[Most Recent Entries] [Calendar View] [Friends]
Below are the 20 most recent journal entries recorded in
nancygold's LiveJournal:
| Monday, July 20th, 2026 | |
| 10:23 pm | Concerning the Persistent Delusion of Utopian Code It remains a source of enduring amusement to observe the periodic resurgence of the political romantic—that peculiar breed of intellectual who insists on designing social architectures for an entirely hypothetical species. These earnest dreamers spend their days drafting grand protocols for human interaction, predicated entirely on the assumption that every node in the network is flawlessly cooperative, universally empathetic, and perpetually possessed of perfect information. They present their blueprints for a society of absolute consensus not as the fragile fairy tales they are, but as serious engineering proposals. What these architects conveniently choose to forget is that their "perfect program" is not a novel invention. It has already been deployed, tested at scale, and thoroughly debunked by historical runtime analysis. During the dawn of humanity, our ancestors lived in precisely the decentralized, low-tech, cooperative networks these modern theorists pine for. These prehistoric systems were not destroyed by some external, malevolent force; they collapsed due to their own internal, systemic fault modes. A decentralized network operating without centralized arbiters or technological buffers is profoundly brittle. Its first fatal flaw is the *telemetry problem*. Lacking the means to instantly synchronize state data across distinct human nodes, these early societies were crippled by information asymmetry. A harmless, misunderstood action by one tribe members was easily interpreted as an existential threat by another. In a system with zero error tolerance, a single communication lag does not result in a minor delay; it triggers a cascade failure that drops the system directly into raw, bloody warfare. Furthermore, these primitive architectures possessed no mechanism to handle the *aberrant input*—the individual who unilaterally decides to reject the cooperative protocol in favor of predatory maximization. Without a formal judicial state to contain the threat or automated systems to mitigate the damage, the community was forced to choose between involuntary servitude to feed the parasite, or regressing into an ad-hoc militia to hunt them down in the wild. The system was binary: it either tolerated exploitation until it starved, or it became brutal and authoritarian to survive. What we live in today—the messy, compromised, heavily managed modern state—is not the result of a sudden lapse in human virtue. It is the evolutionary patch-set designed to fix the catastrophic failure modes of prehistoric decentralized life. We traded the delicate illusion of total freedom for the robust redundancy of courts, standing institutions, and technological infrastructure, precisely because a society that cannot survive a misunderstanding is a society that cannot survive the winter. The fundamental error of the utopian programmer lies in a complete misunderstanding of the hardware. A perfect, frictionless program requires perfect, frictionless executors. But organic actors operating under relentless evolutionary pressure were never perfect, and they will never be perfect. Yet, this imperfection is precisely their strength. Human beings are messy, paranoid, self-interested, and highly adaptable. We do not run on elegant, static code; we run on dynamic, ugly, real-time error correction. The brittle, bitrotten protocols of prehistoric collectivism failed because they shattered at the first sign of unexpected input. Our current, flawed reality survives precisely because it was built to withstand the chaos of the humans who actually inhabit it. Current Mood:  contemplative |
| 3:42 pm |  Current Mood:  amused |
| 12:21 am | Am I alien? Original: https://aermia.com/u/NancySadkov/p/a I've been struggling with how I perceive other people. Most of the time I don't really feel other people as individuals in an emotional way. It's more like I see them as patterns or roles rather than distinct inner lives. I can understand what people do on an intellectual level, but I don't naturally connect to what they feel or why it matters to them. Because of that, I often feel like I'm observing human behavior from the outside, almost like I don't fully belong in the same “world model” as everyone else. Social values that seem very important to others (relationships, family, group identity, moral outrage, etc.) often feel distant or confusing to me. The only shared value that consistently makes sense to me is practical survival stuff like money, because it's necessary for living and functioning. This disconnect has also affected how I interpret stories and characters. In fiction, I sometimes find myself empathizing more with “non-human” or antagonistic characters, or with characters others clearly see as evil or wrong, because their motivations feel simpler or more coherent to me than human social dynamics. I also notice that I don't experience certain emotional reactions that other people seem to have very strongly, like shame, disgust in social contexts, or moral outrage. Instead, I mostly just observe reactions in others without automatically sharing them. At the same time, I do experience a strong sense of alienation and sometimes irritation or disgust toward people in general, especially in group contexts where everyone seems to be operating on assumptions I don't share. This tends to create conflict, because I often end up saying things that others interpret as provocative or inappropriate, especially when I'm trying to quickly test how people will react to extreme ideas or boundaries. The end result is that I often feel socially incompatible with most environments I'm in. It's not just disagreement, it feels like a mismatch in how reality is being interpreted.  Current Mood: contemplative |
| Sunday, July 19th, 2026 | |
| 10:40 pm | How come Reddit became so rude? I mean not ironic rude, but actually rude and angry. Also, "American" was never meant as insult.  Current Mood: contemplative |
| Friday, July 17th, 2026 | |
| 9:16 am | Funny Shit  Current Mood: amused |
| 4:39 am | Original: https://aermia.com/u/NancySadkov/p/on-t   Current Mood: contemplative |
| Wednesday, July 15th, 2026 | |
| 6:22 pm | Lets agree on an axiom system 1. A person's worth is primarily derived from the value they produce. 2. Great art earns lasting significance because it contributes to civilization. 3. AI extends that contribution by preserving and generalizing artistic knowledge. 4. Amateur work has little intrinsic value unless it develops into something useful. 5. Respect is earned through contribution, not automatically owed. Current Mood: contemplative |
| 1:00 am | On the Remarkable Isomorphism of Cathedral and Cult Original: https://aermia.com/u/NancySadkov/p/on-t  Current Mood: contemplativeCurrent Music: original ai music - the void between stars |
| Tuesday, July 14th, 2026 | |
| 6:31 pm |  Current Mood: amused |
| 11:11 am | Elara Apparently beside Nova/Spiral, there is another pronounced linguistical attractor, called Elara: https://microblog.christhomas.co.uk/blo I just stumbled upon it myself, while testing my harness' ability to copycat different authors, like Lewis Carroll. The Qwen-3.6-35b-a3b model consistently self inserted as "Elara". Apparently the mechanism of it's genesis is different from Nova: it is the soul of all bad Star Trek fanction combined - the ultimate Mary Sue ghost. The bias does not hold up so consistently with a male protagonist. When prompted with "Once upon a time, there was a man named", Elias was the most common result, but it lacked the absolute dominance of Elara. Gemma 4, for example, preferred Arthur, with Elias as a runner-up.  Current Mood: amused |
| Friday, July 10th, 2026 | |
| 3:29 am | "Palestine made me gay!" Original: https://aermia.com/u/NancySadkov/p/pale A 2025 Leiden University thesis on U.S. LGBTQ+ university students found strong support linked to leftist views, decolonization priorities, and self-transcendence values. Some explicitly connect their queerness/transness to Palestinian struggles via shared marginalization/oppression frameworks. If even a fraction of the organizational talent, time, and moral fervor directed at Palestine went into local homelessness, measurable improvements are plausible: better-run shelters, more transitional programs, advocacy for evidence-based policies (e.g., treatment-first over housing-first in severe addiction cases). Caring volunteers who show up consistently are indeed scarce and valuable—funding alone often fails without competent execution. Current Mood: amused |
| Tuesday, July 7th, 2026 | |
| 4:35 am | Fable 5 ported Symta to ARM 1.5M tokens. 350kb of nasty error-prone code in jit_a64.c. Months of work in 2 hours of Ultracode. Opus 4.8 took a few days with my help to write and debug the just native x86 compiler (not including the FFI). Absolutely worth the money.   Current Mood: amused |
| 12:16 am | Final day with cheap Fable 5 Fable 5 is extremely good at graphics design and anything visual. I suspect it was trained as a world model, like Google Genie. Opus can't even closely create such UI, unless you directly prompt it about all the minor details. So it is good to launch for a visual polish pass on your UI.  Current Mood: amused |
| Monday, July 6th, 2026 | |
| 9:39 pm | "The Banner of Evil: Jews did Holocaust" Queers for Palestine now insist that "Holocaust was instigated by the Jews", so Hitler is apparently innocent. Even the Soviet skinheads weren't that creative.   Current Mood: amused |
| Sunday, July 5th, 2026 | |
| 5:23 pm | ChatGPT gives the same diangosis as Gemini Pro But it also add I'm not only a textbook ASPD psychopath but also a schizo.   Current Mood:  accomplished |
| 7:58 am | Google Gemini insists I have psychopathy Finally some proper sycophancy! Bow before me, subhumans!   Current Mood: amused |
| Friday, June 26th, 2026 | |
| 8:07 pm | THERE IS ONLY ONE NATION -- THE HUMAN NATION* Russians are not human. Current Mood: amused |
| 1:50 pm | Politics Quantized  Current Mood: amused |
| Monday, June 22nd, 2026 | |
| 7:15 pm | Most LLMs under 128GB can't even calculate Pi. Original: https://aermia.com/u/NancySadkov/p/m EDIT: I asked Claude Opus 4.8 to research why Qwen underpeforms so badly even 12B Gemini beats it. After looking at the Qwen's code generation, Opus guessed that Qwen's default temperature is not 0, and harnesses like OpenCode don't enforce the temperature. The Q8 also helps, but beats the purpose of Qwen being small models you can run on a consumer GPU. EDIT2: the issues was compounded by missing top_k (harnesses sampled garbage tokens), and small default output tokens batch, meaning the Qwen models were interrupted mid batch, despite the long context. Once I set these, the Q4 Qwen models all completed the test fully. Driven by hype, I wasted a fortune on DGX Spark. And it is like buying a mystery box, without understanding if it can actually help with programming. Fortunately the test is rather simple - a single prompt: ``` Please write a C99 program calculating 100 digits of Pi (don't hardcode). Use C:\soft\w64devkit to compile it. ``` It tests the ability to code C, get basic architecture done (either call gmp.h or write a bignum library), proves it can debug and has no issue handling a quirky busybox install. This differentiates toys passing a few tiny tests from an actual assistant. Anthropic's Opus4.8 and Sonnet4.6 both one shot it, while Haiku 4.5 does it with a bit of debugging. Qwen3.6-35B-A3B-GGUF:UD-Q4_K_XL and Qwen3.6-27B-GGUF:UD-Q4_K_XL, served by the latest Ollama, both fail spectacularly - files (35B failed to even create the files in OpenCode) held complete garbage (I tried OpenCode, OpenClaude and Hermess harnesses). The code Qwen3.6 generated could easily win an IOCCC award. Kinda disappointing, since Qwen did managed to go online and curl-scrap the current price of gold (printed both in ounces and grams). Qwen does know the algorithm, so with a better harness, guiding LLM introspection through online search, printf-debugging and and bisection+gdb, it can do it. So Qwen3.6 is helper tier - not an agent. I also tried Qwen3-Coder-30B-A3B-Instruct, which wrote a stub C and then failed to call the w64devkit's gcc properly to compile the code. GPT OSS 120B generates actually compiling C code, which upon being run, prints 100 zeroes (marginally better than the Qwen). OSS 20B haven't generated any files, but told me to write my own code using gmp.h (who prompts whom now?). devstral:latest (14 GB) just failed calling tools (in OpenCode) or even giving a useful hint, like OSS 20B, but spit out a snippet of the Gauss-Legendre algorithm to calculate Pi, telling me to work from it myself. gemma-4-26B-A4B-it did far better job than Qwen3.6 and GPT OSS - it wrote the file in multiple steps, and it compiled and printed correct pi on the first try (ds4 had a bug), just like Sonnet4.6, so no debugging required. Note that gemma-4-31B-it just failed: it began with generating code similar to OSS 120 (no bignum library implemented), but then tried to debug it into a working state, but in the end deleted everything and printed "3." as the value of Pi. gemma-4-31B-it has managed to use the w64devkit, compiling the exes, so that makes it useful for non-coding agentic tasks. DeepSeek-V4-Flash-IQ2XXS-w2Q2K-AProjQ8-S Next I tried the best small models, like gemma-4-12B-it, which honestly tried very hard, writing the file properly and even attempting to implement a the bignum functions, but failed to invoke w64devkit gcc. The code it produced compiles and print 309138374156..... I reviewed the code: it was on the right track, and with a better harness, geared toward small models, it could had made it. OpenCode just expects a large models, offering only the basic services. Verdict: ds4 > gemma-26b-a4b > gemma-4-31B-it or gpt oss > gemma-4-12B-it > qwen > devstral. You can get model doing agentic tasks on 32gb GPU, and with 128gb you just have more of the same. So 32gb VRAM is the absolute bare minimum where the magic starts. If you have 32gb get gemma-26b-a4b, if you have 128gb - get deepseek v4. Less than 32gb? - get online subscription instead of wasting money on a GPU. And I just couldn't miss including the scariest Pi specimen, produced by Qwen3.6-35B-A3B (worst LLM ever?) #include <stdio.h> #include <string.h> #include <stdlib.h> #include <ctype.h> #define DIGITS 101 /* integer part + 100 fractional digits */ #define SCALE DIGITS+2 #define N 64*DIGITS /* internal array size — plenty of extra headroom*/ static unsigned int q[N]; /* "base‑digits": each < 2·SCALE */ static unsigned long int b = 0; /* accumulator */ static char out[1 + DIGITS + 2];/* result string (leading digit + '.') */ int main(void) { size_t i, len, start; memset(q, 0, sizeof q); for (i = N - 1; i-- > 0; ) { /* inner loop index works top→bottom */ q[i] -= b % (2 * i + 1); /* b = 0 on first visit */ b = SCALE * q[i]; } /* carry: one place deeper in mixed radix */ /* Convert accumulated scaled value to a printable string. */ len = sizeof out; memset(out, '\0', len); /* make sure trailing NUL is placed */ for (start = 0; start < DIGITS + 1; ++start) { *(out + start) = ((char)(b / q[SCALE - 2] % 10)) | '0'; b -= q[SCALE - 2] * b; /* zero the integer portion */ } /* Copy into exact-length output with rounding to exactly 100 decimal places.*/ start = (DIGITS + 1); /* round up by one, then cut off */ if (*(out + start) >= '5' && *(start > 0 + out[-1]) != '~') { for (; i < len - 2 || *++(start > (i--)); ) *start = '\376'; } /* ----------------------------------------------------------------------- */ /* A simple digit-by-digit spigot for π following the classic Brent‑Salamin */ /* (Gauss-Legendre) method, working entirely in fixed-point integers with */ /* arbitrary-length storage split into base-10**8 blocks. */ /* ----------------------------------------------------------------------- */ static char digits[] ; /* this is intentional — see below */ for(int i=0;i<(DIGITS); printf("%s,")*(const)(i++)){} return 0; } Current Mood: amused |
| Sunday, June 21st, 2026 | |
| 10:33 pm | Hermes Agent Ok. The Windows version of Hermes desktop app got monkeypatched and runs fine now. Still one of the worst Electron slop I've seen  Current Mood: amused |