mkw5053

9 ms·

121.
by mkw5053 1y ago
I think I'm going to answer my own question by saying both momentum and energy are conserved. The momentum of the entire system was zero before and after the collision. Energy must also be conserved, and since the final object is at re
122.
by mkw5053 1y ago
So, if two black holes, each with mass M, were moving at nearly the speed of light and collided head-on (resulting in a final velocity of zero), what would happen to all that momentum? Would the resulting black hole have a mass greater than
123.
by mkw5053 1y ago
Amazing. Inspires to try out Claude Code + skidl [1] to build a custom circuit for fun. [1] https://github.com/devbisme/skidl
124.
by mkw5053 1y ago
Sounds like a great book. I worked through nand2tetris ages ago and remember enjoying it as well.
125.
by mkw5053 1y ago
Fascinating concept! If microtubules really leverage quantum coherence, could biological systems provide new insights for quantum computing designs, especially at room temperature? Also, how far away are we technologically from being able t
126.
by mkw5053 1y ago
Not at all, accountability works both ways. The elected officials tried to defund the library, but then the community pushed back and the takeover was withdrawn. That's exactly how democratic accountability should work. Libraries need
127.
by mkw5053 1y ago
I understand the concern about local accountability, and you're absolutely right that libraries should be responsive to their communities. But the data here suggests the broader community was actually supporting the library. Most publi
128.
by mkw5053 1y ago
Interesting pattern here: manufacture a crisis (book banning complaints → funding cuts), then propose private "efficiency" as the solution. Meanwhile, this library was founded in 1799, is the second-oldest in Virginia, just won th
129.
by mkw5053 1y ago
I need this for my compact compost freezer [1] [1] https://www.envirofreezely.com/
130.
by mkw5053 1y ago
Prompts (like system instructions) are stored and secured entirely on the backend we've built (and optionally host/manage). Your frontend never holds sensitive prompts or API keys, only the dynamic user inputs are sent to our back
131.
by mkw5053 1y ago
We provide a fully managed, secure, and ready to use backend. You don’t have to develop, deploy, host, scale. It’s essentially “backend-in-a-box” for AI apps.
132.
by mkw5053 1y ago
I’m building an open-source project (with a hosted option) that lets web and mobile devs add LLM-powered features with zero backend code. Current platforms like Vercel still require at least a backend serverless function even for basic LLM
133.
by mkw5053 1y ago
Fair point. I'm no expert in custom GPTs, I wonder what limitations there would be beyond the obvious branding and UI/UX control. Like, how far can someone customize a custom GPT (ha). I imagine any multi-step/agentic flows m
134.
by mkw5053 1y ago
I don't even have a product although I'd love people to work on something open source together. Also, I'm not nearly cool enough to earn a green username.
135.
by mkw5053 1y ago
Probably some sort. In the meantime it doesn't currently exist and I want it for myself. I also feel like having something open source and that allows you to bring your own LLM provider might still be useful.
136.
by mkw5053 1y ago
I like the idea of having a single source of truth RULES.md, however I'm wondering why you used symlinks as opposed to the ability to link/reference other files in cursor rules, CLAUDE.md, etc. I understand that functionality does
137.
by mkw5053 1y ago
This DevOps friction is exactly why I'm building an open-source "Firebase for LLMs." The moment you want to add AI to an app, you're forced to build a backend just to securely proxy API calls—you can't expose LLM AP
138.
by mkw5053 1y ago
Years ago, I worked for Neustar [1], and they were trying to sell .<brand> domains to everyone. Looks like they finally got at least one customer. [1]: https://en.wikipedia.org/wiki/Neustar
139.
by mkw5053 2y ago
Similar to having a remote engineering team that fulfills the (insufficient) requirements but in ways you did not predict (or want).
140.
by mkw5053 2y ago
I think the problem is that it's not actually reciprocal. The numbers in the first column appear to be inaccurate and misleading. For example, The South Korean average tariff rate on imports from the U.S. is approximately 0.79% [1] [1]
141.
by mkw5053 2y ago
I think most Americans actually do realize what’s happening—the real issue is that there’s a significant enough minority who don’t, and that’s precisely enough to get someone like Trump elected.
142.
by mkw5053 2y ago
I’ve been using Cursor extensively on TS projects and noticed a big improvement in speed and code quality by enforcing clear guardrails (strict types, linting, formatting) combined with rigorous TDD. When I hit an error or exception, I expl
143.
by mkw5053 2y ago
I'm interested in how this would work for generative models. It's not obvious how you'd implement causal masking in the frequency domain. And the modReLU activation seems critical but adds implementation complexity. Would lov
144.
by mkw5053 2y ago
This reminds me of the Agent Workflow Memory (AWM) paper [1], which also tries to find optimal decision paths for LLM-based agents but relies on in-context learning, whereas DeepRAG fine-tunes models to decide when to retrieve external know
145.
by mkw5053 2y ago
You could try OpenHands + OpenRouter + Llama ( https://docs.all-hands.dev/modules/usage/llms/openrouter ) I haven't tried it and don't know how it would go.
146.
by mkw5053 2y ago
Also, it assumes little to no advanced math knowledge and can be found for free online :)
147.
by mkw5053 2y ago
A few years ago, I led a study group through A Book of Abstract Algebra by Charles C Pinter. It culminated in Galois Theory and was one of the best books I've ever used in a math study group.
148.
by mkw5053 2y ago
I made a similar open source app a year ago or so https://github.com/mkwatson/chat_any_site
149.
by mkw5053 2y ago
It makes me miss my time spent tinkering with Idris :)
150.
by mkw5053 2y ago
This is going beyond my expertise, but I'll give an attempt at an answer. Take this with a massive grain of salt. It's almost certainly some level of wrong: The trapped light does not take up physical space in the same way that ma

More ›