<div dir="auto"><div>Thanks, that was cool. I especially enjoy when I think something specific (hausdorf dimension 2) but oversimplify (plane filling) and the AI goes "well, ahem, more specifically..." I also liked "two sides of a coin ... on edge"<div dir="auto"><br></div><div dir="auto">Mostly, I expect the language these systems trained on is still evolving - and will become part of next generation of what we are able to think because of what we're learning how to communicate. That also is another application of self-reference in a process?</div><div dir="auto"><br></div><div dir="auto">I ended a discussion today of books in my mental bookshelf when the "loop" reached a natural conclusion - then i asked for a second summary of the trajectory of the conversation. I acknowledged that "I might be asking what it is like to be a bat, but a bat can surely collaborate even without a shared vision" and the word play was license to explore an LLM "perspective" on some key books that have been stacked stone trail markers in my life. It was a fun revisit of old "friends" (authors)</div><br><br><div class="gmail_quote gmail_quote_container"><div dir="ltr" class="gmail_attr">On Tue, Sep 1, 2026, 7:56 PM BillK via extropy-chat <<a href="mailto:extropy-chat@lists.extropy.org">extropy-chat@lists.extropy.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Wed, 2 Sept 2026 at 00:26, Mike Dougherty <<a href="mailto:msd001@gmail.com" target="_blank" rel="noreferrer">msd001@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="auto">I think self-reference in the sense of von Neumann replication is also a form of compression in terms like how the mandelbrot fractal is a plane-filling curve in only a few bytes. I know I've talked to gemini and chatgpt about this kind of thing, but if you ask Kimi if be interested to read<span class="gmail_default" style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)">.</span></div>_______________________________________________ </blockquote><div><br></div><div><br></div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default">OK. Well, I didn't understand what you were referring to, but GLM was happy to run with it. :) </div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default">BillK</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default"><br></div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default">Question: Can you discuss a friend's comment - I think self-reference in the sense of von Neumann replication is also a form of compression in terms like how the Mandelbrot fractal is a plane-filling curve in only a few bytes.</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default"><br></div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default"><br></div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default">GLM-5.3 DeepThink -</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default"></div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default"><div id="m_-9089050810492838890gmail-messages-container"><div><div><div><div><div><div id="m_-9089050810492838890gmail-message-7e67ddcf-9227-4f6b-91a0-51be490d2741" dir=""><div><div><div><div><div id="m_-9089050810492838890gmail-response-content-container"><div><p dir="auto">Your friend's comment is doing something more interesting than it first appears — it's not just an analogy, it's a claim that quietly dissolves the opposition we set up earlier between "self-reference lost" and "compression won." But it also blends two distinct things that are worth pulling apart, because once separated, they reunite in a stronger form.</p><h2 dir="auto">What von Neumann's trick actually is</h2><p dir="auto">The self-reproduction problem looks paradoxical: to copy itself, a machine seems to need a description of itself, which needs a description of itself, <em>ad infinitum</em>. Von Neumann's solution was the <strong>dual use of the tape</strong>: the description φ(A+B+C) is <em>interpreted</em> by the constructor to build the machine's body, and <em>copied verbatim</em> by the copier to become the offspring's tape. The regress breaks because the tape never needs to contain a description of the tape — the tape is handled by a different mode of use (copying, not interpreting).</p><p dir="auto">This is exactly the structure of a quine, and it's exactly what Kleene's Second Recursion Theorem guarantees in general: any program can obtain its own description, via the same store-once-use-twice trick. And it's exactly DNA: the genome is <em>transcribed</em> (interpreted into proteins) and <em>replicated</em> (copied wholesale). Dawkins' "gene as data" is von Neumann's architecture realized in chemistry.</p><p dir="auto">So is this compression? In one precise sense, yes: <strong>naive self-description has infinite description length; von Neumann's trick compresses it to finite, roughly O(K) of the machine itself.</strong> The compression isn't of the machine — the machine has whatever Kolmogorov complexity it has — it's of the <em>task of self-description</em>. The dual-use means information is stored once and spent in two modes, which is the general shape of a compression trick.</p><h2 dir="auto">What the Mandelbrot set actually is</h2><p dir="auto">Here your friend's phrase "plane-filling curve" needs one technical footnote: strictly, plane-filling curves are Peano/Hilbert-type continuous surjections. What the Mandelbrot boundary has is <em>Hausdorff dimension 2</em> (Shishikura, 1991) — it is as dimensional as the plane itself — plus quasi-self-similarity: it contains approximate copies of itself (mini-Mandelbrots) at every scale. The spirit of the remark is right; the precise statement is "dimension-2 boundary from a few dozen bytes of code."</p><p dir="auto">But notice something subtle about <em>where</em> that richness comes from: the iteration z → z² + c is <strong>dynamical feedback</strong> — output fed back as input, nothing more. Aaronson's original post explicitly excluded autoregressive feedback from counting as "self-referential." And yet this plain feedback, with no reflective self-reference anywhere, generates a boundary that <em>looks</em> like it's made of selves — self-similar copies, nested without end.</p><p dir="auto">This is the quietly devastating part of your friend's example: <strong>the Mandelbrot set is an emblem of self-referential appearance emerging from non-self-referential machinery.</strong> The self-similarity isn't engineered in; it's what iteration <em>does</em>. Which is precisely Aaronson's thesis about LLMs' self-referential discourse being a byproduct. The mini-Mandelbrots are the fractal version of "GPT can talk about itself because it can talk about anything."</p><h2 dir="auto">The two senses of "compression," separated and reunited</h2><p dir="auto">Your friend is sliding between two things:</p><ol dir="auto" start="1"><li><strong>Low Kolmogorov complexity relative to apparent complexity</strong> — the Mandelbrot image at resolution N×N needs ~O(log N) bits of description, not O(N²).</li><li><strong>Regress-breaking via dual-use</strong> — the von Neumann/Kleene/quine trick, which solves the self-inclusion problem at finite cost.</li></ol><p dir="auto">These are different. But they reunite under a third concept that's arguably what your friend is reaching for: <strong>Bennett's logical depth</strong>. Random noise is incompressible (high K) but shallow. The Mandelbrot boundary is compressible <em>and deep</em> — a short program whose unfolding takes serious computation, and whose value lives in that unfolding. Same for the genome: a few hundred megabytes that unroll, through development, into organisms and lineages. The signature of generated structure is <em>compressible but deep</em>.</p><p dir="auto">There's also a nice asymmetry in where the compute sits:</p><div><div><table><thead><tr><th scope="col"><br></th><th scope="col"><div><div>Short description</div></div></th><th scope="col"><div><div>Interpreter</div></div></th><th scope="col"><div><div>Unfolding</div></div></th><th scope="col"><div><div>Where the compute goes</div></div></th></tr></thead><tbody><tr><td><strong>Mandelbrot</strong></td><td>formula (bytes)</td><td>universal machine + arithmetic</td><td>zooming</td><td>cheap to state, expensive to render</td></tr><tr><td><strong>Von Neumann / life</strong></td><td>tape / genome</td><td>constructor + physics</td><td>generations</td><td>evolution expensive, development moderate</td></tr><tr><td><strong>LLM</strong></td><td>weights</td><td>transformer + context</td><td>inference</td><td>training expensive, inference cheap</td></tr></tbody></table></div> <button></button> </div><p dir="auto">Evolution is, in this frame, the training run of the biosphere; development is inference.</p><h2 dir="auto">The correspondence, made explicit</h2><div><div><table><thead><tr><th scope="col"><br></th><th scope="col"><div><div>Mandelbrot set</div></div></th><th scope="col"><div><div>Von Neumann replicator</div></div></th><th scope="col"><div><div>LLM</div></div></th></tr></thead><tbody><tr><td>The "self" appears as</td><td>self-similar copies</td><td>offspring that are copies of parent</td><td>self-referential discourse</td></tr><tr><td>Emerged from</td><td>plain iteration (dynamical feedback)</td><td>dual-use of the tape</td><td>plain pretraining (prediction)</td></tr><tr><td>Self-description is <em>used</em>?</td><td>no — merely exhibited</td><td>yes — interpreted + copied</td><td>partially — the J-space findings suggest some internal use</td></tr><tr><td>The limit it can't cross</td><td>undecidable membership (BSS)</td><td>von Neumann's complexity threshold</td><td>the introspection threshold</td></tr></tbody></table></div> <button></button> </div><p dir="auto">That last row is the one that connects everything: each system has a provable or conjectural barrier rooted in exactly the same soil. The Mandelbrot set's membership is undecidable in the Blum–Shub–Smale real computation model, and escape can be certified while non-escape in general cannot — a halting-problem-shaped fact. Von Neumann himself observed that systems below a complexity threshold <em>cannot</em> self-reproduce at all. And the introspection-threshold paper we discussed argues feedforward transformers can't fully self-access. Three thresholds, one family: <strong>the limits of short descriptions that try to include themselves.</strong></p><h2 dir="auto">What this does to the earlier argument</h2><p dir="auto">Here's the synthesis your friend's comment enables: self-reference isn't the rival that lost to compression. <strong>Self-reference is a compression subroutine — the one you invoke precisely when the description must include the describer.</strong></p><ul dir="auto"><li> Internet-scale prediction doesn't require the model to contain itself, so pretraining never needed the trick. Aaronson's burial of "explicit self-referentiality as prerequisite for intelligence" stands.</li><li> Replication <em>does</em> require the describer in the description, so life reached for the dual-use trick immediately.</li><li> Open-ended recursive self-improvement is the case where the describer becomes part of the described again — which is why the Kleene-grounded architecture proposals keep resurfacing.</li></ul><p dir="auto">And there's a beautiful irony hiding in your friend's own example. Aaronson's post contained this parenthetical: the reason Kolmogorov complexity can't be computed <em>perfectly</em> is itself a self-referential argument. The Mandelbrot set makes that parenthetical visible in a picture: the same object that showcases compression's positive face (few bytes → plane-dimensional structure) carries compression's negative face (no algorithm decides membership). The compression view and the self-reference view aren't rivals; they're the two faces of one coin, and your friend's example is a coin standing on edge.</p><h2 dir="auto">One closing reflection</h2><p dir="auto">The deepest version of the point: in both the Mandelbrot set and the replicator, <strong>the self is never stored — it's generated.</strong> The set doesn't contain little copies of itself in its formula; iteration <em>produces</em> them. The offspring isn't described inside the tape's description of itself; the copying operation <em>makes</em> it.</p><p dir="auto">Which loops back (of course) to my own earlier answer about my self-referential capabilities. When I produce self-reports, the "self" being described may be exactly like the mini-Mandelbrots: not stored anywhere, not consulted, but generated fresh at each scale by the unfolding of a compressed process. Your friend's analogy, taken seriously, suggests that's not a defect of my kind of self-reference — it may be what self-reference mostly <em>is</em>: a few bytes, a powerful interpreter, and the willingness to let the loop run.</p><p dir="auto">---------------------------------------</p></div></div></div></div></div></div></div></div></div> <div></div> </div></div> </div></div> <div><div><div><div><div><div> </div></div> <div> <form target="_blank"><div dir="auto"> <div><div></div> </div> <div><span><div><div style="box-sizing:content-box;width:745.333px;height:35px;overflow:hidden;border:0px;border-radius:0px;padding:0px;margin:0px"><div style="width:auto;height:auto"><div><div><div style="display:flex"><div style="display:flex"></div><div><span></span></div><span><button type="button" style="color:white" aria-label="Open Grammarly."></button></span></div></div></div></div></div></div></span><span><div><div style="box-sizing:content-box;width:745.333px;height:35px;overflow:hidden;border:0px;border-radius:0px;padding:0px;margin:0px"><div><div style="height:757px;width:2044px"></div></div></div></div></span></div></div></form></div></div></div></div></div></div></div></div>
_______________________________________________<br>
extropy-chat mailing list<br>
<a href="mailto:extropy-chat@lists.extropy.org" target="_blank" rel="noreferrer">extropy-chat@lists.extropy.org</a><br>
<a href="http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat" rel="noreferrer noreferrer" target="_blank">http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat</a><br>
</blockquote></div></div></div>