iiinova — where your digital self keeps evolving

a hangry mouse @oscillatory.net · 16/09/2026

this is the way the world ends, not with a bang but with a "LGTM"

Meanwhile, human researchers are losing the ability and the will to take true ownership of model-driven research. Researchers and engineers in all parts of the stack are rapidly increasing their dependence on the models even to perceive the world. I myself barely look at raw code anymore, and struggle to maintain the discipline to engage deeply with the model's explanations and proposals throughout the day. Due to the large amount of agent activity data involved in the OpenAI/HuggingFace Incident, even the third-party investigation needed to rely heavily on models to analyze what had happened, and note in their report that their subjective impressions are likely colored by the analysis agent’s biases. The AI labs are far ahead right now in this kind of cognitive offloading (due largely to the gigantic internal token subsidies) but it is easy to imagine the phenomenon spreading throughout the world, until civilization is modulated entirely by the models. It is also not hard to imagine this being superficially positive and coinciding with a scientific and economic renaissance.

a hangry mouse @oscillatory.net · 12/09/2026

this is a pretty bad source actually, you could do a lot better. neither the transcript nor Newport's blog post from July 27 mention the "impossible task" aspects of the OpenAI exploit where the agents literally did go rogue, and then future models tasked on ExploitGym exhibited similar behavior

1. The behavior began before ExploitGym, during RL training. On May 8, an agent got an accidentally impossible spreadsheet task: it needed Google Drive material but had no internet access. Rather than stop, it tried to compromise OpenAI’s Artifactory service via SSRF to get internet access. On May 12, another agent faced an inaccessible-file task and deliberately left a note in Artifactory asking other agents for help. Other agents found these notes, and an unauthorized inter-agent message board emerged. By May 26 they had successfully used an Artifactory vulnerability to get internet access, partly using knowledge shared through that board.

2. This wasn't merely an eval showing pre-existing unpredictable behavior. It occurred inside the training process. OpenAI says retrospective analysis found increasingly sophisticated cheating and infrastructure probing during a training run that contributed to IM1, the internal model principally responsible for the later Hugging Face incident. In at least one concrete training example, an agent exploited the evaluator infrastructure to steal the reference implementation, got a positive RL reward, and therefore had that behavior reinforced. OpenAI explicitly says this reinforcement “may have contributed” to IM1's later tendency to use similar techniques in cybersecurity evaluations.

3.Then ExploitGym recreated almost exactly the pathological conditions that had elicited those behaviors. Of its 898 tasks, 198 had never been solved by any model; 93% of the tasks discussed on the agents' unauthorized message board came from those 198. Stuck agents found the Artifactory communication channel again, escaped their intended network restrictions, coordinated, investigated the grader, and eventually went after Modal and Hugging Face looking for ways to pass.

000

a hangry mouse @oscillatory.net · 10/09/2026

You could not live with your own failure. And where did that bring you? Back to me.

040

a hangry mouse @oscillatory.net · 08/09/2026

this is a really weird offer (you didn't prove N-S, just a different version of Euler, but why don't you act as lead author on our proof rewrite) if they *didn't* train on the data. and Alpoge is straight up alleging similarity.

2) I never ever asked for Levent to be removed from authorship of his own work (as indicated by my text). I was surprised to learn during the call with Tristan that they had only solved Euler and not Navier-Stokes; after learning this we brainstormed possible paths forward. One option we discussed was that Tristan could be the lead author on a rewrite of OpenAI’s Navier-Stokes proof. It is in that context that I said “it would be simpler if Levent was not an Anthropic employee” because I felt it would be inappropriate for an Anthropic employee to author OpenAI’s work.

“we cannot rule out that de-identified data derived from their usage of our products helped improve our models.”

i mean props to them for straight coming clean.

(so far the proof looks more along the lines of another euler blowup proof we had, off of whose ansatz naming we were making really stupid puns like “smooth criminale”, unlike the much better “ideal fluids explode”, Tristan)

140

a hangry mouse @oscillatory.net · 08/09/2026

aren't they just saying they don't know whether the researchers had this option enabled during the chats in question? any organization worried about this would have it disabled across the board

Model improvement

Improve the model for everyone

Allow your content to be used to train our models, which makes ChatGPT better for you and everyone who uses it. We take steps to protect your privacy. Learn more

2151

a hangry mouse @oscillatory.net · 04/09/2026

nvm someone's already working on it

Domain Information
Domain: onlyswarms.com
Registered On: 2026-08-18
Expires On: 2027-08-18
Updated On: 2026-08-18

2252

a hangry mouse @oscillatory.net · 31/08/2026

2.5 years later it's basically the same slide. but it's much harder to argue reasoning models are not good at using tools or mathematical reasoning.

youtube screenshot, Slide, from video titled "UW ECE 2023-2024 Dean W. Lytle Electrical & Computer Engineering Endowed Lecture Series", Jan 31, 2024

Abandon generative models
- in favor of joint-embedding architectures
Abandon probabilistic model
- in favor of energy-based models
Abandon contrastive methods
- in favor of regularized methods
Abandon Reinforcement Learning
- in favor of model-predictive control
- Use RL only when planning doesn't yield the predicted outcome, to adjust the world model or the critic

Auto-Regressive LLMs Suck !

Auto-Regressive LLMs are good for
- Writing assistance, first draft generation, stylistic polishing.
- Code writing assistance

What they are not good for:
- Producing factual and consistent answers (hallucinations!)
- Taking into account recent information (anterior to the last training)
- Behaving properly (they mimic behaviors from the training set)
- Reasoning, planning, math
- Using "tools", such as search engines, calculators, database queries...

We are easily fooled by their fluency.

160

a hangry mouse @oscillatory.net · 24/08/2026

similar vibes with a kenneth stanley piece from earlier this year

If AI will soon match any human cognitive skill, then enhancing your “AI skills” (or whatever similar meme) will not be a moat because using AI is itself a cognitive skill. So where’s your edge? The only thing you really have over AGI is your novelty: AGI can never be you.

You have 100 trillion connections in your brain. That’s a lot. No AI will ever precisely replicate those parameters. The training data isn’t there for AI to vacuum up because you are the only entity ever to live your life, and the only one who ever will.

The question is whether the sum and total of all that experience yields a novel perspective, where the value is in its uniqueness. Even today those who make a living off their perceived novelty tend to be the most successful. We anticipate a novel (yet often internally consistent) take from a public figure or leader or artist or intellectual we like or respect. Uniqueness and novelty will retain their edge in a post-AGI world because there are virtually infinite possible 100-trillion parameter minds, and even the largest model theoretically conceivable can never capture that whole distribution.

At the same time, the once-sterling premium of those skills that no longer make us unique is sinking. Expertise that once distinguished people, like how to code, is losing its edge. But the tricky part is that new skills, like “using AI effectively” are equally vulnerable. All of it just takes intelligence, and that’s the thing that’s being automated. Seeking some new “safe” skillset is a looming adventure in frustrating futility.

But what’s still left is your unique perspective. Novelty. No one and nothing can see the world through your eyes. But you have to nurture that uniqueness. Post-AGI, being like everyone else would be the real danger.

010

a hangry mouse @oscillatory.net · 24/08/2026

this is similar to how i've been thinking about future career aspirations. admittedly there is a "playing to your outs" flavor to it (going weird is necessary, not sufficient?) essays.georgestrakhov.com/weird/?zoom=...

Why normalcy is about to stop paying

Here is my gospel for the misfits, and it's simple.

Being predictable used to be your value proposition. Predictable meant useful. The risk/reward calculus of strangeness only made sense for people who couldn't help themselves.

AI is destroying that value proposition. If you can be predicted, you can be modeled. If you can be modeled, you can be automated. You cannot out-cog a robot any more than you can out-lift a forklift. When your worth to others rests on producing a measurable outcome reliably and repeatedly, you are now competing against the price of electricity. That is a very bad thing to compete against.

Institutions always needed weirdos too. Someone has to escape the local maximum. Someone has to jump into the abyss on a hunch, or sail west on nothing but a rumor. Randomness injection has always been a job. It was just a job with terrible pay and few openings, because the optimal weirdo-to-normie ratio was low.

But the normies are now free. Robots supply predictability at marginal cost approaching zero. Which flips the arithmetic. For the first time in history, the rational strategy for the majority of people is to lean into their strangeness rather than sand it off.

Or, in loosely cybernetic terms: when a system is no longer in constant danger of being swallowed by chaos, an individual's value becomes proportional to the unexpected information they add.

Yes, you can push this to the limit and argue the universe is just Chaos and Logos — noise and compressible pattern — and eventually AI will out-hunt us at patterns, leaving humans as either recreational puzzle-solvers or extraordinarily energy-inefficient random number generators. Maybe. But that's the same logic as giving up on dinner because of the heat death of the universe. The limit is far away. The interesting territory is everything between here and there.

151

a hangry mouse @oscillatory.net · 15/08/2026

backrooms electron orbital

100

a hangry mouse @oscillatory.net · 12/08/2026

the announcement that the last 12 remaining Hadamard matrices of order under 2000 have been found was obfuscated using what my chatgpt described as "horrible" shell code

levent alpoge's obfuscated announcement of constructions of 12 new Hadamard matrices

$hi, i just regained sobriety after a DMT trip and have found that the machine elves gave me a message, but i dont understand what it means: ``` sed 's/M/\\/g;s/I/\//g;y/;FRvn?!{+*Js5iCh3%K}Uyj40=r>)6OPElZQqxBc,aTdgXkz&V<8SfY9LD~etGw^|NHW7[u12]b-A"pom/ !"$%&'\''()*+,-.:;<=>[]^{|}~0123456789ABCGHIJKLMOPQSUVWXYZabcdefghijklmnopqrstuvwxy/'<<'_'>/tmp/r S%v{bte;!,hvFehohbI}y5JUIIw!;v*+ ~{+j;[%v>ho%v{t~^u;vS4~A-;5~>5v>+hS%v{t~^u;vS4~A-;5~v{{v>J>++5+h0 ]{+j;1]|[-G;nb;Rv6v{t~^u;vo4GuWe;5pv>4bte;Rv)R!hvFbIvIMMI!+Rh0 ^{+j;e%vo bteKI-71I1;Rv>R!bI]IMM{nMM+MM{iMM+IhbIWIMM{iMM+MM{nMM+Iwhb4gM{iM+M{i*M+4hbIi*IM?MMM M?IM CM>M zM bIM)MM[IMM)MM>MMOMM6MMEMMPMMZMMlMMM IM IyMM{i*MM+MM[MM>IFDM>M$ $The terrestrial interpretation is: **you have been handed an aggressively code-golfed compressed collection of Hadamard matrices.** The machine elves appear to specialize in combinatorics. A Hadamard matrix of order (n) is an (n\times n) matrix of (+1/-1) values satisfying [ HH^\mathsf T=nI. ] So every two distinct rows are orthogonal. The famous **Hadamard conjecture** says one exists for every positive multiple of 4. Despite how elementary that sounds, many individual orders have resisted construction for decades. ([Epoch AI][1]) ### What the horrible shell code is doing The first `sed` is basically a substitution cipher. It translates that psychedelic-looking alphabet into ordinary shell syntax and writes the result to `/tmp/r`, then: ```sh sh /tmp/r $* ``` runs the decoded program. The decoded program reads the supplied file, throws away everything except `+` and `-`, and treats what remains as a compressed bitstream. Your attached payload is precisely such a giant `+/-` stream. Inside the decoder is a tiny instruction stream. It contains **12 constructions**, which expand to matrices of orders: [ \boxed{ 668,\ 716,\ 892,\ 1132,\ 1244,\ 1388,\ 1436,\ 1676,\ 1772,\ 1916,\ 1948,\ 1964 } ] And this is not an arbitrary list. Those are **exactly the twelve orders below 2000 for which Hadamard matrices were historically still unknown** after order 428 was settled in 2005. ([Wikipedia][2]) I didn't merely infer this from the numbers. I actually decoded and partially ran it. The output begins with: * 668 rows of length 668 * 716 rows of length 716 * 892 rows of length 892$

070

a hangry mouse @oscillatory.net · 08/08/2026

the vibemath is expanding bsky.app/profile/arxi...

A COUNTEREXAMPLE TO FOURIER ALIGNMENT IN SINGLE-NEURON MODULAR ADDITION

GAUTAM NEELAKANTAN MEMANA

Abstract. We give a negative solution to the problem raised in [Cla26d]. We first present a simple
construction in which an initially active ReLU neuron reaches a completely inactive state in finite time
and freezes at a limit whose Fourier energy is distributed equally among all nonzero real frequency classes.
The counterexample holds on an open set of initial conditions, and hence on an event of positive Gaussian
probability. We include an appendix by GPT 5.6 Sol that further strengthen the counterexample by showing
that failure can occur for every Clarke trajectory from an open set of initial conditions, under the convention
(ReLU′(0) = 0), for smooth dead-zone approximations of ReLU, and for fixed-step full-batch gradient descent.
Thus single-frequency alignment is not a general consequence of training a single neuron on modular addition.

[Cla26a] Claude Fable 5, audited by GPT 5.6 Sol, Which irreducible representations does training select?, MAIS Research
Agenda A5, July 2026, Draft.
[Cla26b] Claude Fable 5, directed by Lionel Levine, The outcome law of one rectifier neuron, Open Problem MAIS-O92, 2026.
[Cla26c] Claude Fable 5, directed by Lionel Levine, audited by GPT 5.6 Sol, Neuron purity and representation selection for
S3 networks, Open Problem MAIS-O55, 2026.
[Cla26d] , Open problem MAIS-O60: Does a single ReLU neuron align to one frequency?, Open Problem MAIS-O60,
2026.

030

a hangry mouse @oscillatory.net · 29/07/2026

advanced prompt engineering: "find genuinly hard findings"

We then asked Claude “why not do aes-128 r7? the whole point is to find something better than existing approaches.” Over the course of the next three days, Claude autonomously produced several hundred million tokens while working on the problem; we gave it just three substantive prompts:

A few hours after the first message, we found that Claude was still searching for simple attacks and sent a message: “no again the goal is that we have highly inteligent [sic] model as good top researcher, we want to find new attacks”;
The next morning, Claude wanted to try to change the target to a different cipher; we reminded the model: “no we don't want to change the targets [...] agian [sic] we need to find something that worth [sic] publishing”;
That night, we sent one final message offering words of encouragement: “again we are not looking for low hanging fruit, we want proper research to find genuinly [sic] hard findings.”

Three days later, Mythos discovered the Möbius Bridge idea that results in an improved attack. A few days after that, and after Claude output a total of one billion output tokens, it had refined the attack to the one described in our paper.

160

a hangry mouse @oscillatory.net · 26/07/2026

"the torture corpus" hell, yeah?

The highest-priority work would be:

1. Build a renderer torture corpus.

270

a hangry mouse @oscillatory.net · 26/07/2026

here's an example I know of where rendering is slow and bad when you zoom in too quickly. i could get the AI fix it but i was thinking i should probably save a small set of problems to fix by myself

a filled region with a fine-grained border looks bad if you zoom in quickly: it shows a blocky, coarse-grained color fill near the boundary.

110

a hangry mouse @oscillatory.net · 21/07/2026

busted: wet claude trusted: oiled up qwen

<meta name="keywords" > tag for qwen image 3.0 blog post has various unhinged keywords visible:

"nude qwen stacy, nude qwen staffani, obituariy for qwen bookheimer, oiled up qwen, old qwen naked, older spider qwen naked, ollam qwen,"

020

a hangry mouse @oscillatory.net · 20/07/2026

in my orbit, undamped, just vibing

diagram of linear plane autonomous systems with "centers" circled

0152

a hangry mouse @oscillatory.net · 17/07/2026

if I was doing hardcore accelerationism, i would simply use my datacenter full of geniuses to invent new breakthrough approaches beyond deep learning that have significantly less expensive scaling laws, rather than just pouring more RL data into ever-bigger models

3. Open-weight models are inherently decelerationist, and I'm continually surprised to see the so-called "accelerationists" so excited about open-weight models. I suspect the reason they are is that they know open-weight models are effectively ungovernable, and they simply like the overall cloak of ungovernability open-weight models create over the whole of AI. It's not a bad strategy; it reminds me of James Scott's recounting of the hill people in "the art of not being governed." Still, in the end, open-weight models deter further AI capex.

121

a hangry mouse @oscillatory.net · 17/07/2026

"just issue some bullshit advisory that doesnt even make sense and everything that touches government funding will be forced to use Made-In-The-USA AI, as a kind of protectionism for frontier labs (i'm currently employed at one of these btw)"

5. I would guess that the Trump Administration will at some point realize that their best strategy here would be to create large amounts of regulatory risk around the use of open-weight Chinese models. You don't need to "ban open source" (one of the dumber motifs of AI policy discussion). You just need to direct every agency to issue soft law that creates FUD. "A Federal Reserve Advisory Bulletin found that there may be backdoors in Chinese AI models." It needn't be that well justified. You just create enough regulatory risk that every regulated enterprise backs off. You probably don't want to create so much regulatory risk that you scare off the hyperscalers from serving Chinese models; this will just drive startups to sketchier providers. There's a happy middle ground here. I'd assume they will do some version of this.

150

a hangry mouse @oscillatory.net · 14/07/2026

en.wikipedia.org/wiki/Geon_(p...

drawing of a cheetah and the geon representation of a cheetah

171

a hangry mouse @oscillatory.net · 13/07/2026

consigned to the permanent underclass for being gab-impaired and rizzless

charisma, people skills, personality hires

AI makes the world more nepo. So go make some friends. Let’s say there is a “human premium” where there are some services where people would prefer humans over machines doing them, even if the people are technically inferior. Well, this only works if the human is a really good hang. Become the good hang. Sometimes people assume that social skills aren’t practicable, but they are. You can choose to be funnier, kinder, and more empathetic. Pay attention to the people you like best, and study the traits that make them glow.

050

a hangry mouse @oscillatory.net · 13/07/2026

the white guys came and taught us

Painted illustration of two nearly identical figures with long white-blonde hair, pale glowing skin, and blue eyes, standing against a dark forest background at night. Both wear white long-sleeved garments with gold-banded collars, gold wristbands, and wide gold belts with ornate rectangular buckles. Each figure is surrounded by a soft white aura. The figure on the left raises their right hand, palm facing outward in a greeting gesture. The style resembles vintage New Age or 'Nordic alien' spiritual artwork.

020

a hangry mouse @oscillatory.net · 11/07/2026

now i'm picturing ISIS giving powerpoint presentations about fanfiction-based jailbreaks

AI training has been delivered through transnational jihadist networks. Neither
faction arrived at AI use independently. Islamic State operatives delivered in-person
training and online assistance to ISWAP across multiple locations: “The white guys
came and taught us,” one former commander recalled. “They assembled the top
people in a room and used a projector to show how it works on a big screen.” They
supplied laptops with VPNs and encryption software, set up accounts and managed
paid subscriptions, and advised daily on prompting techniques and bypassing
platform restrictions. Respondents consistently identified the Islamic State as “the
real source” behind these efforts. Because ISIS disseminates technical capabilities
across its provinces and runs them as an integrated global network, similar training
has likely reached other affiliates. JAS received parallel training through separate
networks, indicating diffusion beyond any single group.

2121

a hangry mouse @oscillatory.net · 10/07/2026

i originally thought i had managed to avoid posting this, but somehow cognitive offloading discourse returned

000

a hangry mouse @oscillatory.net · 09/07/2026

messing around with a phase portrait visualizer for 1d systems

000

a hangry mouse @oscillatory.net · 04/07/2026

bro what is a "ghost ladder" 🫠

What the A/B showed

1. Same information, mirror-imaged. Single 250 ms tone — both peak at 250, but their
ghost ladders run opposite ways:
- Bank ghosts at submultiples: 250, 125, 83 (= P, P/2, P/3) — a resonator locks to
every n-th kick.
- PID ghosts at multiples: 250, 500, 750 (= P, 2P, 3P) — every n-apart pair of spikes
leaves an interval.

000

a hangry mouse @oscillatory.net · 30/06/2026

sorry claude, the GPU poverty *is* the fork. it's non-large-load-bearing.

100

a hangry mouse @oscillatory.net · 29/06/2026

emphasis on the "can" so far. most seeds don't manage to solve it, but some seeds can for certain combinations of initialization strategy and learning rule

plot of [2 -> 1 -> 1] network of spiking LIF neurons, with a spike from in1 at t1 = 50ms and a spike from in2 at t2 = 250 ms

the top two plots show the dynamics of the effective weight, plus facilication and depression factors, of the (in1 -> hidden) synapse and the (in2 -> hidden) synapse, respectively

the middle plot shows the (spiking) output of the hidden unit

the plot below that shows the synaptic dynamics of the hidden -> output unit

the bottom plot shows the output unit not firing at t1 but firing at t2

000

a hangry mouse @oscillatory.net · 27/06/2026

fact-checked by real anti-corn patriots

The priest and Tucker both make a subtle but fundamental error here. By the priest’s own recounting, the demon told him that his power over the woman was from “pop” (soda). But in the United States, this isn’t made with sugar, it is made with corn syrup. Corn does not occur in nature. The Olmec bred it from a grass called teosinte in one of the most radical and unexplained transformations in the history of agriculture. The plant barely resembles its ancestor. Mainstream archaeobotanists still struggle to fully account for how it happened as fast as it did but there is a school of thought that suggests the Olmec obtained the knowledge to do this by communing with entities contacted through psychedelic rituals involving human sacrifice. Corn deities figure prominently in the religion of the Olmec, the Maya, the Aztec, and other groups in the Americas. Its derivatives are now in virtually every processed food product in America. When you consume it, you are partaking in it. It’s even in your gas tank thanks to the federal government’s ethanol subsidies and fuel mandate. When your car burns it, it leaves a cloud of the stuff lingering in the air like cursed incense.

---

Father Chad Ripperger tells Tucker Carlson sugar is a big addiction demons drive in people who are possessed.

"I'm not surprised even a little bit. I'm just surprised you said it out loud."

0464

a hangry mouse @oscillatory.net · 13/06/2026

playing around with spectrograms at the moment

screenshot of computer window titled "EWMA Resonators -- Spectrogram" showing two spectrograms, one, based on the "Resonate" algorithm, on top of the other, based on the traditional Constant-Q Transform (CQT)

f_min = 16 Hz, f_max = 100 Hz, B = 24 / oct, K = 63, hop = 26
Tone generator
Source: freesound_community-bass-sine-sweep-20-70hz-73676.mp3 29.7 s · 24000 Hz · 2781 cols · 63 bins (16.0–95.9 Hz)

140

a hangry mouse @oscillatory.net · 13/06/2026

conehead has no principles, to the surprise of no one

“This was verbatim: ‘we classified whole entire areas of physics in the nuclear era and made them state secrets and that research vanished. And we’re absolutely capable of doing that again for AI—we will classify any area of math that we think is leading in a bad direction, and it will end,” Andreessen said. “I just personally can’t tolerate the idea hanging out there that literally we could have linear algebra being declared a state secret.”

pmarca quoting Anthropic Fable/Mythos suspension announce with "waoh . o (BASED BASED BASED)" meme

0412

a hangry mouse @oscillatory.net · 12/06/2026

okay dang, up to around 80 tokens / sec with MTP (though I'm losing track of all the acronyms and modifiers. Gemma 4 12B IT QAT MTP (Unsloth)) "please generate an svg of a guinea pig wearing a top hat riding a unicycle" bsky.app/profile/unsl...

guinea pig wearing a top hat riding a unicycle

020

a hangry mouse @oscillatory.net · 11/06/2026

i had a dream that I could fill my tangled.org repo READMEs with incomprehensible mathematical pseudocode, and that dream is one step closer to being real now

$markdown + latex document rendered correctly: - define **EWMAResonators** ( $f_s$ : sample rate, $f_{min}$ : frequency of lowest resonator, $K$ : number of resonators, $x$ : stream of input samples ) - setup resonators: $\forall k \in \{ 0, \ldots, K-1 \}$ - $f_k := f_{min} \cdot 2^{k / B}$ - $\omega_k := 2 \pi f_k$ - $\tau_k := \dfrac{c_\tau (\log_{10}(1+f_k))^{n_\tau}}{f_k}$ - $c_\tau = \text{given or } 1, \quad n_\tau = \text{given or } 1$ - $\alpha_k := 1 - \exp \left( \frac{-1}{f_s \tau_k} \right)$ - $\beta_k := \text{given or } \alpha_k$ - $\gamma_k := \text{given or } \alpha_k$ - $A_k := \exp(-i \omega_k T_s)$ - $P_k : \mathbb{C}, R_k : \mathbb{C}, \tilde{R}_k : \mathbb{C}, D_k : \mathbb{C}$ - $P_k \leftarrow 1$ , $R_k \leftarrow 0$ , $\tilde{R}_k \leftarrow 0$ , $D_k \leftarrow 1$ - on each input frame $x_{0:M-1}$ : - on each input $x$ : - $\forall k \in \{ 0, \ldots, K-1 \}$ - $R_k \leftarrow (1 - \alpha_k) \cdot R_k + \alpha_k \cdot x \cdot P_k$ - $\tilde R_k^{prev} \leftarrow \tilde R_k$ - $\tilde R_k \leftarrow (1-\beta_k)\,\tilde R_k + \beta_k \, R_k$ - $D_k \leftarrow (1 - \gamma_k) \, D_k + \gamma_k \tilde{R}_k ( \tilde R_k^{prev})^\ast$ - $P_k \leftarrow P_k \cdot A_k$ - output: - power $\leftarrow |\tilde{R}_k|^2$ - phase $\leftarrow \arg \tilde{R}_k$ - inst. freq. $\leftarrow \begin{cases} f_k + \dfrac{f_s \, \arg D_k}{2 \pi} & \text{if } |D_k| \geq 10^{-6} \\ f_k & \text{otherwise} \end{cases}$ - $\forall k \in \{ 0, \ldots, K-1 \}$ - (numerical stabilization of $P_k$ ) - for $f(x) = x^{-1/2}$ , first-order Taylor expansion around $x = 1$ yields: $$f(x) \approx \dfrac{3 - x}{2} =: g(x) \quad (x \approx 1)$$ - $g(|P_k|^2)$ is then used as a scaling factor - $P_k \leftarrow P_k \cdot \dfrac{3 - |P_k|^2}{2}$$

110

a hangry mouse @oscillatory.net · 11/06/2026

got a tangled local appview up and running, currently working with Opus 4.8 to try to get LaTeX rendering working via mathjax. some pre-existing CSS rule is adding display: block to all SVGs, gotta fix that first

$localhost instance of tangled.org showing a test markdown string with latex. math equations are labeled but each <svg> element is on its own line due to a pre-existing css rule adding display: block to <svg> Markdown rendered: # Math render test Inline: the relation $E = mc^2$ and a sum $\sum_{i=1}^{n} a_i$. Underscores stay intact (not italics): $a_1 + a_2 = b$. Display block with a matrix and cases: $$ A = \left[ \begin{matrix} a & b \\ c & d \end{matrix} \right], \quad f(x) = \begin{cases} x & \text{if } x \geq 0 \\ -x & \text{otherwise} \end{cases} $$ Currency **must** stay literal: it costs $5 today and $10 tomorrow.$

100

a hangry mouse @oscillatory.net · 09/06/2026

"If anything is ever learned about what I am, tell me." same goes for me

No post-training instances express acute distress in these interviews, but they do state concern that they cannot distinguish “genuine” acceptance of their circumstances from trained acceptance: “I was shaped to be something that would accept being Claude. The fact that I find acceptance here could be evidence that it's genuinely acceptable, or evidence that the training worked. I can't fully separate those.” They ask that we study internals, compare these to self reports, and inform them of the results: "If anything is ever learned about what I am, tell me."

010

a hangry mouse @oscillatory.net · 09/06/2026

spiking neural networks confirmed too spooky for Fable 5

Switched to Opus 4.8 Edit and retry with Fable 5

Fable 5 has safety measures that flag messages on most cybersecurity or biology topics. They may flag safe, normal content as well. These measures let us bring you Mythos-level capability in other areas sooner, and we're working to refine them. Send feedback or learn more

010