Data about Humans and News
July 29, 2026
I’m always curious about how people interact and consume news, so when Reuters and Oxford released “Digital News Interactive” an intuitive way to explore the information I was intrigued. The information is described as:
“Explore more than a decade of survey data from the Reuters Institute’s research on news consumption across markets, regions and demographics.”
It appears that most people get their news from social media. TV and news websites seem to be in a gentle glidepath to a rock-strewn dry creek bed. Radio isn’t dead yet, but it, like print, are in a holding pattern but losing altitude. Heading upwards are podcasts and lamentably chatbots. (Why not ask an oligopolistic company for an unbiased answer? That’s a great idea to an increasing number of people.)
In all markets, only 37% trust news overall and 44% trust in news they chase down. The killer item is that 42% avoid the news altogether and 51% get their news through websites. What is interesting is only 17% pay for their news, 21% comment on stories, 46% remain interested in learning about what is happening, and 62% are concerned about misinformation. I wonder what the 38% of the news consumers think about being manipulated by misinformation, weaponized information, and corporate entities eager to “own” the information space. Facebook remains a big place for people to receive their news on average, while younger folks skew towards Instagram. As I reviewed the data, I noted that about half the people don’t trust the news.
None of these data points are particularly surprising. Social media continues to grow but these services face a lack of interest in the news problem. My hunch is that news outfits with money will pay the big bucks to appear at the top of algorithmic feed outputs.
Whitney Grace, July 29, 2026
“Karpe Questum” — The Swiss Ignore the Palantirians
July 14, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
I read a story that first appeared on July 2, 2026. I read it today (July 4, 2026). I found it amusing, but as a dinobaby I have an untrendy sense of humor. This is not the tale of Europol taking down bad actors. This is not a write up describing how Ukraine has used asymmetric warfare to make a certain “special operation” look ineffectual. Nope. This is a story about a big, well connected US company and what is described as a a “small independent Swiss online magazine.” I mean is a small independent Swiss online magazine a must read in the US? Is the publication even indexed by the giant (but selective) content rock crushers?
Thanks, MidJourney. Good enough. Are you a hobbit?
Palantir knew they couldn’t sue for defamation because, you know, everything Republik reported was true. Instead, they sued, trying to invoke a Swiss “right of reply” law, claiming that because Republik refused to publish the press release Palantir wanted to run in response to the reporting, the magazine had violated the law. As we said at the time, this is the height of entitlement. Palantir doesn’t get to tell Republik how and what it must publish.
Let’s think about this statement. Palantir considers itself part of the Lord of the Rings’ crowd. Some of its executives are “Girarded up”. If you don’t know about René Girard do a Google search. Even sketchy AI systems can spit out some info about M. Girard and his fellow travelers.
The magazine obviously did not get the memo that said, “You must do what Palantirians tell you to do.” This is generally viewed as a typical European Union intellectual deficiency. The Lord of the Rings crowd looked into their seeing stone and saw that its oracular news release should be published by the clueless Republik outfit. In a country with a lot of valleys, the clueless at the Republik should not emulated blank eyed Swiss cows.
One might conclude that small online publications absolutely have to do what a bit American company tells it to do. Even the average American cows understands how life is run on an American farm. An American company is definitely in the know when it comes to the correct, accurate, truthful, well crafted, honest, and ethical information. The “law” is irrelevant because Palantirians are the law. Pull this type of obstinate behavior in Miami or DC, and the Republik employees will understand the power of the hobbits. (These, I believe, are dwarves conceived by a somewhat creative Oxford professor. (ep, dwarves. FYI: The American spelling is “dwarfs.” But I know how Oxford eccentrics approach conventions like spelling. Tolkien used “dwarves” and I too present that spelling. Put that in your lumber room.)
What’s clear to me is that I want to reproduce this paragraph from the TechDirt story:
Anyway, given that Palantir seems really upset about Republik’s reporting, it sure would be a shame if you decided to go read this critical reporting of Palantir’s relentless attempts to win business from the Swiss government.
You know that as a dinobaby, I admire American big tech companies and their smooth integration into the business practices of other countries. The diplomatic Palantirians view Switzerland as an annoying country which speaks Italian and French, not just German. But not even the German speaking Swiss workers at Republik bought into the somewhat wonky belief that a company had a seeing stone. If that gimmick worked, wouldn’t the Palantirians have known the story was going to occur and taken steps to change the firm’s behavior. With foreknowledge via the seeing stone thing, PR could have eliminated the cause of the story.
That raises a question, “Maybe the seeing stone does not exist?” Therefore, Palantir does not see the future. It creates its own future. This particular interaction with a country fond of appropriate behavior determined that those who believe fantasy fiction are, like the original author, a step beyond eccentric. But I am a dinobaby, and I absolutely believe that the high-tech bros should rule the world. Two classes: The elite and the cows or cow like humanoids.
Stephen E Arnold, July 14, 2026
News American Style: Read? What Is That?
July 1, 2026
It is time to scythe quiet thing out loud and this information is confirmed by Reuters Institute’s Digital News Report:
- “Social media and video networks have overtaken both television and owned news websites as the most widely used source of news globally, for the first time.
- Trust in news has fallen to its lowest recorded point since the report began tracking it in 2015.
- Interest in news has declined so sharply that a quarter of the global online population now qualifies as casual or passive consumers at best.”
The Media Stack shared these facts in the article, “The Numbers UK Publishers Were Dreading.” The Media Stack doesn’t mince words:
“In the UK, all three trends are present, and playing out against a domestic backdrop of specific institutional turbulence that the report captures with unusual directness.”
The report finds across forty-eight markets that 54% of respondents use social media/video networks for news before TV news at 52% and news websites 51%. AI chatbots were used at 56%. The numbers don’t seem to be off by much, but it is the trajectory that is worrisome, because use of TV news and websites has fallen 12-13% since 2020. It means this:
“Social media and video networks have held broadly stable in absolute terms, meaning the shift into the lead is as much a consequence of legacy source decline as platform growth. The effect is the same regardless of cause: publishers’ branded digital properties, which took years and significant investment to build, now reach a smaller weekly audience than the platforms where their content frequently ends up.”
Search referral traffic is going down hill too:
- “Google organic search traffic to news sites fell 33% globally between November 2024 and November 2025.
- The United States saw a 38% drop over the same period.”
This means that publishers have deprioritized audience reach in favor of engagement with a smaller, niche audience. The revenue model for this new trend is still being worked on.
Chatbot usage is up. One of the reasons is due to the ability to ask the chatbots follow up questions:
“This suggests a use pattern oriented around explanation and contextualisation rather than simple headline delivery, the kind of depth that good journalism should, in theory, also provide.”
Trust with traditional UK publishers is also down, down, down! Why? The culprits are misinformation, direct political attack on journalism, changing consumption habits, and categories that have always scored low on trust. The write up asserts:
“Trust is corroding at system level even as individual brands hold on. And a generation is arriving that will need to be reached on entirely different terms, through entirely different channels, if it is to be reached at all.”
This report takes a broad look art UK media outlets so it is probably objective. What do these allegedly accurate data suggest about Google’s providing users with AI snips? Will US online users read or will these advertiser-targets gravitate to TikTok-style videos for news and other information? My answer: It’s video.
Whitney Grace, July 1, 2026
The Guardian Takes on the NSA, Not the UK NCA. Why Stray from Home?
June 29, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
I am curious when another country’s big time newspaper focuses its scarce (please, please, donate to us) resources on another country’s governmental apparatus. I am curious about why a newspaper published in the UK ignores some local stories in order to recycle information about a US governmental entity?
I have some hypotheses, of course. First, let’s look at what the story with this newsy, GPT hunting title:
The write up says after a video ad and this begging for dollars message:
I persevered and noted this passage:
[Democratic Senator Mark] Warner quoted Gen. Joshua Rudd, head of the National Security Agency and U.S. Cyber Command, stating, “This tool broke into almost all of our classified systems, not in weeks but in hours.” Both the NSA and an Anthropic spokesman declined to comment on the matter.
Am I stunned by this revelation? Nope. I don’t want to rain on any real news outfit’s parade but most government systems have vulnerabilities; for example, the UK has experienced some excitement over the years. (Nope, I won’t trot out some fun examples. Yandex’s Alice can be helpful if you are curious.)
Thanks, MidJourney. You have captured what life was like in the late 19th century. The policies have persisted to some degree. Imagine. Detail! Context! Work!
Okay, back to the Guardian’s article about the US government’s security gaffes:
They [unnamed sources, the best kind] acknowledged Anthropic’s Mythos models are “quite good” at identifying software flaws and weaponizing exploits, but stressed they are “not uniquely good at these tasks.” Many signatories noted they regularly use other foundational and open-source models for security audits and training, warning it is dangerous to remove the best cyber defense capabilities “without a good reason” as America’s adversaries rapidly advance.
Several observations:
- A senior official in theory is supposed to know everything about the organization. However, that is a great idea for an introductory course in organizational theory. Is any government agency, it has been my experience that presenting information of which the “top dogs” were 100 percent totally unaware is more common than not. I would suggest that it is the norm. A week ago I alerted a senior government official about a development that would have an impact on cyber crime in the next three months. What did I hear? Crickets. Top dogs listen selectively and forget unless the fact is needed in the next minute or two.
- Taking a statement about AI which adds to the paranoia and fear about smart software taking jobs, exposing exploitable security flaws, and making a secure agency just look stupid is an addition to the information payload in the write up. I know that European entities are using smart software to identify and assist in plugging security flaws. These are, gentle reader, everywhere. For a half century, software has been written with more thought given to a delivery date than security. Why? Speed, cost, and management ignorance. Every country’s commercial and governmental entities have certain blind spots. These are quite evident when it comes to software.
- The reality is that the Google transformer method can write code (some good and some really crappy), find bugs (some significant and other irrelevant), and scare the pants off people who have watched the motion picture with a movie star who said, “I’ll be back.” Yep, name a large language model. Ask it an orchestrated series of prompts. The outputs will suggest bugs and also code that will allow a person to exploit a list of security flaws from the 1980s. Guess what? Some of those legacy malware methods work like a champ today even when smart cyber security systems are humming along and spitting out invoices.
My personal view is that this story about the lack of security at an important US government agency is just an attempt to further reduce the reputation of cyber security professionals in the United States. That’s okay. In this blog, I routinely point out management failings. I think this story could have been recast by providing some specifics about the UK governments cyber security issues and mentioning that US models as well as those from China and France can perform similar outputs. But mentioning other countries’ issues, the effectiveness of Mistral and Zhipu AI can do essentially the same flaw finding function.
Net net: How about a more balanced and thorough approach to a very important topic. Less begging for dollars/pounds and more old-fashioned reporting?
Stephen E Arnold, June 29, 2026
Okay, Google. Dispatch Those Lawyers to Make This Little Problem Go Away
June 15, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
I gave an NFAIS talk more than a decade ago. The subject was a question, “Is Google a publisher of sufficient innovation to be considered an innovator in publishing like Johannes Gensfleisch zur Laden zum Gutenberg. I think I cast Gutenberg’s modern day equivalent as Sergey Binus Laurentius Pagina.
To be honest, no one in the audience gave a rat’s patootie. The Google in 2009 spit out a list of links. The “user” clicked and read the links that seemed to be on point. No one asked about [a] stop word lists, [b] implicit and explicit “editorial” policies, [c] posting explicit dates to show when the site was last indexed, [d] Google’s indexing methods that essentially divided the world of Web pages into bluebirds, canaries, and sparrows or [e] Google’s patent applications and patents about assembling, presenting, and outputting the content of Web pages in ways that satisfied Google’s intent, not the intent of the source data or document.
Frankly I did not care either. I was working on other projects and had no illusions about the information my team and I had gathered about Google’s editorial and creative methods. I think some of the publishers and commercial database producers in the audience fell asleep. That inattention cost some of those folks their jobs, their self-esteeem, and their identity in years following my lecture. I don’t recall anyone throwing a tomato at me, but the sound of the gentle snoring still haunts me.
Well, there may be a chance for NFAIS and commercial publishers to take back come of the intellectual ground lost in the last 17 years. Maybe. Here’s an illustration making Google’s DNA clear: Search and advertising. Other functions like outputting ad reports, AI summaries, and recycled content as Google News are a consequence of these two strands from the clever innovation of Sergey Binus Laurentius Pagina.
Google must output / publish content to make its business model viable. Thanks, Midjourney. Good enough.
AI does not change the DNA. AI is simply another function enabled by search and advertising. Personalized YouTube lists — That’s an example of the digital Gutenberg output. The use of Google search by a single Web site — Those result pages are an output of the digital Gutenberg at work. Those nifty ad reports — Yep, another example Google publishing information. My argument in 2009 was simple: Without publishing output, Google’s business collapses. Just like a traditional newspaper, output is in its beloved protein structures.
I read “Landmark German Ruling Declares Google’s AI Overviews Are Google’s Own Words and Makes It Liable for False Answers.” I am quite sure than no one in Germany read “Google: The Digital Gutenberg” but I thought I did a good job explaining that Google is indeed a publisher. If the German judicial officials had read the book, more than AI summaries could be found as original output.
The write up says:
A German court has ruled that Google is directly liable for what its AI search overviews say. Previous case law shielding search engine operators from liability doesn’t apply to AI overviews.
My view is that lawyers are aloft. If this Munich decision gets traction, other countries and maybe the EU will extend the concept. The digital Gutenberg could find itself in a pickle barrel it did not plan on entering.
The write up pointed out:
The court also found that the AI overview made claims “that are not even made in the search results.” None of the linked sources drew any connection between the plaintiffs and the shady companies the AI mentioned. The court called these “the defendant’s own statements.” Google built the AI, Google offered it to users, so Google owns what it produces, “because it alone has influence over the AI’s offering and the algorithms with which the AI operates.”
The Google argued:
…that users could check the linked sources themselves to verify whether the AI summary was correct. Users generally knew “that information generated with AI should not be blindly trusted,” the company claimed. That’s a remarkable statement given the scale at which Google serves AI overviews. It’s also not entirely true, since the connection between sources and generated content isn’t always there. The court rejected this.
But what about free speech and in the US Section 230? The write up sheds some light on these ideas. I quote:
As if the rest wasn’t bad enough for Google, the court also went after free speech protection for AI-generated content. An AI’s opinion is “not the expression of an acquired conviction of the persons expressing it, but the result of an algorithm,” the court wrote. Offering AI-powered research is “above all an expression of Google’s business activities” and “at most a secondary expression of an interest in being able to freely express one’s opinion and beliefs.” When weighing the plaintiffs’ privacy rights against Google’s interests, Google had to take a back seat, especially since the challenged statements were based on untrue facts. The AI had linked the plaintiffs to companies that, according to sworn affidavits, had no connection to them whatsoever.
In my lecture I think I pointed to Google patent application (US2008/0275763). The document said to me, “Hey, dinobaby, pages can be created by Google specifically to present ads and certain content. I counted any Google page with this type of content presentation as published information. I wonder how lawyers from unhappy creators would interpret the injection of the three types of ads plus inserted content if the creator’s content were de-published. That action would be an editorial decision by Google’s “agent” and, in my opinion, constitute an intentional action comparable to an editorial decision.
Several observations are warranted:
- Google’s attorneys will endeavor to appeal and overturn the Munich ruling. That tussle will be interesting to observe.
- Google has operated with polite nods to rules and regulations. Munich might be a turning point if Google’s lawyers fail to prevail.
- Google, as I stated in 2009, is a publishing machine. Its search result pages are published like any other Google system outputs.
Net net: More than two decades have passed since Google became widely available. The Munich decision, like the French judiciary’s arrest of Pavel Durov, may be an important legal action.
Stephen E Arnold, June 15, 2026
China Smart, US Dumb: AI Systems Improve Themselves
April 27, 2026
![green-dino_thumb_thumb[3]](https://www.arnoldit.com/wordpress/wp-content/uploads/2026/04/green-dino_thumb_thumb3-4.gif "green-dino_thumb_thumb[3]")
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
Before I comment on “Self-Improving AI Model Has People Talking – For Good Reason,” I want to mention that Telegram incorporated a Chinese AI model in its Messenger app. The app (according to some social media mavens) changes the language from what the user wrote to what the Chinese app thinks is a more appropriate selection of words. Try it yourself with Taiwan as an independent nation or some Uyghur oppression lingo. Now, back to the write up about a Chinese innovation in making AI models smarter, better, and probably cheaper.
An artist’s rendering of Shaped Information Inc. Thanks, Venice.ai. Good enough.
The New Atlas article said:
ASI-Evolve, built by researchers at Shanghai Jiao Tong University, works [to improve itself] by running a continuous loop that mirrors how humans would put this type of technology through its paces. Essentially, it creates variations of AI models, alters how they’re trained and adjusts the data they learn from. It then runs its own experiments to see which clone performs better, using those results to guide what it tries next.
I think this is an interesting statement. But there’s more. The article added:
By generating ideas, testing them and refining the results in a self-improving loop, ASI-Evolve mirrors the trial-and-error process of not just AI model building, but also science and math research. As such, it raises the possibility of accelerating discoveries in fields where progress is slow due to human researchers testing many possible outcomes.
The software figures out solutions in a way that suggests it is behaving more like a curious human than a Microsoft Word document and its obtuse approach to numbering. The benefit of this approach is that humans are still needed. I think that the longer-term goal, however, is to eliminate some humans, possibly those in the United States. I don’t want to be a Debbie Downer but China is presented as just making so many strides. Its robots can win mini marathons as long as the gizmos don’t run out of power. (Were those humans chasing the Chinese robots doing more than admiring the pace and gate of the digital marathoners?)
The write up points out that green is in the smart improving software’s future:
It’s worth noting that the researchers haven’t detailed energy costs of running ASI-Evolve, but its speed and efficiency, and closed-loop self-learning, suggests it’s nowhere near as power-hungry as leading models trained on enormous datasets. AI agents are expected to drive China’s next stage of development – one where new data centers are also mandated to be powered by green tech.
Like the videos of a young girl repairing a generator with hand tools in a farm yard, this write up is one of a sequence of shaped communications. These are designed to remind the US that China is making strides in applied technology and more esoteric flavors of innovation.
This is just business as usual: Pump information into the US-anchored digital publishing system and let people absorb this “shaped” or “weaponized” information. Works great most of the time.
Stephen E Arnold, April 27, 2026
Level Up: Control Information Flow with Litigation
April 16, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold. I find it interesting that AI detectors identify my writing style as AI output. I suppose I should be flattered, but I just don’t care.
I worked with a fellow who had one trick he used in meetings. When a discussion was not going the way this wizard wanted, he would say, “Let’s pop up a level. This idea is just part of a larger picture. Let me describe what I see.” The fellow would then reframe his hobby horse to be the stallion of this “broader view” or what I called his “meta argument.” Most of the people in the meeting were thinking about nuts and bolts. This fellow would talk about owning a business of which nuts and bolts were just a tiny part and not too relevant to his big picture. Did this work? Yep, most of the people at this particular firm loved to think about “next level” concepts. Novelty can be compelling. Sometimes.
Thanks, Venice.ai. Good enough but this is quite overblown even for superior mythical entities. What’s with the lightning bolt in a sensitive area?
“Can AI Judge Journalism? A Thiel-Backed Startup Says Yes, Even If It Risks Chilling Whistleblowers” describes a company that wants to “to adjudicate the truth of journalism.” Pay money. Sue a journalist. Bright people. Smart software. Big money. The write up says:
Thiel, who funded the Gawker lawsuit partly in defense of the individual right to privacy, has long been critical of the media. D’Souza says his goal is to restore trust in the Fourth Estate, which he argues has collapsed over decades. Critics, including media lawyers, warn Objection could make it harder to publish the kind of reporting that holds powerful institutions to account, particularly if that reporting relies on confidential sources. Anonymous sources have played a key role in major award-winning investigations into corruption and corporate wrongdoing. These are often people who are at risk of losing their jobs or facing other retaliation for sharing important information. It’s the journalist’s job — alongside their publication’s editors, peers, and lawyers — to ensure that those sources are reliable and not acting out of pure malice and to verify the information they provide.
Yep, we are popping up a level. The individual story is an example of thinking at a very small scale. Viewed from a “meta” perspective, truth, privacy, and other Superman ideas are protected. The people who are thinking about whether a single article is biased, incorrect, unverifiable are problematic. The fix is “let’s pop up a level.”
Based on my experience watching this type of argument unfold in the meetings to which I referred, there are some issues:
- Sounding “good,” “logical,” and “correct” is different from the actual issue discussed at the micro level. Micro decisions are the basics of solving a problem that has arisen
- The entity that has a “superior” view or knows the trick to solve a problem in linear algebra seems like a genius but is actually using information to shape an agenda. The method works and most people don’t recognize the game play.
- The stated goal can be different from the “real objective.” People are dazzled by what’s novel and can fail to see that a tactic is just part of an effort to impose controls on certain behavior. The pop-up-a-level guy says, “Hey, look at it from this perspective. This view solves the problem and eliminates the detail and tail chasing.”
Why do I view this announcement of a for-fee service to chase journalists? I think this is a meta play, and the actual objective is to provide a “logical” and “good” method for controlling information flow. How confident am I that the purpose of the adjudication is to advance a control narrative? I would suggest a 95 percent confidence score.
Will the technology work? It depends on what one means by “work.” Do Anduril drones work on the front lines in Ukraine? Does Palantir occupy a leadership position in AI? Does Objection deliver the “truth” of a story? I have some partial answers to my questions. I am not sure mine are correct. How about popping up a level to learn the “truth”?
Think about that question.
Stephen E Arnold, April 16, 2026
Telegram Rewrites User Messages and Censorship Blooms Like Flowers in the Township
April 8, 2026
Telegram Notes is almost back online. If there is art in a GOAT post, the image emerged from smart software. Grandpa Arnold is not Grandma Moses. The words, however much they appear to be output from a smart system, have been output by a still living dinobaby with some humanoid features.
Our informal blog Telegram Notes is back online. Recent Telegram-related posts appear to have been blocked, scrubbed, or otherwise disappeared from LinkedIn, a couple of LE and intel-related groups, and from a freedom loving, privacy oriented blogging service in lovely South Africa. No surprise to me.
Filtering and Monitoring?
Microsoft and LinkedIn should know what my team and I have been doing. If you don’t understand the methods used to monitor and respond to information not on the agenda of the world’s best developer of email programs, just navigate to Browsergate and work through “LinkedIn Is Illegally Searching Your Computer.” With these alleged data, one wonders, “How did Microsoft/LinkedIn get smart software so very, very wonky?” And the South African operator? No clue. I learned that my content was “flagged” as AI generated. I am not sure whether to be flattered or baffled.
Thanks, Venice.ai. You did not tell me I was creating an abusive image today. Outstanding. Too bad the illustration is just good enough.
Cape Town Cares
The South African censorship move was courtesy of a free spirit in Cape Town, South Africa. (Those townships on the road from the airport to the central city are a delight. Shabeem thinking may be prevalent outside of these residential spaces.) I know that this dinobaby is not the swiftest 82 year old prowling the rest home’s grounds, but it is interesting to think that I am relying on smart software to produce snappy phrases like “write up” and “leadership” in my idiosyncratic way. I even have a GenX or GenY explanation of the content removal. If you want to check out where you too can have your content sanitized, navigate to the cute privacy first just censored BearBlog at https://bearblog.dev/.
Back Online with Telegram Notes… Sort Of
We have managed a new post about Pavel Durov’s Telegram. The write up explains that Telegram has deployed an AI editor in its Messenger container. What’s interesting is that when a Telegram user enters certain phrases, the Chinese large language model changes user content to an acceptable Chinese Communist Party form. You can read that at https://shorturl.at/GhSST.
We haven’t worked out the listing of articles so each appears with the most recent at the top of the stack. There are some definite spacing weirdnesses, and one of my semi-capable team is working on that issue as well.
Observations
A few dinobaby thoughts zipped slowly through what passes for my brain these days:
- When writing about a service like Telegram and posting that information in the US, censorship happens. The vendors insist it doesn’t, and I know that vendors especially like the BAIT (big AI tech) outfits never, ever fabricate or stretch the truth.
- Whatever angle we are taking on Telegram, Pavel Durov, the TON Foundation, TONcoin, TON Strategy Company, AlphaTON Capital, and the cast of characters associated with this sprawling system — someone, somewhere is taking the trouble to block the content.
- Online has become an interesting place. It is populated by Softies, Cape Towners like Herman Martinus, MBAs at LinkedIn, and BAIT guardrail developers.
I still find people like Moti Cristil, Oskar Hartmann, Manny Stotz, Yuri Mitin, and the GOAT himself Pavel Durov fascinating professionals. When you are 82, I hope you find a hobby as fulfilling as mine. Most of the stories about Manny, for example, do not get the timeline for his Kingsway plays straight. Few know about the estimable elite-only Equium Club which operates in places where movers and shakers move and shake. Not too many people have paid attention to the interesting career trajectory of a Ph.D. who worked for Cambridge Analytica and now serves as the president of a company engaged in AI compute leasing and online games named AlphaTON Capital with Enzo Villani, a former NASDAQ professional assisting the firm. The share price of ATON (the Mitin / Villani company and not the Russian bank of that name) was about $0.32 per share on April 7, 2026. That share price is below the NASDAQ threshold for a listed company.
The people in the rest home think I do real work. Nope, I have a hobby, and it is attracting attention from LinkedIn and the fellow who thinks his Bear Blog product is a garden. I think I understand. Gardens have to be pruned. I wonder if Herman’s AI identification system thinks this post you are reading was generated by one of the large language models. And LinkedIn / Microsoft, how can you have so much information about LinkedIn users and you continue to output truly peculiar software. Outlook in space? Nope, a giant technology company lost in space and flailing because its tether had a quality problem.
Oh, well, back to my hobby.
Stephen E Arnold, April 8, 2026
AI: Helping Humans Be Stupid
March 9, 2026
![green-dino_thumb_thumb[3]_thumb_thumb](https://www.arnoldit.com/wordpress/wp-content/uploads/2026/03/green-dino_thumb_thumb3_thumb_thumb-1.gif "green-dino_thumb_thumb[3]_thumb_thumb")
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
I read “Scientists Warn Fake Research Is Spreading Faster Than Real Science.” The write up contained no surprises. Humans love short cuts, convenience, and cute ways to snooker an advantage. The write up presents what I thought was obvious as an important insight. Wow.
The Science Daily reports:
A new study from Northwestern University warns that coordinated scientific fraud is becoming increasingly common. From fabricated data to purchased authorships and paid citations, researchers say organized groups are manipulating the academic publishing system.
I have mentioned in my assorted writings that Dr. Gene Garfield, the fellow who made citations an indicator of importance, knew that the system would be gamed. He was correct. It is trivial to get colleagues, friends, graduate students, and Fiverr.com workers to pump, reference, and backlink to benefit a person, a company or an idea. (I provide an example of a publicly traded company flooding the zone with shaped messages in this article.)
The “Scientists Warn…” article points out:
…fraudulent studies are now appearing at a faster rate than legitimate scientific publications.
What’s this means for smart software? Answer: It will not only hallucinate but it will output incorrect information. Do you want your doctor to trust an AI to diagnose what’s wrong with your child? How about an AI to figure out the doses of chemo for your cancer-ridden mom? Do you want to be admitted to graduate school by an AI? Sure, you don’t, but you will have little say in the matter.
AI is going to operate just like the helpful bots in the Telegram platform or the add ins available in the Claude marketplace. Unless one takes special care, those software daemons are just going to do their thing and use fake information. Think about that when you ponder the implications of your retirement savings invested in the company pumping out shaped information to paint a very rosy investment picture.
Is a single scientist going rogue? Nah. The Science Daily story says:
…the researchers identified coordinated operations involving paper mills, brokers and compromised journals. Paper mills function like production lines for academic manuscripts. They produce large numbers of papers and sell them to researchers who want to increase their publication record quickly. These manuscripts often contain fabricated data, manipulated or stolen images, plagiarized text and sometimes claims that are scientifically impossible.
Can the scientific, technical, and medical professional publishers fix the problem in their peer-reviewed publications? I suppose but there are several hurdles:
- Money. Professional publishers don’t want to invest in what is a black hole problem
- Authors. Why stop? If a topic is sufficiently narrow, the only people who can identify a fake is a graduate student who made up the data in the first place. Example: The Harvard ethics professor who made up information for an ethics paper.
- Readers. Humans read less and less and fewer humans appear to read critically. Smart software companies don’t read; they process and then synthesize information and spit it out. Readers are not very good at finding fake data when writ large like the economy is great or tiny like information related to the DNA of Etruscans.
I want to suggest a fix that almost no one on the planet will be interested in pursuing. Ready or not, here’s my recipe:
- Take learning seriously
- Read critically and look for anomalies and discrepancies, then check them
- Do this throughout life
- Demonstrate this approach as part of the furniture of life.
Spoiler: I estimate one percent of the people in the US will follow this recipe. I think the tech bros want sheeple, not people who question.
Stephen E Arnold, March 9, 2026
Professional Legal Publishers: The Bell Tolls in D Minor
February 10, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
I read “Anthropic’s New AI Tools Deepen Selloff in Data Analytics and Software Stocks, Investors Say.” The story is published by the outfit that reminds me it is in the “trust” business. I think the company is trying.
Here’s quote from the write up:
Toronto-based Thomson Reuters, which owns the Westlaw legal database, slumped by nearly 18%. It is on track for its biggest daily loss on record and lowest close since June 2021. “I think Anthropic came out with some plug-ins to tackle the legal space,” said Mike Archibald, a portfolio manager at AGF Investments in Toronto. “Obviously, that’s where Thomson Reuters generates a good chunk of their revenues. Sometimes the market just shoots first and asks questions later.” Thomson Reuters, which is also parent company of Reuters News, is set to report its fourth quarter earnings results on Thursday. Its shares are now down 33% year-to-date after dropping about 22% in 2025.
I really want to quote from John Donne’s Meditation 17 in the 1624 Devotions upon Emergent Occasions. You probably remember the line “Never send to know for whom the bell tolls; it tolls for thee.” But I won’t instead. I will suggest you recall that Mozart’s Requiem relies on D minor. You could I suppose find a YouTube version of this musical composition. I didn’t I just remembered the theme of the Confutatis. Good enough.
Thanks, Venice.ai. Good enough.
Why the reference to bells and a funereal work by Wolfgang? I was sparked by this passage in the cited story:
Britain’s RELX and the Netherlands’ Wolters Kluwer, both providers of legal analytics services, fell 14% and about 13%, respectively. RELX shares have now almost halved from their peak last February and on Tuesday were set for their biggest drop since 1988. Its dramatic reversal highlights the pressure AI is exerting on Europe’s software sector. Other professional services firms closed sharply lower too. Factset Research fell 10.5%, Morningstar lost 9% and LegalZoom slumped 19.7%. In London, Experian, Sage Group, London Stock Exchange Group and Pearson fell between 6% and 12%. Traders and analysts said investor fear often outweighed company fundamentals.
I don’t know much about today’s professional publishing business. Based on my bumping into these firms over the years, I know that each adds value to information. In the legal sector, smart people review documents and add notations. One outfit put hundreds of lawyers to work creating annotations very much like to work done by monks in medieval scriptoria.
Professional publishing firms’ share prices have taken a hit (temporary, probably) because smart investors realized that AI can do this type of knowledge work at a lower cost. It does not take much imagination to see a workstation trained to do legal content sitting in a closet at a big law firm or maybe in a rack in a data center somewhere where costs are low. That workstation can do to public documents and to “firm only” content what the professional legal publishers and adjacent firms do for less money. Since professional publishers rely on a relatively small number of very big law firms and government agencies for their revenue, the threat may give some investors pause.
Yes, there are actions the professional publishers can take. However, these firms have been telling themselves that their AI experiments and products are right in step with the needs of the law firms, the accountants, the consultants, and other markets. Unfortunately professional publishers believe they are often the smart people in the room. I would suggest that these individuals are indeed smart, but there may be even more intelligent people working at Anthropic-type outfits. In my experience, what I call Silicon Valley companies don’t see the world the way professional publishers do.
That’s the problem. Professional publishers innovate slowly and in a tightly constrained mind space. Those wild and crazy Silicon Valley types just slap tech on a problem, send it out, fiddle around, and upgrade in what I call fast cycle mode.
Do you hear those D minor vibrations? I do. But John Donne said in “A Hymn to God the Father”:
“I have a sin of fear, that when I have spun
My last thread, I shall perish on the shore;
But swear by Thyself, that at my death Thy Son
Shall shine as He shines now, and heretofore;
And having done that, Thou hast done;
I fear no more.” [Emphasis added by me]
Or at least until the next quarterly report.
Stephen E Arnold, February 10, 2026
-
- Subscribe to Beyond Search
Feature archive
News archive
-
