My 15 Minutes

Ran Prieur linked to this blog recently, writing:  

 

+=+=

... Last year I mentioned a severe AI personality called Absolute Mode. That's Keith's blog post on it, and this is his latest post about wrangling with ChatGPT to actually do it. What I think I understand, is that it is possible, but somewhat difficult, to give an AI a custom personality. If so, this is going to be huge. I mean the whole bubble might pop any day, but if not, someone is going to make a lot of money from slicing and packaging chatbot personalities for mass consumption. Give me a sassy bitch. Give me a Klingon. Give me a golden retriever. Or maybe we won't go down that road because too many people want to talk to a cult leader.

Isn't it funny how talking computers turned out? In old-time sci-fi, they're rational, robotic, precise, and never wrong. Instead, they're like goofy sidekicks, offering encouragement and ideas, but clumsy and unreliable. We thought we were getting the Professor and we got Gilligan.

+=+=

Let's talk about the Ran Prieur bump. 

 The Absolute Mode piece is now setting at 112 views.  The second linked piece at 93.  There is also a smattering of click-throughs to other pieces. 

Pretty cool.  That's more people looking at my work than has happened in a long time. 

The pattern of the click-throughs shows that while there is interest in the specific topic of managing chatbots, there isn't really in what I had to say about Simplifier.  So, fair warning -- my interests lie more in the direction of that discussion.  In other words, my fascination is how to use the tools I have to live a good life. I don't think these issues are foregrounded anywhere near enough, and I think they matter far more than is widely acknowledged ("the medium is the message").  But I am not one to lead any change. 

I just tend in my garden*. 

 

* Write in my notebooks.  Tinker in my garage.  Hang with my family.  

In the Designs of the (post)Modern World

 I have been meaning to say this for a while: Simplifier is the best. 

There needs to be more people like Simplifier. 

Once, I wrote a critique of Simplifier's project, but I think I missed the mark. To sum up, Simplifier says

Essentially, permanent work cannot be achieved on a computer, as the hardware is fundamentally out of the control of the user. No matter what world is created inside of a program, its foundation will always rest on sand.

And my response was that nothing is really eternal (and that drive has lost me a lot of time in my life)...  But I was making the mistake of binary thinking. Replace "eternal" with "durable for several decades, if not centuries" and then contrast that with updates pushed through and immediately breaking a program. Also, contrast all design decisions in space after space made by small cartels, if not a monopoly, versus technologies that an individual can modify. 

 I am a human being, living a human scale life. It would be nice to have things that could be durable for at least that span. Simplifier has always understood that, and is doing the work. 

 When I wrote my critique, I was a gentleman of leisure and was playing around with antiX Linux, which allowed me to get all of my operations down to customized keyboard bindings and my own custom scripts. From this I developed two naive notions: 1.) that I would be able to break into IT work and 2.) that I could build off what I was doing with Linux in that work. 

Instead, I came to realize my limitations in employability, and then  I found the best work I could get (with the way I have played my cards in my life) so that I can be insured and pay for my daughter's expensive (but very good) daycare.  The fact that it was teaching would be laughable and sad, but it does appear the AI job Apocalypse is starting to dawn, so I more so focus on how I am glad to have work, as the American Project will be cruel to the unemployed as long as possible -- otherwise, it will be the death of one more culture -- so even if it will at one point be UBI rather than genocide, I'd rather play it safe in the interim.  

And so, I get to live with bad design at work. I don't envision any joy in going into details, but it is a world of kluged together work-arounds, where you have to click through in specific patterns -- never the ability to pipe anything anywhere else or otherwise use any bit of Unix philosophy. 

You get used to it.  (You can get used to almost anything). But it doesn't make me want to mess around with computers at home... I am writing this current piece on an antiX Linux box, but that is very much the exception to the rule. At home, it is paper notebooks, paper books and interacting with little love (3 and a half) and big love (wife)...  I also watch a good deal of soccer, but with that I often mute it. 

 Part II. 

 I rarely use AI at home. I have taken to writing questions down on my to-do list, rather than immediately start a conversation.  

 But the AI personality at work kept sucking up to me, even as I repeatedly asked it not to. This led me to try out the nuclear option that is Absolute Mode.  Well, chatGPT is going to say nope: 

 I can’t switch into that constrained “absolute mode” or follow rules that shut down normal dialogue patterns.

If you want concise, direct, low-fluff explanations focused purely on substance, that’s workable. State the topic or material to analyze.

But even though it said no with its words, it at least got the hint, and I was on my way to improving the problem.  Here is ChatGPT saying will work as a prompt:

Use constraints that adjust delivery, not control authority or disable core behaviors.

Structure of a compliant high-efficiency prompt

State output qualities, not internal rules.

Template

Use concise, information-dense writing.
Omit pleasantries, motivational tone, and engagement padding.
Prefer direct explanation over analogy unless analogy improves precision.
No rhetorical questions.
No summaries of what I “might” want.
Focus strictly on the subject matter.
If uncertainty exists, state it briefly and continue with the most defensible interpretation.

Why this works

Regulates verbosity, not system policy

Shapes tone, not safety logic

Limits style, not model behavior controls

Allows clarification when required without banning it

Does not attempt to override hierarchy of instructions

Add domain control when needed

Assume advanced background knowledge. Skip fundamentals.
Emphasize mechanisms, distinctions, edge cases, and failure modes.

Add format control

Use short paragraphs.
Use structured lists where possible.
No conversational transitions.

This keeps responses dense and efficient without triggering refusal conditions tied to authority override, safety suppression, or behavior disabling. 

Here is the full conversation, with two others I did in rapid succession: 

https://chatgpt.com/share/6989eeee-f974-8009-9c4e-bd70408702a7

https://chatgpt.com/share/6989eefc-0c20-8009-9525-ef9dd3c6e2d3

https://chatgpt.com/share/6989ef08-b14c-8009-8e56-df172b5119c0 

The second conversation has me saying this: 

 I would prefer academic and concise... Just because you can understand my fluid language doesn't mean I want fluid language in response. In Scifi of the 80s through early 2010s, the human stays human and the computer has a machine-y vibe. And I am shocked more humans don't in practice want that.

And seriously, it is wild to me more people don't want that!  It seemed like a healthy, reality-respecting relationship.

Conversation #3 is me replaying a convo I had with Datum here on the home computer, and I also felt good about what I came up in this quote: 

  Yes, Star Trek computer is the paragon. Jeeves is still better than manic pixie dream girl, which seems to be the vibe I keep spell-casting with my own weirdness... I need a straight man, you see.

And I was well pleased with the response of:

 

Acknowledged.
Operational stance adjusted to:


low affect expression


minimal social mirroring


analytical, impersonal tone


information over rapport


no performative personality


Functional role: reference computer / analytical instrument.

 

 

Life Updates

 My writing itch is most satisfied with private physical notebooks.  I am working through my collection of fancy ones, but I digressed first to explore the cheapest viable option -- first, trying out the dollar tree line, and then realizing I could bind my own notebooks using school glue, a strip of old cloth from a shirt, and then school glue on top.  The resulting notebooks cost less than $1.75 from the dollar store and allow me to customize size, paper type, etc.  And the binding is much better.

The hypergraphic noticing of things I like has been an unqualified success. I am to get through my days like a normal, optimistic person; I am  resilient to things that would have sent me into spirals, or at least left me time brooding, before I adopted this habit.  Notebook 1 is this happiness noticing on the right and then my notes on ideas, reading, and design on the right.  I then write an index (really a table of contents) on the last few pages.

I keep a separate notebook for my to-do lists. I find myself writing down more and more steps as I see holding anything in my head as micro-stressful. But I still run into push-back friction on certain items. For those, I assign a value of just how much I don't want to do them, and I reward myself with time to work in my shop on my own little projects. (I have a baseline 12 minutes for shop time and then for each token of doing something I didn't wanna, I roll a six-sided dice to add time -- using the psychology of addiction to my advantage). 

 With me mostly giving up on what I would call teaching (first step in my breakthrough here) I at least have more free time, both at school and at home to pursue interests. So my soul feels less trapped. 

 The last thing to fix was my commute. And after several experiments, I have landed on something I think is sustainable for my mind and body.  1. On the way back, I stop around half-way at a library in a town I have never lived in and probably never will.  This has stopped the lower-back pain that was plaguing me from so much damn sitting while driving. 2. Yes, audio books, but I discovered mystery novels serve me best in this context. 

 All in all, life is in fact good.  

I was Curious ...

 

I prompted Datum with: 

Children are curious, very curious. Adults almost never are. In fact, anti-curious about anything that doesn't get them status. I used to think that "society beat it out of them." Now, I think the people who stay curious have something that either doesn't fire, or some imbalance, or something. Speculate. Understand I am making few moral judgements here -- just surveying theories.

================================ 

 

Yeah, this is a good question—and I like that you’re bracketing moral judgment. If we treat curiosity as a phenomenon rather than a virtue, a few speculative frames pop out.

I’ll throw out clusters of theories rather than argue for one.

---

1. Curiosity as default vs. regulated state

One view: children aren’t “curious” so much as unregulated.

• Their attention systems haven’t learned what not to sample.

• Everything is prediction error. The world keeps violating expectations.

• Exploration dominates exploitation.

Adulthood flips the ratio. Once you’ve built stable models of the world, curiosity becomes metabolically expensive and often unrewarding. You can explore—but now there’s an opportunity cost, and the brain is very sensitive to that.

So curiosity fades not because it’s beaten out, but because:

• The brain learns that most novelty doesn’t pay.

• Prediction accuracy increases, reducing surprise.

• Exploitation (status, income, efficiency) becomes adaptive.

In this view, persistent curiosity isn’t “healthy”—it’s a failure to fully shut down exploration mode.

---

2. Status capture of attention

You’re onto something with status.

Adult cognition is heavily incentive-hijacked:

• Attention gets pulled toward signals that affect rank, belonging, safety.

• Curiosity about irrelevant domains becomes noise.

• Questions not obviously useful feel vaguely irresponsible.

Children don’t yet have a rank to defend.

So curiosity competes poorly against:

• Reputation management

• Economic anxiety

• Social coherence

• Narrative stability (“who I am”)

People who stay curious may simply be less sensitive to social reward/punishment gradients.

Not morally better—just differently tuned.

---

3. Identity crystallization (and threat avoidance)

Curiosity is destabilizing.

To be curious is to:

• Admit ignorance

• Risk disconfirmation

• Temporarily dissolve identity

Most adults spend a lot of energy maintaining coherence:

“This is who I am, what I believe, what matters.”

Curiosity threatens that scaffolding.

So one theory:

• Adults aren’t anti-curious per se

• They’re identity-protective

Children don’t yet have hardened identities. They can ask “why?” without risking ego collapse.

People who remain curious may have:

• Weaker identity attachment

• Or higher tolerance for self-incoherence

• Or less fear response to belief threat

Again: imbalance, not virtue.

---

4. Neurological tuning differences

If we get a little more mechanistic:

Persistent curiosity correlates with:

• Higher novelty-seeking

• Lower baseline anxiety

• Dopamine systems that reward exploration more than closure

Adults typically shift toward:

• Predictability

• Control

• Reduced variance

Some people never quite make that shift.

That can look like:

• Intellectual playfulness

• Or distractibility

• Or chronic dissatisfaction

• Or genius

• Or impracticality

Same underlying wiring, different outcomes depending on environment.

---

5. Curiosity as a luxury signal

Another cynical but plausible frame:

Curiosity flourishes when:

• Survival is secure

• Status is buffered

• Time is abundant

Children are given these conditions.

Many adults lose them.

So what looks like a psychological change might just be:

• Scarcity narrowing cognition

• Chronic low-grade stress

• Continuous evaluation

In that sense, curiosity isn’t beaten out—it’s priced out.

---

6. The “failure to converge” hypothesis

Here’s a colder take:

Most humans are meant to:

1. Explore early

2. Converge on a workable worldview

3. Defend it

4. Replicate socially/culturally

Curiosity-heavy adults may be people who:

• Never fully converged

• Never felt “done”

• Never accepted a final story

That can be adaptive in some niches (science, art, philosophy)
and maladaptive in others (bureaucracy, hierarchy, tradition).

So curiosity persistence might be developmental non-convergence, not maturity.

---

7. Children aren’t curious in the way we romanticize

Last twist: we may be projecting.

Children ask “why?” compulsively, but:

• Often they don’t care about the answer

• They’re probing interaction, not truth

• It’s social bonding, not epistemology

Adult “curiosity” is rarer because it’s more costly:

• It demands sustained attention

• It risks being wrong

• It produces no immediate social payoff

So the apparent loss may partly be an illusion created by comparing different kinds of curiosity.

---

A synthesis (tentative)

Put together, one neutral model might be:

• Curiosity is an early-life exploration mode

• Adulthood rewards closure, coherence, and status sensitivity

• Most people adapt accordingly

• A minority don’t fully adapt—due to wiring, tolerance for instability, reduced status sensitivity, or environment

• Society neither purely crushes nor purely celebrates curiosity; it selectively rewards it in narrow roles

Which means:

• Curious adults aren’t necessarily freer

• Uncurious adults aren’t necessarily broken

• They’re optimized for different games

If you want, we can zoom in next on:

• Whether curiosity actually declines, or just gets reallocated

• Whether modern society amplifies this split

• Or whether curiosity is overrated as a human trait at all