is this a leverage?

https://openai.com/index/hugging-face-incident-and-the-road-...

mission completed

I’d like to understand how this works.

This account was created only 13 days ago.

It has made only one submission, which is this post, and had only two comments before submitting it.

How does this happen?

I have an HN account that is 10 years old, but when I submit something, it often gets flagged almost immediately. So how exactly does HN’s flagging and ranking system work? Does account age or karma matter at all?

For reference:

user: terranivium created: 13 days ago karma: 18 about: vocalslice.com

flagging is entirely driven by user's flagging a post

Which part of this exactly looks like slop to you?

There are lots of AI-sloppy tells, starting from, let's see:

* the web site's layout (eyebrows, glows, em-dashes, middots, that certain breathless tone in the pricing box) (edit: see the source for https://vocalslice.com too)

* the changelog listing, eg. the fact that you can search by language code as a headline feature

* in the app, things like "Click here to choose an audio file to transcribe – swap it any time" are Claude-like copy; I'd be surprised if you _couldn't_ change the audio file at any time?

* looking at https://vocalslice.com/demo.webp, another claudeism is the glowy status indicator blob, it really likes to add that by default.

If OP gets a business going with this, more power to them and all the best, but it really is something that's buildable pretty easily with transformers.js and models like onnx-community/whisper-small_timestamped. (Speaking of which, I hope OP is hosting the models somewhere themselves and not trusting, say, Hugging Face with the continuity of their business.)

Classic link. Not the same thing though. I mean try telling a competent agent something like

> Let's build a webapp that allows the user to choose an audio file, we slice it into timestamped word regions with transformers.js + onnx-community/whisper-base_timestamped (with language selection). The user can then scroll through the waveform (that shows the found word regions), synchronized with the transcript, and select word regions in the transcript or in the waveform, and fine-tune the selection if they need to, and then export the slice as WAV.

and see what pops out.

So I tried this with GLM 5.3 and it kinda worked. I used AssemblyAI cause I don't have the ability to run local models.

The result is pretty ugly and unintuitive, OP's design is much nicer. And there are desync issues, but it's not too bad for a one-shot. (Kinda crazy that we can even do that these days, and then complain about it instead of being amazed!)

the truly important thing from discussion with users is the actual workflow optimisation that the app provides, but i totally agree some of the copy on the website could use improvement and the comments on this show hn have made it clear that i'm not communicating things as well as i could

You might also want to take HH feedback with a grain of salt, it's a very particular demographic.

Classic example:

https://news.ycombinator.com/item?id=9224

HN*

UX-wise, from what I can glean from the .webp, I'd personally (having worked in this same space) appreciate being able to see the slices I already have without having to switch tabs back and forth, for one...

great suggestion - the transcription tab is already quite limited for real estate. i recommend trying out the free trial to get a feel for the workflow.

bun tried hard to be written in zig.

eventually, they fell back to rust.

there's a lesson here: subset languages are fine for experiments and small projects, but they're rarely the right choice for production software.

...and C still can't be replaced. it's still the language of choice for mission-critical systems and low-level problem solving.

Good point! Regarding C being the language of choice for mission-critical systems, probably you want to avoid memory bugs that might result ruining the mission. I understand many projects have large C codebases or interact with other projects in C, but migrating to Rust when possible (like replacing lead pipes as you find them for plastic pipes) sounds like a smart choice to build mission-critical low level systems.

If with "production software" you mean large pieces of software I'd generally agree.

I personally like Zig but I'd never write a huge 200k loc project with it, I see it more fit for low level stuff like small embedded devices where you must interface with C code.

What's the lesson to be learned from Roc migrating from Rust to Zig?

I think the next movement is heading to multi model orchestration.

https://developer.nvidia.com/blog/train-small-orchestration-...

Yeah, this model actually seems like it could handle the job now with good prompting.

when will WASM integration become widely accepted? we need realtime canvas draw like oldschool desktop ui/ux.

Do you think the new WASI 0.3 stuff will help? I hope so. I'm always super impressed when I see an awesome WASM + WebGPU implementation that doesn't spin up my laptop fan.

For canvas/WASM-style apps, I actually think explicit agent-facing APIs become even more important. If the UI is mostly pixels, an agent has less useful DOM/accessibility structure to infer from. WebMCP gives the app a way to expose semantic actions like “create shape”, “select layer”, “export”, “run simulation”, etc. regardless of whether the UI underneath is React, vanilla DOM, canvas, or WASM.

IMO there's a lot more we can do around hooking into the DOM directly, with more efficient data formats over the wire, before we absolutely need WASM

We'd like to experiment more with this but haven't got their yet. Just some internal prototyping

Tegmark's 2017 piece on AGI and existential risk — still very relevant given recent developments in AI.

waiting for the anouncement: cursor for grok heavy users.

article is too old. who is using gpt-4o today?

That's a valid concern, given the paper makes clear that the effect over the polite/impolite scale seems to be model dependent (it finds the reverse correlation of earlier studies on even older models).