As AI companies search for more training data to improve their models, one company is offering old, printed books as an ideal source because they are guaranteed to be free of the very AI slop AI companies are producing. 

“The world's best AI training data is sitting on a shelf,” ISBNdb, a company that produces what it claims is “the world’s largest book database,” and that offers high-volume book acquisition services for AI companies, says on its site. “Books represent curated, peer-reviewed, domain-specific human knowledge, structured in a way no web crawl can replicate. Dense, edited, authoritative.” 

In one article on its site, ISBNdb explains that printed books published before 2022 are ideal for AI training data because they don’t include AI generated text. As the article correctly notes, much of the data that AI companies can scrape from the internet today is likely to include AI generated text, which could result in “model collapse,” a process by which AI models that are trained on AI generated data results in worse models that are more prone to errors. The article also notes that book authors who object to their writing being scraped for training purposes can now easily poison AI models by producing writing designed to manipulate and sabotage the resulting AI models. 

“Print books from the pre-LLM era are structurally guaranteed to be free of this contamination. That alone is a significant advantage [...]  “Physical books published before this date [pre-2022] are structurally clean of modern poisoning tools.”

ISBN stands for International Standard Book Number, the numerical commercial book identifier and barcode on the back of most books. For years, ISBNdb helped book sellers, libraries, and distributors manage their inventory and find and sell books, but the generative AI boom has made it valuable to AI companies. In addition to selling access to book metadata, ISBNdb now helps AI labs source bulk printed book purchases of between 1,000 to 1 million books per order. ISBNdb’s data makes it easier for AI companies to methodically acquire, scan, and turn printed books into training data while avoiding duplication. 

AI companies’ attempts to hoover up printed books for training data got wide attention in January after a copyright lawsuit from book authors against Anthropic revealed internal documents detailing its plan to obtain and scan millions of printed books, and destroy them in the process. The Washington Post article found that Anthropic was buying books from one company called Better World Books, one of several marketplaces where libraries, retailers, and individuals can sell their books. Google was recently sued by book publishers for similarly training Google Gemini on copyrighted books.  

ISBNdb advertises that it can keep the identity of AI companies secret. 

“Strict NDA [non-disclosure agreement] on every engagement,” ISBNdb’s site says. “Every project begins with a legally binding non-disclosure agreement. Your identity, strategy, and acquisition targets are never disclosed.”

ISBNdb notes that AI companies may not want to be caught destroying printed books during the scanning process. 

“The optics problem is real,” ISBNdb’s site says. “‘AI company destroys two million books’ is not a headline that generates sympathy.”

One professional bookseller who specializes in selling foreign language books on these marketplaces told me that, starting in April, he and other booksellers noticed a historic spike in sales. This bookseller asked to remain anonymous so he can continue to do business on these platforms. 

“I personally have mixed feelings about all of this,” the bookseller, who suspects he’s sold hundreds of books to AI companies for training data, told me. “It benefits me financially as well as by clearing out old inventory that is otherwise unlikely to sell. I’ve been well-suited for these sales with inventory from overseas and foreign language books. On the other hand, I don’t like the end-use, and I don’t like that uncommon books are being pulped.”

This bookseller said his inventory is full of rare, foreign language, and low circulation books, meaning that if they are destroyed in the process of becoming training data, they’ll be even harder to obtain. 

The seller told me that, normally, on a good week, he’d sell about 20 books. Since April, he has regularly sold hundreds of books a week. While the seller didn’t have clear evidence that the purchases were being made by AI companies, the purchases made him suspect that they were. First of all, he said, the kind of books he sells are specialized and are usually bought by schools and libraries. Purchases from these organizations have been trending downward because of reduced funding, he said. Bulk purchases also usually reflect interest in a specific topic, whereas the recent, very large purchases were of books that had little in common, except for the fact that they all had ISBNs. This seller also sells rare books that do not have ISBNs, and none of those were part of the bulk purchases. I have not seen any evidence that this bookseller’s recent sales were facilitated by ISBNdb or that the client was an Anthropic or another AI company.  

“It's not just the quantity, but the weirdness of the orders,” the bookseller told me. “I've had library orders before, and usually they're mostly confined to a single subject or maybe a slightly broader range of subjects. But basically, almost every library in the world has lost their budget. I know all the U.S. college libraries don't buy much anymore. The Australian libraries don't buy much anymore. The type of books [...] there's no rhyme or reason to it. Also, there's a total disregard for the price of the book. I've had some books that sold through this way that were [...] greatly overpriced. That's kind of a tell for AI because they have just so much money.” 

“Is it just me, or has there been an uptick in the number of AutoBuy orders since the tail end of last year?” one bookseller wrote on the forums for Alibris, another marketplace for selling books, in February. The AutoBuy function allows a customer to flag books they want to automatically purchase once they become available for sale on Alibris. “Any comment on what is happening? Is an AI going to read every single book? Any insight into how the selections are made? They seem to vary quite a bit in condition, format (hardcover and softcover), price and so on.”

“We have a couple of new bulk buyers that are scooping up trade books so lots of sellers are getting lots of orders,” Mike Feldman, director of client services at Alibris, responded. 

One bookseller told me that similarly large orders of books were coming through another marketplace called Biblio. Customers can provide Biblio with a spreadsheet of ISBNs they want to purchase and the company takes it from there. 

In June, a publication in the Netherlands talked to several rare booksellers who reported similar large bulk purchases they assumed were coming from AI companies. 

It’s hard to say for a fact that the books are being bought for training data and possibly being destroyed by AI companies because ISBNdb and book marketplaces like Biblio and Alibris keep the identity of the buyer hidden. Large bulk purchases of books are first sent to distribution centers where, for example, Alibris checks the quality of the books before sending them off to the client.  

Internal Anthropic documents about its plan to scan millions of books, revealed in the copyright lawsuit, don’t make clear why the company wanted to destroy the books in the process. A deposition of Tom Harvey, who Anthropic hired to lead the project and who previously helped create Google Books, shows that one company Anthropic contracted to scan the books was Datamation, which offers both “high volume destructive and non-destructive book scanning” services. In a destructive book scanning process, the spine of the book is cut so the pages can be fed into a scanning machine, which is faster and cheaper than non-destructive book scanning.

Regardless of its original intentions, the federal judge in the copyright lawsuit from authors against Anthropic, William Alsup, found that Anthropic’s creation of digital copies of the books was legal specifically because the books were destroyed. 

“Here, every purchased print copy was copied in order to save storage space and to enable searchability as a digital copy,” Alsup wrote in his ruling. “The print original was destroyed. One replaced the other. And, there is no evidence that the new, digital copy was shown, shared, or sold outside the company.” 

This, Alsup said, was “clearly transformative” and therefore qualified as fair use under Section 107 of the Copyright Act. 

ISBNdb’s site advertises this legal argument to AI companies as well. 

“Purchasing paper books at scale from the secondary market does not deprive any creator of income they would otherwise have received,” ISBNdb’s site says. “These are books that have already fully discharged their financial obligation to their creators.”

“Responsible physical sourcing is not book burning,” ISBNdb’s site in a section about why it’s crucial to recycle the destroyed books. “It is the completion of a book’s lifecycle: from tree to knowledge to tree again.”

ISBNdb and Anthropic did not respond to a request for comment.

Apple says it has fixed a vulnerability in its Hide My Email feature which let essentially anyone figure out a user’s real email address which was supposed to be protected by the feature. Apple only fixed the vulnerability after 404 Media wrote about it at the start of July, despite Apple knowing about the issue for more than a year.

The news also follows the filing of a class action lawsuit against Apple over the vulnerability.

On Wednesday Apple told 404 Media it deployed a patch for the issue on July 3, which the company says has fully resolved the issue. 

Hide My Email is part of Apple’s paid iCloud+ product. It lets customers quickly create a new, anonymous email address they can then use to sign up to websites, services, or email people with. The generated email addresses typically contain two random words followed by a number and the @icloud.com domain. I use it heavily so hackers may have a harder time cross-referencing my activity and accounts across data breaches, for example. 

Tyler Murphy, co-founder of EasyOptOuts, discovered he was able to find the real email address of Hide My Email users. At the time, Murphy said, “We don't know the full scope of the issue, but in our limited tests with volunteers, 100% of Hide My Email addresses were exploitable.” That included mine, which we tested.

Murphy first reported the issue to Apple in June 2025. Over the subsequent months, Apple said it was looking into the issue and said it had fixed it; Murphy found it was still exploitable; and Apple again said it was looking into it. Murphy, thinking Apple may not fix the issue at all, then contacted 404 Media, around a year after Apple learned of the vulnerability.

When 404 Media first covered the issue several weeks ago, we did not include any details on how it worked because Apple had not fixed it. Meaning, if we published more specifics, third parties might figure out how to exploit it and reveal peoples’ real email addresses.

Now Apple says it has been fixed, we can add that, in simple terms, it required sending a target Hide My Email user a message that got rejected as spam. “We don't know how often hidden email addresses were leaked in email logs. For many major email hosts, the leak was triggered simply by an email being automatically rejected as spam, even if it was a legitimate message. Such emails probably didn't make it to your inbox, so you can’t review your spam folder to learn whether you were affected,” Murphy and EasyOptOut co-founder Ben Weiner said in a new statement.

“The bug that caused Apple's Hide My Email to leak hidden email addresses to senders has been fixed. However, we don't think the risk to Hide My Email users has been eliminated. Because non-malicious emails could bounce, revealing your hidden email address, and because mail transfer logs are often retained, we'd assume that any hidden email address linked to a Hide My Email address created before July 7, 2026, may have been exposed and could still be in third-party logs,” they added.

The class action lawsuit against Apple seeks full recovery of the subscription costs customers paid for the feature and an injunction against Apple for its “deceptive conduct,” PCMag reported.

When this blog started, back in 2008, the idea was modest: collect anecdotes. Stories from the trenches of a housing market that had, even then, begun to detach from anything a sober person could call fundamentals. “He said, she said.” “Profiting from the boom.” “Where do buyers get the money?” We built categories for them and filled those categories, over the years, with thousands of posts and something like a quarter of a million comments. The blog became, almost by accident, a diary of a mania — a slow-motion social document of what a whole city does when it convinces itself that shelter is a lottery ticket.

At the end of 2009 we broke our own rule. Instead of archiving somebody else’s story, we posted a prediction: that a real estate bear market would be Vancouver’s defining social and economic event of the coming decade. We laid it out like a play in three acts — the background stories (“Best Place On Earth,” “Running Out Of Land,” wealthy foreign buyers, “Real Estate Always Goes Up”), the stage-setting conditions (rent-to-price and income-to-price ratios at historic extremes, both suggesting housing was more than twice overvalued), and the coming action (rising supply, tightening credit, price falls, seller panic, a multi-year grind back toward fundamentals).

We gave that scenario an 80% probability. We were, as we’ve wholeheartedly admitted since, spectacularly early. This post is an attempt to ‘mark our own homework’: out loud, against the numbers as they actually stand in the summer of 2026.

First, the confession we’ve already made

There’s no dressing this up, and we won’t try. The “coming decade” we named was 2010–2019. Prices did not collapse in that window. They roughly doubled. The very crisis that should have pricked the bubble — the 2008 global financial meltdown — instead delivered a bailout Vancouver didn’t need, in the form of emergency-low interest rates in early 2009. That was the first booster shot. The COVID free-money era of 2020–2021 was the last one, and it launched prices into what we called, at the time, the stratosphere.

So the bears — us included — spent a decade and a half being early, which in market terms is indistinguishable from being wrong. We take our lumps. As we put it when the blog stirred back to life in late 2025: our bearish calls were early “by something between 6 and 15 years, depending on how you measure.” A commenter in 2018 wrote in to tell us flatly that “all the doomsayers have been proven wrong.” At the time, he had a point. Anyone who sat out the 2010s waiting for our crash paid dearly for the privilege — in rent, in anxiety, and in the corrosive “coulda-shoulda-woulda” that we warned all along was the second lesson every market participant has to learn to survive.

We say all of this first because a retrospective that only counts the hits is propaganda, not analysis.

And now, the part where the thesis holds up

Here is where it gets interesting — because getting the timing catastrophically wrong is not the same as getting the structure wrong. And on structure, the record reads very differently.

The core claim of this blog was never really “prices crash next year.” It was something more durable: that Vancouver housing was priced two to three times above the level that local incomes and rental yields could justify, that this gap was the product of a speculative mania rather than genuine scarcity, and that the reconciliation with fundamentals — whenever it came — would be the defining economic story of the city. Let’s test each piece against 2026 data.

On overvaluation. The 2026 Demographia International Housing Affordability report gives Vancouver a “median multiple” — median house price divided by median household income — that has hovered around 11 to 12 in recent years, placing it among the four least affordable major markets on earth, behind only the likes of Hong Kong and Sydney. Demographia calls anything above nine “impossibly unaffordable.” The historically-normal, genuinely-affordable multiple is three to five. In other words, the independent international benchmark now says Vancouver is roughly two-and-a-half to three times overpriced relative to income — which is, almost to the decimal, the number this blog was shouting into the void back in 2009 and 2010. National Bank’s affordability monitor tells the same story from another angle: as of early 2026, servicing a mortgage on a typical Vancouver home still eats about 82% of median household income, against a long-run norm closer to 40%. The overvaluation we described was real. It was never a hallucination of “permabears.” The metrics were always there; we just read them a decade too early on timing.

On the reconciliation, now underway. This is the headline. As of June 2026, the Real Estate Board of Greater Vancouver’s composite benchmark sits at $1,099,100 — down 6.0% year-over-year, and roughly 12% below the April 2022 peak of $1,252,800. That top-line number understates what’s happening underneath it:

– Condos have been sliding for more than four years. Rennie’s economists peg the segment at around 11% off its spring-2022 peak and still falling — “more than 48 months into this downturn.”

– The Fraser Valley — the suburban frontier where the mania spread last and hardest — now sits about 26% below its 2022 peak.

– Presale buyers who signed at peak prices are facing $100,000 to $500,000+ losses on assignment at completion, with some projects marked down 15–25% from original pricing.

– Sales hit 25-year lows in 2025, inventory has climbed to multi-year highs, over 80% of homes sold below asking, and days-on-market have stretched out. This is not a healthy, tight market taking a breather. It is a market where, as we wrote in point 29 of last December’s “33 Concepts” post, the buyers’/sellers’ indicator will “stick to the wall all the way down.”

And the West Side detached mansions — the trophy assets, the ones most freighted with 2022 optimism — are the tell. Houses selling for less in nominal dollars than they fetched in 2013, 2015, and 2016. Adjust for the roughly 28–30% of compound inflation over that decade, and a buyer who paid $4.5 million in 2016 and sold for $3.1 million in 2026 has lost something like 47% in real terms — before carrying costs. The reconciliation with fundamentals we always said would be “brutal, in all segments” is no longer a forecast. It’s a data series.

The most remarkable development of all

If you had told the 2009 version of this blog that one day the Bank of Canada itself would say, out loud, that “there is no path to affordability in Canada without home prices coming down” — we would not have believed you. The entire architecture of the bubble, as we catalogued it for seventeen years, depended on vested interests insisting the opposite: that prices must always rise, that any softening was a buying opportunity, that a “soft landing” was not only possible but assured.

Yet here we are in 2026, with the central bank conceding the quiet part, with the CMHC having earlier called for prices to fall, and with mainstream columnists writing matter-of-factly about a decade of stagnation at the top end. The narrative has flipped. That flip — the culture finally releasing housing from its grip on the collective mental space, “RE vacating social dialogue” as we phrased it in Act Three — may be the single most important thing to have “come to pass” since we started writing.

So where does that leave us — and the thesis?

The bubble did not “pop” in the cinematic sense we half-expected in 2010. There was no single 2008-style crash. Instead, after two undeserved bailouts extended the mania by fifteen years, we are now several years into exactly the kind of grinding, multi-year, sentiment-driven deflation that we described in Act Three and again in last winter’s 33 points — prices reverting toward fundamentals not in a weekend but in a long, wave-upon-wave descent. Rates have been cut (the overnight rate is down to the mid-2% range from its 2023 peak), which in any prior cycle would have relit the fire. This time it has produced only stabilization and sideways drift, not a new boom. That, more than any single price print, is the evidence that something is structurally different — that the psychological engine of “buy now or be priced out forever” has, for now, stalled.

Are we declaring victory? No. We got the timing so wrong that we forfeited the right to any victory lap, and we said as much in point 33 of the concepts post: not so fast — we could be wrong, yet again. Perhaps prices turn on a dime and rocket from stratosphere to orbit.

But the shape of what we described — overvaluation of roughly 2–3x, driven by local leveraged speculation dressed up in scarcity narratives, unwinding into a prolonged reconciliation that reshapes the city — is no longer a fringe bear fantasy. It is, increasingly, the base case of the Bank of Canada, the CMHC, and the Vancouver Sun.

To the old regulars, the bears who somehow made it through all of this: how ya doin’? Pull up a chair. After seventeen years, the beast has finally woken up. It was never going to be pretty, and it isn’t. But for the first time, the numbers are moving in the direction we always argued gravity demanded.

As ever: a home is first and foremost a place to live. If this whole long, painful cycle teaches the city that one thing, it will have been worth documenting.

— vreaa

—

A note on sources: current figures in this post are drawn from the Real Estate Board of Greater Vancouver / Greater Vancouver Realtors June 2026 release, WOWA.ca’s market summary, the National Bank Housing Affordability Monitor (Q1 2026), the 2026 Demographia International Housing Affordability report, Rennie market commentary, and reporting in the Vancouver Sun and Business in Vancouver. Figures are benchmark (HPI) prices unless otherwise noted; averages differ and can be skewed by sales mix.

Click here to go see the bonus panel!

Hovertext:
It's not an uncanny valley, it's an uncanny mountain that you're afraid to climb.

Today's News:

I would say that this is kind of the point where I started to recognize my own tendency toward not taking pictures, and where I started to try to actively remind myself to work on that. As such, there will be more to see here, but I don’t guarantee that the pictures will all be good, or even worth it.