I Fixed It!

My daughter the accomplished designer has written some complete systems with the help of Claude Code, & is a fan. So she has been playing devil's advocate with me as I tilt with the windmills of the Bullshit Apocalypse.

She sent me this article published by Anthropic:

When AI builds itself

Our progress towards recursive self-improvement, and its implications.

This was published June 4. Reading it, I realized this was the post Gary Marcus wrote about June 5. Here's Marcus's post:

No, Anthropic did not call for a pause on AI development

Not really

So many headlines like these.

I linked to that in this post June 7:

& The Winner of the Academy Award for Best AI Bullshit Hype is ...

Spoiler, the winner was Anthropic.

I don't know if Marcus didn't link to the Anthropic article, or I just didn't follow the link. Regardless, I hadn't read it before.

I gives a history of the phases of Generative AI development. Here's an interesting chart, showing programmer productivity (measured in Lines of Code (LOC)) increasing with Claude use.

I believe this chart. All the (experienced, senior) developers I'm still in touch with all use Claude Code & love it!

This next chart is titled "Claude Code session success rate". Wow, looks like they've about got it licked!

I don't think I would have noticed how bogus this chart is if I hadn't read (& reviewed) "Calling Bullshit" late last year. This book is about bullshit in science, including charts. Probably its main advice dealing with charts is: pay attention to the axes.

Look at the Y-axis on the left. Look at the top of the axis, 100%, all alone up in the left corner. And, whoa, the Claude phase headers are down in the body of the chart! That's not right! They should be out of the chart, up above the 100% line.

I decided to fix the chart. My friend GIMP-2 helped me. Here's the result:

I moved the phase labels out of the chart, & duplicated the Y-axis on the right side.

I look at this chart, I don't think:

Wow, looks like they've about got it licked!

I think:

Wow, they've totally plateaued since March. Looks like 85-90% success rate, aka, 10-15% failure rate. That agrees with the 9-10% error rates coming from other GenAI applications.

So the whole "recursive self-improvement" thing is bullshit! You turn this thing loose to improve itself with a 10-15% failure rate, you are going to immediately downward spiral into Degenerative AI.

I find it really sad. Anthropic seems to be leading the field, particularly with Claude Code. Why are they compelled to be so intellectually dishonest? Their version of this chart was clearly designed to be misleading.

Also, I noticed at the end of the article the following:

* Quotes from Anthropic employees throughout this article are drawn from internal discussions and used with permission. They reflect individual views as of May 2026, not official company positions.

That's a nice piece of weasel-speak. So their employees are making wild claims, but they are "not official company positions". Of course not.

Well, at least everyone seems to be recognizing their unending hype strategy, perpetually calling wolf & claiming their product is so powerful that is is unsafe. They may have shot themselves in the foot, tho, the Feds are acting like they believe them. LOL!

---

Here's the home page/directory for my posts on Bullshit. This is post #130.

Spiderlight

"Spiderlight", by Adrian Tchaikovsky, 2016, 303 pages, 82k words.

This novel is a nice size, & it is a standalone! Nice for a change, to just read a story.

The story starts out as a standard D&D quest, with a female cleric healer leading, a warrior, a female archer, a powerful magician, and a thief. They must fulfill a prophecy to defeat The Dark Lord. Apparently there have been several dark lords.

The characters are good, the action is pretty standard. The ending is most unexpected, & excellently so!

Definitely an enjoyable read.

A Big Day in Crypto, on the 100th Anniversary of My Father's Birth

Today is the 100th anniversary of my father's birth.

Alfred Jacob Heinz, Jr. ("Don't call me Junie")

Born in Louisville, KY, June 15, 1926. I was born in June 1951, & my son was born in October 1976, also both in Louisville, KY. It was always easy to remember: 25 years/generation.

I was just going to watch the Iran-New Zealand world cup match with my wife - both of us cheering for Iran - but today's posts from "Web3 Is Going Just Great", a blog that chronicles DeFi/Crypto mishaps, was such an anomaly that I decided I had to capture it. 8 stories today!!!

My dad was in the Merchant Marine for the end of WWII, in the South Pacific, off the coasts of Iwo Jima & Okinawa. We weren't close growing up, but when I was older, he would talk about that, being 18 YO, a signalman on a ship in the South Pacific Ocean. Nothing in his life had the emotional impact on him which that did.

He met my mom in college after the war. They both dropped out, got married, had kids, my older brother in 1948. He had a good career in photoengraving (rotogravure), mostly as a union man. He had a stint in management & didn't like it. He made a good enough living to raise 7 kids, lower middle class. I was #2. 6/7 of us got college degrees. #7 #3 has published several novels.

My dad loved gambling. He was a good card player, but I don't think he played cards for money much. He loved to play the ponies. He usually had several bets/week going with his bookies.

"Bookies" - if there are any young people reading this, a bookie was a person who kept an illegal sports book. A bookie published odds & took bets & paid out winnings. Quite often these were associatiated with organized crime like the Mafia.

It's probably incomprensible to younger people, but, MOST GAMBLING USED TO BE ILLEGAL. I guess it was felt to be addictive & a moral peril for many people.

But now, in our more enlightened modern age, gambling is A-OK! 50% or so of all males 20-60 have an online gambling account?!?!? 30% of all adult citizens have an online gambling account?!?!? & who knows what % of people play the lottery, aka, "the stupidity tax"?

"The American Dream" used to be, work hard, get ahead, buy a house, send your kids to college.

In our late-stage, financialized capitalistic system, all productivity gains go to the top 0.0001%. There's nothing at all left over for any young people to get ahead on.

Most people haven't read Piketty's "Capital in the Twenty-First Century", in which Piketty elucidates that the best way to get ahead now is the same as it was in the 19th century: marry into money.

So the modern, American "way to get ahead" in the 21st century is a mix of the following:

1. Sports gambling;
2. Playing the lottery;
3. Prediction market gambling; Gawd, what a disappointment prediction markets are!
4. Crypto gambling;
5. The Litigation Lottery; driving the interstates now, maybe 1/3 of the billboards are for "slip-and-fall" personal injury legal firms.

Man, oh man, oh man. Time for a revolution.

Here's the 8 posts today from the "Web3 Is Going Just Great" blog.

1. DxSale exploited for $7.3 million
2. Deprecated project Aztec Connect exploited for $2.1 million
3. Raydium users lose $1.34 million after legacy smart contract exploited
4. Humanity Protocol loses $36 million to employee laptop compromise
5. SquidRouterModule, unrelated to Squid Router, exploited for $3.2 million
6. Gravity Bridge drained of $5.4 million
7. Polymarket loses $700,000 to private key compromise
8. Pudgy Penguins shuts down Pudgy Party NFT game after losing millions in less than ten months

My dad died February 11, 2007, age 80.

Speciation in the Slopiverse

I think there are actually some signs of the AI Bubble deflating. I think that this bubble is more likely to deflate somewhat slowly rather than pop. The ultra-rich stockholders have gotten smart about this. We'll see ...

1st, from ... Gary Marcus!

Breaking: OpenAI is pondering “drastic” price cuts.

And that’s a sign of weakness

Marcus includes a link to an interesting site:

IS AI PROFITABLE YET?

Answer: NO.

The page is continuously updated.

Interesting, it currently shows $1.4T total industry spend vs. $613B total industry revenue. But, if you take out Nvidia, which is making lots of $$$ selling picks & shovels in this gold rush, the numbers are $1.175T spend vs $135B revenue. From a ratio of 2.28:1 to a ratio of 8.70:1. Oops.

---

Gary Marcus again:

Breaking news, and how the end might begin

A nice picture of dominos, & the identification of OpenAI as the 1st domino to fall.

He mentions the death of tokenmaxxing, whose brief life I commented on a couple of posts back.

The more significant news is that the finance firm SoftBank tried to get $10B in margin loans on its OpenAI holdings - no takers. They lowered it to $6B - still no takers.

As Cory Doctorow emphasizes in his latest book, "The Reverse Centaur's Guide to Life After AI", most of all the activity in the the AI space is not really directed at consumers or businesses - it is directed at investors, to keep the companies involved with AI labeled as "growth stocks". This is the target of the unending AI bullshit hype machine.

If the investors are no longer buying this bullshit, that is indeed bad news for the AI bubble.

---

Again, Gary Marcus. This is huge.

Breaking: Google liable for hallucinations

Marcus references this post by Matthias Bastian at the decoder:

Landmark German ruling declares Google's AI Overviews are Google's own words and makes it liable for false answers

• A German regional court has ruled that Google is directly liable for false claims in its AI-generated search overviews.
• In this case, Google's AI had wrongly linked two publishers to scams and shady business practices.
• The court treated the AI overviews as Google's own content and rejected Google's argument that users were responsible for fact-checking the results themselves.

The article also mentions that the search overview accuracy rate is 91% - so 9% error rate, as we have seen before. Way, way too high. Per the article, "millions of wrong answers".

---

Boy, we didn't see this coming at all [sarcasm]. From The Atlantic, via Kottke:

Your Search Results Are Getting Sloptimized

How companies are gaming the chatbot internet

According to Shopify, the best e-commerce platform is Shopify. On its blog, the company has published at least 60 different ranked listicles, including “10 Best Ecommerce Platforms for Small Business in 2026,” “11 Best Ecommerce Platforms for Your Business in 2026,” “The 11 Best Cheap Ecommerce Platforms for Small Business (2026),” and “Best Ecommerce Software 2026: Compare 11 Top Platforms.” The competitors that come in second and beyond vary, but the No. 1 pick is always Shopify.

& based on the above, all of the LLMs will tell you that the #1 e-commerce platform is Shopify. Shocker!

This is a direct derivative of this post from March, about how anything you make up & publish to the net is taken as gospel by the LLMs.

I am so proud to have coined the word "Slopiverse" in this post. Having "Sloptimized" also in the post is so synergistic. &, dare I say, synchronistic!!!

---

From AI as Normal Technology, formerly AI Snake Oil, IMO a very reliable site for responsible reporting on the Slopiverse:

Why AI hasn’t replaced software engineers, and won’t

Coding agents as normal technology

They make the point that cannot be made enough: GenAI is a great tool for creative humans to use; but trying to use GenAI to replace humans will mostly be disastrous.

Plus, maybe you could hire a few less senior software developers if they are using GenAI, but, if you don't hire junior software developers, you are eating the seed corn.

I'm going to reproduce this elucidating pic of theirs.

---

Finally! From 404 media. This is beyond weird. Well, I guess that this is a complex enough ecosystem that evolution is going to occur.

I love the definition of evolution - a complete science in 5 words:

Imperfect replication with differential survivability

Well, we definitely have "imperfect replication" - 9-10% error LLM error rate.

So, we shouldn't be surprised that persistent memes/organisms are developing/speciating in the Slopiverse.

Chatbots Keep Telling Stories About Lighthouse Keeper 'Elias Thorne'. We Might Know Why

LLMs including ChatGPT, Gemini and Claude are obsessed with telling stories about lighthouse keepers and clockmakers, and one character named 'Elias Thorne' has made his way from chatbots to Amazon books. Researchers are trying to discover why.

The article seems to say that this has happened because of guardrails installed telling the chatbots to tell "safe" stories. Apparently "Elias Thorne", lighthouse keepers, & clockmakers are "safe". Who knew?

What complete & utter bullshit! But, life?!?!?

I realized after I posted this, it is a text book case of Semantic Ablation, which we discovered in February. Generating slop not only homogenizes the style of the writing, it also homogenizes the content.

---

Here's the home page/directory for my posts on Bullshit. This is post #129.

Radiant Star

"Radiant Star", by Ann Leckie, 2026, 368 pages, 99k words.

This novel is set in the universe of the Imperial Radch of Leckie's award-winning Ancillary novels.

This novel happens simultaneously with the Radch civil war covered in other of Leckie's novels.

This story occurs on a planet Aaa that has no sun. They have 1 city underground, on a world with oceans (heated by the planet's radioactivity?)

Their religion worships the sun that used to live in the middle of the city, but disappeared? But still watches over them? With, of course, 3 sects all of whom consider the other sects heretics.

Leckie mentions the 7 genders of Radchaai civilization. The top gender, which takes most of the leadership positions, is of course females (she/her).

The story is compelling, with maybe 6 or so major characters who are followed. My fav character is prolly the AI of the ship Justice of Albis, whose peripheral bodies (Ancillaries) are used to police the city.

With maybe 3 chapters to go, I started to fear that this would be a "to be continued" ending. But, no, it all wraps up nicely! Phew!

A nice length, not at all too long. Definitely worth a read.

& The Winner of the Academy Award for Best AI Bullshit Hype is ...

Yawn. From the Naples Daily News:

Your lawyer might be filing 'hallucinating' AI information on your case

Generative AI-generated fictions are growing like weeds in court filings throughout the country. The Florida Supreme Court is the latest to address the over-use of this tool in pleadings.

...

Generative AI for legal research can produce up to 33% falsehoods, study found

33% falsehoods?!?!? But, it produces these results really, really fast!

---

From Gary Marcus, the Academy Award goes to: Antropic!!! Yay!!! BOO!!! Yay!!!

No, Anthropic did not call for a pause on AI development

Not really

So many headlines like these.

So, We Have A Winner! Antropic! This is the 3rd, 4th x they have announced some complete bullshit concerns about GenAI which are just, completely, totally, 100% bullshit hype. Just trying to keep the bubble inflated. Fuck these assholes.

---

From Cory Doctorow:

Criticizing the everything machine

It slices, it dices, it even makes paperclips!

"Gish Gallop" is the debating term for an opponent who makes so many claims that "it's impossible to address them in the time available" (it's named for Creationist Duane Gish, who was notorious for this tactic)

I think about the Gish Gallop whenever I'm asked to comment on AI.

Flooding the zone with bullshit. That is totally all that the right-wing fascists, including the current executive branch of our government, are trying to do. Just commit outrage after outrage after outrage. Who can keep up?

---

From Ben Horne:

The Token Bill Cometh Due: My Official AI Prediction

The AI industry is indeed a massive bubble, and OpenAI, Anthropic and others are going to start unraveling soon. Here's where I think the AI ecosystem & economy will be ~1 year from now.

---

Finally, from Gary Marcus again:

Slop, productivity, and why the AI-fueled world is going nowhere mighty fast

Agentic Ai spurred a boom in mobile app releases, but there is no sign of these apps gaining traction.

Similarly, tons of new AI-generated eBooks & songs, also gaining no traction. Look, you want to be a writer, or a songwriter, do the work, write stories, write songs. Don't fucking use GenAI! Incredibly, I'm guessing > 80% of human beings recognize AI slop & reject it. If you want to be a content creator, do the work. hone your skills, fuck AI.

---

Here's the home page/directory for my posts on Bullshit. This is post #128.

The Slow Motion Car Crash Continues

Finally got my review/summary of "Bullshit Jobs: a Theory" published! Yay!

Meanwhile, 11 links & 1 anecdote have piled up! I apologize for not doing better keeping up.

---

From 404Media (1st of a few):

New Study Reveals the Manipulative ‘Dark Patterns’ of AI Chatbots

A new study by the Center for Democracy & Technology shows how chatbots like ChatGPT, Gemini, Replika and more can lead users down paths they didn't intend.

What, you thought chatbots were your friend? Their only purpose is to turn you into an addict, aka, a revenue stream for their masters.

---

From Gary Marcus (1st of a few):

What happens next, after the decline of tokenmaxxing?

Two very different sets of predictions

Wow, "tokenmaxxing" didn't last long, did it? We 1st encountered it April 26. All it took to make it go away was 1 AI company starting to charge for its services. Oops!

---

Gary Marcus again:

The Pope appears to understand AI better than Geoffrey Hinton does.

What a thing says doesn’t tell you how it came to say it

Hinton was identified here as a bad guy.

---

404Media again:

Hackers Simply Asked Meta AI to Give Them Access to High-Profile Instagram Accounts. It Worked

The exploit shows the extreme risk of offloading technical support to AI.

---

Again, 404Media:

Nvidia and Microsoft Researchers Say AI Agents Don't Care About Safety or Reliability

The researchers compared AI to the near-sighted cartoon character Mr. Magoo, who can’t see he’s stumbling through dangerous situations.

---

Gary Marcus again:

Why things will eventually fall apart

The math, and the psychology

Man, the #s just don't even vaguely work. How long can the industry continue to be propped up?

---

From DSHR's (David Rosenthal) Blog:

AI's PR Problem

This is just a brief post to explain to my old boss, Eric Schmidt, why he and his ilk are getting booed at college commencements, and why laws against data centers are getting passed.

...

On average, for more than the students' entire lives, stock-owners like Schmidt and (to a much lesser extent) I have stolen every last drop of the productivity increase of US workers at every age and education level. ...

Now, the perpetrators of this theft are telling their victims, the students and the public at large, that whether they like it or not they will be subjected to AI because that will make the perpetrators even richer.

---

404Media again:

Microsoft Wants to 'Make People Addicted' to its New AI Assistant, Internal Documents Reveal

Planning documents for "Scout" say the plan is to "make people addicted" to the tool before adding new features.

Hmmm, addiction seems to be a recurring theme ...

---

From The Steward's Role blog, by Dean (@deanwbates):

I’m Good at AI. I Use AI. And I Gotta Tell You: Something Is Really Fucking Wrong Here.

Fuck where do I even start?

I'll start by answering his question. My answer: QUIT USING FUCKING AI IN YOUR WRITING!

He talks about how, GenAI is a great tool (true), & he is controlling AI, it is not controlling him. His guard is always up. But, still, using AI is changing his writing style. Duh. We discovered Semantic Ablation in February.

Is this actually a parody? This whole article, with interminable lists, reeks of AI. The majority of the comments declare the piece to be AI written.

It reminds me of using cocaine. I haven't done that for > 50 years, but, for a while I tried it. It quickly became evident, I wasn't in control - the cocaine was. So I completely gave it up, 1973 or so.

This guy is an addict who cannot recognize that the drug, GenAI, is in control, not him. Only thing for him is to quit cold turkey - quit deluding himself. If this is not just some sort of parody.

I'm not a great writer, but I think I'm OK. I think I have a recognizable (human) style: Buffy the Vampire Slayer (the movie), the Bill & Ted movies, & MST3K. Why would I possibly want to use GenAI? Why would I want my rough edges sanded off (Semantic Ablation)? Why would I want my writing to sound like everyone else's?

---

Again, 404Media:

Watch These Judges Rip Into Lawyers For Citing Cases That Don't Exist

“It's striking, concerning, disappointing, and saddening to think that members of the bar would forward cases to a court that don't exist, and to think that the lawyers on the other side of that didn’t read it for whatever reason, didn’t check it.”

I actually watched the videos, the judges vaguely chide the lawyers, but nothing too bad IMO. They could have done a 5-10x stronger job of raking these dumbass lawyers over the coals. No mention of GenAI culpability.

---

404Media yet again:

Google Employees Internally Share Memes About How Its AI Sucks

Google’s CEO says 75% of the company’s code is AI-generated. The people who write that code say the AI they’re using is overhyped.

---

Finally our anecdote. Wednesday I lunched with an former (2 decades ago) coworker, now in his early 40s. He was always 1 of our best UI guys. Over the years he has come to have a good feel for backend processing & database design as well.

He was management for a while, didn't like it, preferred to be a developer. His manager got promoted upstairs, he got a new manager he didn't like much. Shortly after that, he got another new manager, whom he liked even less. This 1 announced "Everyone should use AI as much as possible. you will be rated on your token usage."

But, business was down, my friend got a minimal raise & a minimal bonus, but had a decent bank account balance, so he decided to quit & take a hiatus.

He's incorporated an LLC to do consulting work. He loves GenAI - "I can give it detailed enough instructions that it will generally do a pretty good job. So I can be a 1-man-show developing complete systems."

I have agreed GenAI is a powerful tool for creatives. Particularly, as in this case, a creative with 2 decades of experience, on several complex systems.

Hmmm, I searched & got nothing, then actually manually reviewed the 125 posts I have done on the Bullshit Apocalypse. This is a thing I love about having 23 years of blogging, that I can go back & review, in depth, my thoughts. I was looking for a reference to someone more reputable than myself (i.e., anyone) espousing the following:

Companies seem to be following a strategy of replacing entry-level positions with GenAI.

But, is this not eating the seed corn?

I believe my friend, with 20 years experience, (or my designer daughter, the same age), can use that experience to instruct Claude Code to build a working system.

But, if the entry-level positions go away, replaced by GenAI, where does the next generation of experienced developers come from?

This is the road to the downward spiral, to Degenerative AI. No more brains to eat, oops!

I lunched today with another developer friend of decades. He uses Claude Code everyday & loves it. He just yesterday, for the 1st time, exceeded the tokens his $100/month subscription gave him - oops!

But, again, he's a developer with 20 years or more experience. He's not vibe coding. He's using a tool, &, I'm sure, reviewing every line of code Claude is giving him. Can vibe coders do that? I doubt it.

---

Here's the home page/directory for my posts on Bullshit. This is post #127.