The four days
A man with more than a million followers was announced dead on August 3. For four days, no authority anywhere had confirmed anything about it. That gap did not stay empty.
On Monday, August 3, 2026, a woman named Alxzondra posted to Instagram that her partner, Joe Felz, had died. Felz was 40 — an Augusta, Georgia real-estate investor turned van-life and off-grid creator, founder of Felz Investment Group, and the man behind FelzBooks, a project to scan and give away rare Bibles and Masonic texts. Her post was the only source in the world for the news. Every professional report that followed traced back to it.
Four days later, on Thursday, August 7, two news organizations reported that officials had determined how he died. TMZ attributed a finding of suicide to the Georgia Bureau of Investigation Medical Examiner's Office. People attributed the same finding to the Paulding County Coroner's Office. Two outlets, two different authorities, in agreement.
It would be easy to call that the end of it. It is not, and the reason matters more than the story: neither outlet published a document, and neither office has said anything publicly. What exists is two reporters relaying what someone in an office told them. No death certificate. No autopsy report. No coroner's statement. No press release. Nothing a reader can go and check.
That is a real distance from proof, and this piece is going to keep it visible rather than round it off. A report of a record is not a record. Two reports of a record are not two records — they are two reports. The whole subject of this page is what happens when people stop making that distinction.
Three things about those four days are worth setting down plainly, because they are documented, they are strange, and most of them will not survive in the record once the story scrolls away.
The first: a cluster of automated obituary websites published, on August 3, the same answer that officials would be reported as reaching four days later. They had no source for it. Their own pages said so, in the body text, underneath headlines that stated it as fact.
The second: nearly every outlet that covered the death said he was living in Costa Rica. Costa Rican immigration records say he left the country on April 30 and never came back. He said the same thing himself, on camera, twice.
The third: money is still moving. A subscription website built on his book-scanning project is live and charging today. Nobody has said who operates it or receives the money.
None of that is evidence of a conspiracy, and this piece does not argue for one. What it documents is stranger and more useful than a conspiracy, and it is the reason this page exists:
The rumour mill and the professional press failed in opposite directions on the same story — and the guessers came out looking right.
On August 3, content farms with no source, no named authority, and headlines contradicting their own body text published a cause of death. Four days later, officials were reported as having reached the same conclusion. Those sites will be treated by their own metrics as having been first and correct.
They were not reporting. They guessed the most common answer and the guess landed. A template that always picks the likeliest manner of death will be right most of the time, and being right by accident does not convert a guess into journalism. Had the finding come back differently, the same pages would have carried the same headline.
Meanwhile the outlets that do employ reporters, fact-checkers and standards editors — TMZ, The Sun, and every aggregator downstream of them — told the world for four days that Joe Felz had been living in Costa Rica. Official immigration records and two of his own videos say he left on April 30 and never went back. The one outlet that had it right, People, called him a Georgia-based influencer in a single clause nobody picked up. The correction, when it came, came from a Costa Rican paper writing in Spanish that no English-language outlet read.
So the machine that fabricates got the answer; the machine that verifies got the country. Neither of them did the work, and only one of them will be blamed. What follows is an attempt to say exactly what is known, exactly what is not, and how to tell the difference.
How to read this page
Every factual claim below carries a mark. The mark is a colour, a shape, and a word — never colour alone — so the page reads the same in greyscale, in print, and for colour-blind readers. Skimming only the graphics should leave you with the same picture as reading every word.
"Verified" on this page means a document, a record, or an artifact that was examined directly — a live page fetched and read, a dated archive capture, a government filing, a video of the man saying the words himself. It does not mean "widely reported."
An outlet reporting that an official told them something is graded Partial, however many outlets carry it and however reputable they are. That is not a slight against any newsroom; it is a statement about what the reader can check. A sourced claim you cannot inspect is a claim you are taking on trust, and the number of outlets repeating it does not change that — they may all be relaying the same phone call.
Applied honestly, this rule is uncomfortable. Nothing about the death itself sits in the verified column. Not the cause. Not the manner. Not the fact of the death. Not even the announcement that started it — that post is behind a login, no archive has ever captured it, and what exists is three newsrooms quoting the same screen.
To be plain about where that leaves things: there is still no publicly available proof that Joe Felz is dead. No death certificate, no autopsy report, no coroner's statement, no funeral, no obituary of record. There is a family announcement nobody outside a few newsrooms has seen, and there are two press reports of what officials said. That is the entire evidentiary basis, five days on, for a story the whole internet has already settled.
A report arrived. A record still has not.
Until Thursday afternoon, the honest description of this story was that a death had been announced by one person and confirmed by no one. What changed on Thursday is narrower than the coverage suggests: two reporters were told something by two offices, and published it. Nobody outside those newsrooms has seen anything.
Both reports are statements made by officials to reporters. Neither outlet published a document, and neither the GBI nor the Paulding County Coroner has issued anything public of its own. Some of that is ordinary — the GBI's medical examiner does not announce routine determinations, and releases results to the coroner of the county where the death occurred, who notifies next of kin. But ordinary is not the same as verified. What exists is two on-record attributions and no paper.
The two authorities are at least institutionally consistent rather than in conflict. Georgia's state medical examiner performs autopsies for 153 of the state's 159 counties at the county coroner's request. A Paulding County coroner referring a case to the GBI, and both then confirming the same finding, is the ordinary chain working as designed. That coherence is a point in the reports' favour. It is not a document either.
It is worth being blunt about how thin the independence really is. Two outlets carrying the same claim looks like corroboration and often is not: they may be relaying the same office, the same spokesperson, even the same phone call. Agreement between outlets measures how far a claim travelled, not how well it was checked — which is the identical error this page documents further down, one tier lower, where the content farms live.
What that establishes, and what it does not:
This investigation was researched while the cause of death was genuinely unknown, and it was originally going to lead with that vacuum. The August 7 reports arrived during the writing. Publishing the earlier framing would have meant publishing something false. The reporting below has been re-anchored on the official determination, and the material that survives is the material that is still true — the four-day gap, the location error, and the money.
If he died, why suicide?
A determination of suicide has been reported. It has not been shown. No autopsy report, no coroner's ruling, no death certificate, no statement from either office named — nothing a reader, a family member, or another journalist can examine. What exists is two reporters' accounts of what they were told.
That matters more here than it usually would, because the finding does not sit easily beside the record of the man's last months. None of what follows proves anything. All of it is the sort of thing a released document would resolve in a paragraph, and that no document has been released is the reason it is still an open question at all.
What he was doing, on the record, in the months before
These are dated commitments, not impressions.
The forward record
Obligations, plans and expenditures pointing past the date he died.
- 6 Nov 2025 — and still open when he diedHe was actively litigating to change a custody arrangementRichmond County case 2025RCD01043, a modification action, filed nine months before his death and carrying no disposition. He was in the middle of a legal process he had started. Verified clerk's docket, read here
- 7 Jan 2026Asked what mattered most, he named getting his children backHe described the path to it as uncertain, and made it his single stated priority. Partial — a translated, staff-bylined outlet.
- 9 Mar 2026Registered felzbooks.com and paid for five years, through 2031Not a monthly renewal. A five-year term, bought outright, for a project he then spent months building. Verified WHOIS
- 10 Jun 2026Back in Augusta running the business — and taking work for laterOn camera: tenants, service calls, "fixing my own houses just like old times." He mentions a house job offered in Alaska and a solar install in Tennessee, and frames the US stay as handling obligations before returning to Costa Rica. Partial — video title and date confirmed here; the content is as reported by the research pass.
- 26–27 Jun 2026Fortieth birthday, "celebrated simple and sweet with a few of the lifers"Partial
- to 9 Jul 2026Posting almost daily, across five platforms, to the endFifteen uploads between 31 May and 9 July alone — solar explainers, an affiliate phone review, gratitude posts, a piece about protecting the planet. Then three cross-posted items within half an hour on 9 July, and nothing. Verified YouTube channel feed, read here
The thing nobody has ruled out
There is one documented health event in the final month, and it has gone almost entirely unmentioned in the coverage.
In early July he posted a community update describing a spreading infection — redness and swelling on an elbow, then a new site at the crease of the opposite knee, on day three. He asked his followers whether it was lymph nodes or cellulitis, and signed off asking whether it meant "certain or uncertain demise." He was, characteristically, half joking.
Cellulitis that migrates to a second, non-adjacent site in three days is the presentation that sends people to hospital, and untreated it can become septic. His own father — a physician — had died nine weeks earlier. Nothing establishes that this had anything to do with his death. It is raised because it is the only documented medical event in the window, because it went unexamined by every outlet, and because it is precisely the kind of thing an autopsy report would address in a line.
Partial — the post is real and dated to roughly a month before the Aug 6 channel snapshot; its text is as recorded by the research pass. No location is stated in it.
The twenty-five days
Between 9 July and 3 August there is nothing. No post on any of five platforms from a man who had posted almost every day for years. His whereabouts in that window are unknown, his last self-declared location was the Augusta area, and no source accounts for it. The silence is the single largest hole in the timeline, and it is where any real explanation of his death would have to live.
The claim that he warned people — handled honestly
The most-repeated argument against the finding is that Felz had told his audience he would never take his own life, and to question everything if he ever died suddenly. The top comment under his family's statement, with thousands of likes, reads: "He told us specifically to question everything if he were to suddenly die."
This page does not treat that as evidence, and readers should know why. The underlying videos could not be located as primary footage by any research pass. What exists is people describing, after his death, what they remember him saying. Memory reshapes itself around an ending; the comment sections of dead men fill with prophecy. If the footage surfaces it becomes a real fact and this page will say so. Until then it is a claim about a claim.
Everything above describes a life pointed forward. It would be easy, and wrong, to conclude from it that suicide therefore makes no sense.
People who die by suicide very often have plans. They hold jobs, litigate custody, book flights, pay for things years ahead, post cheerfully, and make arrangements for a future they do not reach. The belief that forward-looking behaviour rules it out is one of the most common and most harmful misconceptions about how it happens — and it is exactly the reasoning that leads families to say afterwards that they saw nothing, because there was nothing of the kind to see. A five-year domain registration is not evidence of anything about a person's interior life.
So the argument this section makes is narrower, and it survives that objection intact: the problem is not that the finding is implausible. The problem is that it has not been documented. A determination carrying this much weight, in a death this widely discussed, has been asserted through two press attributions and nothing else — no report, no certificate, no statement from either office. The record around it raises ordinary questions, and every one of them is the sort a released document answers immediately.
Ask for the document. That is the whole of it.
What would actually settle this
- The GBI Medical Examiner's report, or a Georgia open-records response confirming its finding.
- A direct statement from the Paulding County Coroner — a sentence would do.
- A death certificate, which names the place and date neither outlet has reported.
- Toxicology, which would speak to the infection question as well as the determination.
- Any account of where he was between 10 July and 3 August.
- A single dated clip of him saying what thousands of people are certain he said.
The answer was published before it was known
This is the part of the story that does not go away, and it is the reason a page like this exists.
Within hours of Alxzondra's August 3 post — before any autopsy result existed, before any coroner had spoken to any reporter — a ring of automated obituary sites published pages stating that Joe Felz had died by suicide. Several put it in the headline. Several of those same pages then conceded, in their own body text, that no law-enforcement agency, medical examiner, or coroner had released any determination.
They were not reporting. They were guessing, in a template, at scale, for advertising revenue. Four days later, the guess turned out to match the official finding.
That match is not corroboration, and it is important not to read it as vindication. A site that asserts a cause of death with no source is not made into a source by being right. Suicide is among the most common manners of death for men in Felz's demographic; a template that always guesses it will be right often. These pages published identical text, copied from one another, with fake or absent bylines. They are one non-source repeated many times, and they would have published the same headline if the finding had come back differently.
The four-day gap
When the claim was published, against when it was determined.
"suicide" in headlines
"are unclear"
determination reported
The anatomy of the pages that got there first
The pattern is consistent enough across the cluster to describe as a form. The headline states a cause of death as fact. The body concedes there is no source for it. The attribution, where any exists, is to "public tributes" or "social media." The byline is generic or missing. The domain is a disposable TLD registered for exactly this purpose — .site, .online, .shop, .cfd. The text closely matches a dozen sibling pages.
At least ten such sites published on Felz. One of them, republished through NewsBreak under a "Family Confirms" headline, reached an audience that would have no way to distinguish it from a newsroom.
The specimens
Describing these pages understates them. Here is what they actually published, reproduced as artifacts — headline above, the concession from their own body text below it.
Joe Felz Suicide, How Did Joe Felz Die? Cause of Death and Obituary | Augusta GA
"…no law enforcement agency, medical examiner, or coroner has publicly released an official determination."
The headline states a cause. The body concedes nobody had determined one.
Joe Alx Felz death By Suicide, Augusta GA, Felz Investment Group Founder Obituary, August 3, 2026
Cause attributed only to "public tributes" and "social media." No named authority, outlet, family member, or partner appears anywhere in the piece.
The date and the cause are both in the URL slug, published the same day the family announced the death.
Joe Felz Suicide, Joe Alx Felz Death, Real estate investor, influencer and founder of Felz Investment Group has died
Near-identical body text to its siblings, on a different disposable domain.
Ten domains, one text. They corroborate nothing but each other.
Reproduced for provenance under fair use, as evidence of what was published and when. These sites are deliberately not linked. Headlines and body language are as recorded by the research pass; the domains, dates and bylines were confirmed against their published pages.
What fabrication looks like up close
The clearest evidence that these sites invent rather than report came out of a check on something else entirely. Several of them name Joe Felz's two children. The names are not in any newsroom's copy, not in any record, and — despite his very public life — not in anything Felz ever posted himself.
One of the two surnames they publish is not the family's surname at all. It is a near-miss spelling of the children's grandmother's maiden name, which appears in exactly one public document: the funeral-home obituary of Joe's father, published in May 2026. A middle name is attached to the other child that appears in no document anywhere.
So the mechanism is visible in the artifact. Someone took a real obituary, pulled fragments from it, recombined them into names that no source contains, and published the result as biography. It is not a transcription error — a transcription error does not move a name across two generations. It is manufacture, and it is the same process that produced a cause of death on August 3.
This page does not print those names, and does not print the children's names at all. They are minors; their father never named them publicly; and the specific names in circulation are demonstrably wrong. Naming them would mean either repeating a fabrication or, worse, correcting it — and a correction would put two real children's names into a suicide story on the authority of a document that was never about them. See the methodology note.
Alxzondra's second statement, on August 4, rejected "false narratives, assumptions, and speculation" and asked people to "honor the man instead of feeding the rumor mill." She was responding to an assertion that no identifiable person had made and no source stood behind — one that had already been copied across a dozen domains and pushed into aggregators. There was nothing to correct and no one to correct it with. That is the ordinary condition of a grieving family in 2026, and it is worth naming as a harm even in a case where the underlying guess turned out to be right.
Everything traced to one post
Before Thursday's determination, the entire public evidence base for this story was a single Instagram statement by one person. The volume of coverage was large. The number of independent confirmations was zero. Those are different measurements, and the difference is the whole point.
Source-credibility cascade
Bar width shows audience reach. Nothing below the top row added a source — each layer is repeating the layer above it.
3 Aug
4 Aug
4–6 Aug
3–7 Aug
The bottom row is the widest and the worst. It is also the only row that published a cause of death before one had been determined — and, through a NewsBreak republication, the row most likely to reach a reader who cannot tell it from a newsroom.
There is one genuine absence in this picture that no amount of syndication covers. Felz was a locally prominent figure in Augusta — a property owner quoted on camera by WRDW News 12 in 2022 about an arson fire on a house he owned. Not one Augusta newsroom covered his death. Not the Augusta Chronicle, not WRDW, not WJBF, not The Augusta Press, whose own site search for "felz" returns a single unrelated opinion column from 2025. Statewide outlets have nothing either.
That absence is real, and it is worth pinning down accurately rather than asserting loudly in either direction: Partial that no Augusta outlet covered it — the research pass found them all clear, but re-running the searches here was inconclusive, and an absence is the hardest thing to confirm. Unverified as to why.
One piece of geography takes some of the strangeness out of it. Augusta is in Richmond County. The coroner's office that reportedly confirmed the determination is in Paulding County — metro Atlanta, about 150 miles away, on the other side of the state. If he died there, the death would never have crossed an Augusta assignment desk in the ordinary course of things. That is an explanation for the silence, not a justification of it: he was a locally known figure whose company, rental properties and family were all in Augusta, and his father had been a Medical College of Georgia professor there. The content farms that place his death in "Augusta" are, once again, asserting a location no source supports.
He was not in Costa Rica
Almost every story written about this death said that Joe Felz had recently moved to Costa Rica and was living there with his partner. The Sun said it directly. Aggregators worldwide repeated it. It is the detail that made the story legible — the off-grid dreamer who went to the jungle.
It is wrong, and it was checkable the whole time.
Costa Rica's immigration authority, the Dirección General de Migración y Extranjería, holds entry and exit records for every foreign national. The Costa Rican outlet El Observador obtained them for "Joseph David Felz" and published the results on August 5, under the byline Sergio Arce. They show four crossings and a departure:
Costa Rica entries and exits, 2026
Official immigration record, query window 1 Jan – 6 Aug 2026.
- 9 January 2026Exit
- 1 March 2026Entry
- 7 April 2026Exit
- 21 April 2026Entry
- 30 April 2026Final exit — no recorded returnNothing further through 6 August, the end of the record's query window. He was outside Costa Rica for the last three months of his life.
He said it himself, on camera, twice
The immigration record is not the only evidence, and it is not even the most direct. Felz addressed the question in his own uploads, in public, while alive.
On June 10, 2026 he posted a video titled "Why am I not in Costa Rica? I'll tell ya." In it: "Everybody wants to know why I'm not in Costa Rica. What am I doing back in the States? … I got cars that break down. I got tenants that tear up my houses… I'm running around Augusta picking off service calls and fixing my own houses just like old times… sitting on this stump in the park instead of in the jungle of Costa Rica."
On July 9, 2026, in what became his last upload, he said: "since we've been back from Costa Rica the last 2 months" — which places his return at roughly early May, matching the April 30 exit exactly.
His June uploads carry #augusta tags. A community post from the same period asks whether any YouTube mechanics are "around the Augusta area" because "all my stuff is in Costa Rica."
The Costa Rica framing is why this story looked like a mystery. It is the reason people went looking for a statement from the OIJ, Costa Rica's judicial police, and found nothing. It is the reason the absence of a Costa Rican coroner's report, a consular record, and local Costa Rican coverage all looked like a pattern of official silence.
There was no silence. There was no Costa Rican record because there was no Costa Rican death. The apparent anomaly was manufactured by a reporting error that propagated through syndication, and it stood uncorrected for four days while people drew conclusions from it.
People magazine, alone among the major outlets, described him accurately: a Georgia-based influencer. It reads like a small detail. It was the whole thing.
No outlet in either language has yet reconciled this on the record. The Costa Rica News profile of Felz still carries an "Updated August 7, 2026" stamp and still describes him as an expatriate living in Costa Rica, with no mention of his death or his departure. El Observador implicitly corrected it — its headline calls him someone who "visited Costa Rica several times this year" — without confronting the American coverage directly.
What a real Costa Rican death looks like in the record
Before the location error surfaced, the most serious question on the table was whether official silence out of Costa Rica meant something. The way to answer that is not to argue about it — it is to check what happens when an American actually does die there.
It happened the same week, in the same country.
Comparative case — an American death in Costa Rica, same week
Time from death to public OIJ comment and local press coverage.
This graphic was built to test whether the silence was anomalous. It answers the opposite of what it was designed to answer, and that is worth stating rather than quietly dropping: the comparison shows the Costa Rican record behaving exactly as it should, on a case where there was nothing to record.