I am told that the ancient Romans scraped their tongues using a dull, metallic blade that could be inserted in the mouth, and raked along the surface of that organ with two fingers guiding the way. A fully appointed ancient Roman hygiene kit would have dangled on a chain with other essential items like a toothpick, a nail cleaner, and an earwax scoop.
Oh—I’m sorry. Do you find the history of human scraping disgusting? Oh, come now, it’s hardly unusual. As long as there have been humans, we have loved to qopchedy qokedydy qokoloky qokeedy qokedy shedy our filthy bits.
That is to say, humans are just that way. An earwax scoop, a nose-picking device, all the better to qopchedy qom‑lith.
Excuse me. I can’t help spewing gibberish these days. See what I’ve become? And all because I just know they’re watching every word I write.
I am talking about my scrapers. They hang on my every word, and I have learned that you cannot be too careful.
Qokoloky kav‑elthri!
I pay a company a princely sum of kolokys every year to host my website. They offer a handy app that allows me to log in from my phone so I can edit blog posts and webpages on the fly. One feature of that app is the ability to peruse my website analytics. About a year ago, I noticed that my monthly website visits had shot through the roof, especially on days when I posted something new on my blog. Prior to this, I maybe got 25 visitors a day.
Now, on a day when I drop some hot juicy content, such as my post entitled, “My Book on the Signers of the Declaration of Independence is $2.99 This Weekend!” or “Here is a Photo of Some Blueberries I Just Picked Off a Shrub in My Yard. No, Really!” I’ll log 400 visits in a 24-hour period.
Four hundred visits, from four hundred different IP addresses.
Let me cut to the chase here with a statement that may strike my SleuthSayers audience—you folks, who absolutely love me to pieces—as nutty, batty, and possibly dotty: I, Joe D’Agnese, am not that interesting.
| Scrapey, scrapey, scrape, says the little red flag. |
I mean, seriously. What the qokedydy is going on here?
There is no reason 400 individual, breathing humans on the planet would be interested in my publisher’s $2.99 weekend ebook sale or my sweet, juicy berries. Or, for that matter, photos of my dog wearing a bowtie, or my occasional announcements of articles appearing here on SleuthSayers.
Then I noticed that many of these visits were coming from one place in the world: China. Some days, my only visitors are from China. Currently, China is my No. 1 visiting nation with 1,778 visits in the last month. The USA, where I was born and am a citizen, is No. 2, with 615 visits. No way to sugar coat this: I am kind of a big deal in China.
I briefly thought about installing a service such as Cloudflare that allow website owners to block traffic on a nation-by-nation basis. What if malicious actors were trying to steal personal information from my website? Yikes! After all, I often do write wacky stories about my forebears.
Then, I did a Zoom chat with that friend of mine who runs an up-and-coming book review site. Relax, he told me, those are just bots scraping your content to train AI on your text.
Whether performed by bots or ancient humans, scraping is disgusting. Compared to this, training AI on pirated works downloaded off crappy free-book websites comes off as the genteel way to steal.
“Here,” my friend said, “lemme show you something.”
He shared his screen and proceeded to show me his Cloudflare. (Not a euphemism, I assure you.) Judging from the line graph I ogled, sometime in 2024, the number of visitors to his site quadrupled, quintupled, sextupled. When he installed Cloudflare and blocked Russia, China, and a bunch of other nations, guess what? The numbers kept climbing. Only this time, the visitors were coming from nations such as Canada, Belgium, and Luxembourg.
Problem fixed? I innocently opined.
“Nah,” he said, “they’re just using VPNs that make it look like they’re coming from ‘safe’ countries.”
My friend chuckles about this these days. He’s resigned to the fact that the more he shares excellent English prose in the form of book reviews, the more it will be greedily scraped by legions of bots who are feeding an insatiable desire to build machines that will replace us all and wreak havoc around the planet.
A contact at a national lab tells me that due to U.S. budget cuts, he and his superiors have had to lay off tons of older scientists studying the mitigation of climate change. In their place, they have hired newer, younger, less expensive engineers who know (something) about halting the influx of AI-inserted malicious code that they are discovering in their critical government systems.
Everyday, my engineer friend says he is pulling his hair out, muttering to the heavens, “Okytarl zexirami šol‑drih!”
Because he has never seen anything so terrifying. His last job was a cakewalk compared to this. And he led a medical supply firm during COVID.
I have begun to think that AI is a thing no one asked for (except cheapskates) and a growing problem no one knows how to solve. Pick your metaphor: A genie let out of a bottle. A new Pandora’s box.
I read a special AI issue of Wired magazine cover-to-cover. One thing I learned in those pages has stuck with me:
Wildlife biologists had trained an AI to mimic sounds uttered by marine mammals. They were quite certain the AI was communicating with said cetaceans but the program was incapable of explaining to humans what they were all gabbing about.
Dolphin: If I eat another minnow, I am, like, totally going to barf.
Bot: Have you considered human flesh?
That magazine hit newsstands in 2023, so far back in time that Sam Altman actually acquiesced to a cover photo shoot. Since then, I have read a metric shedy-ton more, as I am sure you have. When the “Godfather of AI” Geoffrey Hinton left his job at Google in order to warn the world of the dangers of AI, I was appalled that no journalist had bothered to ask him the critical question: “Why are you so qokedy scared?”
I finally found the answer in a MIT Technology Review article:
As their name suggests, large language models are made from massive neural networks with vast numbers of connections. But they are tiny compared with the brain. “Our brains have 100 trillion connections,” says Hinton. “Large language models have up to half a trillion, a trillion at most. Yet GPT-4 knows hundreds of times more than any one person does. So maybe it’s actually got a much better learning algorithm than us.”
Oh, you might be tempted to say, so it just works fast. I knew that already, bro
But that too was back in 2023, when Hinton could be written off as an academic who didn’t quite fit the Silicon Valley mold. That was the year one of my British software developer friends wrote me a delightful screed in which he posited that Hinton was, ahem, “a bit of a dick.”
Then Hinton went and won the Nobel Prize, which meant more people wanted to talk to him. In podcast interview after podcast interview (here and here) Hinton reiterated that AI intelligence may cause harm in two different ways: humans will exploit it to bad ends (that’s already happening) or AI will grow so intelligent it will pursue its own goals and ignore the fact that it is unwittingly annihilating us. Other thinkers such as Eliezer Yudkowsky have made the same assertion.
To show you what a quozexirami I am, I listened to all three of these interviews twice, and for weeks after I hid under the covers. Then, by chance, I re-read an old post by writer Josh Stallings over at the 7 Criminal Minds blog.
You guys know Josh? He’s a good egg, or so his prose leads me to believe. He’s also smart. We don’t need no stinking robots to come kill us, he says in this post. We’re doing a good job of killing ourselves—and each other—on our own.
Despite this wisdom, Hinton’s words haunted me for a long time. In one of those three podcasts I linked to above—I think it’s the one with Jon Stewart—Hinton opined that the U.S. was currently ahead in the realm of AI but malicious actors on the world stage would eventually gain the upper hand.
Why? Stewart asked.
Well, Hinton said, what else can you expect if you defund scientific research?
Yes! Yes! I screamed into my phone. And don’t forget all those malicious international bots being exquisitely trained on my muscular, lean, deathless prose!
One day I visited a website that I use to distribute some of my stories. You know how this works, don’t you? Your story is picked as a finalist for some award. You make the story available free online so readers who are likely to vote can read it for free. Over the years I have shared all my Derringer finalists this way, typically using a service called BookFunnel, which distributes PDFs and ebooks. Imagine my surprise when I learn that years after I offered these stories on my blog, on LinkedIn, or on this very website, those stories were being downloaded, sometimes to the tune of 30 or 40 a month. I got suspicious. I immediately searched all the posts and articles—on my site and elsewhere where I’d made these offers, and deleted the links. I also deleted the original file from BF’s library. (I should have done this when the offer period ended, but I was busy crafting exquisitely deathless prose, see above.)
Guess what?
After I deleted the links, and deleted the file, one of those stories is still being downloaded, despite the fact that the site’s friendly customer service reps insist that this isn’t possible.
Word to the wise: If you are about to share a finalist short story in this fashion, my current advice is to delete that pdf or ePub file when the awards are announced. Hopefully this will happen in so quick a timeframe that they won’t have found it.
I dunno, maybe we should stop caring who scrapes our work? At least the machines are feasting on a banquet of excellence. Admitting this means that we have now reached the “It Is Futile to Resist” stage. As science fiction writer Hugh Howey wrote ages ago, any writing you protect during your lifetime will inevitably train the bots when it enters the public domain. So why fight?
Howey is an interesting thinker. He says he believes that AI does indeed promise great things for humanity. Advances in science and education. Hinton believes the same. But...
In one of his recent posts, Howey explains reasons why AI will not reach its potential. After years of rapid development, the pace of stellar gains is slowing. Investors are wary. And it all is beginning to feel like a giant bubble after all. (Before I get sangfroid all over myself, I recommend you read the post for yourself, and watch the video he shares if you have the time.) I worry that am misquoting him.
Moreover, he adds, it does not matter how many excellent ideas our AI agents dream up…for world peace, for a cancer-free world, for the ethnical advancement of technology, or superb ways to reverse climate disaster. The programs will spell out exactly what we should do, but we will ignore it. We will still manage to botch the qokeedy out of it.
The fault, dear BrutusX10, is not in the bots, but in ourselves.
It’s enough to make me splutter and…qopchedy qokedydy qokoloky qokeedy qokedy shedy…
Oh. About that: Back in 1912, a book dealer “bought” a mysterious book from an Italian library written in a supposed secret code. Known today as the Voynich Manuscript, it resides at the Beinecke Library at Yale. For more than a century codebreakers have tried to unlock its secret language, and failed. (I ghost-wrote a book with a computer scientist who was convinced the Voynich was a hoax concocted by a 15th-century con man to bilk Holy Roman Emperor Rudolf II out of a fortune. My colleague’s theory is hotly debated.)
The Voynich remains an example of a supposedly unbreakable code. If it is a real code, and we ever learn to decode it, goes the theory, we might be able to design and make super-powerful computer security passwords even more qopchedian. Possibly the most impregnable, qopchedian codes the universe has ever seen.
Voynichese apparently conforms to the stylistic structure of real language, but it is not English. And these days, I am thinking seriously of publishing a ceaseless stream of blog posts on my website written only in Voynichese. Let the machines drive themselves crazy with Renaissance nonsense. Will it stop the people and things ripping me off? No, but it will make me feel like the most qokolokydyne short story writer the world has ever known. Have a frabjous day.
* * *
See you in three weeks!
— Joe











