< Back to the archive

Like what you see? Subscribe here and get it every week in your inbox!

Issue #317 - April 6, 2025

If you are looking for work, check out this month's Who is hiring? and Who wants to be hired? threads.

Here are the top threads of the week, happy reading!

Top comment by enisberk

Finishing up my PhD thesis on low-resource audio classification for ecoacoustics. Our partners deployed 98 recorders in remote Arctic/sub-Arctic regions, collecting a massive (~19.5 years) dataset to monitor wildlife and human noise.

Labeled data is the bottleneck, so my work focuses on getting good results with less data. Key parts:

- Created EDANSA [1], the first public dataset of its kind from these areas, using a improved active learning method (ensemble disagreement) to efficiently find rare sounds.

- Explored other low-resource ML: transfer learning, data valuation (using Shapley values), cross-modal learning (using satellite weather data to train audio models), and testing the reasoning abilities of MLLMs on audio (spoiler: they struggle!).

  Happy to discuss any part!
[1]https://scholar.google.com/citations?user=AH-sLEkAAAAJ&hl=en

Top comment by fancyfredbot

There is more than one way to answer this.

They have made an alternative to the CUDA language with HIP, which can do most of the things the CUDA language can.

You could say that they haven't released supporting libraries like cuDNN, but they are making progress on this with AiTer for example.

You could say that they have fragmented their efforts across too many different paradigms but I don't think this is it because Nvidia also support a lot of different programming models.

I think the reason is that they have not prioritised support for ROCm across all of their products. There are too many different architectures with varying levels of support. This isn't just historical. There is no ROCm support for their latest AI Max 395 APU. There is no nice cross architecture ISA like PTX. The drivers are buggy. It's just all a pain to use. And for that reason "the community" doesn't really want to use it, and so it's a second class citizen.

This is a management and leadership problem. They need to make using their hardware easy. They need to support all of their hardware. They need to fix their driver bugs.

Top comment by nickjj

I am guessing it's a hard thing to do unless you slam dunk something that hits absolute critical mass.

I have a bunch of repos with 500-1000+ stars. I've gotten folks who emailed me entire stories on how a project I created helped them get over being blocked on something, or how it kick started them into getting more involved with programming. I've on many occasions had folks email me asking me why I put some of the things I do up for free on GitHub.

All in all I've made around $17 in almost 10 years through donations (GitHub sponsorships).

But I don't do it for the money, although I'll admit it would be nice to receive income for doing something you'd happily do for free anyways. I create almost all of my projects based on a personal need and the idea of openly sharing what I can is built into how I operate given how much I've learned from others, I feel very strongly about returning the favor when I can with no strings attached.

Top comment by LinuxBender

Instead of learning software engineering I would want to learn:

Chemically induced reprogramming to reverse cellular aging [1] a.k.a. OSKM

My first experiments would be on some really old horses. I could probably buy a 30 year old horse from a neighbor. She is on her last leg. I want to make her younger again and then just let her have many more years of chilling and not having to make babies every year. If can learn this well enough to reverse the age of a dozen horses then my second test subject would be myself. If I get that right then my friends could optionally do the same.

[1] - https://www.aging-us.com/article/204896/text

Top comment by bionhoward

Avoid it! Perplexity rules prohibit commercial use, so it’s effectively a toy for hobby stuff like gardening and recipes. ChatGPT rules prohibit sharing the output with other AI, and they’re training on everything you pass in, so you’re literally paying to get brain raped, they hold all the cards, and that’s as dumb as it sounds.

Anybody coding with OpenAI at this point ought to be publicly mocked for being a sheep. All you’re doing when you code with AI is becoming dependent on something where you’re paying them to train their model to imitate you. It’s not good for anyone but OpenAI.

Better off learning a better (compiled) programming language. If you have to use AI, use Groq or Ollama as you can keep these outputs to train or fine tune your own AI one day.

Why pay OpenAI for the privilege to help them take your job? Why outsource mental activity to a service with a customer noncompete?

Top comment by FloorEgg

You are missing something really really important.

Most startups don't have product market fit. Further, most startups make things no one wants. The product might sound good to you and you will think "oh I can sell this", but you will work your ass off for no results.

They you may go back to the client and say "Hey the market doesn't want your product, they want this instead". Now you are doing free market research for them, and most likely, they won't listen to you anyways.

You may think "okay I need to be good at identifying startups struggling to sell who have products people want" and then you may realize there's a whole industry of VCs trying to do this, that mostly suck at it.

I think you'll find that young companies with products people want who are struggling to sell is extremely rare.

There's a reason the business model your suggesting isn't more prevalent with startups, and it's because sales skills matter far less than market fit.

Top comment by 0xCE0

Here are couple of thoughts:

- Whenever anxiety fills my mind, I like to look deep space images to remind myself that all the human problems here on Earth are just made up. No other animal cares "economy", "startups", "performance reviews", "jiras" etc. What we humans have created here is truly form of "matrix" that forces to play by its rules, but it is illusion.

- If you think what other people think of you, the most probable answer is, they don't think anything of you, or if they think, one such thinking might last ~10 seconds. There are just too many people and too little time to think other people. No one probably cares, at least that is my own experience.

- I have always been the slowest performer in my jobs (and many times got special coaching/discussions because of that), yet it usually have been me who got promotions. It is because faster people usually just churn the same stuff, while my slow work has given me time to think what the work is actually about, and produce something completely new line of thinking and valuables that other people just don't do, because they just keep churching and churning the same.

- Keep up with the most important basics: sleep, healthy food, and physical exercise, to have energy for mental work.

Top comment by solardev

I frequently recommend Wix to freelance clients who just need a basic site. Once they set it up, it basically keeps going for years and years, which is not true of most other stacks, including Wordpress. It's an easy service from a single vendor, so no need to deal with different hosting/CDN/SSL/etc providers. I think it's a wonderful thing for clients with simpler needs.

The benefit for clients is that they can pay you once, for a few hours, to help them set it up (if they even need that)... and then they basically don't need you anymore. I've "lost" several happy clients this way, but I'd rather they just use that service than waste their money on a developer they don't really need. It's very easy to use, reliable, and cheap. And they have a single vendor to go for any sort of support they might need for their website.

In contrast to many of the over-engineered Next.js or Gatsby sites I've seen, Wix is far, FAR easier to maintain and I get pretty much zero complaints about it after initial setup. All the other stacks I've ever made for clients, whether they were in Next, React, Angular, vanilla HTML, Wordpress, Drupal, Joomla, other CMSes... all became a maintenance headache after 2-3 years, and usually obsolete, unusable, and completely rewritten within 4-5. Not so with the Wix sites; they just keep going year after year and the client never worries about it again, logging in to post an occasional update every week or so but otherwise letting it do its thing.

I wouldn't choose to use it for a personal project anything more advanced than a personal blog or a very simple marketing site. But it's fine for what it is, and the web is better off for having services like this for regular people to choose from. Not everything needs a super-heavy JS frontend.

Top comment by igouy

"ENVY/Manager augments this model by providing configuration management and version control facilities. All code is stored in a central database rather than in files associated with a particular image. Developers are continuously connected to this database; therefore changes are immediately visible to all developers.""

https://www.google.com/books/edition/Mastering_ENVY_Develope...

~

pdf 1992 Product Review: Object Technology’s ENVY Developer

http://archive.esug.org/HistoricalDocuments/TheSmalltalkRepo...

Top comment by talkingtab

My unfounded belief is that the environment is changing dramatically, and will continue to change. And not because of AI. The reason is that the current internet business model if depleted. Users are weary of the lack of reasonable stuff on the internet. Personally, I use my browser less and less each day. It has become, in my opinion, pretty much a wasteland.

Given that situation, what can you do? One could just change careers. Or perhaps consider why things are this way. One path would be diagnose the current situation. Beginning with the predominant internet business model - tracking users. The number of obvious problems with this is large. I once read (not sure where) that the customer of the internet is advertisers, not users.

Things will now quickly get worse in this version of the internet. The prey (users) will now be inundated with generated content.

Yeah, yeah, I know this is dark and you don't want to read it.

However, perhaps if we consider what it is possible to do with the internet if it wasn't so polluted there is quite a bright light. Fundamentally the internet provides an enhanced way for people to interact and communicate. There are indeed examples of this happening.

I sometimes think We people are in a situation similar to a prehistoric culture that finds a steel axe. So we go around hacking everything in sight. But of course the thing about the axe is the metallurgy.

Currently it appears to me that we have pretty much hacked everything to death with the internet and created a very nice waste land. Here's looking at you Google!!! So perhaps time to consider the equivalent of the metallurgy.