily high level of deference. This deference demands more than mere lip service to the standard of review and allows us to reverse agency decisions in only the narrowest of circumstances—if the record compels a different conclusion. Wang v. Sessions, 861 F.3d 1003, 1007 (9th Cir. 2017). To properly and faithfully apply this extreme deference, we must search the record as a whole for evidence that supports the agency’s decision—and in doing so, resist the temptation to reweigh evidence or rigidly quantify the qualitative inquiries that the agency alone is tasked with answering. Our immigration decisions should recognize (as the Supreme Court recently reaffirmed in Dai) the wide discretion that Congress has afforded the agency in weighing how sufficient, credible, or persuasive it finds each part of the evidence and a petitioner’s testimony. See Garland v. Dai, 141 S. Ct. 1669 , 1680 (2021). Here, because the agency’s decision falls 32 FLORES MOLINA V. GARLAND within the wide parameters of its discretion and the record does not compel a different conclusion, we have no authorization to remand to reach a different outcome (even one that may be more palatable or equally reasonable). This is the type of case where the agency certainly could have granted relief from removal. If I were the BIA-for-aday, as our court so often likes to play, I may have been persuaded that Molina’s hardships amounted to persecution and found him eligible for asylum. 1 But the close cases are where our deferential standard of review is most important and where, unfortunately, this court too often disregards it in favor of a preferred result. Once again, our court asks and answers the wrong question—not whether the record or parts therein could permit a different result—but whether the record as a whole compels a different result. Reviewing the agency’s decision in light of its wide discretion and the extraordinary amount of deference we owe, I cannot conclude the record here compels a different conclusion. So I must respectfully dissent. I. DISCUSSION Before turning to the specifics of this case, it’s important to review some fundamental features of immigration law and our limited role as a reviewing court. To fully appreciate just how far our court has strayed from the narrow role Congress prescribed for us, we must first look back—far — 1 — Judge Korman in his concurrence purports to “agree” with me that, if we (inappropriately) applied a de novo standard of review, the BIA’s “decision that Flores Molina’s hardships did not amount to persecution was legally erroneous.” I nowhere say that. To be clear, and consistent with my intentional use of the word “may,” I remain studiously agnostic because I can (and should) under the appropriately deferential standard of review. FLORES MOLINA V. GARLAND 33 back—from our court’s wayward precedents to the deferential standard of review we are supposed to apply. After limning the various factors affecting the extraordinarily deferential review that our court should be giving to agency immigration decisions, I then relate how we have instead built a remarkably unbalanced edifice of circuit precedents that enable and mask our frequent unwillingness to properly defer to the agency—just like in this case. A. The Extreme Deference Required in Immigration Cases In our review of immigration decisions, three interconnected layers of deference combine to form what should be one of the most deferential standards of review in our legal system. First, the scope of our review is tightly circumscribed by the extraordinarily deferential standard that Congress has commanded—yielding to the agency’s determinations unless a different conclusion is compelled. Second, we apply that extreme deference to the extraordinarily difficult and often indeterminate factual inquiries that the agency alone is charged with making. And third, as the Supreme Court recently reiterated in Dai, the agency enjoys extraordinary discretion in making the difficult determinations of how much credibility, weight, and persuasiveness to afford different parts of the record in reaching its factual conclusions. 1. Extraordinary Judicial Deference In the Immigration and Nationality Act (INA), Congress codified the highly deferential substantial evidence test and established what should be our court’s guiding star in the review of immigration decisions: that “administrative findings of fact are conclusive unless any reasonable adjudicator would be compelled to conclude to the contrary.” 34 FLORES MOLINA V. GARLAND INA § 242(b)(4)(B) (codified as 8 U.S.C. § 1252 (b)(4)(B) (emphasis added)). Congress later amended the INA by passing the REAL ID Act, further reining in our role and discretion as a reviewing court and stripping federal courts of jurisdiction to hear certain immigration claims. See Nasrallah v. Barr, 140 S. Ct. 1683, 1698 (2020) (Thomas, J., dissenting). Over time, however, this court’s decisions have chipped away at these statutory standards—broadening the scope and standard of our review far beyond the limited and deferential posture that Congress unmistakably set out in the INA. See id. To properly apply our deferential standard of review, we are supposed to scour the record to answer a single question: could any reasonable adjudicator have agreed with the agency’s result, or does the record as a whole compel a different conclusion? See INS v. Elias-Zacarias, 502 U.S. 478, 481 (1992) (explaining that substantial evidence review requires that we review “the record considered as a whole” and reverse the agency only if no reasonable factfinder could agree with its conclusion); see also Prasad v. INS, 47 F.3d 336, 339 (9th Cir. 1995) (describing Elias-Zacarias as “the touchstone” and “definitive statement of ‘substantial evidence’ in the context of . . . factual determinations in asylum cases”). On its face, this is an exceptionally deferential standard of review. But there’s more. 2. Inherently Indeterminate Inquiries Our deferential standard of review becomes even more significant in light of what our court is tasked with reviewing: agency answers to inherently imprecise and difficult factual inquiries. The factual determinations involved in immigration law are innately indeterminate— difficult to answer quantitatively or reduce to precise legal categorization—which underscores the importance of our FLORES MOLINA V. GARLAND 35 deferential posture and limited role in reviewing agency decisions. I illustrate this with just one example, but others abound. The inquiries at issue in this case—whether Molina’s past hardships constitute “persecution” under the INA or whether his fear of future persecution is “well-founded”—are anything but self-evident and could be answered with a range of reasonable views (any one of which we must defer to if selected by the agency). There is no one objective answer to the questions of whether a petitioner has suffered real past “persecution,” or if his fear of future persecution is “well-founded.” Congress established the relief Molina seeks with the Refugee Act of 1980, which gives the Attorney General discretionary authority to grant asylum to an alien who is unable or unwilling to return to his home country “because of persecution or a well-founded fear of persecution on account of race, religion, nationality, membership in a particular social group, or political opinion.” 8 U.S.C. § 1101 (a)(42). Unsurprisingly, courts have struggled to articulate what constitutes persecution or a well-founded fear of it, as they are inherently obscurant inquiries left largely undefined by the INA or its implementing regulations. See, e.g., INS v. Cardoza-Fonseca, 480 U.S. 421, 448 (1987) (recognizing “[t]here is obviously some ambiguity in a term like ‘well-founded fear’” left undefined by the Act); Mikhailevitch v. INS, 146 F.3d 384, 389 (6th Cir. 1998) (observing that “[t]he Act provides no definition of ‘persecution’”). Despite this inherent ambiguity—or perhaps more accurately, because of this inherent ambiguity (in order to portray our review of these difficult questions as more objective or unbiased)—our court has latched onto the 36 FLORES MOLINA V. GARLAND concept that a 10% possibility of future persecution constitutes a “well-founded fear” under the statute (when accompanied by a genuine, subjective fear of future persecution). See, e.g., Al-Harbi v. INS, 242 F.3d 882, 888 (9th Cir. 2001) (“[E]ven a ten percent chance of persecution may establish a well-founded fear.”) (citing CardozaFonseca, 480 U.S. at 431 ). 2 But there are several problems with this artificial threshold. The first is that slapping a percentage threshold on an inherently indeterminate question does not somehow magically make it more determinate, because deciding whether or not someone has a higher than 10% possibility of persecution still remains an entirely qualitative inquiry. By pretending to resolve these questions with numerical exactitude, we are merely masking the nonquantifiable nature of the inquiry. It is a barely disguised rhetorical ploy, and nothing more. Even putting that aside, our 10% test is itself a misreading of Supreme Court precedent. It doesn’t come from the statutory text or any implementing regulation, but from a theoretical scenario from a law journal that the Supreme Court cited in Cardoza Fonseca. It was an academic thought-experiment for considering when a petitioner may have a well-founded fear of future persecution, even when the mathematical possibility of persecution was less than 50%. See Cardoza-Fonseca, 480 U.S. at 448 . — 2 — Because the majority concludes that Molina’s past hardships amount to past persecution, it relies primarily on the rebuttable presumption that arises in favor of a well-founded fear of future persecution without directly examining whether Molina satisfied this 10% threshold. But the example illustrates my point nonetheless. FLORES MOLINA V. GARLAND 37 The Supreme Court found that for a fear to be “wellfounded” in support of asylum eligibility there had to be a “reasonable possibility” of future persecution—which it decided (after looking at the statute’s structure, its legislative history, and international law) was something less than the more-likely-than-not standard used in withholding proceedings. Cardoza-Fonseca, 480 U.S. at 427–49. The Court stopped short of filling the statutory gap and declined to dictate a specific percentage of persecution-possibility that would be reasonable and therefore sufficient to establish a well-founded fear—instead leaving it for the agency to fill on a case-by-case basis. See id. at 448 . But the Court did offer an example, a hypothetical plucked from a law journal article (published years before Congress passed the statutory language at issue), to illustrate that, if a petitioner’s country of origin was executing one out of every ten adult males, the petitioner could still show a well-founded fear of future persecution even though the mathematical possibility of persecution did not meet the greater than 50% standard used in withholding proceedings. Cardoza-Fonseca, 480 U.S. at 431 (quoting 1A Grahl-Madsen, The Status of Refugees in International Law). Latching onto the Court’s academic musings in Cardoza-Fonseca, our court wasted no time stepping over the agency to fill in the statutory gap with a new categorical rule—just as Justice Scalia had feared— thereby (again) expanding both asylum eligibility and our role beyond that authorized by Congress. See id. at 453 (Scalia, J., concurring); Blanco-Comarribas v. INS, 830 F.2d 1039, 1042 (9th Cir. 1987) (citing Cardoza-Fonesca for the 10% test); Al-Harbi v. INS, 242 F.3d 882, 888 (9th Cir. 2001) (same). Our reliance on this fictitious rule is problematic first and foremost because it finds no basis or support in the statute we are supposedly applying—and therefore expands the 38 FLORES MOLINA V. GARLAND class of petitioners eligible for asylum beyond what Congress authorized. The 10% standard is also problematic because it essentially lowers the bar that petitioners must surpass to establish eligibility for asylum—which, in turn, effectively widens this court’s discretion. Today, under our errant rule, we no longer ask whether the agency has considered what the statute commands (i.e., if the petitioner’s fear of persecution is well-founded), or what the Supreme Court actually ruled in Cardoza-Fonseca (i.e., if the petitioner faces a “reasonable possibility” of persecution). 8 U.S.C. § 1101 (a)(42)(A) (emphasis added); Cardoza-Fonseca, 480 U.S. at 440 (emphasis added). Instead, we ask if there is even a marginal, 10% chance that the petitioner may face persecution. Our reliance on these judge-made rules that purport to quantify the unquantifiable questions in immigration law is particularly problematic because it obfuscates what is actually happening: the substitution of our own discretion in place of the marked deference we owe the agency. Immigration cases are both complicated and indeterminate. But our court cannot resist the urge to gloss over these complexities with a quantitative sheen that gives the appearance of a mere technical application of the law when, in reality, this faux quantification only aggrandizes our discretion at the expense of the agency’s. We cannot and should not ignore the fundamental complexities of immigration inquiries—which produce a far wider range of “reasonable” views to which we owe more deference than our precedents suggest. But that’s still not all. 3. Extraordinary Agency Discretion Lastly, it is not only how we are to review agency decisions (with a very limited scope and deferential posture) or what we are reviewing (answers to inherently FLORES MOLINA V. GARLAND 39 indeterminate questions) that undergird the super-deference Congress has commanded we apply. The Supreme Court also recently, and unanimously, reemphasized that Immigration Judges and the BIA enjoy wide discretion in how they arrive at, and articulate, the conclusion and supporting grounds that we review. See Dai, 141 S. Ct. at 1677–81. Before being reversed in Dai, our court followed a selfmade rule that severely limited the agency’s discretion in answering the indeterminate inquiries presented in immigration applications. See id. Our court, in its “deemedtrue-or-credible-rule,” decided that any agency decision issued without an explicit adverse credibility determination must mean that the agency not only found the petitioner’s testimony 100% credible but also gave it 100% weight and found it 100% persuasive. In other words, whatever the agency decided on credibility, so too (we assumed) went the determinations on weight and persuasiveness. This presumption essentially narrowed the agency’s wide discretion to two options: either find the petitioner explicitly noncredible and reject 100% of his testimony, or our court would assume the agency accepted 100% of the petitioner’s testimony as true. See id. In Dai, the Supreme Court stated the obvious: that our judge-made rule was wholly irreconcilable with the INA and had “no proper place in a reviewing court’s analysis.” Id. at 1677. Dai emphasized that in answering the indeterminate and ambiguous inquiries described above, the agency does so with an extraordinary amount of discretion as to how credible, weighty, and persuasive it finds different parts of the record. Id. at 1680–81. In other words, how much weight and persuasive value the agency affords various parts of the record is done on a dial with a range of reasonable 40 FLORES MOLINA V. GARLAND allocations; it does not always follow or correlate to the on/off switch of credibility as our pre-Dai regime wrongly assumed. The agency is afforded this wide discretion not only because of the nature of inquiries being assessed but also because of the practical realities at play in immigration cases. A petitioner’s testimony about what happened in a foreign country is often practically impossible to verify, and petitioners are usually strongly motivated to avoid deportation, which predictably often results in the embellishment of their own past hardships and other testimony. Separating the wheat from the chaff under these conditions is more of an art than a science, which is just one reason why an Immigration Judge may ultimately conclude that an alien is generally credible, but still not give full weight to his testimony. These fact-heavy determinations are complex and murky and present precisely the kind of questions that Congress has entrusted the agency alone to answer. And as Dai illustrated, we are supposed to consider the record as a whole, and diligently search for grounds to affirm the agency’s determination. We are bound to uphold agency decisions, even those of “less than ideal clarity if the agency’s path may be reasonably discerned.” Zamorano v. Garland, 2 F.4th 1213, 1222 (9th Cir. 2021) (quoting Dai, 141 S. Ct. at 1679). The overworked agency is not required to show every step of its work, or “follow a particular formula or incant ‘magic words’” in exercising its discretion. It is not required to disclose exactly what parts of the record it found persuasive or quantify how much of the petitioner’s testimony may have been discounted or even why. See id. Indeed, Dai makes clear that the agency may exercise its wide discretion in making these determinations implicitly, FLORES MOLINA V. GARLAND 41 and we must consider this possibility in searching through the record for support of the agency’s decision. See Dai, 141 S. Ct. at 1680–81. 3 — 3 — Recognizing the difficulty in justifying the majority’s decision under substantial evidence review, Judge Korman’s separate concurrence proposes a solution both simple and elegant: just don’t defer. The concurrence acknowledges a divide within our circuit (and amongst the circuits) as to what standard of review should apply when determining whether certain facts constitute persecution, and expresses a preference for characterizing them as legal questions that we review de novo. Judge Korman argues for de novo review because “the BIA did not base its past persecution finding on a rejection of the veracity of Flores Molina’s description of his past experiences [i.e., an explicit adverse credibility determination]. Rather, the BIA held that Flores Molina’s ‘claimed past harm, cumulatively considered, does not rise to the level of past persecution.’” But Judge Korman seems to have missed the Supreme Court’s clear message in Dai. As explained above, Dai unmistakably directed that when reviewing agency decisions, even if no adverse credibility determination was explicitly made, we must account for the possibility that the agency did not give every factual assertion made by the petitioner full weight or persuasive value. Converting every question of what facts constitute persecution into a legal one that we review de novo does not allow for this possibility because it wrongly assumes the agency found a petitioner’s version of facts fully persuasive. Here, as in other cases where the agency did not explicitly reject the veracity of a petitioner’s testimony, we don’t know if the agency discounted Molina’s factual account of his past experiences. De novo review, as Judge Korman would prefer, cannot account for this possibility. Nor can Judge Korman avoid Dai by attempting to constrain its implications to only those cases that present a wholly “different set of facts than the one put forward by” the petitioner. The Supreme Court in Dai repeatedly chastised our court for failing to consider whether the agency had implicitly determined that some of the evidence in that case was “outweighed” by other facts in the record. Dai, 141 S. Ct. at 1681. Dai thus requires us to consider whether the agency may have implicitly given reduced weight to some of the facts in the record vis-à-vis others, and if so, defer to the agency’s conclusion. There is no way to do that under Judge Korman’s de novo approach—it is, in fact, just our old, 42 FLORES MOLINA V. GARLAND B. Our Court’s Replacement of Extraordinary Deference with an Imposing Edifice of Lopsided Immigration Precedents The three nested tiers of deference just discussed reinforce one another when we properly apply the substantial evidence test and ask only if the record compels a contrary conclusion. Viewed synergistically, (1) our extraordinarily deferential standard of review, carried out with (2) a recognition that we are reviewing inherently indeterminate and fact-intensive questions, in light of (3) the wide range of discretion Congress has given the agency in answering these difficult inquiries, should combine to present one of the most deferential standards of review we apply as judges. But not in the Ninth Circuit. Our failure to defer begets more (and worsening) failures to defer, as our court keeps relying on its own distorted immigration precedent to justify a downward spiral. Every once in a while, the Supreme Court corrects us in a decision like Dai, where anyone with any common sense (including the unanimous Supreme Court) wonders how we could have strayed so far afield from our statutory mandate. But like a meandering elephant being smacked with a flyswatter, our court lumbers on. Unaffected by the occasional reversal, our ever-growing pile of perfidious immigration precedents make it harder and harder for judges to properly defer to the agency without seemingly conflicting with some precedent (while still following other discredited “deemed-credible-or-true” approach under a different guise. As my dissent explains, reviewing agency determinations as to what facts constitute persecution for substantial evidence is the only practical way of effectuating the Supreme Court’s instructions in Dai and not substituting judges’ own weighing of the factual record for the agency’s. FLORES MOLINA V. GARLAND 43 precedent—more on that below). The mechanics of how this works merits further elaboration. Despite the importance of this highly deferential and narrow scope of review to which Congress has anchored us, our court simply cannot resist the sirens of sympathy. So we frequently set our discretion free from the constraints of this hyper-deferential regime and reverse the agency in favor of a more palatable result. Each time a panel breaks free from the INA’s anchor of deference, it publishes a precedential decision that elevates its own discretion over the agency’s. The first panel to do so erected a buoy of caselaw that is only a stone’s throw away from the statutory anchor that should have tied the panel to the agency’s decision and commanded deference. But the next panel, drawn again to the sirens of sympathy or the lure of its own discretion, no longer starts from the original anchor of deference that Congress set. It instead starts from the bobbing buoy of our last wayward precedent. The process is foreseeable and plays out once again in this case—the majority latches onto a handful of similar facts from previous panel decisions that overruled the agency, and now erects its own new buoy of precedent, just a little further out to sea for the next sympathetic panel to analogize to (and stray further from). Eventually we are surrounded by precedential buoys that make any decision to reverse the agency appear like just routine reliance on precedent, even if it means the INA’s statutory anchor of deference is by this point far out of sight. Unmoored from the extremely deferential standard of review Congress has tied us to, we are now essentially lost at sea in our review of agency immigration decisions and tend to grab onto the nearest buoys of friendly caselaw closest to our facts (irrespective of whether the prior precedent was properly deferential). This is not how our 44 FLORES MOLINA V. GARLAND “extremely” deferential standard of review should work. We should remain anchored to the standard of review Congress has dictated, and under that deferential regime I cannot say the record compels a different result with respect to Molina’s claims here. Instead of respecting our deferential role and looking for ways to affirm agency decisions, our court has covered over the statutory standard of review with a common-law edifice of our own precedent that continues to obfuscate what proper application of the INA’s substantial evidence review should look like. This case follows and perpetuates the trend I’ve described. The majority recounts the standard of review, but then purports to follow it by cherry-picking immigration cases that justify its interpretation and preferred weight of the evidence (i.e., that Molina’s past harms amount to persecution). The mostly unspoken reality—because our court maintains a large immigration docket and sometimes applies the deferential standard but often does not—is that for the facts of almost any immigration case there is usually somewhat “analogous” precedent that supports both denying and granting the petition. Justice Scalia’s oft-cited concerns with respect to the indeterminate use of legislative history aptly apply in this context—that our huge court’s inconsistent application of the deferential standard of review has resulted in immigration precedent that offers a little “something for everybody.” ANTONIN SCALIA, A MATTER OF INTERPRETATION 36 (1997). Given that reality, judges can point to our precedent as justification for either result. The majority here analogizes to precedent that it argues shows the agency erred. But as I demonstrate below, our court’s precedents (including those cited by the BIA) can be just as easily applied to support the agency’s decision. Of course, in this situation—and in light of our extremely FLORES MOLINA V. GARLAND 45 deferential standard of review—a tie should go to the agency. But it often doesn’t in our court. One problem exacerbating this troubling trend is that decisions from our court that properly defer to the agency are usually resolved in unpublished dispositions with no precedential value. See, e.g., Islam v. Sessions, 743 Fed. App’x 734, 736 (9th Cir. 2018) (finding, in an unpublished disposition, that the petitioner’s past harm, including a beating and threats similar to what Molina suffered, “d[id] not evince actions so severe as to compel a finding of past persecution,” and that the petitioner failed to establish a well-founded fear of future persecution because he failed to show a “reasonable possibility” of persecution) (citing Hoxha v. Ashcroft, 319 F.3d 1179, 1182 (9th Cir. 2003)). By contrast, decisions granting the petition by extending prior precedents are typically published. So even if one was inclined to engage in this inappropriate game of picking friendly precedents from a crowded and diverse room, the game is rigged and lopsided against proper deference. 4 Looking to our circuit’s enormous, slanted edifice of deviant immigration precedents perpetuates our court’s regretful trend of granting ourselves massive discretion instead of granting massive deference to the agency. The — 4 — It shouldn’t matter much for purposes of precedent whether a circuit decision ruling on behalf of the BIA is published or not. The proper question is whether this record compels a result different than the agency’s. And if in a similar case the panel concluded that the record did not compel a different result, that conclusion—whether published or not—is a strong indication that this record also does not compel otherwise. The very fact that two or more judges previously reached that conclusion in a comparable case is powerful evidence that the REAL ID Act’s high standard hasn’t been met, whether or not the previous decision is technically binding on subsequent panels. 46 FLORES MOLINA V. GARLAND remnants of our overturned Dai decision, and other clearly wrong and manufactured rules that we have etched into this edifice, echo throughout our caselaw, blatantly favoring immigration relief and disguising our disdain for the properly deferential role that Congress prescribed because we can always say we are just following (or slightly extending) precedent. We cite the obligatory deferential language at the beginning of each new recalcitrant opinion, masquerading as if we’re being deferential even though we’re not. Deference to our own precedents? I suppose. But not to the agency’s discretion that Congress has authorized it to exercise and commanded that we defer to. Unfortunately, this case will no doubt be another layer in our leaning tower of precedents, encouraging further future deviation from our properly deferential role. II. ANALYSIS A. Proper Deference in This Case In this case, proper deference requires the panel to resist reweighing the evidence (i.e., giving greater weight to Molina’s claimed past harms in order to find they amount to past persecution) and resist cherry-picking precedent to justify a conclusion that, even if reasonable, is not compelled by the record. Because, in applying the three-nested tiers of deference described above, I cannot say the record before us compels a contrary conclusion, I would affirm the agency’s decision to dismiss Molina’s asylum claim. The agency’s decision to deny Molina’s asylum claim centered on his failure to demonstrate past persecution. Navas v. INS, 217 F.3d 646 , 655–56 (9th Cir. 2000) (“In order to establish eligibility for asylum on the basis of past persecution, an applicant must show: (1) an incident, or incidents, that rise to the level of persecution; (2) that is on FLORES MOLINA V. GARLAND 47 account of one of the statutorily-protected grounds; and (3) is committed by the government or forces the government is either unable or unwilling to control.”) (quotation marks omitted). The difficulty of reviewing agency determinations on credibility, weight, and persuasiveness—and the consequences of not properly deferring to those determinations—is well-illustrated here. There is no formula for how much weight or persuasive value to give Molina’s testimony that six hooded assailants threatened his life after giving him a bruised lip, and no formula for how much to discount his fear of future persecution in light of the evidence that conditions in Nicaragua seemed to be improving when the agency considered this case. In accordance with Dai, we must recognize that the absence of an adverse credibility determination does not mean that the agency also gave full weight and persuasive value to Molina’s testimony. The majority’s decision does not recognize or allow for this possibility, and instead implicitly assumes that because Molina’s testimony was found to be consistent, the agency must have also accorded full weight and persuasive value to his testimony on past harms. 1. The Record Doesn’t Compel that Molina’s Past Harms Evince Past Persecution. Persecution is “an extreme concept that does not include every sort of treatment our society regards as offensive.” Duran-Rodriguez v. Barr, 918 F.3d 1025, 1028 (9th Cir. 2019) (quoting Nagoulko v. INS, 333 F.3d 1012, 1016 (9th Cir. 2003)). Whether harm rises to the level of persecution necessarily requires weighing the evidence—a task only the agency is authorized to complete. Similarly, whether threats like the kind Molina received rise to the level of persecution (considering how direct, severe, repeated, etc.) is also a core 48 FLORES MOLINA V. GARLAND function of the agency because it requires weighing the evidence. 5 In the agency’s shoes, I may have allocated more weight to the harm Molina experienced and found that he suffered past persecution, or a well-founded fear of future persecution. But the evidence does not allow for only that view, as the majority claims. These fact-bound inquiries involve judgment calls that only the agency can make and we cannot reweigh the evidence as my colleagues in the majority have implicitly done here. See Guo v. Sessions, 897 F.3d 1208, 1212 (9th Cir. 2018); see also LeonHernandez v. INS, 926 F.2d 902, 904 (9th Cir. 1991). a. The Majority’s Discretion Trumps Proper Deference The majority identifies three bases for its conclusion that Molina’s past experiences amount to persecution: (1) he was forced to flee; (2) he received death threats; and (3) cumulatively, the effect of those incidents amount to persecution. With respect to the first two bases, as is often true, we have caselaw pointing in both directions, making the majority’s conclusion possible but not compelled since there are also circuit precedents evincing that threats combined with a minor assault are not enough to compel a finding of persecution. As to the third basis of the majority’s decision, the agency explicitly conducted a cumulative — 5 — Sharma v. Garland, 9 F.4th 1052 , 1061 (9th Cir. 2021) (“Determining whether the facts compel a conclusion of past persecution is ultimately a fact-bound endeavor that is not reducible to a set formula. . . . Under our cases, [relevant facts] must be evaluated in combination with each other to form a sufficiently negative portrait of the petitioner’s experience in his or her own country that not only allows a finding of past persecution but requires it.”). FLORES MOLINA V. GARLAND 49 review of the evidence, which requires our deference even if the majority disagrees with the result of that cumulative analysis. i. Forced to Flee As the majority correctly notes, our court has recognized that under certain circumstances “being forced to flee from one’s home in the face of an immediate threat of severe physical violence or death” can constitute persecution. Mendoza-Pablo v. Holder, 667 F.3d 1308, 1314 (9th Cir. 2012). But the circumstances in which our court has found that fleeing evidenced persecution were much more severe than those faced by Molina. For example, in MendozaPablo, the petitioner’s village was burned to the ground and its inhabitants massacred, including his immediate family members who were locked in their homes and burned alive—forcing his mother to flee to the mountains with him while eight-months pregnant. Id. at 1311–13. And in Knezevic, our court emphasized the “ethnic cleansing” that forced petitioner to flee after his home and business were destroyed while his hometown was shelled and bombarded by hostile forces. Knezevic v. Ashcroft, 367 F.3d 1206 , 1208–12 (9th Cir. 2004). Mendoza-Pablo and Knezevic do not stand for the proposition that fleeing always, or even usually, constitutes persecution, and those cases cannot be read isolated from their extreme facts, which differ markedly from Molina’s decision to relocate. Molina chose to relocate after a threat was spray painted on his house, and later after paramilitary members came looking for him. But our precedent sets a higher bar for persecution, including in the cases cited by the majority on fleeing one’s home. The petitioners in Mendoza-Pablo and Knezevic fled their homes, because, unlike Molina, their homes were completely destroyed. At the end of the day, 50 FLORES MOLINA V. GARLAND fleeing is one of many factors the agency considered, and viewed in light of the record as a whole, one reasonable conclusion (which the agency drew) is that Molina did not suffer persecution even though he relocated after receiving threats. The cases cited by the majority do not compel otherwise. ii. Death Threats Just as with being forced to flee, our court has also recognized that death threats can constitute persecution in a “small category of cases,” but only “where threats are repeated, specific and combined with confrontation or other mistreatment.” Duran-Rodriguez, 918 F.3d at 1028 (citation and quotation marks omitted) (determining that receiving death threats from armed men and fleeing to another town was not sufficient to compel a conclusion of past persecution); Lim v. INS, 224 F.3d 929, 936 (9th Cir. 2000) (holding that threats standing alone rarely constitute past persecution and “only when the threats are so menacing as to cause significant actual ‘suffering or harm’”) (citations omitted). Here, the record could be weighed, as the agency did, to find that Molina’s threats fall outside the small category of cases where threats constitute persecution, because the threats were not repeated, specific, or severe enough. Other than one instance, Molina’s threats were mostly indirect and detached from physical harm. 6 — 6 — While Molina was identified along with other protestors in social media posts as a violence instigator, called despicable things like a “rabid dog,” and told that “Chipote awaits,” those posts did not contain an explicit death threat and were not sent to him directly. Molina did receive a direct threat when “Bullets to the strikers (protesters)” was FLORES MOLINA V. GARLAND 51 It is true that our court has pushed the boundary further and further away from the core of “extreme” harm and suffering. But even under our expansive caselaw, the severity and frequency of Molina’s past harm does not compel a conclusion different than the agency’s. See, e.g., Mashiri v. Ashcroft, 383 F.3d 1112 , 1119–20 (9th Cir. 2004) (petitioner and his family were plagued with death threats and physical assaults for months, accompanied by slashed tires and a ransacked home); Ruano v. Ashcroft, 301 F.3d 1155 , 1160–61 (9th Cir. 2002) (petitioner received dozens of death threats over the course of six years, was chased by armed men on multiple occasions, and frequently followed to his home and work); Navas v. INS, 217 F.3d 646, 658 (9th Cir. 2000) (petitioner was threatened with death after two members of his family were murdered, shot at by the same perpetrators, and his mother beaten); Smolniakova v. Gonzales, 422 F.3d 1037 , 1041–42 (9th Cir. 2005) (petitioner was attacked and almost strangled to death by assailants who called her a “Jewish Bitch,” had her wrist slashed, and received multiple death threats and anti-Semitic harassment that the police refused to stop, including profanities and human feces smeared on her apartment, fires set in her mailbox, and repeated slashings of her front door). So even under our oft-wayward precedents, whether Molina’s threats and singular assault amount to persecution may be a close call, but it is not compelled by the record or our caselaw. The majority distinguishes the cases relied on by the BIA, claiming they are “not on-point” with the harm Molina suffered. But the petitioner in Lim (though he was not spray painted on his house. But—and without minimizing the gravity of the spray-painted threat—it only occurred once and was not accompanied by any physical harm. 52 FLORES MOLINA V. GARLAND physically attacked) had colleagues murdered by the dissident political group he infiltrated undercover, appeared on their death list, and suffered two years of death threats that prompted him to hire a personal bodyguard—a much longer period of harm and much more direct threats than Molina experienced. Lim, 224 F.3d at 932–35. Similarly, the petitioner in Gu (though he did not receive an explicit death threat) suffered more severe physical harm than Molina’s lost tooth and bruised lip—he was struck ten times with a rod when he was detained by the police. Gu v. Gonzales, 454 F.3d 1014, 1020 (9th Cir. 2006). Yet our court agreed that a finding of persecution was not compelled in either case. Just as with the precedents relied on by the majority, Lim and Gu are no doubt distinguishable in some ways, but they illustrate how our court has affirmed findings of no persecution despite more severe harm or threats than Molina encountered, just as the BIA did here. Again, none of the cases the majority cites compel a finding of persecution, and Lim and Gu are more than enough to support the BIA’s conclusion. iii. Cumulative Effect Finally, the majority argues that even if none of Molina’s past incidents rise to the level of persecution, a finding of persecution should be compelled given the “cumulative effect” and “escalating fashion” of his harms. Overturning an agency decision based on “cumulative effect” reasoning is always particularly undeferential, because whether or not disparate and independently insufficient harms cobbled together somehow suffice to meet an already ambiguous standard is really nothing more than the application of a “we know it when we see it” standard—reminiscent of Justice Stewart’s infamous search for the bounds of obscenity. See Jacobellis v. Ohio, 378 U.S. 184, 197 (1964) (Stewart, J., FLORES MOLINA V. GARLAND 53 concurring). The only circumstance in which I could imagine a court remanding an immigration case on “cumulative effect” without abandoning its deferential role would be if the agency obviously and completely failed to consider cumulative effects altogether. But here, the agency explicitly recounted and “cumulatively considered” Molina’s past harms. And while the escalating nature of his harms could support a finding of persecution, I do not see how it compels a finding of persecution—and the majority cites no case to show otherwise. The most factually similar precedent to this case is Hoxha v. Ashcroft, which was cited by the BIA in support of its determination that Molina’s past harm, even cumulatively considered, did not constitute past persecution. 319 F.3d 1179 (9th Cir. 2003). Hoxha, an ethnic Albanian from Kosovo, received multiple threats since his early childhood “that he would be harmed or killed if he chose to stay [in Kosovo],” and he was beaten on one occasion when a group of Serbs overheard he and a friend speaking Albanian. Hoxha, 319 F.3d at 1180–81. Hoxha suffered two broken ribs and extensive facial bruises from the beating. Id. at 1181 . This court acknowledged that “[a]lthough Hoxha’s experiences are disturbing and regrettable, they do not evince actions so severe as to compel a finding of past persecution,” and concluded that substantial evidence supported the BIA’s finding of no past persecution. Id. at 1182 . While there are some immaterial factual differences between Hoxha and this case (as no two immigration cases are the same), it is instructive here that evidence of multiple 54 FLORES MOLINA V. GARLAND threats, together with a severe beating, did not compel a conclusion that past persecution occurred in Hoxha. 7 The majority’s contention that the agency did not conduct a cumulative analysis is, at bottom, a disagreement with the result of the agency’s cumulative consideration of Molina’s past harms. Could the agency have concluded that Molina’s past harm cumulatively amounted to persecution? Sure. And the majority does an excellent job of outlining how our precedent could justify such a conclusion. But neither the record, nor even our precedent, compels a finding of past persecution. The majority places a heavy emphasis on the escalating nature of Molina’s harassment. But the reality, which after Dai we must account for, is that the agency presumably placed less weight on that consideration, and in reviewing the agency’s decision we must consider the range of permissible weight-allocations to see if any reasonable adjudicator could find as the agency did. As the majority recognizes, determining whether the facts compel a conclusion of past persecution is ultimately “a fact-bound endeavor that is not reducible to a set formula.” Because — 7 — One difference between Hoxha and the present case is that in Hoxha an active summons had been issued for the petitioner to report to the Serbian government and the country conditions evidence indicated “grisly documentation of numerous atrocities committed against ethnic Albanians” like Hoxha, that were not improving. Hoxha, 319 F.3d at 1181–84. Those facts led the court to conclude that Hoxha had a wellfounded fear of future persecution. See id. Here, no summons has been issued for Molina and the record is devoid of any other evidence indicating that the government or government-aligned groups have an active, ongoing interest in harming him. In the present case, the country conditions evidence before the agency indicated that conditions for political protestors in Nicaragua were improving. FLORES MOLINA V. GARLAND 55 here the formula used by the agency in weighing the evidence finds support in the record, I must defer to it. b. How Proper Deference Should Work in This Case Mimicking the majority, I could hop from buoy to buoy plucking different cases that support the agency’s decision, given that there are many decisions that illustrate neither death threats nor a single instance of physical harm, nor even a combination of the two, compel a finding of persecution. 8 Indeed, the instances that our court has recognized as rising to the extreme level of past persecution generally far surpass — 8 — See, e.g., Fuyong Cui v. Barr, 806 Fed. App’x 588, 590 (9th Cir. 2020) (petitioner failed to demonstrate past persecution even though he was arrested while participating in a protest, detained, kicked, punched in his face causing a tooth to fall out, and, after a subsequent protest, was hit several times with a baton and shocked with an electric baton); Saenz Martinez v. Barr, 818 Fed. App’x 767, 767–68 (9th Cir. 2020) (single beating at the hands of Sandinista supporters, which did not require medical treatment, and five or six unfulfilled threats did not rise to the level of persecution); Wakkary v. Holder, 558 F.3d 1049 , 1059–60 (9th Cir. 2009) (petitioner’s past experiences, including two beatings, even considered cumulatively, did not compel a finding of past persecution); Gu, 454 F.3d at 1021–22 (a three-day detention, two hour interrogation, and beating with a rod did not compel a conclusion of past persecution); Hoxha, 319 F.3d at 1182 (harassment, threats, and “one incident of physical violence” did not compel a finding of past persecution); Prasad, 47 F.3d at 339–40 (arrest, interrogation, beating, and other forms of harassment including rocks thrown at his house and attempts to steal his property were not enough to compel a finding of past persecution); see also Samad v. Whitaker, 759 Fed. App’x 634, 636 (9th Cir. 2019) (threats and beating did not rise to the level of persecution); Argieta-Chavarria v. Barr, 780 Fed. App’x 519, 520 (9th Cir. 2019) (single beating and subsequent harassment and threats did not rise to the level of persecution); Dong v. Barr, 830 Fed. App’x 239, 239–40 (9th Cir. 2020) (arrest, interrogation, beating, and 48-hour detention did not rise to the level of persecution). 56 FLORES MOLINA V. GARLAND what Molina experienced in Nicaragua. 9 But as already explained, that is not the proper application of substantial evidence review. This case is a snapshot of how difficult it is to properly defer under our court’s immigration precedent because we have strayed so far from our limited role—gaining confidence, even expertise, at second-guessing the agency and reweighing the evidence—which is what the majority reflexively does here. In short, the majority’s three bases in support of its decision are simply three bases that could support a finding of past persecution under a different weighing of the evidence and emphasis on caselaw. But none of the bases for the majority’s decision compels that conclusion in the face of the agency’s authoritative decision otherwise. When analogous precedent goes both ways on an issue (as it frequently does in our circuit’s immigration caselaw, and does here on the scope of harms that constitute persecution), deference should dictate our decision and we — 9 — See Parada v. Sessions, 902 F.3d 901 , 909–10 (9th Cir. 2018) (petitioner’s brother was assassinated, his neighbor murdered, and he was captured and beaten to the point of unconsciousness, repeatedly subjected to forced home invasions, and specific death threats toward his family—which collectively rose to the level of persecution); Bondarenko v. Holder, 733 F.3d 899 , 908–09 (9th Cir. 2013) (three detentions and one severe beating constituted past persecution); Guo v. Ashcroft, 361 F.3d 1194 , 1197–98 (9th Cir. 2004) (multiple arrests, detentions (including one for fifteen days), beatings, and inability to find work after being fired rose to the level of persecution); Guo, 897 F.3d at 1215 (beating which left petitioner unable to stand on his own and required medical attention, coupled with being unable to practice his faith constituted persecution). Accordingly, while Molina’s “experiences are disturbing and regrettable, they do not evince actions so severe as to compel a finding of past persecution.” Hoxha, 319 F.3d at 1182 . FLORES MOLINA V. GARLAND 57 should rely on precedent that supports the agency’s decision. Just as Justice Scalia warned with respect to legislative history, we should avoid the temptation to just “look over the heads of the crowd and pick out your friends.” ANTONIN SCALIA, A MATTER OF INTERPRETATION 36 (1997) (quoting Judge Leventhal). Unless we are guided by some North Star, our precedents will often support whatever conclusion we want. That guidance—the tie-breaker between conflicting precedents—is deference to the agency. The majority cannot point to a single precedent that would require any reasonable adjudicator to find that Molina suffered persecution. Given the short duration of his harm, the singular (and relatively minor) physical encounter, vague threats, and evidence before the agency of improving country conditions—a reasonable adjudicator could have weighed the record as a whole to find that Molina suffered harassment, but not persecution, which we reserve for “extreme” suffering or harm. See Donchev v. Mukasey, 553 F.3d 1206, 1213 (9th Cir. 2009). “The possibility of drawing two inconsistent conclusions from the evidence does not prevent an administrative agency’s finding from being supported by substantial evidence.” Leon-Hernandez, 926 F.2d at 904 (citation omitted). The agency’s conclusion may not be the result I would have reached or prefer, but again, that’s not the question before us. Applying the very deferential standard of review we should be anchored to, the record does not compel the conclusion that various indirect threats, combined with a single physical attack resulting in minor injuries, constitutes past persecution. See Hoxha, 319 F.3d at 1181–82. 58 FLORES MOLINA V. GARLAND 2. The Record Doesn’t Compel that Molina’s Fear of Future Persecution is Well-Founded. Because the record as a whole does not compel a finding of past persecution, Molina is not entitled to the presumption of future persecution that the majority awards him. Davila v. Barr, 968 F.3d 1136 , 1141–42 (9th Cir. 2020). That means Molina was required to demonstrate to the agency a well-founded fear of future persecution. As already discussed, what constitutes persecution or a well-founded fear of it is an inherently ambiguous concept—underscoring the importance of deferring to the agency’s determination on those issues. The majority claims that the BIA “selectively” cited portions of the record and “ignored” other evidence of the country conditions in Nicaragua that would have supported Molina’s fear of future persecution. The majority’s position is a classic demonstration of what I’ve outlined above— looking for ways to reverse the agency’s decision instead of looking for ways to uphold it. Several reports on the conditions in Nicaragua were offered to the agency, most from 2018, and some more recent reports from 2019. The agency emphasized the more recent reports that indicated conditions in Nicaragua were improving—political prisoners were being released, and the government had announced its intention to release “all remaining political prisoners.” Whether correct or not, the agency reasonably concluded based on the evidence in front of it at the time that the political tide was turning in Nicaragua, and given that “upwards of a million people” participated in the protests but only a small fraction were pursued by the government, Molina’s fear of future persecution was too speculative to be well-founded. FLORES MOLINA V. GARLAND 59 The majority claims that because the earlier reports, which detailed the human rights abuses in Nicaragua during 2018, were not explicitly mentioned in the agency’s decision, they were ignored—justifying reversal. But it is the majority that ignores parts of the BIA’s decision in its quest to overturn it. The agency’s actual analysis expressly said it reviewed the entire record and explicitly acknowledged that “human rights abuses occur in Nicaragua.” It gleaned this fact from the parts of the record the majority says the agency ignored. The agency didn’t ignore any part of the record; it simply emphasized and gave more weight to the more recent accounts that showed conditions were improving, precisely what the agency is permitted to do. Additionally, much of the country conditions evidence that the majority claims was not considered by the agency was largely focused on the treatment of detainees in prison—not political protestors such as Molina. The majority once again misconstrues the agency’s obligations. It is not obligated to spell out every piece of evidence it relies on or rejects, nor is it required to explicitly state how much weight it gives various pieces of the record or “incant ‘magic words’” in the exercise of its discretion. Zamorano, 2 F.4th at 1222 (quoting Dai, 141 S. Ct. at 1679). III. CONCLUSION Ultimately, my view of Molina’s past harms is not far from that of my colleagues—as I noted above, the facts present a close call, and I am sympathetic to the majority’s view that Molina may have suffered past persecution. Where we diverge is our approach to the agency’s decision and the record. The majority admittedly travels a welltrodden path in its approach: looking for a basis to overturn the agency instead of scouring the record as a whole looking 60 FLORES MOLINA V. GARLAND for a way to uphold the agency if even a single reasonable factfinder could agree with its ultimate conclusion. Our court’s edifice of immigration caselaw has obfuscated the correct standard of review, making the proper approach harder to see and even harder to execute. These small differences of opinion, as illustrated in this case between my position and that of my colleagues in the majority, have been multiplied over time in many decisions, leading to the lopsided edifice that is currently improperly driving much of our court’s immigration caselaw. Because the record does not compel the conclusion that (1) the past harassment suffered by Molina rises to the level of past persecution, or that (2) such harassment—together with the most recent country conditions evidence that was before the agency—demonstrates a well-founded fear of future persecution, Molina’s petition for review of his asylum claim should be denied. Likewise, the record does not compel a contrary conclusion with respect to Molina’s remaining applications for withholding of removal, humanitarian asylum, or protection under CAT, and his petition with respect to those claims should also be denied. And finally, the BIA did not abuse its discretion in denying Molina’s motion to reopen. I would therefore deny Molina’s petitions for review.