HNHacker News
TopNewBestAskShowJobs

cultofmetatron

4,628 karma · joined March 27, 2007

submissionscomments
cultofmetatron··on India's surprise baby bust
plenty of women in LATAM love possessive jealous men. I've seen it first hand.
cultofmetatron··on The Pentagon is running an AI propaganda mill targeting Latin America
millions died due to exploitative labor practices by colonialist invaders exploiting the resources and cheap labor potential of the people in it yet I don't see you aknowledge that.

communism wasn't behind american slavery or the Belgian occupation of the congo.

Not a fan of communism but I dont' think unfettered capitalism is much better. both systems benefit a minority of people at the expense of the majority largely because they allow power to be concentrated in the hands of a few.

At this point, anyone spouting vitriole about communism and socialism like they are the same thing just come off as lacking basic capacity to understand nuance at best and mentally ill at worst.

cultofmetatron··on Self-hosted dev sandboxes with preview URLs (Docker, Go, no K8s)
firecraker is optimized for lambda. ie: fire and forget. not so much live systems that maintain long running state. also, I dont' think you can run it on top of a hypervisor.
cultofmetatron··on Elixir v1.20: Now a gradually typed language
yes, they do. its minimal though
cultofmetatron··on Failing grades soar with AI usage, dwindling math skills in Berkeley CS classes
obligatory https://smbc-wiki.com/index.php/2014-12-17
cultofmetatron··on Nvidia RTX Spark
can these do training or only inference? currently working on learning machine learning and I'd love to have a physical machine I could aim to build real workloads on in a few years.
cultofmetatron··on Why are large language models so terrible at video games?
cough JEPA cough
cultofmetatron··on Ask HN: Entrepreneurs, how long did it take you to succeed?
> I am a middle aged man who lives on a sailing boat.

literally my goal. hopefully we can meet up on the strait of malaca at some point

cultofmetatron··on The Lone Lisp Heap
this is my favorite thing about agentic coding. its super easy to build a v1 and get the system to a point where it does what you want which means you learn what you need for the v2 much faster
cultofmetatron··on Building durable workflows on Postgres
conversely, startups that start scaling for tbs of data never make it to needing tbs of data. They burn too much energy scaling when they don't yet have a product people want yet.
cultofmetatron··on What I've Learned (So Far) Building Online Mini Games with Elixir and Swift
not op either but I built my entire startup using elixir. We have a realtime sync component across a range of devices that sync data between them.

It (phoenix channels) worked out of the box. in the last 6 years of development and growth, not once have we ever had to even discuss scaling issues related to it.

It just works and when you're in a startup, the more you can focus on growing your product instead of scaling issues, the more you're winning.

cultofmetatron··on The User Is Visibly Frustrated
frustration detection and model escalation aside. there's probably also a lot of training data between two interacting people in which swearing escalates the goal oriented behavior of the yelled at individual.
cultofmetatron··on Time to talk about my writerdeck
I feel so called out ^_^
cultofmetatron··on .NET (OK, C#) finally gets union types
Rust and Zig brought new ideas for memory management that Haskell, OCaml, Erlang sidestep having garbage control. its honestly amazing to me that they managed to get the adoption they have while being so innovative. I say this as a fulltime elixir dev.
cultofmetatron··on Show HN: Rmux – A programmable terminal multiplexer with a Playwright-style SDK
so far really liking it. if I had one critisism, I'd say a less awkward meta combo would be nice. ctrl-b feels awkward and I have caps remapped to ctrl. something on the same row as the caps lock would be way easier to learn from a muscle memory perspective.

that said, really appreciating the world you're putting into it!

cultofmetatron··on DeepSeek makes the V4 Pro price discount permanent
open code works with them today. I've been using it fulltime for 2 weeks so far.
cultofmetatron··on OpenAI Is Preparing to File for an IPO Soon
> Deepseekv4 is freely available to download you can host your own and sell it keeping the proceeds and it rivals Claude Opus 4.7.

good luck getting a machine that can run its specs though. Even flash is goign to require ponying up 5-10 grand to run the minimal specs for it. The vast majority of people will find their machine falls behind as tech progresses long before they get a return on that investment. That said, it does mean there will be a healthy market for "generic providers" in the AI landscape with these open weight models.

cultofmetatron··on Show HN: Rmux – A programmable terminal multiplexer with a Playwright-style SDK
well it looks great so far but Its a no go for me for now until you have at least a cli that a coding harness can use to drive it. thats going to be the killer feature going forward for all these multiplexers. last week I vibecoded a plugin that lets me open a line specific input box that sends the context of my code along with my prompt into my opencode session running on a seperate pane. its a game changer when it comes to moving faster.

excited to check back in on the progress as you work on it though! I think these kinds of tools are going to be much more relevant going forward as we all move to agentic workflows.

cultofmetatron··on Show HN: Rmux – A programmable terminal multiplexer with a Playwright-style SDK
my plan is to eventually move my entire workflow to a remote workmachine over ssh. with kernal, I can automate all my smoketests entirely from a headless machine rendering my laptop overkill to run agentic workflows. I'd rather be using it to watch movies, learn new stuff or hand code in my downtime. if your workflow requires cmux, you're stuck running your agents on your local machine.

To be clear, That doesn't make cmux bad. if you WANT to do everythign on your mac, cmux is fantastic. its just not designed with the remote ssh control use case in mind.

cultofmetatron··on Show HN: Rmux – A programmable terminal multiplexer with a Playwright-style SDK
commonlisp (sbcl) is one of the best sleeper techs out there. one of those if you're in the know type industry secrets. part of it is because their website looks like something out of the 90's. clearly having a nice website and some level of marketing is necessary to get adoption. Im more than happy to see devs use AI to automate that so they have more time to focus on their creations.
cultofmetatron··on Show HN: Rmux – A programmable terminal multiplexer with a Playwright-style SDK
a week ago I was using cmux but its osx only and doen't work on remote terminals. then I switched to herdr which is great so far except its not s great at managing panes. I can't move them around or change ordering. now another terminal multiplexer. I'm getting whiplash.

all that said, none of the existing solutions are perfect and rust codebases are nice. how easy is it to reorder panes? is there a cli that lets me control the panel layout via a skill file and allow my opencode session to target and send data to other panes?

cultofmetatron··on DeepSeek V4: The Open-Source Model Frontier Labs Feared
$10/month on opencode go and I have yet to hit the limits using just flash. literally I'll abuse it for 2 hours straight and my rolling usage goes up 5%. kimik2 will use that much in about 10 min of use
cultofmetatron··on Unknowable Math Can Help Hide Secrets
> to create a powerful new tool in cryptography.

I thought that was THE ENTIRE PREMISE of cryptography

cultofmetatron··on DeepSeek V4: The Open-Source Model Frontier Labs Feared
deepseek v4 flash is basically the anthropic killer. I've been able to offload the vast majority of my workflow to that using opencode go. between that and the occasional use of pro and kimi k.26, I don't understand what the big deal is about claudecode.
cultofmetatron··on Parents say ChatGPT got their son killed with bad advice on party drugs
> should have pointed their kid to Erowid.

solid advice. I know several people alive in spite of their efforts because of that site

cultofmetatron··on If AI writes your code, why use Python?
built my startup in elixir and can concur. elixir has a relatively consistent syntax that makes for a pretty good target for llms.

In my opinion, the only thing holding elixir back as an llm deliverable is that there's not as much training data for llms to work with.

Of course if we had a new AI that could be trained on a minimum of existing training data, common lisp would absolutely beat out everything else. everything you mentioned about elixir (repl, runtime, and ability to hot reload / directly test functions) are possible and were invented in lisp with an AST instead of a syntactic language as the ultimate build artifact. CL lets you recover from exceptions and rewind the stack before reloading your fixes and continuing. I can't even fathom the workloads an LLM could conceive of working with that.

cultofmetatron··on If AI writes your code, why use Python?
I can't imagine a better output for llms than python. not because its particularly good. far from it, its got dynamic typing and more or less sets you up for runtime failure. however, it has probably the largest corpus of training data aside from javascript.

Part of my worries that all this push to LLMs will marginalize niche programming languages from being used in startups since the lack of training data means falling back to hardcoding. a skill that I have a feeling will get increasingly niche overtime. I feel capitalism will basically render programming languages into a build artifact overtime.

cultofmetatron··on Singapore introduces caning for boys who bully others at school
> "we need to respect the differences between the genders" as your word salad reflects.

describing my points as "word salad" seems a bit dismissive and unnecessarily pejorative - but I digress. I'll ignore your obvious ideological axe to grind and see if we can at least reach common ground.

Watch any group of boys and girls at play. sure you have outliers but by and large they do tend to conform to one of two specific behavioral groups that historically we associated as ways boys and girls act. And I say this as someone who is fairy noncompetitive which itself made me somewhat of an outlier among boys when I was a kid. it doesn't change that the vast majority of boys found motivation in being competitive and historically pedagogy used that fact in order to motivate boys in their matriculation.

In my own experience, I tried to get my nieces more into tech and programming. trying to motivate them from a "hey isn't systems based thinking about how these things interact is cool" did absolutely nothing. Showing them how they can make a cool website design to show their friends (with a bit of vibe coding) absolutely got them more motivated.

But if you want to go further, lets start with the entire generation of boys who were told to sit still and listen and got medicated when they couldn't. We expected them to behave like girls and because we changed pedagogy to favor girls, we didn't notice when boys fell behind.

consider the two statements

- boys and girls exhibit dimorphism in behavior, motivations and interests.

- two boys can be more different from each other than between a boy and a girl.

These are only contradictory on a 2 dimensional graph. at 3 or higher, you can definitely have groupings of traits heavily skewed towards one group or another while having a different distribution on other traits.

I don't think the issue is us focusing on getting women into higher education. We changed the way we pick our educators and tuned our methodology primarily to benefit women. Its only a problem because we expect boys to excel in that system. its no more fair than what we had before when women were excluded at every level. treating them the same means shifting our methodology will inevitably benefit one at the expense of the other.

cultofmetatron··on I'm going back to writing code by hand
I've been relying primarily on deepseek-v4-flash for 90% of my work. It sips tokens. that model will run on 128gb. not a cheap configuration for a consumer but within the budget of a developer relying on it for work.

Ive only been using kimi 2.5 and deepseek pro for reviewing PRs for security issues. less than 10% of my workflow requires a full powered frontier model.

I think the issue is overblown by people who think claude code is a good harness and use opus for everything. opencode is objectively better. its much more verbose about what its doing, you have more control when it comes to offloading to subagents with targeted context (crucial for running through larger jobs) and I can swap between codex and open weight models.

cultofmetatron··on I'm going back to writing code by hand
the ship has sailed on my handcoding at work. the AI is producing stuff thats more bulletproof than what I can do in the same timeframe and if my competitors are using it, the pressure to ship is that much higher.

Personally, I've taken the time its freed up to spend more time on mathacademy and reading more theory oriented books on data structures and algorithms. AI coding systems are at their best when paired with someone with broad knowledge. knowing what to ask for and knowing the vocabulary to be specific about what you want to be built is going to be a much more valuable job skill going forward.

One example is a small AI based learning system I have been developing in my free time to help me learn. the mvp stored an entire knowledge graph and progress in markdown files. being an engineer, I knew this wouldn't scale so once I proved the concept viable, I moved everything into sqlite with a graphdb. then I decided to wrap some parts of teh functionality in to rust and put everything behind a small rust layer with the progress tracking logic still being in python.

someone with no knowlege of graph databases or dependncy graphs or heuristics would not be able to build this even if they had AI. they simply don't know what they dont' know and AI wont' save you there.

That said, I think its important to also spend time in the dirt. I've recently started pickign up zig as my NO AI langauge just to keep. those skills sharp.

← PreviousPage 3 of 34Next →