Rick Rosner On, 3: AI, Archival Identity, and Informational Cosmology

Author(s): Scott Douglas Jacobsen

Publication (Outlet/Website): Personal SubStack

Publication Date (yyyy/mm/dd): 2026/06/10

In this interview, Rick Rosner and Scott Douglas Jacobsen discuss whether years of dialogue retain value in an era when AI can simulate voices, extract themes, and generate speculative models. They examine archival identity, high-IQ culture, Decree 770, reproductive coercion, and Rosner’s “informational cosmology.” The exchange turns to AI-generated mathematics, dark matter, redshift, and whether machines can help formalize philosophical physics without replacing human originality as both tool and cultural force in intellectual history today.

Rick Rosner: We have been doing this for 12 years. We have generated millions of words: three million, four million, some crazy amount. We get dozens to hundreds of hits every time you publish, which is not nothing, but it is not a lot, and it has tended not to lead to anything else.

I have been hoping that somebody could go through everything we have said together, maybe even posthumously, posthumously for me; in this version, you are still alive, and see whether there is any wisdom from this IQ guy, whether any of it was worth it. I feel like I have had some decent things to say, and that has all changed. The idea of slogging through millions of words from some guy, looking for wisdom, when now you can go to AI and ask it almost any question and likely get an answer that is better informed, and maybe smarter than anything I have to say on the subject, means that the significance of what we have been saying has been reduced.

There still might be some historical value: this is one of the last guys from before machines became smarter than humans, or before we merged with smart machines. But it is a little bit of a kick in the balls. Any comments?

Scott Douglas Jacobsen: Generally, my take on this is akin to the holodeck without being the holodeck. It is more like a text deck, where you can recreate a textual representation of a person’s edited speech or written work based on millions of words. That is a large and significant sample, more than most people would ever leave behind in a lifetime.

From that, you can get a curated, image-like representation of the individual. I think that will probably play some role in archival work in the future, because you are a minor-to-moderate public figure based on your television work, especially with Jimmy Kimmel, and on the eccentric public record around the high-IQ material. I think that is a generally consistent theme in the commentary.

Your results on the Mega and Titan tests, which were among the better-known experimental ultra-high-range tests, are also part of that record. From that, your personality is likely to be represented textually in some form. It will not be like a celebrity AI chat app in your voice and likeness, but it will be a point of interest among a number of people.

Rosner: I do not know about “large,” but that is nice in its way. It is also disheartening in its way, because people do not even need to ask me questions anymore. They can ask an AI version of me questions, and it might get reasonably close. Now, I probably have not exhausted the store of ridiculous, grotesque personal details. AI would not know that I was going to talk to you about developing a more mathematized way of wiping my ass.

Jacobsen: We did not talk about that.

Rosner: But when it comes to asking for an opinion, we should test it out. We should ask me a question, and then, before the answer is published, ask AI the same question and have it respond in the manner of me, and see how close the two answers are. I think there would be quite a bit of overlap.

Jacobsen: My take is that you can ask an AI, using the segmented work between us. So asking as if a third party about “Rick Rosner and Scott Jacobsen,” or you independently through the Mega site or In-Sight Publishing or rickrosner.org, or related material, “What are the key quotes?”

For instance, there was an interview I did with another semi-prominent member of the high-IQ community, Kirk Kirkpatrick, who scored well on some childhood tests. It has been the most viewed article of the last day or two on the website, for some reason.

It was based on commentary that we did for The Huffington Post, back when it still had a blog system. The commentary revolved around two separate phrases that you both said independently to me, but used to describe the same phenomenon. One was “superempowerment,” and the other was “American disease,” in Kirkpatrick’s phrase.

You can ask AI, “What are the key quotes from this?” and it will find them thematically within the material. Similarly, if you increase the search range to millions of words rather than a few thousand in one interview, it can find things like that.

I think the fewer points of analysis the AI has, in terms of available material and computational context, the more superficial the search will be. But at some point, these systems will have immense capacity to review material, and even a million words will not be much. They are trained on immense corpora, but search is a different thing, because you are feeding that search into weightings that are not themselves the full corpus. They are the result of iterated training on those corpora.

I think you can get a reasonably good textual analysis. You can find good quotes from people. You can do it in clever ways with LLMs now, too. You can do a search such as: “Find me quotes by Margaret Atwood specifically around Ceaușescu’s Decree 770, or around real-world precedents for reproductive coercion in The Handmaid’s Tale and The Testaments.”

Rosner: What is Decree 770?

Jacobsen: Decree 770 was Ceaușescu’s 1966 Romanian pronatalist decree. It severely restricted abortion and contraception. Abortion was permitted only under narrow exemptions, including women over a certain age, women who had already borne four children, later five, or cases involving serious medical risk, rape, or incest. Women were also subject to intrusive gynecological monitoring. The larger point is that the state was trying to force births.

That is generally the picture. There was a major baby boom, followed by severe social and economic consequences, including children being born into families and institutions without adequate resources to care for them. So, that is a very significant issue.

The contrast today is that more women are choosing to become single mothers by choice, through sperm donors or other routes, as an individual decision rather than as a result of religious or state pressure. That is not the same phenomenon; it is almost the moral inverse. Individual empowerment of women is much more robust than coercion by the state or by religion, typically.

You can find quotes along those lines. I think the same could be done for you if you asked about math or physics. I think the big weakness of information cosmology, as you have framed it, and as I have discussed with you, is that, and this is your point anyway, it lacks formal mathematics.

You can frame it as philosophy of physics. I think that is fair. I do not think you are simply hand-waving, and you are using more concrete terms, but the math is the primary determinant. You may be able to get an AI to develop a mathematical formulation based on the conversations. That probably could be done soon. I could try.

Rosner: We should really try that.

Jacobsen: What question would you ask the AI in terms of developing a mathematical formula for IC?

Rosner: I do not know. I would have to think about it.

Jacobsen: Do you have any responses to that before we go to the next session?

Rosner: I would see whether it could put the big five particles into some kind of mathematical framework, or at least take stabs at it. I do not know if this is still philosophical rather than mathematical. I would want it to take stabs at some of the constants that we have.

I would want constants, such as the electron-proton mass ratio and things like that, to arise from first principles plus the size of the universe. I have always suspected that part of that ratio involves how much of the universe is collapsed matter near T = 0. The more collapsed matter you have, the more precisely defined your universe should be. The collapsed matter acts as tent poles to hold the universe open and to define the matter within it more precisely.

So, I would see whether I could set AI loose on that.

Jacobsen: I said, “Based on our prior conversations between these two individuals, please develop a simplified mathematical form that would describe what is meant by informational cosmology.”

Rosner: Did you just ask?

Jacobsen: Here it is:

A simplified Rosner–Jacobsen informational cosmology formula can be written as a toy model of durable relational information:

dtdIR=e∈E(t)∑DeCe[−log2P(e∣Ht)]−Λ(t)

and, if we want the cosmological-scale claim in its simplest form:

aIC(t)=a0(IR(t0)IR(t))α

Where:

IR(t)

means the universe’s effective relational information at time t.

−log2P(e∣Ht)

is the Shannon-style information gained when event e occurs against the background of the universe’s prior history Ht. In plain English: the less expected the outcome, the more information it contributes. Shannon’s original framework treats entropy/information as a measure of uncertainty, choice, and information generated per symbol or per second when logs are base 2.

De

is a durability/readability factor from 0 to 1. This reflects Rosner and Jacobsen’s point that not every microscopic event “counts” cosmologically unless it leaves some persistent trace in the system. In their recent framing, information is not just any event, but a selected outcome that leaves a durable, distinguishable record.

Ce

is a coherence or relational-integration factor from 0 to 1. This says an event matters more if it becomes woven into the universe’s network of relations: particles, memories, records, galaxies, civilizations, black-hole-like remnants, and other persistent structures. That fits their language of the universe as a relational information-processing system, not merely a digital register flipping isolated zeros and ones.

Λ(t)

is informational loss, degradation, noise, collapse, or erasure. Rosner’s IC commentary explicitly links loss of information-bearing structure with the universe becoming less well-defined, possibly looking like collapse or a “big bang in reverse.”

aIC(t)

is an IC-style scale factor: not standard ΛCDM expansion, but the apparent size/scale/definition of the universe as a function of accumulated relational information. Rosner states the scale and size of the universe should be roughly proportional to the amount of information it contains, with local scale depending on local information/matter distribution.

So, in ordinary language:

The universe becomes more real, larger, older, and more structured to the extent that possible outcomes become actual outcomes, leave durable records, and become integrated into the relational history of the whole system.

A very compressed version would be:

Cosmic reality∝durable, self-consistent relational information

Or, even more aphoristically:

U(t)∼IR(t)

That is probably the cleanest mathematical translation of the idea. It is not a standard accepted cosmological equation; it is a simplified formalization of Rosner and Jacobsen’s speculative IC language. The closest mainstream guardrail is that physics already recognizes links between entropy, information, energy, and finite regions of space through information/entropy bounds such as Bekenstein’s, where entropy and information capacity are bounded by size and energy.

Rosner: It looks a little bit like the Drake equation. The Drake equation is a simple framework for estimating the number of communicative civilizations in the galaxy: you string together a bunch of factors, many of them highly uncertain.

This thing similarly strings together a bunch of factors in a basic, workmanlike way. It does not offer any further mathematical insight. It is what a couple of people in a hallway with a whiteboard might jot down initially as some of the factors, before they really start deploying expertise on the problem.

It is like cops setting up a murder board, but they are really just putting up the pictures of the victims and suspects, and they have not added the yarn connecting anything yet. It is real basic.

That does not mean AI cannot act in the manner of Einstein’s collaborators. Marcel Grossmann, for example, helped point Einstein toward the tensor mathematics and differential geometry that became central to general relativity. AI has not really done that here.

But based on AI’s increasing fluency in math, that does not mean AI could not do it. We would have to lay out what we are thinking much more systematically than we have. We would have to emphasize the mechanics. In my IC picture, it would need to take into account that as photons are redshifted over cosmic distances, the energy they lose is not simply gone, but is transformed into information.

So, in that picture, the energy is not really lost. It increases the precision with which the universe is defined, and that precision is proportionate to the amount of information in the universe. Photons losing energy through cosmic redshift or gravitational effects would be treated as a transformation of one form of energy into another. The energy it is transformed into would be informational energy encoded tacitly into the universe.

We could ask for a mathematical treatment of how the universe increases information by becoming more unlikely. With entropy, closed systems move toward more probable configurations, less specially arranged configurations. A universe that is increasing in information would, in this framework, move toward more unlikely, less random configurations. It would be nice to have some math for that.

It would also be nice to have some mathematical and physical picture of what can happen around a blackish hole in terms of information. We could ask AI, under Rosner’s eyes, what some possible things are that can happen around a good-sized hole.

Rotten Tomatoes? Or we could ask AI: Rosner suspects that one major line of evidence for dark matter comes from galactic rotation curves, where orbital speeds do not drop off as quickly as expected. That implies missing mass around many galaxies. Rosner posits that it is not exotic matter, but very old collapsed matter. Is that physically plausible if we allow for the universe to be much, much older than it appears to be?

Can AI take a shot at that? I do not know. There is a lot of tricky physics involved. But maybe people have done work on that. Probably. It is a big world full of people doing physics. Anyway, we could ask AI a ton of questions.

And now you are muted.

Jacobsen: True. When I finish transcription today, I can put that section into AI and see what it does. That is similar to what OpenAI reported with an internal model on the Erdős unit-distance problem.

Rosner: What is it? GPT-5.0?

Jacobsen: Not necessarily. OpenAI described it as an internal model. The Erdős problem is almost 80 years old.

Rosner: Not just one problem, but a lot of problems, quite a few problems.

Jacobsen: The Erdős one in particular was reportedly put to the model without detailed mathematical scaffolding, and it produced a construction and proof that external mathematicians checked. More precisely, OpenAI said the model disproved a longstanding conjecture in the planar unit-distance problem, rather than solving every aspect of the broader problem.

Last updated May 3, 2025. These terms govern all In-Sight Publishing content—past, present, and future—and supersede any prior notices.  In-Sight Publishing by Scott  Douglas  Jacobsen is licensed under a Creative Commons BY‑NC‑ND 4.0; © In-Sight Publishing by Scott  Douglas  Jacobsen 2012–Present. All trademarks, performances, databases & branding are owned by their rights holders; no use without permission. Unauthorized copying, modification, framing or public communication is prohibited. External links are not endorsed. Cookies & tracking require consent, and data processing complies with PIPEDA & GDPR; no data from children < 13 (COPPA). Content meets WCAG 2.1 AA under the Accessible Canada Act & is preserved in open archival formats with backups. Excerpts & links require full credit & hyperlink; limited quoting under fair-dealing & fair-use. All content is informational; no liability for errors or omissions: Feedback welcome, and verified errors corrected promptly. For permissions or DMCA notices, email: scott.jacobsen2025@gmail.com. Site use is governed by BC laws; content is “as‑is,” liability limited, users indemnify us; moral, performers’ & database sui generis rights reserved.