Two asylum seekers cross the southern border six months apart. Same country, same persecution, same statute. A clerk neither of them has ever met routes one to the regular docket and the other to an expedited docket. One gets heard in eight months. The other is waiting until 2028. That single routing decision is now six years of difference.
In the first of a two-part deep dive on immigration adjudication, Gwen and Marc walk through the system that will decide both of their cases:
- 3.2 million pending cases, 2.3 million of them asylum, divided among roughly 700 immigration judges — about 3,200 cases per judge, each requiring hours of fact-intensive testimony through interpreters about events in another country
- The particular social group doctrine — the single most contested category in asylum law, which has flipped four times in eleven years through Attorney General certification. Same statute, same words, opposite results depending on who holds the office when your case is heard
- How the AG cleared 767,000 cases in 2025 — the highest single-year total in the system's history — through a mix of procedural exits, compressed dockets, and doctrinal changes that foreclosed categories of claims that were viable when applicants filed
- The MSPB decision that stripped immigration judges of civil service protections, holding they're inferior officers removable at will — meaning the judge deciding whether to grant asylum on a borderline theory now knows she can be fired tomorrow with no notice and no appeal
- The denial rate climbing from roughly fifty percent in 2023 to eighty percent in February 2026, without a single statutory change
Rosa's hearing is in 2028. She'll argue a social group theory the Attorney General has already foreclosed, with six-year-old evidence about a country she fled six years ago, in front of a judge who can be removed at will. The statute hasn't changed. Everything else has.
Next episode: the person standing inside the machinery — detention, representation, expedited removal, and why the Constitution doesn't fix any of it.
00:00:14 --> 00:00:17 Hello, and welcome to Administrative Remedies,
00:00:17 --> 00:00:19 because you can't fix what you don't understand.
00:00:19 --> 00:00:21 Brought to you in part by the University of Tulsa
00:00:21 --> 00:00:24 College of Law. I'm Gwendolyn Savitz, an associate
00:00:24 --> 00:00:26 professor here at TU and the associate dean of
00:00:26 --> 00:00:28 research and intellectual life. And I'm Mark
00:00:28 --> 00:00:31 Rourke. I'm the dean of the College of Law. We'll
00:00:31 --> 00:00:33 be breaking down complex doctrines with real
00:00:33 --> 00:00:35 -life analogies and examples to demystify the
00:00:35 --> 00:00:37 world of administrative law for everyone trying
00:00:37 --> 00:00:39 to understand how government actually works.
00:00:40 --> 00:00:41 Agencies are the main way the federal government
00:00:41 --> 00:00:44 gets things done. It's not through Congress for
00:00:44 --> 00:00:46 reasons we'll be addressing over the course of
00:00:46 --> 00:00:49 this series. Mark, this was supposed to be the
00:00:49 --> 00:00:51 last episode of the season, but I think we'll
00:00:51 --> 00:00:54 need to break it into two. You can't just put
00:00:54 --> 00:00:58 a good thing down. So this is part one. Let's
00:00:58 --> 00:01:01 say in early 2022, you have two asylum seekers
00:01:01 --> 00:01:03 crossing the southern border six months apart.
00:01:03 --> 00:01:07 We'll call them Rosa and Elena. Within their
00:01:07 --> 00:01:09 first month in the system, they get routed differently.
00:01:10 --> 00:01:13 One goes to a dedicated docket. The other goes
00:01:13 --> 00:01:16 to the regular docket. And that single routing
00:01:16 --> 00:01:18 decision that's made by a clerk neither one of
00:01:18 --> 00:01:20 them has ever met is now the reason one of them
00:01:20 --> 00:01:22 was heard in eight months and the other is now
00:01:22 --> 00:01:26 going to be waiting until 2028. So the same statute,
00:01:26 --> 00:01:29 the same courthouse, but six years of difference
00:01:29 --> 00:01:31 based on which docket somebody dropped them into.
00:01:32 --> 00:01:34 Yes. That's the immigration court version of
00:01:34 --> 00:01:37 the problem we've been tracing all season. how
00:01:37 --> 00:01:38 much time the person has, what doctrine applies,
00:01:39 --> 00:01:41 which judge hears the case, and how much pressure
00:01:41 --> 00:01:43 the judge is under. But today is immigration
00:01:43 --> 00:01:46 court. Right. Today is the machinery, the volume,
00:01:47 --> 00:01:50 how asylum works, the attorney general certification,
00:01:50 --> 00:01:53 immigration judges, and judge -to -judge variation.
00:01:54 --> 00:01:59 And then next time, we will be turning to the
00:01:59 --> 00:02:02 person standing inside the machinery. the impact
00:02:02 --> 00:02:05 of representation and detention, expedited removal,
00:02:05 --> 00:02:08 and the very limited constitutional floor that
00:02:08 --> 00:02:10 applies. We have a lot to cover, so let's get
00:02:10 --> 00:02:13 started. All right. Rosa's the first one. She
00:02:13 --> 00:02:15 arrives in early 2022. She's fleeing gang violence
00:02:15 --> 00:02:17 in Honduras. They're both going to be fleeing
00:02:17 --> 00:02:20 gang violence in Honduras. So they both pass
00:02:20 --> 00:02:23 a credible fear interview. And for Rosa, DHS
00:02:23 --> 00:02:26 files a notice to appear, and she gets a master
00:02:26 --> 00:02:30 calendar hearing date in 2026. And that's not
00:02:30 --> 00:02:33 the merits one. That's just an initial hearing.
00:02:33 --> 00:02:35 The merits hearing, the one that decides whether
00:02:35 --> 00:02:38 she gets asylum, is scheduled for 2028. And the
00:02:38 --> 00:02:41 judge has to assess her fear of return as of
00:02:41 --> 00:02:44 when? It is as of the hearing date. Asylum law
00:02:44 --> 00:02:46 is forward looking. The question isn't whether
00:02:46 --> 00:02:49 she had a well -founded fear in 2022 when she
00:02:49 --> 00:02:51 came here. The question is whether she'd have
00:02:51 --> 00:02:53 a well -founded fear if she were removed the
00:02:53 --> 00:02:56 day the hearing is happening. It's the country
00:02:56 --> 00:02:58 conditions at the moment of decision. That's
00:02:58 --> 00:03:00 what matters. OK, so in some ways, the delay
00:03:00 --> 00:03:03 could help her. If conditions in Honduras have
00:03:03 --> 00:03:05 gotten worse, that's evidence supporting her
00:03:05 --> 00:03:08 present fear. It could cut either way. If she
00:03:08 --> 00:03:11 can show past persecution in 2022, she gets a
00:03:11 --> 00:03:14 regulatory presumption of well -grounded fear
00:03:14 --> 00:03:16 going forward. And the government has to rebut
00:03:16 --> 00:03:19 that with evidence of changed conditions. But
00:03:19 --> 00:03:22 the longer the delay, the more rebuttal material
00:03:22 --> 00:03:24 the government has. State Department country
00:03:24 --> 00:03:30 reports from 2023, 24, 25, 26. Each one is potential
00:03:30 --> 00:03:33 ammunition. If the gang she fled has been suppressed
00:03:33 --> 00:03:36 or fractured or moved on, that's the government's
00:03:36 --> 00:03:37 affirmative argument that she no longer has a
00:03:37 --> 00:03:40 well -founded fear. So the delay doesn't make
00:03:40 --> 00:03:43 her old fear stale. It just shifts the evidentiary
00:03:43 --> 00:03:45 fight from what happened in Honduras in 2022
00:03:45 --> 00:03:49 to what's happening now. Right. And there's also
00:03:49 --> 00:03:51 a separate problem, which is evidence degradation.
00:03:51 --> 00:03:54 So witnesses to this 2022 persecution may be
00:03:54 --> 00:03:57 dead or unreachable by that point. Medical records
00:03:57 --> 00:03:59 of the assault might have been destroyed. Her
00:03:59 --> 00:04:02 own memory for specific dates is going to be
00:04:02 --> 00:04:04 now six years older. None of that's doctrine.
00:04:04 --> 00:04:06 That's just the reality of what happens to evidence
00:04:06 --> 00:04:09 over time. And on top of that, she'll have been
00:04:09 --> 00:04:11 living for six years in the U .S., potentially
00:04:11 --> 00:04:13 with a work permit. She might have had kids who
00:04:13 --> 00:04:16 were born here. She's presumably formed community
00:04:16 --> 00:04:19 ties. Removal six years after arrival doesn't
00:04:19 --> 00:04:21 undo six years of having lived here. Okay. Well,
00:04:22 --> 00:04:24 what about Elena? All right. So Elena crossed
00:04:24 --> 00:04:27 six months after Rosa. She got assigned to an
00:04:27 --> 00:04:30 expedited docket, and her merits hearing was
00:04:30 --> 00:04:33 just eight months out. That gave her eight months
00:04:33 --> 00:04:35 to find a lawyer, document her fear, gather evidence
00:04:35 --> 00:04:38 from the country she fled, build a case, and
00:04:38 --> 00:04:41 grant rates on expedited dockets run substantially
00:04:41 --> 00:04:44 below the regular docket. So the variation is
00:04:44 --> 00:04:46 not just in the outcomes. It's how much time
00:04:46 --> 00:04:50 the system gives you to prepare. Right. Last
00:04:50 --> 00:04:51 time we talked about Social Security disability,
00:04:52 --> 00:04:54 the largest adjudication system in the federal
00:04:54 --> 00:04:56 government, two and a half million applications
00:04:56 --> 00:04:58 a year. Today is the second part, immigration.
00:04:59 --> 00:05:03 It's a slightly smaller volume than SSA and has
00:05:03 --> 00:05:05 a different framework. But most of the stuff
00:05:05 --> 00:05:07 we talked about before shows up here again, just
00:05:07 --> 00:05:11 stronger. Stronger how? The variation is wider.
00:05:11 --> 00:05:13 The independence protections for the adjudicator
00:05:13 --> 00:05:16 are weaker. The doctrinal rules shift more aggressively
00:05:16 --> 00:05:19 more often and through a more direct mechanism.
00:05:19 --> 00:05:22 And the consequences are different in kind. Not
00:05:22 --> 00:05:24 denial of benefits, but removal from the country.
00:05:26 --> 00:05:28 Let's do this by the numbers, because I don't
00:05:28 --> 00:05:30 think people understand the scale of this issue.
00:05:31 --> 00:05:33 All right. The pending caseload at immigration
00:05:33 --> 00:05:36 courts is currently around 3 .2 million cases.
00:05:36 --> 00:05:38 That's not the annual, so we're not directly
00:05:38 --> 00:05:40 comparing it to Social Security. That's just
00:05:40 --> 00:05:42 what's pending. That's down from a peak of 3
00:05:42 --> 00:05:47 .7 million in 2024, but it's still virtually
00:05:47 --> 00:05:49 the highest in the system's history since we
00:05:49 --> 00:05:51 started coming down from that peak. And in comparison,
00:05:51 --> 00:05:54 in 2022, it was 1 .9 million. So we're still
00:05:54 --> 00:05:58 way up. These aren't all asylum cases, but...
00:05:58 --> 00:06:01 Of this 3 .2 million, 2 .3 million of them are
00:06:01 --> 00:06:04 asylum cases. They're like 70 % of the docket.
00:06:04 --> 00:06:06 And what about the immigration judges? We currently
00:06:06 --> 00:06:08 have about 700. When Trump took office, that
00:06:08 --> 00:06:12 was more like 735. But they got down to around
00:06:12 --> 00:06:16 600 since so many people left or were asked to
00:06:16 --> 00:06:20 leave or just not there for some reason. So the
00:06:20 --> 00:06:24 backlog has come down, but so has the decision
00:06:24 --> 00:06:27 -making capacity. If you divide even the 2 .3
00:06:27 --> 00:06:31 million pending asylum cases by 700 judges, you
00:06:31 --> 00:06:35 get about 3 cases per judge. And new ones
00:06:35 --> 00:06:38 keep coming. So when you compare it to Social
00:06:38 --> 00:06:42 Security, which has about 1 ALJs handling
00:06:42 --> 00:06:45 about 2 .5 million new applications a year. Right.
00:06:45 --> 00:06:48 Immigration has way fewer judges handling cases
00:06:48 --> 00:06:51 of a similar complexity. And asylum cases in
00:06:51 --> 00:06:54 particular are really fact intensive, more so
00:06:54 --> 00:06:56 than the average disability case. The credibility
00:06:56 --> 00:06:59 determinations turn on testimony about events
00:06:59 --> 00:07:02 in another country. We need evidence of the country
00:07:02 --> 00:07:05 conditions, expert testimony, interpreters. A
00:07:05 --> 00:07:07 merits hearing can take three to four hours if
00:07:07 --> 00:07:09 it's done properly. Not every Social Security
00:07:09 --> 00:07:11 case also has to be seen by an ALJ. Some of them
00:07:11 --> 00:07:14 end before that point. And the immigration judge
00:07:14 --> 00:07:17 has 3 cases that's waiting for him. Right.
00:07:17 --> 00:07:19 So this creates the kind of structural pressure
00:07:19 --> 00:07:22 we talked about a couple episodes ago. This volume
00:07:22 --> 00:07:24 isn't neutral. When you have that many cases
00:07:24 --> 00:07:26 per judge, you can't conduct each one at the
00:07:26 --> 00:07:29 pace the stakes require. Something has to give.
00:07:29 --> 00:07:31 And what gives is some combination of time per
00:07:31 --> 00:07:34 case, depth of factual development, quality of
00:07:34 --> 00:07:36 credibility assessment, and access to representation.
00:07:37 --> 00:07:39 And when leadership wants to reduce the backlog...
00:07:39 --> 00:07:43 They have ways to make it happen. So EOIR completed
00:07:43 --> 00:07:48 more than... 767 cases in 2025. That was
00:07:48 --> 00:07:51 the highest single year total in the agency's
00:07:51 --> 00:07:53 history. And the pending caseload dropped by
00:07:53 --> 00:07:57 about 447 between January and September.
00:07:57 --> 00:08:00 So these were agency productivity records. And
00:08:00 --> 00:08:02 you're going to tell me that the productivity
00:08:02 --> 00:08:05 is something different than what it sounds like.
00:08:05 --> 00:08:08 Right. They're really different kinds of cleared
00:08:08 --> 00:08:11 cases. First, we can have procedural. These are
00:08:11 --> 00:08:13 terminations or dismissals in absentia removals
00:08:13 --> 00:08:16 if the respondents don't even appear. Those cases
00:08:16 --> 00:08:18 exit the system without an individualized merits
00:08:18 --> 00:08:22 decision at all. OK, so it's not adjudication.
00:08:22 --> 00:08:25 It's just simply the case is over. Yeah. These
00:08:25 --> 00:08:27 respondents never get a merits hearing. Maybe
00:08:27 --> 00:08:28 they were detained and transferred and didn't
00:08:28 --> 00:08:31 get notice. Maybe DHS moved them to expedited
00:08:31 --> 00:08:33 removal. Maybe the case got dismissed because
00:08:33 --> 00:08:36 the government decided not to pursue it. Regardless,
00:08:36 --> 00:08:38 we have no judge spending time weighing the evidence.
00:08:39 --> 00:08:41 Okay. What's the second mechanism? We can accelerate
00:08:41 --> 00:08:43 it with these rocket dockets. We still have a
00:08:43 --> 00:08:46 merits hearing, but everything is on a compressed
00:08:46 --> 00:08:48 timeline. Eight months from notice to appear
00:08:48 --> 00:08:51 to a final hearing in some cases, or even less.
00:08:51 --> 00:08:54 This is running under the same statute, same
00:08:54 --> 00:08:56 doctrine, but you will probably have a thinner
00:08:56 --> 00:08:59 record going in. And the third mechanism? All
00:08:59 --> 00:09:02 right. We can just change the laws that are handling
00:09:02 --> 00:09:04 these. When we were talking about appeals, about
00:09:04 --> 00:09:06 how the attorney general can certify and decide
00:09:06 --> 00:09:09 a case. And they can do that to narrow what counts
00:09:09 --> 00:09:11 as a viable claim. Those cases can still get
00:09:11 --> 00:09:14 a full merits hearing. The applicant can testify.
00:09:14 --> 00:09:16 The judge applies the law. The decision is on
00:09:16 --> 00:09:18 the merits. But the doctrinal rule the judge
00:09:18 --> 00:09:20 is applying has been rewritten to make a category
00:09:20 --> 00:09:23 of cases non -viable. So someone could lose on
00:09:23 --> 00:09:25 the merits here on the same facts that they might
00:09:25 --> 00:09:28 have won on 18 months earlier. Okay. Do you have
00:09:28 --> 00:09:30 a concrete example to illustrate this? All right.
00:09:30 --> 00:09:35 So let's talk matter of SSFM from September 25.
00:09:35 --> 00:09:38 Here, the attorney general certified a case,
00:09:38 --> 00:09:41 overruled the Biden -Garland era framework, and
00:09:41 --> 00:09:44 reinstated a rule from the Sessions era. There
00:09:44 --> 00:09:46 were no statutes that had changed, no regulation
00:09:46 --> 00:09:49 that went through notice and comment, but categories
00:09:49 --> 00:09:51 of claims that were viable when some applicants
00:09:51 --> 00:09:53 filed stopped being viable when the certification
00:09:53 --> 00:09:58 came down. EOIR press release counts all three
00:09:58 --> 00:10:01 of those. The procedural exits, the rocket docket
00:10:01 --> 00:10:04 merits decisions, the doctrinally foreclosed
00:10:04 --> 00:10:07 merits decisions, all of those ends completions.
00:10:07 --> 00:10:10 Right. The agency's own metrics don't distinguish.
00:10:10 --> 00:10:14 From the dashboard, 667 cases got cleared.
00:10:14 --> 00:10:17 The completions aren't equivalent. The numerical
00:10:17 --> 00:10:20 reduction in the backlog is real, but what got
00:10:20 --> 00:10:22 reduced in some categories wasn't just. Delay,
00:10:22 --> 00:10:24 for some respondents, it was the availability
00:10:24 --> 00:10:26 of the merits decision under the doctrine that
00:10:26 --> 00:10:29 existed when they filed. Okay. In the last episode,
00:10:29 --> 00:10:32 you walked us through the SSA's five -step sequential
00:10:32 --> 00:10:35 evaluation. Can you do the same thing in the
00:10:35 --> 00:10:42 immigration context? Sure. All right. We don't
00:10:42 --> 00:10:44 have a numbered checklist here like we would
00:10:44 --> 00:10:47 with disability, but every asylum case is running
00:10:47 --> 00:10:50 through the same basic framework. The applicant
00:10:50 --> 00:10:53 has to prove a few different elements. First,
00:10:53 --> 00:10:55 that they have a genuine and well -founded fear
00:10:55 --> 00:10:58 of persecution. This means serious harm or threats
00:10:58 --> 00:11:01 if they return. Second, that persecution has
00:11:01 --> 00:11:03 to be on account of one of five protected grounds,
00:11:03 --> 00:11:05 their race, religion, nationality, political
00:11:05 --> 00:11:08 opinion, or membership in what the statute calls
00:11:08 --> 00:11:11 a particular social group. Third, the threat
00:11:11 --> 00:11:13 has to come from the government or from people
00:11:13 --> 00:11:16 the government cannot or will not control. Fourth,
00:11:16 --> 00:11:18 the court looks at whether the person could be
00:11:18 --> 00:11:21 safely relocated within their own country. So
00:11:21 --> 00:11:23 maybe they're not safe in that city, but they
00:11:23 --> 00:11:26 could be safe in a different city. And finally,
00:11:26 --> 00:11:28 all of this needs to be supported with credible
00:11:28 --> 00:11:31 testimony and corroborating evidence. OK, let's
00:11:31 --> 00:11:35 put this to Rosa. All right. Reza's fleeing gang
00:11:35 --> 00:11:38 violence in Honduras. The gang has extorted her
00:11:38 --> 00:11:40 family. They murdered her brother after he refused
00:11:40 --> 00:11:42 to pay, and they've threatened her directly.
00:11:42 --> 00:11:44 So on that first element, whether she has a well
00:11:44 --> 00:11:47 -founded fear, she has a clear case. People in
00:11:47 --> 00:11:49 her position have been killed. The threat is
00:11:49 --> 00:11:53 documented. What's then the problem? It hits
00:11:53 --> 00:11:56 when we get to that second element, that this
00:11:56 --> 00:11:59 needs to be on account of her race, religion,
00:11:59 --> 00:12:02 nationality, political opinion. None of those
00:12:02 --> 00:12:04 apply. So her case, like most of the Central
00:12:04 --> 00:12:06 American asylum cases, has to go through membership
00:12:06 --> 00:12:08 in a particular social group. And that is the
00:12:08 --> 00:12:11 single most contested doctrinal category in immigration
00:12:11 --> 00:12:14 law. So why is that so contested? OK, well, what
00:12:14 --> 00:12:17 does particular social group mean? Particular
00:12:17 --> 00:12:22 means specific, identifiable. Social means defined
00:12:22 --> 00:12:25 by some relationship to other people, not biology
00:12:25 --> 00:12:28 or geography alone. And group is just a set of
00:12:28 --> 00:12:31 people with something in common. So a particular
00:12:31 --> 00:12:33 social group is some identifiable set of people
00:12:33 --> 00:12:37 defined by a social rather than a purely individual
00:12:37 --> 00:12:40 characteristic. That's broad enough to cover
00:12:40 --> 00:12:43 almost anything. I mean, women in a particular
00:12:43 --> 00:12:46 village, former gang members, children of a certain
00:12:46 --> 00:12:48 family, people who testified against a cartel.
00:12:49 --> 00:12:51 That's the problem the doctrine has been wrestling
00:12:51 --> 00:12:53 with for 40 years. If particular social group
00:12:53 --> 00:12:55 means everything you just listed, it swallows
00:12:55 --> 00:12:58 the rest of the statute. Asylum becomes available
00:12:58 --> 00:13:00 to almost anyone who can describe their persecutors
00:13:00 --> 00:13:03 in social terms. So the doctrine has narrowed
00:13:03 --> 00:13:07 it. How so? The phrase itself isn't defined in
00:13:07 --> 00:13:10 the statute, but the Board of Immigration Appeals,
00:13:10 --> 00:13:12 the Attorney General, and the federal courts
00:13:12 --> 00:13:14 have all been interpreting it. And these interpretations
00:13:14 --> 00:13:18 have shifted back and forth. The current frame
00:13:18 --> 00:13:20 is that a protected social group has to be composed
00:13:20 --> 00:13:22 of members who share a common, immutable characteristic.
00:13:23 --> 00:13:25 It has to be defined with particularity and it
00:13:25 --> 00:13:27 has to be socially distinct within the society
00:13:27 --> 00:13:30 in question. So Rosa's lawyer has to construct
00:13:30 --> 00:13:33 a social group that fits those three requirements.
00:13:33 --> 00:13:36 Right. Needs to construct it and do it as narrowly
00:13:36 --> 00:13:38 as possible. Rosa can't just be a member of Hondurans
00:13:38 --> 00:13:41 Afraid of Gangs. That's not socially distinct.
00:13:41 --> 00:13:49 Her lawyer might try something like, It's narrow,
00:13:49 --> 00:13:52 particular, immutable, socially distinct. And
00:13:52 --> 00:13:55 whether that group is cognizable. Yeah, that's
00:13:55 --> 00:13:57 been litigated for a long time. And the answer
00:13:57 --> 00:13:59 keeps changing depending on who's in the attorney
00:13:59 --> 00:14:02 general's office. We've had four reversals in
00:14:02 --> 00:14:05 11 years on what the social group doctrine for
00:14:05 --> 00:14:07 survivors of domestic violence should be. Same
00:14:07 --> 00:14:10 statute the entire time. And what that actually
00:14:10 --> 00:14:13 changes between those readings. On the same statutory
00:14:13 --> 00:14:16 words, what's a Sessions reading versus a Garland
00:14:16 --> 00:14:18 reading, for example? There's two different ways
00:14:18 --> 00:14:21 these change. First, the level of generality
00:14:21 --> 00:14:23 the group can be described at. The Sessions reading,
00:14:23 --> 00:14:26 the stricter one, insists on extremely narrow
00:14:26 --> 00:14:28 groups that aren't defined by the persecution
00:14:28 --> 00:14:31 itself. So women who suffer domestic violence
00:14:31 --> 00:14:33 can't be a social group because the persecution
00:14:33 --> 00:14:36 itself is doing the defining. A Garland reading
00:14:36 --> 00:14:38 allows more room. The group exists in society
00:14:38 --> 00:14:40 and the persecution targets it because of what
00:14:40 --> 00:14:44 it is. Women who've suffered domestic violence
00:14:44 --> 00:14:47 could be a social group then. Second, where this
00:14:47 --> 00:14:50 violence needs to come from. Can it be gangs,
00:14:50 --> 00:14:53 family members, husbands? Is that going to count
00:14:53 --> 00:14:57 as persecution or is only specific state violence
00:14:57 --> 00:14:59 going to be? A Sessions reading is pushing more
00:14:59 --> 00:15:01 towards that state only. A Garland reading more
00:15:01 --> 00:15:03 easily lets private violence count if the state
00:15:03 --> 00:15:06 won't protect the individual. And those two readings
00:15:06 --> 00:15:09 produce opposite results on the same facts. Yes.
00:15:09 --> 00:15:12 Rosa's case in 2028 will turn on which reading
00:15:12 --> 00:15:15 is in effect when she's heard. And whoever's
00:15:15 --> 00:15:17 AG when she's heard is the one whose view of
00:15:17 --> 00:15:20 that rule then controls. Right. So under the
00:15:20 --> 00:15:23 current version, it was decided on facts almost
00:15:23 --> 00:15:27 identical to Rosa's. The respondent in SSFM was
00:15:27 --> 00:15:29 a Honduran woman pressing social gang theories
00:15:29 --> 00:15:31 on the same types of facts that Rosa's going
00:15:31 --> 00:15:36 to need in 2028. And the decision in SSFM will
00:15:36 --> 00:15:39 foreclose those theories. Her lawyer in 2028
00:15:39 --> 00:15:42 is going to be arguing a social group that we
00:15:42 --> 00:15:44 already have an attorney general decision says
00:15:44 --> 00:15:47 doesn't qualify on facts that the attorney general
00:15:47 --> 00:15:49 has basically already addressed. OK, hold on.
00:15:50 --> 00:15:52 Courts reversed themselves, too, though. You
00:15:52 --> 00:15:55 know, Loper Bright overruled Chevron. Dobbs overruled
00:15:55 --> 00:15:58 Roe. You're expecting Humphrey's executor to
00:15:58 --> 00:16:02 be overturned later this month. This isn't only
00:16:02 --> 00:16:05 an agency phenomenon. Thanks so much for the
00:16:05 --> 00:16:08 reminder. And you're right, it isn't. Courts
00:16:08 --> 00:16:10 do reverse. But the thing that's different here
00:16:10 --> 00:16:13 isn't whether the doctrine flips, it's who flips
00:16:13 --> 00:16:16 it and how fast. Rowe lasted 49 years before
00:16:16 --> 00:16:20 it was overturned. Chevron was 40 years. This
00:16:20 --> 00:16:22 line has flipped four times in 11 years. And
00:16:22 --> 00:16:25 every flip... is a single official acting alone,
00:16:25 --> 00:16:27 picking a case, writing the decision, and binding
00:16:27 --> 00:16:29 every immigration judge in the country immediately.
00:16:30 --> 00:16:33 So there's no five -vote majority, no concurrences,
00:16:33 --> 00:16:36 no dissents. Right. We've got one signature.
00:16:36 --> 00:16:38 And stare decisis, as the courts apply it, is
00:16:38 --> 00:16:41 a self -imposed constraint that asks for some
00:16:41 --> 00:16:43 institutional consensus before overruling something.
00:16:44 --> 00:16:47 As we've established, it's not perfectly observed,
00:16:47 --> 00:16:50 a little less so recently than it used to be,
00:16:50 --> 00:16:53 but it still does exist as a concept. The attorney
00:16:53 --> 00:16:56 general certifications don't have that same constraint.
00:16:56 --> 00:16:58 The doctrine flips on the timeline of the official,
00:16:58 --> 00:17:01 not the timeline of the institution. And this
00:17:01 --> 00:17:04 isn't just an anomaly. Right. We see the same
00:17:04 --> 00:17:07 thing for family -based social groups. There's
00:17:07 --> 00:17:09 a decision called matter of LEA, which has been
00:17:09 --> 00:17:11 certified and recertified back and forth across
00:17:11 --> 00:17:14 administrations. We get shifts on credible fear
00:17:14 --> 00:17:17 standards. We get shifts on what counts as persecution
00:17:17 --> 00:17:19 versus discrimination. The substantive rules.
00:17:20 --> 00:17:22 governing whether someone qualifies aren't stable.
00:17:23 --> 00:17:25 They're actively contested, and the contest gets
00:17:25 --> 00:17:27 resolved case by case through AG certification.
00:17:27 --> 00:17:30 You're saying immigration judge and not ALJ.
00:17:30 --> 00:17:33 Why? All right. Immigration judges are Department
00:17:33 --> 00:17:36 of Justice employees. They are not ALJs. They're
00:17:36 --> 00:17:39 appointed by the attorney general. Because they're
00:17:39 --> 00:17:41 not ALJs, they don't have the protections under
00:17:41 --> 00:17:44 the Administrative Procedure Act that the ALJs
00:17:44 --> 00:17:46 do. There's no removal protection. There's no
00:17:46 --> 00:17:49 separation of function protection there. If they
00:17:49 --> 00:17:51 don't have those things, what kind of protections
00:17:51 --> 00:17:54 do they have? Until March of this year. The answer
00:17:54 --> 00:17:55 was they at least had civil service protections.
00:17:56 --> 00:17:58 IJs, which is what we generally call these immigration
00:17:58 --> 00:18:01 judges, are in the accepted service. They're
00:18:01 --> 00:18:05 covered by 5 U .S .C. 7513, which is the ordinary
00:18:05 --> 00:18:09 adverse action framework that applies to every
00:18:09 --> 00:18:11 other general federal employee. A proposed removal
00:18:11 --> 00:18:13 had to go through pre -termination procedures.
00:18:14 --> 00:18:17 The IJ could appeal to the Merit System Protection
00:18:17 --> 00:18:19 Board. The agency had to show cause for why they
00:18:19 --> 00:18:22 were doing it. This wasn't the heightened ALJ
00:18:22 --> 00:18:25 standard. The Merit System Protection Board case
00:18:25 --> 00:18:28 law had long held that for ALJs, you can't be
00:18:28 --> 00:18:30 removed based on how you've decided a case. And
00:18:30 --> 00:18:32 that was never the same for IJs. They could be.
00:18:32 --> 00:18:35 But there was still some sort of ordinary civil
00:18:35 --> 00:18:39 service protection framework. And AGs had historically
00:18:39 --> 00:18:41 pressured the system through assignment, through
00:18:41 --> 00:18:44 certification, and through selective firing of
00:18:44 --> 00:18:47 judges when they could find cause. The formal
00:18:47 --> 00:18:50 protections were at least a partial protection.
00:18:50 --> 00:18:54 What changed, though, in March? The MSPB decided
00:18:54 --> 00:18:56 a case where two immigration judges had been
00:18:56 --> 00:19:01 terminated by UIR on February 14th. So awesome
00:19:01 --> 00:19:04 Valentine's Day present for them. That was same
00:19:04 --> 00:19:06 day notice. There was no pre -termination procedures,
00:19:07 --> 00:19:10 no processes. They appealed to the Merit Assistance
00:19:10 --> 00:19:12 Protection Board. And initially it was heard
00:19:12 --> 00:19:15 by an administrative judge who ordered them reinstated,
00:19:15 --> 00:19:18 saying that they were entitled to those sort
00:19:18 --> 00:19:20 of civil service protections. that the agency
00:19:20 --> 00:19:22 hadn't provided them. But then the full board
00:19:22 --> 00:19:25 took it for review, reversed and dismissed the
00:19:25 --> 00:19:51 appeal. Here's what they said. It's pretty shitty.
00:19:52 --> 00:19:55 The protections existed on paper. The MSPB held
00:19:55 --> 00:19:57 the Constitution prevents them from being applied
00:19:57 --> 00:20:02 to IJs. Right. So this is building off of these
00:20:02 --> 00:20:04 cases we've had on inferior officers and presidential
00:20:04 --> 00:20:08 removal for a while. The board said that IJs
00:20:08 --> 00:20:10 are inferior officers and that they exercise
00:20:10 --> 00:20:12 too much policymaking authority to fit within
00:20:12 --> 00:20:15 an exemption that permitted removal restrictions
00:20:15 --> 00:20:19 on certain kinds of inferior officers. So what
00:20:19 --> 00:20:22 that means is that. They are now going to be
00:20:22 --> 00:20:24 considered at -will employees of the attorney
00:20:24 --> 00:20:26 general. This is the same effect as if they'd
00:20:26 --> 00:20:29 never had any protection. All right. But this
00:20:29 --> 00:20:33 is one MSPB decision. Is this where the law actually
00:20:33 --> 00:20:36 is or is this where the MSPB happens to think
00:20:36 --> 00:20:39 the law is? It's kind of both. Circuit courts
00:20:39 --> 00:20:42 have already found IJs to be inferior officers.
00:20:42 --> 00:20:44 That's the predicate move that the MSPB is relying
00:20:44 --> 00:20:49 on. And that can be appealed. But as of late
00:20:49 --> 00:20:53 May 2026. The operative law for adverse actions
00:20:53 --> 00:20:56 against IJs is that the MSPB has no jurisdiction
00:20:56 --> 00:20:58 over them. Whether that survives is a different
00:20:58 --> 00:21:01 question, but this is the Attorney General's
00:21:01 --> 00:21:04 practical authority. It's been exercised, and
00:21:04 --> 00:21:06 these people do not have a job. What does an
00:21:06 --> 00:21:09 at -will IJ actually do differently than a protected
00:21:09 --> 00:21:11 IJ, then? So these differences are going to show
00:21:11 --> 00:21:14 up in marginal cases. If we have an IJ who's
00:21:14 --> 00:21:16 leaning towards granting asylum on a borderline
00:21:16 --> 00:21:19 social group theory, but she knows she can be
00:21:19 --> 00:21:22 fired tomorrow with zero notice and no opportunity
00:21:22 --> 00:21:25 to respond. There's no formal mechanism through
00:21:25 --> 00:21:26 which the attorney general fires her because
00:21:26 --> 00:21:29 of the grant. There doesn't have to be. The grant
00:21:29 --> 00:21:31 is in the docket. The firing happens. And the
00:21:31 --> 00:21:34 next IJ in the courtroom watches and adjusts.
00:21:34 --> 00:21:36 So the doctrinal pressure doesn't have to be
00:21:36 --> 00:21:38 applied case by case. The structural fact that
00:21:38 --> 00:21:40 she can be removed tomorrow is the pressure.
00:21:41 --> 00:21:43 And that feels like a chilling effect that courts
00:21:43 --> 00:21:45 have been describing in the removal power of
00:21:45 --> 00:21:49 cases for hundreds of years. So it is exactly
00:21:49 --> 00:21:51 the reason that the protections existed. But
00:21:51 --> 00:21:55 there's also another parallel move with the ALJs.
00:21:55 --> 00:21:58 In February 2025, the Department of Justice filed
00:21:58 --> 00:22:01 a notice in pending litigation saying that it
00:22:01 --> 00:22:03 would no longer defend those ALJ protections
00:22:03 --> 00:22:08 against constitutional challenges. So the Justice
00:22:08 --> 00:22:10 Department was telling a court it won't defend
00:22:10 --> 00:22:14 the statute on the books. Yes. If we put these
00:22:14 --> 00:22:16 together, IJs have already lost their protection
00:22:16 --> 00:22:20 through this MSPB decision. The ALJs are losing
00:22:20 --> 00:22:22 theirs because the Department of Justice won't
00:22:22 --> 00:22:24 defend it anymore. And so the executive branch's
00:22:24 --> 00:22:27 position across the board is that 10 -year protections
00:22:27 --> 00:22:30 for adjudicators are unconstitutional. And the
00:22:30 --> 00:22:33 gap between IJs and ALJs that used to matter.
00:22:34 --> 00:22:37 is potentially already gone, at least if this
00:22:37 --> 00:22:40 is going before the MSPB. The only question is
00:22:40 --> 00:22:41 what the federal courts are going to do with
00:22:41 --> 00:22:45 it. And on top of all this, the performance evaluations
00:22:45 --> 00:22:48 for IJs can be and have been tied to case completion
00:22:48 --> 00:22:51 numbers. A couple of episodes ago, you mentioned
00:22:51 --> 00:22:55 the 2018 EOIR performance metrics, though. Right.
00:22:55 --> 00:22:59 So those were... temporary. They're not in effect
00:22:59 --> 00:23:01 now. But at the time, they were trying to make
00:23:01 --> 00:23:03 them decide 700 cases a year to get a satisfactory
00:23:03 --> 00:23:07 rating. And if they had fewer than 560, they
00:23:07 --> 00:23:10 were officially unsatisfactory. Those quotas
00:23:10 --> 00:23:12 were explicit. They applied to IJs personally,
00:23:12 --> 00:23:14 and they made case completion a primary factor
00:23:14 --> 00:23:17 in performance evaluations. The quotas were modified
00:23:17 --> 00:23:19 after litigation, but the underlying pressure
00:23:19 --> 00:23:22 towards completion has stayed. And it's taking
00:23:22 --> 00:23:24 different explicit forms in different administrations.
00:23:25 --> 00:23:28 And from the IJ's perspective, how do those two
00:23:28 --> 00:23:30 pressures then interact? They're simultaneously
00:23:30 --> 00:23:33 expected to be producing careful, legally defensible
00:23:33 --> 00:23:35 decisions on cases where the stakes are enormous.
00:23:35 --> 00:23:38 They might be life or death in some cases. And
00:23:38 --> 00:23:39 to complete enough of them to meet performance
00:23:39 --> 00:23:43 expectations. And like I said earlier, a careful
00:23:43 --> 00:23:46 asylum hearing takes hours. A less careful one
00:23:46 --> 00:23:49 takes less time. So if you're doing three -hour
00:23:49 --> 00:23:51 hearings for everything, you're not going to
00:23:51 --> 00:23:53 meet your numbers. If you're doing one -hour
00:23:53 --> 00:23:55 hearings, you're going to miss a lot of factual
00:23:55 --> 00:23:57 development. You might miss credibility nuances.
00:23:58 --> 00:24:00 You might miss things on federal review that
00:24:00 --> 00:24:02 could be sent back for further processing. But
00:24:02 --> 00:24:04 you're at least going to get through more of
00:24:04 --> 00:24:08 them. So these incentives don't line up. Right.
00:24:08 --> 00:24:10 And the IJ has to navigate them with essentially
00:24:10 --> 00:24:14 no institutional buffer. Social Security ALJs
00:24:14 --> 00:24:17 navigate similar pressures. And they used to
00:24:17 --> 00:24:19 have stronger removal protections. There was
00:24:19 --> 00:24:22 a union that kept litigating these issues. And
00:24:22 --> 00:24:25 a case mix where credibility stakes, while real,
00:24:25 --> 00:24:27 it really does matter if people get social security
00:24:27 --> 00:24:30 disability, are not life or death in the immediate
00:24:30 --> 00:24:33 sense that asylum cases can be. We've also seen
00:24:33 --> 00:24:36 overall denial rates increase dramatically. In
00:24:36 --> 00:24:39 February 2026, the denial rate had climbed to...
00:24:40 --> 00:24:44 about 80 % up from about 50 % in 2023. That has
00:24:44 --> 00:24:46 already shifted dramatically against granting
00:24:46 --> 00:24:50 it in the last year. Let's go back to Rosa, though.
00:24:50 --> 00:24:53 2028 hearing, random immigration judge draw,
00:24:53 --> 00:24:58 SSFM is the rule. Okay, so she walks into a courtroom
00:24:58 --> 00:25:00 in 2028 to argue a social group theory that the
00:25:00 --> 00:25:02 attorney general has already foreclosed on facts
00:25:02 --> 00:25:06 similar to hers. The doctrinal ceiling on her
00:25:06 --> 00:25:08 particular theory has been lowered. If the judge
00:25:08 --> 00:25:10 denies it, she can appeal to the BIA, which is
00:25:10 --> 00:25:13 going to be bound by this attorney general decision.
00:25:13 --> 00:25:15 And she may eventually get to the Ninth Circuit
00:25:15 --> 00:25:17 or the Fifth, depending on the venue she's in.
00:25:17 --> 00:25:20 But she's litigating on the agency's terms under
00:25:20 --> 00:25:23 the agency's rule with six -year -old evidence
00:25:23 --> 00:25:26 about a country she fled six years ago. So the
00:25:26 --> 00:25:29 same statute her lawyer would have argued in
00:25:29 --> 00:25:32 2022 created a different rule with a different
00:25:32 --> 00:25:36 country conditions and a different judge. Yes,
00:25:36 --> 00:25:39 that's the machinery. And we will stop here.
00:25:39 --> 00:25:42 Next time, we'll move from the machinery to the
00:25:42 --> 00:25:44 person standing inside it. So we're going to
00:25:44 --> 00:25:47 look at Elena, who had just eight months to prepare
00:25:47 --> 00:25:50 in a Louisiana detention facility, what that
00:25:50 --> 00:25:52 case looks like, why representation matters so
00:25:52 --> 00:25:55 much, potentially more than anything else in
00:25:55 --> 00:25:57 the federal system, and why the Constitution
00:25:57 --> 00:26:01 doesn't fix any of it. So that does it for today's
00:26:01 --> 00:26:03 episode on administrative remedies. Thank you
00:26:03 --> 00:26:06 for joining us today. Please, if you enjoy this
00:26:06 --> 00:26:09 podcast, enjoy this episode, give us a like on
00:26:09 --> 00:26:12 Spotify, iTunes, or whatever platform you're
00:26:12 --> 00:26:15 listening on. And be sure to tune in next time
00:26:15 --> 00:26:17 where we'll continue to dive into the contours
00:26:17 --> 00:26:19 of administrative law. Because remember, you
00:26:19 --> 00:26:20 can't fix what you don't understand.