Kevin Flanagan and the Displacement Nobody Wants to Name

Vice President JD Vance stood at a podium in Middletown, Ohio last week and told a story: a computer programmer named Kevin Flanagan, ordered to train the foreign worker who would replace him just to collect his own severance — who went to work, finished the job, walked to his truck, and killed himself.

Within a day, the story had done what these stories usually do online: it became a fight about tone. Columnist John Podhoretz responded by asking, sarcastically, what explains “the 2 billion other people” who face career setbacks and don’t take their own lives. Donald Trump Jr. fired back, framing the dismissal as proof of “soulless globalism.” By Saturday afternoon the tweet had close to half a million views, and the actual policy question — the one that should be the story — had mostly disappeared into a personality fight on Twitter/X.

We’re not interested in re-litigating who was ruder to whom. We’re interested in the thing that keeps getting lost every time this happens: the practice at the center of the story is real, it’s legal, and it happens constantly.

The training-your-replacement pipeline is not a myth

Search our H-1B databases for any given year and you’ll find thousands of Labor Condition Applications tied to IT and back-office roles at banks, insurers, and Fortune 500 firms — filed by staffing and consulting firms that specialize in exactly this transition: bring in visa labor, have the outgoing American employee document their own job during the handover period, then let the position go. It’s been documented in Congressional testimony, in Department of Labor disclosure data, and in firsthand accounts from displaced IT workers going back over a decade — Disney’s 2015 layoffs being the most widely reported example, but far from the only one.

None of this requires believing every employer acts in bad faith, or that every foreign worker on an H-1B or L-1 visa is complicit in displacing an American colleague — most are not, and the visa holder rarely has any say in how their employer structures a transition. The problem is structural, not personal: a legal framework that lets a company reduce labor costs by rotating out its existing workforce, using disclosure loopholes and prevailing-wage calculations that consistently favor the incoming hire over the outgoing one.

Kevin Flanagan’s case is not new, and it is not a rumor

Vance didn’t invent this story, and it isn’t from 2026. Kevin Flanagan died in 2003, in the parking garage of Bank of America’s Concord Technology Center in Walnut Creek, California, after the bank required him and his colleagues to train the H-1B workers replacing them in order to qualify for severance. His father, Tom Flanagan, later said losing that job was the defining event behind his son’s decision to end his life. The case drew protests outside the same office that year, and it’s documented in Donald Barlett and James Steele’s book Betrayal of the American Dream, among other sources.

It also wasn’t a one-time event at that company. Roughly three years later, Bank of America ran the same process on another wave of tech support workers — reportedly telling them severance was contingent on training their own replacements first. And it isn’t unique to Bank of America: the same structure — American workers required to train foreign replacements as a condition of severance — has been documented at Siemens, Disney, and elsewhere over the past two decades, generally involving the H-1B and its lesser-known cousin, the L-1 visa.

What the data actually shows

This is the part of the story that doesn’t fit in a tweet, and it’s the part we try to make visible. When you can search actual LCA filings by employer, by wage level, by worksite — not press releases, not anecdotes, but what companies filed with the Department of Labor under their own signature — the “isolated incident” framing collapses. Flanagan’s case in 2003 and the layoffs happening right now describe the same pattern: a wage tier low enough to make replacement financially attractive, a transition period that doubles as an unpaid handover, and a severance structure that quietly makes cooperation a condition of the payout. That pattern didn’t end when the cameras left Concord, California in 2003 — it’s still visible in the filings today, if anyone bothers to look.

This isn’t abstract to me

 

I didn’t come to this story as a journalist. I lived the front half of it.

I went to work in software in 1976 and never missed a day until 2003. From 2003 to 2010, the only work I could find was small, scattered projects. From 2010 to 2016, nobody would hire me at all — not for a job I’d spent decades getting good at. I did what I was told to do: I completed two federal retraining programs, the VRAP and the VRRAP, both funded by the Department of Labor and the Department of Veterans Affairs, both of which came with a guarantee of employment assistance afterward. It didn’t help. I finished those programs and went right back to unemployment.

Here’s what I couldn’t get past: I’m an American software developer who can’t find work in his own field, while the same federal government that ran my retraining program certifies thousands of H-1B visas a year for the exact kind of job I used to do — sometimes at the very agencies that trained me.

So I built a database to prove it wasn’t just me. That database became GuestWorkerVisas.com. I’ve been researching H-1B, H-2A, and H-2B filings independently, one person, for going on two decades now. Journalist Tanul Thakur documented my story and others like it in his book Wild Wild East, and back in 2020 a writer for The American Mind, Pedro Gonzalez, interviewed me for a piece connecting my experience to Kevin Flanagan’s — Gonzalez described me, not unkindly, as feeling like the “Rodney Dangerfield” of this fight: the guy who’s been saying this for twenty years and getting no respect for it.

I bring this up not to make the story about me, but because Kevin Flanagan isn’t a one-off, and neither am I. The pattern he died over in 2003 is the same pattern that quietly ended my career the same year, and it’s the same pattern still showing up in the H-1B filings I comb through today.

The reform question buried under the tweet fight

Whatever you think of how Podhoretz or Trump Jr. framed their exchange, both of them are, in their own way, reacting to something real: a legal mechanism, still in active use, that turns severance pay into leverage for a company’s own convenience. The question worth asking isn’t who was ruder online — it’s why, more than twenty years after Kevin Flanagan died, the same disclosure gaps that let his story go unnoticed until his family and coworkers spoke up are still standing. LCA filings are public, but they’re not easy to search or connect to real outcomes, and severance agreements that condition pay on training a replacement aren’t required to be disclosed at all.

Worth noting

Not everyone who works on this issue agrees on the fix. Some economists argue the wage and displacement effects of H-1B and offshoring programs are smaller and more contested than advocates on either side claim, and that visa reform is a poor substitute for broader labor-market policy. That’s a fair empirical debate, and one we’ll keep covering with the data as it comes in — but it’s a different question from whether displaced workers deserve to see the paper trail of their own displacement. On that narrower point, the case for disclosure is hard to argue against.

Discussing H-1B on Twitter and YouTube: A Year in the Analytics

Discussing H-1B on Twitter and YouTube: a year in the analytics

A year of posting H-1B/guest-worker-visa reform content on X and YouTube felt like it stopped landing sometime in early 2026. Rather than guess why, this pulls the platforms’ own analytics — account and video-level exports from X, and full lifetime YouTube Studio data — to see what actually happened. Kept here as a reference to check back against as the next round of posting (including bringing back “jobs direct”) gets underway.

X reach per post, Aug 2025 vs. Jul 2026
8.0K → 1.7K
Impressions per post, at nearly identical posting volume (452 vs. 454 posts/month).
YouTube lifetime (since May 2022)
25,845 views · 301 subs
227 videos published. Growth concentrated almost entirely in Nov 2025–Feb 2026.
Best subscriber-converting video
51 of 301 subs
One video, “jobs direct,” drove 17% of the channel’s entire lifetime subscriber count.
X (Twitter) reach per post, by month

Impressions per post, not just total impressions — so this isn’t a story about posting less. August 2025 and July 2026 both ran almost exactly 450 posts for the month; reach per post fell roughly 5x anyway. The biggest single-day spikes (300–460K impressions) landed on days tracking real national H-1B news moments — the fee/lottery reform fights in September 2025 and January 2026 in particular — which is a “riding the news cycle” signature more than a steady baseline getting throttled. The renewed dip from February through May 2026 coincides with running for U.S. Senate and a stretch spent upgrading recording equipment rather than posting — both of which pulled time away from the account directly, separate from anything the platform’s algorithm was doing.

0K2K4K6K8K8.0KAug7.1KSep5.5KOct3.5KNov1.9KDec5.0KJan2.7KFeb1.8KMar1.2KApr1.4KMay2.4KJun1.7KJul
Impressions per post (thousands), Aug 2025 – Jul 2026
X-native video views, by month

X tracks video separately from overall post impressions. The pattern is the same shape, but the mechanism is clearer here: 19 of the year’s 30 X-native videos were posted in December 2025 alone, then posting nearly stopped (1 in May, 2 in June, 4 in July). The two best individual clips — 10,539 and 5,098 views — are an 11-second and a 17-second clip posted the same day, Jan 25, 2026, which is real proof that short vertical clips are the one format that has reliably outperformed everything else tried on the account. This is a supply problem more than a reach problem: the format works when used, it just wasn’t kept up.

0K10K20K30K40K50K8.7KAug52.6KSep32.3KOct28.4KNov41.2KDec15.9KJan1.7KFeb4.0KMar49Apr412May1.0KJun2.2KJul
X-native video views (thousands), Aug 2025 – Jul 2026
Plan: resurrect “jobs direct” in the next few days — it’s the single best subscriber-converting piece of content across either platform (see below), and reviving it deliberately, rather than letting it sit, is the highest-leverage next move available.
YouTube: where views actually come from

Lifetime totals since the channel started May 12, 2022: 25,845 views, 448 watch hours, 301 subscribers, across 227 videos. Nearly all of that growth landed in the same Nov 2025–Feb 2026 window as the X surge — the same campaign-driven pattern showing up independently on a second platform. The traffic-source breakdown below is the more useful finding: YouTube’s own recommendation system (“Browse features,” the home feed) is the second-largest source at 34.6% of all views, run at a healthy 5.1% click-through rate — the algorithm is actively distributing this content, not burying it. “External” (39.1%, the largest single source) is traffic from links shared elsewhere — meaning reach still depends heavily on personally driving people there. The one genuine weak point: “Suggested videos” gets shown 25,089 times but converts at only 0.9% click-through, the lowest of any source — the algorithm is offering the exposure, the thumbnail/title isn’t winning the click in that context.

External (links you share)10,109 (39.1%)Browse features (home feed)8,954 (34.6%)Channel pages1,914 (7.4%)Direct or unknown1,809 (7.0%)Shorts feed1,128 (4.4%)YouTube search971 (3.8%)Other YouTube features348 (1.4%)Suggested videos334 (1.3%)Notifications160 (0.6%)Playlists115 (0.4%)
YouTube: top 10 videos by lifetime views

Four things stand out. “jobs direct” (#2) drove 51 of the channel’s 301 total subscribers — 17% of every subscriber the channel has ever had, from one video, despite a mediocre 2.2% click-through rate; whatever is converting viewers there is worth understanding and repeating on purpose. “346,000 Texas Kids” (#6) has the best retention by far at 70% average view completion on a 20-second clip — a hook that holds people to the end. “How to use the state map…” (#1) has the best click-through rate at 9.0%, double the channel average. And “Between Foreign Workers and American Workers” (#10) shows the opposite problem: 457 views but only 5.4% retention — people click and leave almost immediately, the clearest example on the channel of a title/thumbnail overpromising what the video delivers in its first few seconds.

# Title Published Views Watch hrs Subs gained Avg view % Impressions CTR
1 How to use the state map to find employers who hire H-1B’s in any state Jan 8, 2026 2,372 66.4 5 24.9% 16,749 9.0%
2 jobs direct Aug 17, 2025 2,321 28.2 51 21.5% 835 2.2%
3 Ken Paxton says he’s tough on immigration and putting Americans first Feb 26, 2026 1,634 7.3 2 32.0% 628 4.0%
4 If you want jobs in Texas & America, please share with everybody including those not on social media Feb 3, 2026 1,553 14.0 2 51.7% 2,590 3.5%
5 H-1B History Files, Jobs and Sharia Law in Houston TX Jun 26, 2026 1,382 12.0 8 47.4% 23,735 4.4%
6 346,000 Texas Kids. Where are their jobs? Nov 15, 2025 957 5.4 3 70.0% 14,290 5.3%
7 Amazing how our population growth has been decreasing since we started sending our jobs overseas Dec 7, 2025 646 8.6 0 21.6% 8,202 5.8%
8 Lets talk about the United States Navy Mar 1, 2026 591 10.9%
9 How can I survive on social security of $1,380 per month? Dec 1, 2025 475 25.6%
10 Between Foreign Workers and American Workers, who gets the jobs? May 29, 2026 457 5.4%
What this actually means

Reach on both platforms tracks two things more than anything else: how much time was personally spent posting, and whether H-1B happened to be a national story that week. Neither platform’s data supports “the algorithm is hiding this” as the main explanation — YouTube’s own recommendation system is actively distributing this content at a solid click-through rate, and X’s biggest spikes line up with real news moments, not a stable audience getting cut off. What the data does show clearly: short video is the highest-leverage format on both platforms, a small number of specific videos already prove a winning formula (hook, thumbnail, subject) exists, and the 1,000-subscriber threshold that’s felt so far away is much more a function of posting consistency than of any ceiling being imposed from outside. That’s a more encouraging problem to have — it’s one a steady posting cadence can actually solve.

Data: X (Twitter) account and video analytics exports, Jul 31, 2025 – Jul 30, 2026 · YouTube Studio channel analytics, lifetime (since May 12, 2022), pulled Jul 30, 2026.