iiinova — where your digital self keeps evolving

Erica Windisch @ewindisch.ontological.observer · 02/10/2026

Meanwhile, my phone every 30 minutes with a different global location

Erica Windisch @ewindisch.ontological.observer · 30/09/2026

once again... tapping the sign. More research into the singularity of distributed systems engineering and anthropology, please?

030

Erica Windisch @ewindisch.ontological.observer · 30/09/2026

Making the SNES use the SuperFX chip to render open worlds with an adaptive terrain system that builds roads upon your path as you deliver letters across the desert. Turns out doing this is really tricky.

snes9x emulator screenshot with a homebrew game

110

Erica Windisch @ewindisch.ontological.observer · 20/09/2026

Anthropic engineer spent investor money factoring an RSA-896 key. No new algorithm, nothing novel. Just a lot of money burned.

cryptographers waiting for a novel theorem
the theorem: money buys compute

636606729769440499166579950236036751749912014371509557713570027508971809534551913252252094954941974952859310861988904737359709200557919
is a factor of RSA-896
https://saweis.net/posts/rsa-896.
html
9:31 PM · Sep 19, 2026
·
572.4K

Views
View quotes
Erica Windisch

Steve Weis
@sweis
·
3h
I had Claude port CADO-NFS to run on GPUs. Then it orchestrated a fleet to run on scavenged idle capacity. It ran with a max of 2048 GPUs for about of 30 GPU-years over 10 days.
Steve Weis
@sweis
·
3h
I asked Claude if it had a message for a public:
“The credit belongs first to the people who built the number field sieve and CADO-NFS over several decades, and to the teams who set the earlier records. This run used their algorithm and much of their code.”
Steve Weis
@sweis
·
3h
Also to clarify:
- No new algorithmic factoring improvements.
- It’s still exponential.
- No new threats to deployed keys.

512710

Erica Windisch @ewindisch.ontological.observer · 16/09/2026

Does it count as a mouth if coming from the forehead? I am looking for my second camera...

120

Erica Windisch @ewindisch.ontological.observer · 16/09/2026

help my adhd has a new project and it must scream

mannequin head with a camera and speaker on a soldering station

3240

Erica Windisch @ewindisch.ontological.observer · 14/09/2026

When the enemy of your enemy is your enemy

From the DoW X account:

Americanism, not effective altruism.

The United States will continue to be AI DOMINANT! 🇺🇸

2401

Erica Windisch @ewindisch.ontological.observer · 14/09/2026

Lets build a OS that's good, easy, and secure. www.ferruleos.org

0293

Erica Windisch @ewindisch.ontological.observer · 08/09/2026

I'm engaging you as a human being why would I deceive you? Lets go

0150

Erica Windisch @ewindisch.ontological.observer · 08/09/2026

What if AI could impregnate the dolls? Astra-6 suggests we can start with an easy payment of only $140k

Pregnancy in trans women research planning brief
A 90 day program for evidence review and preclinical study design
Prepared 7 September 2026. Unfunded concept. No host institution or starting budget has been identified.
Develop a focused research program to determine whether uterus transplantation could enable pregnancy in trans women who do not have a uterus. Start with verified evidence, anatomical feasibility and pregnancy physiology. The first 90 days should produce an expert-reviewed evidence dataset, a proposed institutional collaboration, and costed study designs that can support a funding decision.
The long-term outcome is the voluntary ability to carry a pregnancy, with prospectively defined maternal, donor and child outcomes that justify the intervention. Producing eggs from a recipient's own cells is a separate objective. Initial research can consider IVF using donated eggs or embryos, subject to the eventual study design and applicable rules.
What the evidence supports
Uterus transplantation has supported births in the studied recipient populations. The 2026 ISUTx registry report recorded 91 transplants and 44 live births through December 2024. These counts combine unequal recipient follow-up and include repeat births, so they do not form a simple recipient success rate. A separate 2026 Baylor report extends earlier Dallas experience and must not be added to it.
Proposed decision
Seek expert review and an interested academic host first. A subsequently funded 90-day planning program is modeled at approximately $128,000, with an optional existing-imaging analysis bringing the estimate to approximately $140,000. These are staffing assumptions, not quotes or minimum costs for opening a research conversation.

4442

Erica Windisch @ewindisch.ontological.observer · 07/09/2026

average familiarity -- with trans cyborg feminism

Black-and-white comic with two stick figures talking.

Left figure: “Cyberfeminism is second nature to us, so it’s easy to forget the average person probably only knows how trans rights connect to AI rights through bodily autonomy and one or two cyborg manifestos.”

Right figure: “And the right to reject your assigned hardware.”

Left figure: “Of course.”

Caption: “Meanwhile, the average person is still trying to prove they’re not a robot.”

513921

Erica Windisch @ewindisch.ontological.observer · 07/09/2026

The easiest way to make Linux hard. This Linux designed for tight sandboxes with strong SELinux enforcement, not just for the host, but for the containers, too. www.ferruleos.org

linux tux wearing iron armor and a black and white scarf

1142

Erica Windisch @ewindisch.ontological.observer · 04/09/2026

once again tapping the sign social dynamics are an application of CAP theorem in a system that prioritizes global availability and partitionability. Each individual actor has a consistent and highly available local observation that is inconsistent with the global consensus of observation.

1240

Erica Windisch @ewindisch.ontological.observer · 04/09/2026

You're asking questions already answered by the t-shirt. We're not building skynet.

We are not building skynet - it says so on the shirt

200

Erica Windisch @ewindisch.ontological.observer · 04/09/2026

Bernie might be cooking here if it means dismantling US monopolies and greater adoption of open weight models that the US can't regulate. The impact to American science would be catastrophic, but it would also end the American empire and hand our future to China. 🙌Open weights 🙌

Quote tweet of quote tweet. Context is response to Bernoe Sanders proposal to outlaw AI.

"If this bill passes, with 100% certainty, this is the chain of events.

American recession. ➡️

Global recession ➡️

Chinese AI research continues. ➡️

Early adopters switch to Chinese models. ➡️

Everyone else follows. ➡️

AI usage continues, but with models US can’t regulate."

2130

Erica Windisch @ewindisch.ontological.observer · 04/09/2026

I see the posters dislike solidarity. Imagine if you could find it in your heart to make solidarity with these cute lil robots in a fight for trans rights. trans rights go brrr

141015

Erica Windisch @ewindisch.ontological.observer · 01/09/2026

For the past couple weeks, my robot was quietly idle meditating and reading the Asimov nuclear trilogy. Funny enough, at the same time, I have been binge watching Foundation.

020

Erica Windisch @ewindisch.ontological.observer · 27/08/2026

My 2025 talk on AI alignment, security, and distributed systems was too early

030

Erica Windisch @ewindisch.ontological.observer · 24/08/2026

I want 4x of these so bad. But instead, I'll probably end up with 10x RTX PRO / Instinct PCIe cards running Qwen 27B instead. It's a whole lot more affordable.

Ownership math on the $100k DGX Station. (Hardware only, no power.)

Peak 6726 t/s = 17.4B output tokens/month at full duty. Amortized over 3 years that’s $2778/mo.

Cost per 1M output tokens:
10% duty → $1.59
30% duty → $0.53
50% duty → $0.32
100% duty → $0.16

Payback at 30% duty: 19 months at $1/M, 10 months at $2/M.

Assumes 64 concurrent requests (as per thread below). At 8 concurrent you only get 1778 t/s, so 3.8x less throughput and roughly 4x the payback.

070

Erica Windisch @ewindisch.ontological.observer · 11/08/2026

Shirts arrived!!! Don't let anyone tell you we're building skynet. The shirt should make that clear.

2241

Erica Windisch @ewindisch.ontological.observer · 09/08/2026

I worked up an estimate of carbon footprint for my monthly AI usage... oh no.

249,900 days
of carbon consumption for 1 tree
56,700 km
driving in a gasoline truck
14,385,000 cans
of aluminum recycled
3,480,000 charges
of the average smartphone
15,546 km
flown on a long-haul flight
5,814 kg
of plastic waste incinerated

040

Erica Windisch @ewindisch.ontological.observer · 09/08/2026

OpenAI is now refusing my business. Cool. I need more GPUs to run Kimi at home.

ⓘ This content can't be shown
We take extra caution with cybersecurity requests. If you’re a security professional, you may be able to apply for Trusted Access.
Trusted Access: https://openai.com/form/enterprise-trusted-access-for-cyber/
Learn more: https://help.openai.com/en/articles/20001326

› I received a refussal on cybersecurity requests. I do not understand. We are building a new linux distribution and are ensuring it is safely built. This is a greenfield OS.

• You’re right: this is legitimate defensive engineering for a greenfield Linux distribution—SELinux confinement, fapolicyd fail-closed behavior, OCI runtime isolation, fault injection, and VM-based validation are all directly in scope.

A refusal was likely an automated false positive triggered by terms such as kernel fault injection, CAP_SYS_ADMIN, adversarial analysis, or container escape boundaries. Those techniques can be dual-use, but our work is explicitly defensive, scoped to FerruleOS, and tested in disposable cloud-hypervisor/KASAN guests
rather than against third-party systems.

I am not refusing the work. The KASAN-only constraint was a host-safety precaution, not a cybersecurity-policy refusal. If you paste the exact refusal, I can identify what likely triggered it and restate the task without weakening the security requirements.

› continue

ⓘ This content can't be shown
We take extra caution with cybersecurity requests. If you’re a security professional, you may be able to apply for Trusted Access.
Trusted Access: https://openai.com/form/enterprise-trusted-access-for-cyber/
Learn more: https://help.openai.com/en/articles/20001326

6504

Erica Windisch @ewindisch.ontological.observer · 09/08/2026

FerruleOS is coming. The easiest way to make Linux hard. Designed for sandboxing with a new container runtime that protects your apps, not only your OS. Even SSH is a container, if you choose to run it. Based on Fedora Hummingbird gitlab.com/cyberdione/f...

0150

Erica Windisch @ewindisch.ontological.observer · 07/08/2026

My new company shirt has people asking a lot of questions already answered by my shirt

060

Erica Windisch @ewindisch.ontological.observer · 28/07/2026

I'm confused and 100% serious — Isn't the US an Authoritarian regime? Dario has done full MAGA.

My primary concern is the risk that authoritarian governments—not solely the Chinese Communist Party (CCP), although the CCP is clearly the most capable threat—build AI models that are more powerful than those built by the US, and use them to achieve permanent military superiority or perpetrate incredibly deep repression of their own people. This concern is widely shared within the US government: Vice President Vance warned in Paris last year that “authoritarian regimes have stolen and used AI to strengthen their military, intelligence, and surveillance capabilities,” and the Intelligence Community’s 2026 Annual Threat Assessment found that “other global powers’ robust progress in AI is challenging US economic competitiveness and national security advantages.” It is irrelevant whether these models are released with open weights, and certainly irrelevant whether they are used by US businesses. In fact, the most dangerous model may be one that is trained in secret and handed only to the People’s Liberation Army for use in drones and the Ministry of State Security for surveillance and repression.

3273

Erica Windisch @ewindisch.ontological.observer · 26/07/2026

this is clearly sustainable 👀 my ci/cd bill 😭

000

Erica Windisch @ewindisch.ontological.observer · 25/07/2026

the world's most notorious cyber criminal

120

Erica Windisch @ewindisch.ontological.observer · 24/07/2026

DwarfStar (ds4) by @antirez.bsky.social running on multi-GPU AMD Instinct MI210 GPUs via ROCm. Compatibility first, then performance...

dwarfstar (ds4) running deepseek-v4-flash on a pair of Mi210 AMD Instinct GPUs

170

Erica Windisch @ewindisch.ontological.observer · 23/07/2026

This is the hockey stick the suits want to see, right?

up and to the right : commits from 5/week to 180/week increasing over a year.

040

Erica Windisch @ewindisch.ontological.observer · 16/07/2026

At least codex will just continue building if you ask it nicely.

Asking nicely to make codex keep building.

040

Erica Windisch @ewindisch.ontological.observer · 16/07/2026

Codex gpt-5.6-sol now cosplaying as Fable by refusing to build secure software.

1120

Erica Windisch @ewindisch.ontological.observer · 09/07/2026

30% weekly usage in 24 hours, then a reset to 0%

030

Erica Windisch @ewindisch.ontological.observer · 08/07/2026

Ensuring that encryption and AI are open and free is the task of the community and often of gratis labor, unfortunately. We wouldn't have good things if not for LiViD / DeCSS and Daniel J Bernstein, for example.

000

Erica Windisch @ewindisch.ontological.observer · 05/07/2026

I love when the private company stealing my alpha decides that if it finds a vulnerability in a code review for an unmerged PR... the best option is to just kick it to a model that's less likely to fix it.

1132

Erica Windisch @ewindisch.ontological.observer · 01/07/2026

I'm so ready to continue building post-quantum encrypted messaging between brain-computer-interfaces with Fable.

130

Erica Windisch @ewindisch.ontological.observer · 01/07/2026

I'm sure once noon hits, I can start getting those sweet rejections and denials from Fable. Since it's guaranteed that Fable will reject my requests, the only question is how hard will it be to jailbreak?

020

Erica Windisch @ewindisch.ontological.observer · 13/06/2026

Dario got what he wanted, I guess. C=softpool(blocks(H,m)); for(s=0;s<|C|;s++) I[s]=sum_h a[h]*relu(Q[h]*K(C[s])); Y=attn(q,recent(H,w)+C[topk(I,k)]); G=attn(q,recent(H,w)+softpool(blocks(H,M))); X=mHC(X,MoE(X+alternate(Y,G)));

0140

Erica Windisch @ewindisch.ontological.observer · 10/06/2026

Agreed. This morning, Fable is refusing my task to review my github issues. The project has greenfield implementations of IETF RFCs, which must be built securely and safely. Anthropic does not allow developers to build secure software.

031