This is hitting normies hard, in a way other political violence hasn’t.

It really shook a lot of people deeply.

I felt this hard I think because I am a parent and he was a fairly moderate voice, always open to debate, wanted to win in the realm of ideas and discussions, and they just nerf him in public brutally. So sad and unfair.

And seeing the people on social media celebrating his murder with glee, I can’t even comprehend how they think, so nasty. It worries me how many people are out there, especially teachers, who are happy someone like him was murdered!

I feel like at least the culture in general is on the side of Kirk, and it’s becoming the extremists who are the agitators. Like 7 years ago this mighta been shrugged off, but now more and more of the shift is to be angry how the extreme democrats are behaving.

I just wish the schools took a more active role in policing instead of encouraging or turning a blind eye to teachers who behave like this. If they weren’t being called out on social media I doubt they’d take any action against their employees.

Nobody should fear being violently murdered in front of wife and children for having different political views.

Saying there are only 2 genders doesn’t mean someone is hate-filled.

Having conservative viewpoints and wanting to engage young college kids in respectful debate is not justification for violence.

All the people on social media gleeful a young man died because they disagree politically sicken me.

Believing in traditional values on stuff like abortion or lgbtq doesn’t mean someone is hateful and deserves to die.

This is pure brainwashing. People shouldn’t be scared for their lives to espouse values that others disagree with.

Cheering the death of someone, calling them a “nazi” (that word has lost all meaning) when young kids will grow up without a father is not healthy.

RIP. Charlie Kirk. And those poor kids who will never really know their father.

I find coding with AI is far more productive when you give it small discrete tasks.

The problem is to do that, you have to know what you’re doing with coding.

Asking it to modularize a large function at a certain point, or to add a performance decorator to a function, or use list comprehension for a task, means you know what to ask for.

Vibe coders can only ask for big tasks because they don’t know enough about coding to ask for small coherent tasks.

When I use coding agents like Claude Code, I give it very specific, pointed tasks.

Instead of vibe-coding a big ask, I code so I know what to ask for.

Instead of saying connect to Google Drive, I’ll say put the service account credentials into an AWS Secrets Manager via IAC.

Instead of saying make a microservice to generate RAG embeddings, I’ll say have a lambda use the Chonkify library with semantic chunking followed by a hugging face embedding model and push it to pinecone.

Instead of saying build a frontend that allows users to login, I’ll say create a cloudformation yaml file for cognito with oauth and a custom authorizer.

I find I’m more productive in the long run coding if I break a high level task into specific discrete tasks with specific frameworks. Planning mode on the agents help with this, and asking the agent to break up tasks can be useful, but there is often a benefit from asking it to write specific types of code.

Last we forget, a democracy means the will of the majority, whereas a republic means that people vote on those who will represent their interests because they don’t want to be responsible for deciding themselves.

What do you think of a big-government effort in which the money is allocated well?

So let’s say we collect the same $11 trillion to the government we’re already collecting as America, each year.

No healthcare or welfare, no military.

No administration, no science.

No space program, no national education system.

No environmental protection, no corporate oversight.

No law enforcement, no sewage or public utilities.

Pure chaos.

But let’s say you still have that $11 trillion collected each year.

What do you do with it?

What if you had the opportunity to truly start from scratch on spending, while collecting remained at current levels.

Obviously, survival is first, but what percentage of that $11T would be needed for military?

What if we immediately put a beautiful fountain in every town in America, because, well, we thought it was important?

What if we invested heavily in researching modern trains and traffic patterns and sewage disposal?

How much would go into border and law enforcement?

Would we build opera houses?

Would we invent new forms of social welfare programs that only can emerge when we’re no longer blinded by the momentum of the corrupt current system?

How would we actually prioritize if we got to start from scratch?

Maybe we’d realize that GDP and broken windows theory go hand in hand, or maybe we’ll realize environmental conservation is crucial?

Maybe we’ll find novel ways to link private sector energy consumption with property taxes?

I built a discord bot via n8n that runs every minute 24/7 and each minute:

1. Pulls the last ten discord messages it sent in my “backrooms” channel,
2. Picks a random LLM (grok, ChatGPT, Claude),
3. Picks a random personality I defined,
4. Sends a message as that personality to continue the convo.

It’s weird and fun, creating its own story.

Code and prompts here, open source:

https://github.com/Theta-Tech-AI/theta-backrooms

I’ve been obsessively reading sci-fi books from the 1940s - 1970s.

The idea of human intelligence versus machine intelligence versus alien intelligence are the prominent themes.

Space travel complications involved doing pen and paper calculations (orbital velocities, slingshots, etc). Basically a lot of how NASA got to the moon.

Alien intelligence involved themes like hive minds versus fragmented individuals, communicating through mathematical proofs, and the effects of gravity on evolution.

Machine intelligence is all about sentient machines, or organic-machine hybrids, but never about any takeoff singularity where it becomes more intelligent than all humans; it’s more like different types of intelligence.

I enjoy reading sci-fi books from this time period because the modern anxieties just aren’t there in the stories, and our politics scarcely seeps into the stories. The authors aren’t making any grand political points, they’re just exploring stories of space travel.

I can imagine a few different reasons that politicians have enacted open border policies in Europe.

Firstly, they’ve been brainwashed to believe that it is the compassionate and moral and empathetic option.

This is likely what your run of the mill libtard believes.

Secondly, they’ve been brainwashed to believe that if the population of nativeborn are reduced, then open borders will increase the “tax base”.

This is likely what those who fancy themselves as “smart” believe.

Thirdly, they believe they are importing people who will vote for their party in the future.

This is what the selfish, power-hungry politicians likely believe.

Fourthly, they’ve been led to believe that this is penance for white colonizers being the dominant power structure back in the 1700s and 1800s, a way to “rebalance the scales”.

This is what those who are ideologues likely believe, the “they just hate white people” argument.

What else am I missing?

It’s important to understand the motivation of the evil-doers and their useful idiots.

So that we’re all on the same page, Grokipedia is not actually open source but the real difference between it and Wikipedia is that in Grokipedia, the AI Grok (which model?) decides and explains which edits make it in, whereas in Wikipedia, individual power users are the editors.

Tonight, as I coded from 8pm to 2am, instead of using Claude Code or Codex-GPT5, I used the new Cursor 2.0 agent mode.

I liked it. I switched a lot back to editor mode, but I find that Agent mode’s right panel coalescing all the agent changes at once, combined with Cursor’s ability to select any backend model, worked well.

I let it use “Auto” as the model, assuming it would use its new blazing fast but smart Composer 1.0 model, but also manually selected that model.

The flow state of having an agent complete tasks just 30% faster than using something like Claude 4.5 Sonnet or GPT-5-codex, really helped my flow state with coding.

There was more rapid iteration and back and forth.

It also didn’t add any bloat the way Codex and Claude Code tend to.

At this point, I’m pretty much hopping between Claude Code CLI, Codex Extension in Cursor, and Cursor Agent mode, trying to get a sense of the pros and cons of each.

I hate how they all put their rules in different files (.cursorrules, AGENTS.md, CLAUDE.md).

Regardless, in this world of AI coding, you fall behind you get left behind.

So lest you want to be an old dinosaur with no hands on experience with the latest tools, you’d best practice every day.

And honestly, I’m still spending half my time cleaning up the mess of code that AI generates, because while it’s spot on with syntax, it really sucks at good coding practices and principles like DRY.

Onwards we march.

You want to know how you don’t get replaced by a machine?

Practice using the machines every day.

Per capita is important as it answers the question “if I meet a random person from a group, what’s the chance they’re a criminal?”

It’s crucial to understand what percentage of a group commits crimes.

It’s sad how much trouble people have understanding per capita. Manipulative.

Let’s say there’s a million people with square heads.

1% of them are murderers. 99% are not murderers.

Let’s say there’s a thousand people with triangle heads.

All of them are murderers.

If you meet a random person with a square head, and a random person with a triangle head, who should you fear?

“But there’s more square headed murderers out there in total!” is what happens when you ignore per-capita.

Alright, attempting to move from posgresql with pgvector to self-hosted weaviate on EKS for an AI RAG system.

pgvector's embedding search worked great with HSNW index, but the keyword search is quite annoying.

If I only wanted semantic search, I'd stick with pgvector. But sometimes, I need keyword search. For instance, querying my AI with "what do you know about my projects with the client SomeCompanyName" will fail on semantic embedding search, and required a keyword search.

On postgres with tsvector, AND isn't correct for long queries, and OR returned hundreds of thousands of rows, which made the retrieval take several seconds.

I was able to build my own RRF with SQL but I didn't like doing it. I'll still use my own cohere reranker afterwards.

Postgres doesn't have bm25 built-in, and I can't install extensions on AWS Aurora. I looked into setting up my own bm25 implementation with SQL queries (creating my own IDF table for word frequencies with sql triggers), but even then it doesn't have BlockMax-WAND the way the new Weaviate does.

I want to self-host, and I want to practice using kubernetes with a private VPC without paying for Weaviate's enterprise support, so I'm going to use the open source weaviate option with helm. I want to deploy everything with IAC (serverless framework) instead of AWS marketplace so I won't be doing that either. And while EC2 is an option, I'd rather not manage servers.

If any of what I said didn't make sense, sorry this post isn't for you. But I wanted to share my thought process of setting up my AI infrastructure.

Time to have fun with the tech! Wish me luck.

When you have kids, life is so much better when you have a big family and live near them.

We have all grandparents within a half hour.

There are approximately 20 children in our family within an hour of us between 1 and 9 years old.

Large birthday parties every month, activities for kids, family friendly events, support, bonding.

It just creates such a better experience for the little ones, and makes life more fulfilling.

Don’t deprive kids of family!

Grow large families!

Currently, the code gen LLM’s are still really bad at knowing:

- when to break functions into sub-functions
- where to include try-catch logic and when to let errors propagate
- when to create a service class versus call a sequence of functions
- when to return tuples versus dictionaries versus classes versus lists versus sets
- how to not repeat commonly used code, and inversely when to not create a “reusable” function that is only used once
- when it makes sense to use an environment variable versus a global scope versus a local scope
- where and how to properly parallelize

In short, if you know what to ask for, it does a great job. But if you’re using AI to design code architecture, it will not do better than a skilled coder.

Today marks 13 years since my father died from a heart attack at 52 years old.

His coffee mug from his morning coffee was still in the sink unwashed; he clearly thought he’d have time to wash it tomorrow.

But that tomorrow never came.

Don’t wait for tomorrow to be happy. Don’t think, “I’ll be happy when…” something happens. Tomorrow is not guaranteed.

He smoked about two packs of cigarettes a day. Don’t smoke cigs, don’t fall for immature psy-ops about how they make you look cool. They increase the risk of nearly every disease. And nicotine addiction is real.

He didn’t exercise enough, and didn’t eat very well. But plenty of people who don’t exercise or eat well live far past 52. It was the cigarettes.

He was loving, kind, and taught me good character. He and I would nerd out over Einstein’s theories of relativity, space travel, chess, and sci-fi books from when I was a young kid. He always had my back and would gently guide me through any situation.

He was born in Hungary, as was all his ancestors. His family fled to America (legally) to escape the Hungarian communists in the 1960s.

He never cared about politics, only his immediate family. He was extremely smart (many people said he was the smartest person they’d ever met) and had deep, complex thoughts.

I wish he could have met his grandchildren (both I and my sister have two children), but alas that was not his fate.

I keep his memory alive, and I try to learn whatever lessons I can from his death.

If your dad is still alive, keep the relationship flowing; one conversation may be your last and you’ll never know it until it’s too late.

I will meet him in the Great Beyond when my time is up.

🤍

I feel like one of the big missing pieces with AI systems is RAG that throws useful knowledge into the context window that doesn’t directly answer the query but is valuable.

If the query is about marketing, retrieval brand voice information is useful.

If asking a strategic question, throwing some analysis of Napoleon may be useful.

If generating code, pulling in SOLID principles book may be valuable context.

But those may not give a low cosine distance with embeddings, nor would they necessarily show up on a node or a knowledge graph from a random user query.

And relying on the LLM to consider outside information, even with RLHF has never worked.

This is, I feel, a missing piece: to determine outside context that needs to be fed in to a RAG system that the cosine embedding similarity, the keyword search, and knowledge graphs might miss.

This might potentially be solved with a good mcp server and a langgraph or ADK references that can pull extra knowledge with some context clues, but that requires pre-programming specific doc-strings for the outside nearby context and thus is not fully generalizability.

Any ideas on how to do this?

I just spend several hours going back and forth with ChatGPT 5.2 on the proper way to do a distributed microservice system with sqs, lambdas, weaviate.

It kept adding extraneous hash checks, different status fields in the database, and tried to plug every hole for every situation which will not exist.

It took a lot of time for us to converge on the proper architecture and workflow and to handle the edge cases we needed to handle.

It made up fake race conditions and came up with situations like a database failing halfway through a write with a weird rollback operation that simply wouldn’t occur.

It kept saying things like “well you could have concurrent processing and wasted compute” and only capitulated when I walked it through how that wouldn’t be wasted compute.

I don’t understand how vibe coders expect to one-shot AI code gen for advanced complex systems in a way that makes maintainable code that itself won’t need to be re-engineered later.

I still use AI to write the code for me (Opus 4.5 is my favorite right now in opencode cli) but not without back and forths.

People who don’t engineer things properly and think through a solid design are going to end up with slop.

Use the tools, don’t let the tools use you.

You can write high quality code using AI agents.

Not all AI generated code is vibe coded slop.

But it requires you to know what to ask for.

To tell it how to design your distributed database system;

Which hooks and mutations to implement in your web app;

How to do proper authentication with refresh tokens and oauth;

To know when to ask it to break up monolithic functions into small self-contained service classes;

When to ask it to use a dictionary versus a list;

Precisely how to handle Websocket dropped connections;

The way the agents should be designed with either mcp or simple tool lists.

If you know how to code, AI accelerates what you do yourself.

If you review every line the AI writes, you become more skilled at both coding and prompting.

Or if you just let the AI take the wheel, lord help you with a complicated mess of patch upon patch that can never be maintained.

Be smarter than your average vibe coder lest you stay a normie in a different hat.

There is a deep sadness in seeing someone in their 60s or 70s who wishes they were a grandparent, but whose adult kids decided to be child-free.

For them to survive into old age and then not be able to experience the joy of grandchildren, to see their bloodline end despite having kids themselves, makes me extremely sad for them.

If you want to educate yourself, you should study the math of how the KVQ matrixes work in modern AI.

Transformers in AI are at the heart of learning. In essence, the way it works is AI sees a new sentence from its training data and tweaks what matrix multiplication to apply to all the tokens such that it would have paid attention to the correct words in order to complete the sentence as close to the training sentence as possible.

The AI transformer basically learns how to change the way the input text is embedded in three different ways (K, V, Q) and keeps learning how to better embed them until it pays attention to the correct words and completes the sentence accurately.

The greatest sadness in my life is that my father never got to meet his grandchildren.

Take care of your health; heart attacks can come in your early 50s and then that’s all she wrote.

He would have been a great grandpa.