barrell

What is this about “the agrolobby”? I haven’t eaten processed foods in decades. I started when I was a kid because I didn’t like eating things where I didn’t know all the ingredients. Not from marketing or lobbying or trends. I just stopped eating processed foods, felt better, and now if I eat any processed foods I get ill, so I don’t eat them.

I don’t know anyone who doesn’t eat processed foods because of marketing or advertising or lobbying they’ve been exposed to. There is a solid rational to eat whole foods, and anyone I know who does not eat processed foods does it because they’d rather eat whole foods.

There is a ton of research to suggest processed foods are safe to eat. One could also make the argument these are all funded by their own lobbying groups. The truth of the matter is nutrition is complicated, there is likely more than one answer, and we definitely do not know them.

Not everyone who disagreed has been swindled by some corporation.

HCl is toxic when ingested btw. The fact it is in your body does not mean it’s safe to consume.

There are a lot of people who _thrive_ off of a 100% beef diet, I don’t think there is anyone who could _survive_ off of 100% beyond meat burgers. I don’t think you can say they are way healthier than beef by any stretch of the imagination.

And to extract pure protein from a pea is exactly what I would consider ultra-processed. The checmicals used to separate the protein from the pea are included in the final product. At its purest, you’re at least drenching it in HCl. At its worst, it’s being soaked with who knows what.

Sure maybe it’s cleaned well enough to “not matter” but I think it’s perfectly reasonable to find that a concern and not want to consume it.

And that’s just pea protein, I don’t even want to know the aggregate of all the ingredients and manufacturing process of the “patties”.

Thank you. Bean burgers are delicious. I don’t eat them as part of my normal diet, but have no qualms with them and could always share a meal with my vegan friends.

Nowadays it’s all fake meat products which I would never put in my body, and there’s this weird social pressure where I’m being silly by “refusing to eat vegan foods”.

Fruits and vegetables and legumes are delicious, I will eat all of them.

Bring back bean burgers!

I haven’t been eating processed foods for several decades now. Just because it’s trendy at the moment doesn’t make it wrong, nor does it make those who abstrain game players.

I would say veganism is more trendy at the moment. That doesnt discredit anything about the vegan diet.

I wrapped my headband in some macrame yarn and it resolved all of the discomfort. I can’t say it’s exactly stylish, but I do get a lot of comments on it.

I only really wear them at home or when traveling though, so they’re not a fashion accessory for me.

Still wish they would have improved the headband, but if anyone else is struggling with discomfort, I’d recommend wrapping it in yarn XD

This is such a weird comment.

It’s ultra processed food devoid of micronutrients with low quality protein and poor bioavailability.

Health conscious folks would definitely not choose this. In fact, it’s all the things you try to avoid as soon as you start being health conscious. Folks who want to believe they are being health conscious may be convinced via marketing to buy it, but anyone seriously invested in their nutrition would steer very clear of these.

While I’d love there to be a European frontier model, I do very much enjoy mistral. For the price and speed it outperforms any other model for my use cases (language learning related formatting, non-code non-research).

Does this not count as evidence? I would agree that it sounds a little shaky, but I would not say there is no evidence.

https://www.wheresyoured.at/oai_docs/

Privacy is definitely not a cern for the layman, but it is for lots of people, especially pro users. I also haven’t made a google search in years.

I also haven’t seen any improvements in the frontier models in years, and I’m anxiously awaiting local models to catch up.

Similar use case as well. I use erl ports to spawn a python process as well. Error handling is a mess, but using python as a short scripting language and elixir for all the database/application/architecture has been very ideal

Two counters:

1. I'm not sure China caught up so quickly due to any lack of nationalism.

2. There's an allure to working with an EU business because it's in the EU because they're less likely to jerk you around. You have no idea how many times I get told their in nothing they can do, then have to drop the 'I live in the EU and this is illegal' card, and magically the problem is resolved by the next email.

I feel like my experience with Show HN has been pretty confusing.

Two years ago, I started work on https://phrasing.app. Extreme MVP, and a few people using the buggiest software I’ve ever seen. The Show HN took off and I collected thousands of emails.

I then spent two years grinding. I put in over 10k hours into the project. I ended up turning that crappy mvp into a product I’m genuinely proud of. I know have actual users I don’t know who pay me money and LOVE the product. I myself am using it to learn several languages, most of which I was unable to learn before due to a lack of resources.

I then posted multiple times in Show HN. Crickets.

Yet in the past few months, I’ve seen multiple vibe-coded fraudulent Show HN posts in the language learning space take off. They were eventually flagged but still, it’s just weird to see low effort projects get such massive interest.

I’m sure it’s a skill issue, but my experience is kind of painting show hn in the opposite light you do.

Hope this didn’t come across jaded, it’s just been on my mind (and confusing me) for a while now and felt quite relevant to share here.

I wasn’t claiming fluency in these languages, just making a point that comprehension is normally very over-exaggerated, and that “full comprehension” is a long way off from just average “comprehension”, and in most cases not needed to converse/listen/read. A big part of fluency is being able to deal with a certain level of ambiguity

I’ve got some plans around that :) currently there’s a really naive implementation of what I call “Language islands”. They’re collections of expressions/sentences. The idea here is to be able to share them, discuss them, correct them, and end up with more curated lists of learning materials.

So not post based like lang8, but more granular. I do expect people to write little entries/anecdotes piece by piece though and share them. There’s a pretty thriving telegram community so I’m hoping that solves the community aspect and gives people a place to exchange. (There’s also a subreddit but it’s not yet very active)

Finishing them is not my next task, but the one after. So soon!

Not yet! I speak some Italian and want to learn Arabic. I found out there was a semitic language with Italian vocabulary in the latin script and that's what started my obsession.

Actually to be completely honest, I learned about Maltese, checked my app, and saw it supported Maltese. That night I was learning a language with an app I'd previously built without even knowing the language existed. That's what started my obsession with it.

Then turns out it's an island in an awesome location with an even cooler history. And it uses the letter Ħ

What made you start learning Maltese?

I would say "full comprehension" would mean you don't need words and phrases explained to you on a daily basis.

And to each their own. Fluency is a bad metric because it means something different to everyone. If you live in a language, work in a language, and have friends in a language, most people would consider that fluent. I've met many, many people who qualify with a much lower comprehension level than 90%.

Also, speaking from experience, I'll often "comprehend a sentence 100% in another language". Then I'll really listen to it again and realize I'm not really sure about half of the words. I have a vague idea of most of them and in context my brain get's it and self-reports full comprehension.

I think "full comprehension" is a substantially higher bar than "fluency".

My app works in a bring your own content kind of way. Most users AI generate their expressions, which gets you pretty far, especially in the beginning. But for languages I speak, I curate my expressions with things I read or hear or get corrected on.

I've also tried to use a more sentence-mining approach for Japanese, given the cultural differences, but tbh I haven't found much benefit at my level (A1/JLPT5).

I'm pretty sure it's impossible to avoid awkward situations in other languages, 10x so with culturally foreign languages. Best I can do is help you learn from them :)

Please get in touch if you have any questions! The setup process is hit and miss, but there's a telegram community for members with many happy customers. I'm there all the time and we're all happy to help!

Plus it would be awesome to get more Maltese learners. I have no one to gush about this language with :D

I built an app that offers 90-ish languages, and it is not vibe coded. I mean it's taken me over 10,000 hours, but it is possible :)

Not really. Fluency is probably closer to 70-95% comprehension, combined with an ability to assume the rest. I assume the comment is talking about native level comprehension, which is still only like 99.99%

Source: native English speaker in Europe. I have to explain/reword several words/expressions per day to people who would be by all means considered fluent.

(all numbers in this comment were estimated based on experience)

Not OP, but I also coincidentally built a language learning app to learn Turkish. I don't think you can create courses without experience in the language, but it is possible to build analytical tools to make language learning possible. Tech + linguistics can take you pretty far.

This isn't to hijack the thread, but wanted to comment because honestly one of the coolest feelings in my life has been learning a language I don't know from an app I built.

Għadni kemm bdejt nitgħallem il-Malti! I'm using my own app, and I don't want to link it here given it's someone else's launch post, but you can find many links in my profile. But if you're a self-directed type of learner, I've been making sure it works well with Maltese :)

Thanks for the kind words! The promised walk through will maybe come hopefully this weekend, give or take a few days.

Can I ask if the new page explained it better? The “demo/how it works” section is still missing the graphics, which I think is pretty critical. But I’m hoping the new page is a bit improvement!

One of the pieces of feedback I get most often from users is “oh this is actually really simple”. But explaining it simply has always eluded me

Yeah every expression costs money to extract and it's fully bootstrapped. The trial is I give you 10 expressions at cost

Visually I’m working on a new landing page for phrasing. It’s almost done, just need to record a few videos: https://phrasing.app/next

Behind the scenes I’m rebuilding the sync engine to properly support offline mode. Trying to get to instant opens for the app (and of course work offline). It’s probably my 5th sync engine. It’s been really fun to see how much easier, faster, better, etc each new iteration is.

(And the project at large is https://phrasing.app - a language learning app for polyglots. It’s like anki but designed to be enjoyed)

> It’s very plausible (and increasingly likely) that OpenAI/Anthropic are profitable on a per-token marginal basis

Can you provide some numbers/sources please? Any reporting I’ve seen shows that frontier labs are spending ~2x on inference than they are making.

Also making the same query on a smaller provider (aka mistral) will cost the same amount as on a larger provider (aka gpt-5-mini) despite the query taking 10-100x longer on OpenAI.

I can only imagine that is OpenAI subsidizing the spend. GPUs cost by the second for inference. Either that or OpenAI hasn’t figured out how to scale but I find that much less likely

Very happy with all the mistral work. I feel like I'm always one release behind theirs. Last time they released Mistral 3 I commented saying how excited I was to try it out [1]

Well, I'm happy to report I integrated the new Mistral 3 and have been truly astounded by the results. I still am not a big fan of the model wrt factual information - it seems to be especially confident and especially wrong if left to it's own devices - but with http://phrasing.app I do most of the data aggregation myself and just use an LLM to format it. Mistral 3 was a drop-in replacement for 3x the quality (it was already very very good), 0% error rate for my use case (I had an issue for it occasionally going off the rails that was entirely solved), and sticks to my formatting guidelines perfectly (which even gpt-5-pro failed on). Plus it was somehow even cheaper.

I'm using Scribe v2 at the moment for TTS, but I'm very excited now to try integrating Voxtral Transcribe. The language support is a little lacking for my use cases, but I can always fall back to Scribe and amatorize the cost across languages. I actually was due to work on the transcription of phrasing very soon so I guess look forward to my (hopefully) glowing review on their next hn launch! XD

[1] https://news.ycombinator.com/item?id=46121889#46122612

I’m building a language learning app [https://phrasing.app] and this is really good advice. I’ve not had any interest in SST for the application, and have no plans to integrate it. In my experience, I’ve never seen them be truly beneficial in the language learning process.

What has been extremely beneficial has been having the text and audio forced aligned and highlighted, kareoke-style, every time I hear the audio. It has improved my phoneme recognition remarkably well with remarkably little content. Several users also report the same thing - that even native speech feels a lot more like separate words than just a slew of sounds. I attribute this in large part just due to this kareoke style audio. It works better for phonetic scripts, so I would recommend using this with pinyin/jyutping/furigana for character based languages.

For production, when I was at Regina Coeli (world-class language institute) their main thing was just 1. you hear a short passage in Dutch, 10-40 words 2. you record yourself reading the same passage and 3. you play back the two audio tracks on top of one another and listen for the difference. Optional step 4. Re-record and replay until it’s close enough.

There was no grading, no teacher checking recordings, no right or wrong; just hundreds of random sentences and a simple app to layer them. You needed to learn to hear the differences yourself and experiment until you no longer could. (fwiw this is not present in phrasing, I just found it relevant. One day soon I hope to add it!)

I recently started learning a tonal language, and so far have not struggled too much wrt tones when everything is slow. There was an original strangeness and refusal for my vocal cords to want to work that way, but probably only for the first month or so.

At least, this is the case for slow text. Once the text is sped up it’s amazing how my brain just stops processing that information. Both listening and speaking.

I’m sure this will come with practice and time but for now I find it fascinating

I mean it’s closer, but I don’t think it’s right to equate commissioning an artist with paying a multi-billion dollar corporation to steal from artists.

These tools are just lazy shortcuts. And that’s fine, there’s no problem with taking the lazy way. I’m never going to put in the time to learn to draw, so it’s cool there’s an option for me.

I just take ire with pretending it’s something grand and refined, or spitting in the face of the ones who are willing to put in the work