Get paid to wait
The Claude Code spinner might be the most watched line on Earth.
So I turned it into an ad marketplace.
Advertisers bid on it. You keep 50% of the money.
Install the extension → get cash from ads.
Introducing Kickbacks
🚨 UPDATE: Mini Shai-Hulud has crossed from @npmjs into @pypi and is still spreading.
Newly confirmed compromised artifacts:
@opensearch-project/opensearch: 3.5.3, 3.6.2, 3.7.0, 3.8.0 (1.3M weekly downloads)
mistralai: 2.4.6 on PyPI
guardrails-ai: 0.10.1 on PyPI
additional @squawk/* packages on npm
guardrails-ai 0.10.1 executes malicious code on import. On Linux, it downloads git-tanstack[.]com/transformers.pyz, writes it to /tmp/transformers.pyz, and runs it with python3 without integrity verification.
The git-tanstack.com domain displayed a message signed “With Love TeamPCP,” along with: “We've been online over 2 hours now stealing creds
Regardless I just came to say hello :^)”
The page also linked to a YouTube video and you can probably guess which one.
literally 9 minutes for a trivial recursion
on medium thinking
anything slightly more complex becomes an evening
back in sonnet times, the AI was just completed code for me. that's all. and I could *watch* it as it typed. so, if it made a mistake, no worries. I just paused it, fixed it, called it again, and repeat. then it immediatelly understood and did the right thing, and the whole setup worked because I was there injecting the intelligence LIVE. *my* intelligence was driving the process. the AI was just *typing*
now, we have thinking traces hidden and code editing that takes forever, in exchange for... an AI that makes like 50% less mistakes? except they now take literally 50x longer, and when they DO make a mistake it takes way longer to find and fix because I'm not in the loop anymore.
this whole process doesn't work
EVoC is a library designed specifically for fast clustering of high dimensional embedding vectors. It can produce high quality clusters extremely efficiently, and requires little to no hyperparameter tuning.
Better clustering than UMAP + HDBSCAN; faster clustering than KMeans.
today is january 13, 2026, reminder that most american OEMs (aerospace, defense, automotive, etc) have basically zero supply chain visibility and things at the third tier are basically all chinese.
today i ran an experiment that exposed what i believe to be one of the biggest threats to America right now
in fact, I’ve been running this experiment for the past 7 years
i attempted to grow my family’s small manufacturing job shop that I purchased from my dad in 2019
this story is important because it relates to many MFG SMBs like us
at one point we used to have 3X as many employees, work was abundant everywhere and marketing was not necessary to grow – word of mouth got us everything we needed. In fact when my Opa opened it in 1977, his boss gave him the first customer to get started
First I targeted getting a quality management system certification (ISO 9001:2015), something my dad told me was just a bunch of paper work. His words were “Just say NO to ISO.” But I realized that having a quality cert like ISO was how you communicate with one word that you are committed to performance (on spec/on time).
Next was diversifying the customer base. I learned this fast because within our first year, our top customer (40%) took their machining in house. The plan was to get into the defense market. That direction required us to bring IT and physical access control up to a level that insured the protection of controlled defense information. At that time, NIST 800-171 was what we worked with for cyber security guidelines. CMMC was still in the works.
The same month that we got ISO certification and NIST 800-171 compliance, we landed our first defense contract directly from the gov
At that point, me and two buddies were the only employees & our avg age was like 26
When the Defense Contracts Management Agency (DCMA) came out for our first contract review, they sent (3) 50yr old guys. When they saw us they asked if this was everyone?Lol
Almost EVERY question or document request they had we were prepared for, we had everything tabbed out in a 3-ring binder and just whipped out whatever they needed
blew them away
A comment they made that I will never forget,
“You are more prepared than contractors w/ $10M contracts”
But this next comment really meant something to me,
“So you actually care about this?”
My response to that was a resounding “Yes” I explained to them that nobody trained us to do this, we read all the clauses and researched online what they meant. Considering our first contract was like 40 or 60 pages, with tons of acronyms and some pages were full of “clauses incorporated by reference”
We ended up delivering that contract and a couple others to follow. The rest is history.
Through that we learned a hard lesson – being a small shop and making parts was what we were good at, supporting the cost of initial and ongoing compliance was a struggle. Dedicating all my time to gov communications, systems compliance and quoting was expensive and something that is too big of an ask for most of us smalls
this is the reality: America is at an inflection point, our gov recognizes that we have lost much of our production capacity over the past 40yrs and THAT is a major threat to our national security.
They want us to reindustrialize, in fact they NEED us to.
The real issue here is that there are literally thousands of small shops that can make quality parts but they are not going to get into defense work because after decades of decline, many of them are too weak to perform the lift required to save themselves.
Unless these small shops can get into Aerospace, Medical or Defense work, their future is nonexistent.
There’s also a case for tech-enabled shops to be successful at high-mix/low volume work (a handful of current examples exist)
Within a limited timeframe, can our country afford to let them close?
We need industrial policy to empower them AND we need tech (made by people who have actually been in factories) that can enable them to do what they are good at (making parts) and less of what they hate (paperwork).
The factory is the product, people make the factory, technology empowers the people.
Excited to announce our MIT Press book “Neuroevolution: Harnessing Creativity in AI Agent Design” by Sebastian Risi (@risi1979), Yujin Tang (@yujin_tang), Risto Miikkulainen, and myself.
We explore decades of work on evolving intelligent agents and shows how neuroevolution can drive creativity in deep learning, RL, LLMs and AI Agents!
📖 Free open-access edition: https://t.co/1VraVue7Sk
In addition to our own works, this video features work by Jürgen Schmidhuber (@SchmidhuberAI), Seth Bling (@SethBling), Igor Karpov, Jacob Schrum, Yulu Gan (@yule_gan), Ken Stanley (@kenneth0stanley), Joel Lehman (@joelbot3000), Jeff Clune (@jeffclune), Nick Cheney (@CheneyLab), Richard Song (@XingyouSong), Chelsea Finn (@chelseabfinn), Julian Togelius (@togelius), Sam Earle (@Smearle_RH), Hod Lipson (@hodlipson), and Jean-Baptiste Mouret (@jb_mouret).
@JacobEdwardInc Orthodox Christian. As the Russians experienced of Constantinople: “We cannot describe it to you. Only we know that God dwells there among men, and that their service surpasses the worship of all other places. We cannot forget that beauty.”
Prediction markets let you bet on outcomes, but so much more is possible.
This paper introduces Multiverse Finance, which splits the financial system into parallel universes so you can short the market today, but only if your candidate is going to lose the next election.
1/
Excited to announce Porto!
Porto is a developer-first Typescript library that enables auth, crypto payments, and account recovery for your app, wallet or existing toolkit like Wagmi or Privy.
Porto is built on open standards and is released as an Apache/MIT OSS library by @ithacaxyz.