An Excellent Bullshit Detection Primer

In my recent review/summary of "Calling Bullshit", we ran across a reference to one of the greatest bullshit fighters of all time: Carl Sagan, particularly in his last book "The Demon-Haunted World", 1995.

Maria Popova, of "The Marginalian", does an in-depth look at Sagan's bullshit fighting weapons as detailed in Chapter 12 of the book, "The Fine Art of Baloney Detection":

The Baloney Detection Kit: Carl Sagan’s Rules for Bullshit-Busting and Critical Thinking

I really like that Popova includes both the list of 9 bullshit-detecting tools, & the list of the 20 "most common pitfalls of common sense". Good stuff, I love lists! I thought we all did, but I have heard otherwise recently?!?!?

I met Popova a decade or more ago, entering (or leaving) the DUMBO building where she worked at the time, as did my oldest daughter, designer Erica Heinz, who knew her. I have followed "The Marginalian" (formerly "Brain Pickings") since then & have enjoyed many of her articles. They are always very well thought out & presented, very professional work.

---

Here's the home page/directory for my posts on Bullshit. This is post #89.

Don’t Say Mean Things

LOL, from McSweeney, some comic relief from the Bullshit Apocalypse. But, as seems to be standard now, the comic recounting of the situation is mostly accurate ... What do it mean? We are now post-humor???

Please Don’t Say Mean Things about the AI That I Just Invested a Billion Dollars In

“[Nvidia CEO] Jensen Huang Is Begging You to Stop Being So Negative About AI” — Headline from Gizmodo

---

Here's the home page/directory for my posts on Bullshit. This is post #88.

Necessary Evil

"Necessary Evil", by Ian Tregillis, 2013, 423 pages, 131k words. Book 3 of The Milkweed Triptych.

This is a really, really well done trilogy. Logically intricate, multiverse & time-travel both, nazi artificially created superheroes vs cthulhu class extradimensional demons summoned by British warlocks' blood sacrifices.

As I recounted in my review of the 2nd book of the triptych, I read the 1st volume of this over 10 years ago, & then lost it. I read the 2nd volume late last year, & now have finished the 3rd book.

So the entirety of this book, we have 2 versions of the main character involved in the 1940 action: the young version from the current timeline, & the older version from the 1960s. I don't remember that particularly happening in any prior reads??? Seems like I would have encountered that before. Maybe I will remember instances of it now that I have questioned it. Get to work, subconscious.

As I discovered writing the review of book 2, Tregillis retired from writing sci-fi as of 2016 - too much distraction from his day job as a physicist at Los Alamos. I still have his 2015-2016 trilogy The Alchemy Wars going for me. I think I will wait a while & then read those 3 novels all at once. I think the Milkweed Triptych would have been kickass experienced that way.

I will make that my official recommendation: buy all 3 books of the Milkweed Triptych & binge read the 3 of them! Trust me, you won't regret it!

Dead Hand Rule

"Dead Hand Rule", by Max Gladstone, 2025, 428 pages. Book 3 of The Craft Wars series. Here's Book 1. Here's Book 2.

I have enjoyed this series. It has quite the cast of characters, I'd guess 10-20 people not wearing red shirts. Some interesting depth on some who we had met only cursorily before.

The 1st 2/3 of the book is setup, an alliance forms to combat the incursion of extra-dimensional spiders, worse than Cthulhu. There is some side human interest with the Donnie entitled princess teenage character & her prince/heir soon-to-be boyfriend. Then, all at once, lots & lots of fighting.

I thought this would be it, but then, "TO BE CONTINUED". Oh well. A little long, but still a good read.

The Bullshit is Piling Up

I've had lovely house guests this weekend, so I was unable to incrementally clear the Bullshit Apocalypse stack. Here's the dump:

1. Of course a stochastic parrot/semiotic zombie can't dance. From The Markup:

   Generative AI is eating culture. See how close it’s getting to disrupting dance

   Dancers say their craft can’t be duplicated by AI. Our tests show they’re right — for now.

   I'm going to call bullshit on the "for now".

2. AI is easy to break: see, for previous examples, here, & here. IMO, you cannot build anything close to industrial-strength software when your front-end architecture is "prompt engineering". From Bruce Schneier:

   Why AI Keeps Falling for Prompt Injection Attacks

3. AI breaks things: see, for a previous example, here. From Bruce Schneier:

   AIs are Getting Better at Finding and Exploiting Internet Vulnerabilities

4. This is reaching the "piling on" point. Sir Demis Hassabis is "Google DeepMind CEO". From Gary Marcus:

   Breaking: Sir Demis Hassabis becomes the latest to say that ChatGPT is a dead-end and that we must turn our focus to world models

   Gary Marcus taught us a little about World Models here.

5. The latest "thing broken by Generative AI": democracy. We saw this earlier. From Gary Marcus et al:

   AI bot swarms threaten to undermine democracy

   When AI Can Fake Majorities, Democracy Slips Away

6. Yet another hit on human culture. From "naked capitalism":

   AI-Induced Cultural Stagnation Is No Longer Speculation − It’s Already Happening

   The researchers called the outcomes “visual elevator music” – pleasant and polished, yet devoid of any real meaning.

7. And finally, another one we've seen before. From The Markup:

   We posted a job. Then came the AI slop, impersonator and recruiter scam

   I like how the applicants who submitted fraudulent AI-generated resumes or applications, when questioned on their work experience claimed that NDAs prevented them from discussing their prior jobs. Good one!

I like doing these better atomically. I think where there are multiple items, I will start numbering them. 1 of my brothers recently reminded me, I'm a numbers kinda guy.

---

Here's the home page/directory for my posts on Bullshit. This is post #87.

Old Age

I wanted to share this story, and realized I needed a new label for the topic: old age.

I must have an Old Soul, when I started this blog in 2003 I was only 52 YO, but I apparently thought of myself as an "old man". And don't get me started on what a marketing disaster it is for your brand to be "dumbass"!

Anyway, here's the story:

So I'm reading I think a newspaper article talking about how old folk alcoholics get off of or are denied access to liquor, but then wind up drinking Listerine. I think it said Listerine was 27% alcohol! 54 proof! But Listerine has other stuff in it that you should not drink, I don't think they detailed what substances.

They then have the human interest part:

My grandmother drinks Listerine every day! & she's got no business doing that, she's 93 years old!

??? I think they've got the wrong message here. I'm betting Listerine will start an ad campaign:

Drink Listerine and Live To Be 93 Years Old!

It's all in your perspective ...

OK, Sure, Brain Damage, But, Primo Slop!

This is the 86th post I have done on the (generative AI & social media induced + general) Bullshit Apocalypse, mostly since late spring 2025. This is the 1st time I have gone looking for material. Prior to this, posts were based on articles that showed up in my newspaper or RSS feeds - I did not go looking for the dirt.

But my last post included this statement from the "naked capitalism" post under discussion:

these studies are showing cognitive decline in adults, people with fully developed brains.

This statement was not supported by any links or references, so I decided to check it out. You would expect brain damage in young people, I found the asssertion of cognitive decline in adults surprising.

DuckDuckGo was, as expected, helpful:

1. MIT Study Finds Artificial Intelligence Use Reprograms the Brain, Leading to Cognitive Decline

   A detailed summary of an MIT study. Reminds me of ... me! Here's the MIT study:

   Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task

   "Cognitive Debt", nice term! Also, "Cognitive Offloading".

   This study was students only, so no support re decline in adults.

2. Is AI dulling our minds? - The Harvard Gazette

   Another discussion of the MIT study.

3. From tools to threats: a reflection on the impact of artificial-intelligence chatbots on cognitive health

   This opinion article investigates the concept that an overreliance on AICs [Artificial Intelligence Chatbot] could potentially lead to cognitive decline [i.e., cognitive atrophy (CA)], which we labeled here as “AICs induced cognitive atrophy, AICICA.”

   This article says "AICICA may disproportionately affect the younger generation". So no support re decline in adults.

4. Excessive Use of AI: One of the Main Causes of Cognitive Disorders in the near future?

   Recent research and clinical observations suggest that excessive dependence on AI may be contributing to an increase in cognitive disorders, especially among young and middle-aged adults.

   OK, "middle-aged adults" mentioned, but no links to any of the "recent research and clinical observations".

5. AI’s cognitive implications: the decline of our thinking skills?

   A recent study by Gerlich (2025) explores the relationship between AI usage and cognitive skills, highlighting several key concerns.

   OK, let's check that out.

   AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking

   we conducted surveys and in-depth interviews with 666 participants across diverse age groups[my bold] and educational backgrounds.

   ...

   The findings revealed a significant negative correlation between frequent AI tool usage and critical thinking abilities, mediated by increased cognitive offloading.

   More significant for younger people. Lots of charts. Looked like Real Science.

I think we have a winner, => No "calling bullshit" on the reference in yesterday's post.

In conclusion: excessive LLM use makes you stupid. The younger you are, the stupider it makes you.

---

Here's the home page/directory for my posts on Bullshit. This is post #86.

Think of the Children

I'm a 1 issue kind of voter. My 1 issue is: the goal of every government should be to optimize the outcome of every child. If it helps children, I'm for it. If it harms children, it is anathema.

So this next is heart-breaking. Our ridiculous system still has 20% of children hungry. Every day a child is hungry is a day they grow fewer brain cells. Now we have found a new way to stunt mental growth. From "naked capitalism":

“We Are Watching Critical Thinking Disappear in Real Time” Due to AI Addiction: 40% of Kids Can’t Read, Teachers Quitting in Droves

A fourth grader asked their teacher, “Why do I need to learn how to read if AI can read for me?”

Right now, 40% of fourth graders can’t read. We’re witnessing the collapse of the American education system.

...

these studies are showing cognitive decline in adults, people with fully developed brains. If AI is doing this to adults, then what is it doing to a brain that’s still forming?

---

Here's the home page/directory for my posts on Bullshit. This is post #85.

You Should Need a License - Again

Generative AIs, aka Superhuman Bullshit Generators, are too powerful a tool weapon to be unregulated. Criminals should at least have to break the law to have access to them. Providers should be held responsible when they don't rigorously enforce licensing.

From the San Diego Union-Tribune via the Lexington Herald-Leader:

4 in 5 small businesses had cyberscams last year and almost half of attacks were AI powered

Note, if the Herald-Leader is paywalled, DuckDuckGo the article title ("4 in 5 ...") & numerous hits will come up, including MSN.

---

Here's the home page/directory for my posts on Bullshit. This is post #84.

Generative AI & AI Slop: Threat or Menace?

1st, from Project Syndicate:

Europe Enters the Slop War

Inundated with manipulative slop that has warped public debate, the European Union has begun to confront the threat that social media platforms pose to democratic life.

"The Slop War" is an apt turn of phrase. Why didn't I think of that?

The Slop War is people battling against the ongoing Bullshit Apocalypse. I wish them luck. It will be a battle for the ages, for the soul of our culture.

This is a short & hard-hitting read, check it out.

---

Gary Marcus points us at a recent paper by 2 law professors also loudly sounding an alarm.

How Generative AI is destroying society

The paper is titled "How AI Destroys Institutions"

Both articles use the word "threat" but not "menace". So the answer to our title is, "threat".

---

Here's the home page/directory for my posts on Bullshit. This is post #83.

1 Very Good Idea; 1 Very Bad Idea

Our 1 Very Good Idea comes from Cory Doctorow. It is for Canada to repeal its version of the DCMA anti-circumvention clause, section 1201. Canadian Bill C-11, the Copyright Modernization Act, bans Canadian companies from modifying America's digital tech exports.

Canada then becomes the device jail-breaking capital of the world.

Before he gets to that he delivers a nice, short takedown of the generative AI hyper-bubble.

Pluralistic: Sorry, eh (13 Jan 2026)

---

Here's our 1 Very Bad Idea. Who in this world could ever imagine that having generative AI influencers would be a good idea? Oh, it's one of those $$$ deals. Capitalism, FTL. From 404 Media:

Instagram AI Influencers Are Defaming Celebrities With Sex Scandals

Fake images of LeBron James, iShowSpeed, Dwayne “The Rock” Johnson, and even Nicolás Maduro show them in bed with AI-generated influencers.

---

Here's the home page/directory for my posts on Bullshit. This is post #82.

A Bad Report Card; An Apology to Dogs

Gary Marcus links to 4 articles, none of which have good news about LLMs.

Let’s be honest, Generative AI isn’t going all that well

Update: This is consistent with the finding of the Remote Labor Index that AI could only do about 2.5% of jobs, reported recently by the Washington Post.

---

I introduced the recent post "It’s Band-Aids All The Way Down" on SubStack with this:

Calling an LLM a Talking Dog is an insult to dogs.

Dogs are much more intelligent than LLMs.

I've thought about this some more.

This post of mine, from last July, talks about a fable told by Seth Godin, comparing an LLM to a talking dog:

The Talking Dog

I liked that, it seemed appropriate. Later I expanded it to "black-box talking dog".

But as time passed, I realized: Godin is a marketing guy. Comparing an LLM to a dog, talking or not, is way overselling its ability. It's like calling LLM errors & confabulations "hallucinations", to conflate LLMs with the human mind - marketing hype, marketing bullshit. I called that out in this post:

Hallucinations? No, Bullshit!

and this one:

Hallucinations My Ass!

Dogs have been paired up with humans for 14,000 years. Their vocabulary may top out at around 1000 or so words, but they recognize human tone-of-voice and human body language, which an LLM obviously doesn't. They have far more of the Morality™ module implemented than any LLM: they know when they are a good dog or a bad dog. I'm pretty sure LLMs don't know when they are good or bad, right or wrong. In fact, I'm pretty sure the terms don't even vaguely apply to LLMs.

I look at the 20 human emotions (from "The Cognitive Structure of Emotion") and find myself thinking I have seen most if not all of these in dogs.

Dogs also have an incredible sense of smell, which I'm pretty sure LLMs don't have.

Most importantly, dogs live in the same physical world that humans do, unlike LLMs which are creatures only of words.

So I won't insult dogs anymore by calling an LLM a "talking dog".

I could use "stocastic parrot". Parrots are a lot less capable than dogs, but, now that I think about it, they are probably more intelligent than LLMs.

I still like my own "Semiotic Zombie". The Bullshit Apocalypse, covering the world with various forms of AI slop, is really not that different from a Zombie Apocalypse covering the world with zombies.

To conclude, I formally apologize to all dogs for comparing them to LLMs.

---

Here's the home page/directory for my posts on Bullshit. This is post #81.

Scale

"Scale", by Greg Egan, 2023, 269 pages, 83k words.

Greg Egan is a mathematician, and has been a great favorite of mine over the years. He is also great on computing & physics, including quantum.

He's done several stories in a settings where the laws of physics are different. In this case, 8 different leptons with masses differing by powers of 2 create 8 kinds of matter - the heavier leptons create smaller atoms & denser everything. We have a civilization with people in 8 different sizes: let's say 64, 32, 16, 8, 4, 2, 1, 0.5" tall. But they all weigh about the same. In addition to everything being denser, internal clocks including mental also run faster & lifespans are shorter the smaller you are.

A bit of a stretch. Some plot added on top, also somewhat far-fetched. I went back through this blog searching Greg Egan, he has been prominently featured & bragged on since inception, 2003. But, his days of edge may be over. He is 64 YO now, 10 years younger than me. This was his 2nd most recent novel, I have his latest on the iPad, it'll be a while before I get to it, but we'll see how it goes.

The Greks Bring Gifts

"The Greks Bring Gifts", by Murray Leinster, 1964, 143 pages.

My younger brother Mark, the author, asked me via email if I remembered Murray Leinster; his novel "The Greks Bring Gifts"; and had I read it? I replied: yes, yes, & maybe.

I was surprised when a week or so later when there appeared in my mailbox an original 1964 paperback copy of said novel, sent by my brother Mark. Thanks Mark!

I finished reading it yesterday. An easy and fun read. Aliens come to earth bearing gifts. Beware!

I think I did read it Back In The Day.

The story is narrated by a human from some time after most of the story happens. Things get wrapped up nicely, you can tell this book was written in simpler times.

1 thing I noticed, re 1964: if was before unleaded gas & catalytic converters. Several instances where auto exhaust becomes pretty nasty. A reminder, the good old days weren't always so good ...

There is a strong female character, the fiance of the main male character, the PhD physicist who starts the process of figuring out the aliens' deliberately misleading technology. She is an MD, & a lot of her boyfriend's ideas originate with her. He tells her several times how she is the smarter one. But, she still at times tries to hold back, because men don't like women smarter than they are. Dated, but still a lot better than most of the libertarian, white male classic science fiction. Way to go Murray!

Check out his Wikipedia page. He never met a genre he didn't like. He wrote "Fantasy, science fiction, horror fiction, mystery fiction, Western fiction, pulp fiction".

{Mis/Dis}information; Maintenance Debt

1st up, Gary Marcus links to 2 articles, the 2nd a piece in The Atlantic which he co-wrote, excerpt below.

The (possibly) coming AI backlash and information warfare

Current AI] systems are patient, amoral, and fantastic at mimicry, making them among the greatest tools in history for generating mis- and disinformation—the latter of which is a tremendous weapon, not necessarily for its ability to persuade and convince, but for its ability to sow chaos. This, rather than some intelligence breakthrough, may well be the legacy of generative AI.

---

Cory Doctorow's post today is so spot on. Is there anything this guy doesn't get?

Pluralistic: Code is a liability (not an asset) (06 Jan 2026)

Code is a liability (not an asset). Tech bosses don't understand this. They think AI is great because it produces 10,000 times more code than a programmer, but that just means it's producing 10,000 times more liabilities. AI is the asbestos we're shoveling into the walls of our high-tech society:

...

Code is a liability. Code's capabilities are assets. The goal of a tech shop is to have code whose capabilities generate more revenue than the costs associated with keeping that code running.

I ran software development shops for decades.

1. Writing code fast is great. But, the most important things about code are readability & maintainability. Doctorow mentions that it's maybe OK to cut corners for a 1-off (throw-away code), but, you'd be surprised how often you come back to something like that - and it is really nice if you can read the code.

2. Whenever we were discussing a new product, module, feature, etc., we always talked about the "Maintenance Debt". Doctorow nails this, code is indeed a liability. When you ship code, there is an implicit commitment to the people you ship it to that it will stay running, despite the world changing around & under it. This is the Maintenance Debt.

LLM-generated code is just another modality of AI slop. Maybe OK for prototypes or demos, but nothing you want underpinning a real world software product.

---

Here's the home page/directory for my posts on Bullshit. This is post #80.

The Doors of Eden

"The Doors of Eden", by Adrian Tchaikovsky, 2020, 596 pages, 162k words.

I'm not sure how I missed this book. I believe it is 1 of Tchaikovsky's best. It predates Marvel-induced multiverse burnout (2020). It is also a portal fantasy. The turns he posits leading to completely different development of species on earth are very, very creative.

I like his definition of evolution, slightly different from mine:

imperfect self-replicating system in an environment of limited resources.

I replace "in an environment of limited resources" with "with differiential survivability".

I wound up staying up to 12:40 am finishing this book last night. I almost never do that any more. But, I thought I was about at the ending. But this book ends, and ends, and ends, and ... Be warned.

Again, highly recommended, check it out.

It's Band-Aids All The Way Down

LOL! So ChatGPT doesn't read the newspaper or watch the news! Of course it doesn't. It is a computer program that only exists when it is interacting with a user. It has no persistence, no background tasks, no continuity, no conscious mind, no subconscious mind, no world model, no ... the list goes on and on and on and ...

From Gary Marcus:

Why ChatGPT can’t be trusted with breaking news

Using massive numbers of humans (exact numbers unknown), GenAI companies perpetually put band-aids that they never disclose on top of their LLMs, based on human corrections, fixing each new problem as it arises. But the cycle repeats, endlessly.

---

Here's the home page/directory for my posts on Bullshit. This is post #79.

The Best Science Fiction of the Year Volume 8

"The Best Science Fiction of the Year Volume 8" edited by Neil Clarke, 2024, 807 pages, 250k words, 31 stories: 19 short stories (< 7500 words), 11 novelettes (< 17500 words), 1 novella (< 40000 words).

This volume covers 2022. It was released September 10, 2024. I guess this is Clarke's timetable now.

Wow. I went 9 stories in, not really getting into any of them. I was thinking "Am I aging out of the genre?" It reminded me of a few years ago when, after 10 or so years of interviewing HS seniors for admission to MIT, I started finding almost all of them so immature & clueless. I decided to quit, let someone closer in age to them interview them. Because I figured the problem was with me, not with them.

I finally did like the somewhat snarky Tobias S. Bucknell story "The Plastic People". A sneering look at inequality and the ultra-wealthy.

Note, I just noticed that story #6, "Termination Stories for the Cyberpunk Dystopia Protagonist" was by the author that our editor, Clarke, called out as "one to watch": Isabel J. Kim. The story is very meta, but a little bit too contrived for me, I think.

The Ray Nayler story, "Mender of Sparrows" was of course good. I think I had read it before.

The Suzanne Palmer story, "Falling Off the Edge of the World", was interesting. A astro-archeology story with a kind of incidental 1st contact story thrown in.

The Sarah Gailey story, "When the Tide Rises", was memorable, but somewhat shallow IMO.

---

The Rich Larson story,"Quandary Aminu vs The Butterfly Man", is the high point of the collection. Why do I say that? The story is really edgy. What makes a story edgy?

1. It reminds me of "Neuromancer" or "Snow Crash".
2. It has characters that remind me of cheap detective / noir stories or films.
3. It has John Wick levels of (stylistic) violence.

Apparently I'm a man of simple tastes.

---

Our 2nd story (Bucknell was the 1st) of Caribbean origins is "Bishop's Opening", by R.S.A. Garcia. An odd, quirky story, with a weird society based on chess pieces & no mercy. Garcia has recently struggled with cancer, I GoFundedMe her for $200, last update was that she was better. I (selfishly) hope she has a long life, I enjoy her work.

The Alistair Reynolds story, "Things To Do In Deimos When You're Dead", is very interesting. Very creative setting & back story, relatable characters.

"By Those Hands", by Congyun "Mu Ming" Gu, is an interesting story of betrayal - or is it?

The Greg Egan story, "Solidity", is rather odd. Egan has been very reliable for the last 2-3 decades. I noticed I had missed his last 2 novels, now on my iPad. This is a very creative multiverse story.

"Optimist Cleaver's Last Transmission", by J. C. Hsyu is edgy (see above). I revised #1 above and added 'or "Snow Crash"', because this story features delivery people.

The Tade Thompson story, "Down and Out in Exile Park", has some nice post-capitalism memes in it.

"Two Spacesuits", by Leonard Richardson, is an odd story.

"A Hole in the Light", by Annalee Newitz, features an interesting alien life form.

I seem to remember the Dozois "Year's Best" usually ending with a longish, very strong story. That didn't happen here. The last few stories are shortish, & not particularly strong.

---

OK, so I'm finished with this collection, covering 2022. It was released September 10, 2024 - 21 months after the end of the target year. The last few Dozois "Year's Best" came out in July of the following year - a 7 month lag. I would always look forward to it.

It is now January, 2026, and no sign of the Clarke 2023 "Year's Best" - 25 months delay & counting! This is NOT helping to keep me current.

It looks like Clarke has been having health issues since 2012 when he had a severe heart attack. He's 60 YO now, so he would have been 47 then? His website has a 2021 blog post mentioning kidney stones after 8 years of health problems. I hope he's doing alright.

But, meanwhile, crap, do I need to switch to a different "Year's Best"? There are a few others out there. Any recommendations?