foxfired

I don't use the icon, but at the end of every article I have the "Follow me via RSS Feed" as a direct link to the RSS. As far as tracking the rss traffic, this graph is generated from my server logs. It is literally cat apache logs | grep my feed url | awk daily traffic | sort.

Note this shows me how many RSS readers have accessed my RSS daily. I can't actually track each person, although I have a report I'm working on for the end of the year.

I don't know about being back, but it certainly isn't dead. A few years back, I used to get at least 10k readers a day. That number went down to less than 100 a day at it's lowest, I was writing 10 entries a year at most. Last year, I wrote just 4.

One thing I failed to notice was that RSS was still active. So this year, I started consistently contributing, over 150 so far, and I see RSS picking up right where it left off [0]. A lot of my blog post suck, but I write them as an observation and my current understanding of a subject. Readers have agency to skip what they don't like and only read what they like.

[0]: https://imgur.com/a/RSVtD1W

I remember when ChatGPT 3.5 was going to be AGI, then 4, then 4o, etc. It's kinda like the dooms day predictions, even if they fail it's ok. Because the next one though, oh that's the real doomsday. I, for one, am waiting for a true AI Scotsman [0].

[0]: https://idiallo.com/byte-size/ai-scotsman

I'm not gonna make assumptions on behalf of OP, but if you have domain knowledge, you can quickly tell when a PR is vibe-coded. In a real world scenario, it would be pretty rare for someone to generate this much code in a single PR.

And if they did in fact spend 6 months painstakingly building it, it wouldn't hurt to break it down into multiple PRs. There is just so much room for error reviewing such a giant PR.

The title says it is vibe-coded. By definition, it means they didn't write it.

It's funny just today I published an article with the solution to this problem.

If they don't bother writing the code, why should you bother reading it? Use an LLM to review it, and eventually approve it. Then of course, wait for the customer to complain, and feed the complaint back to the LLM. /s

Large LLM generated PRs are not a solution. They just shift the problem to the next person in the chain.

We used to say "stop copying and pasting from stackoverflow without reading it first". Everything changed, yet everything is the same.

There was a post just a few hours ago on the frontpage asking not to use AI for writing [0]. I copied the content and pasted it on multiple "AI detection" tools. It scored from 0% and to up to 80%. This is not gonna cut it. As someone who used LLMs to "improve" my writing, after a while, no matter the prompt, you will find the exact same patterns. "Here's the kicker" or "here is the most disturbing part" those expressions and many more come up no matter how your engineer the prompt. But here's the kicker, real people also use these expressions, just at a lesser rate.

Detection is not what is going to solve the problem. We need to go back and reevaluate why we are asking students to write in the first place. And how we can still achieve the goal of teaching even when these modern tools are one click away.

[0]: https://news.ycombinator.com/item?id=45722069

Earlier this year, I used AI to help me improve some of my writing on my blog. It just has a better way of phrasing ideas than me. But when I came back to read those same blog posts a couple months later, you know after I've encountered a lot more blog posts that I didn't know were AI generated at the time, I saw the pattern. It sounds like the exact same author, +- some degree of obligatory humor, writing all over the web with the same voice.

I've found a better approach to using AI for writing. First, if I don't bother writing it, why should you bother reading it? LLMs can be great soundboards. Treat them as teachers, not assistants. Your teacher is not gonna write your essay for you, but he will teach you how to write, and spot the parts that need clarification. I will share my process in the coming days, hopefully it will get some traction.

This isn't just enshittification, it's hostile software [0]. When you dominate a space, for example being the only vendor in the stadium, you can impose whatever you want on customers. There are no options, you can tell customers to select between yes or ok, this is the only way to pay for hotdogs now. As an LA resident, the rate at which we implement these broken and invasive services is alarming.

[0]: https://idiallo.com/blog/hostile-not-enshittification

I'm so happy mine is gone. It came in waves for me and funny it enough, it sounded like ocean waves crashing [0]. I used to play ocean sound for my kids to sleep, so I didn't notice that a good part of the sound wasn't coming from the device, but from me. It was unbearable for several weeks before it just disappeared.

[0]:https://idiallo.com/byte-size/nightmare-on-ocean-street

Not sure the one you are referring to, but I found a video for the H200. But the connectors seem reasonable for a desktop machine.

https://www.youtube.com/watch?v=lNumJwHpXIA

At some point, Vimeo was known for having higher quality video formats. But I'm pretty sure you get the same if not better on YouTube today.

A few months back I've downloaded the beta version and was kinda disappointed. I understand that switching to Qt will make future development much easier, but it wasn't addressing my main concern with Audacity.

When I decided to record an audio version of my book, I started with Audacity. But it quickly failed me when I started losing track of what files I was using, and if I deleted a file from the timeline, it was impossible to track what was what. All this because Audacity doesn't have a media manager. Instead, you have to open a folder on the side and manage media yourself. I've written a blog post about it, didn't get much traction. But as of a few days ago, one of the developers commented that a media me is "on our shortlist of new features to include".

I am really looking forward to version 4.0, I hope Audacity gets its Blender moment.

Yes, they show a tiny link behind a collapsed menu that very few people bother clicking. For example, my blog has gone from being prominently taking first spot on Google for some queries. Now with AI overviews, there is a sharp drop in traffic. However, it still showed higher impressions then ever. This means I'm appearing in search, even in AI overview, it's just that very few people click.

As of last week, impressions have also dropped. Maybe people not clicking on my links anymore is the result?

I think there is a problem of incentive here. When we made our websites Search Engine Optimized, the incentive was for google to understand our content, and bring traffic our way. When you make your content optimized for LLM, it only improves their product, and you get nothing in return.

I'm going through this right now where all expectations have been reversed after an acquisition. Ex: I'm not big on metrics, I rather have direct communication with my team to understand issues we are facing and any challenge an individual is struggling with. Looking at metrics hardly tells you the full story. Well, after the acquisition, metrics are in! story points, number of comments on PR, number of PRs, etc.

I don't believe in these methods, but the company as a whole is going to align. I do not pretend I'm excited about it, but I remember that I am in a room with full grown adults. I've addressed the issue, and made sure to frame it with "we are aligning with the rest of the company" as opposed to just saying this is the way forward, deal with it.

Edit: Coincidentally one of my blog posts is on the front page right now and addresses similar issues -> https://news.ycombinator.com/item?id=45359604

Blender went from being the least impressive 3d software when I first downloaded it in ~2003, to disrupting the industry. In the 2010s, you could still hear people would say in forums, "but no company uses blender in the industry". That's not the case anymore. The only limitation with blender now is your own creativity. I worked with several 3D artists, and they wouldn't have had their career without starting with that blender donut tutorial.

A big thanks to Ton. And don't forget that you can support the blender foundation.

This past year, I noticed people reaching out to me whenever my employer is hiring. These seem to be new graduates, and there is a pattern in the way ask for a reference or a job:

  - "Hey, I applied, could you hire me?"
  - "I have a compsci, I'm qualified, I sent a resume"
  - "can I use you as a reference?"

These are people I've never met, yet they are so direct to the point of being rude. But to the best of my knowledge, they are real people. And what it looks like is that I'm contact #258 in a spreadsheet, because they have to cast an extremely wide net to hope for a single response. When I respond, they are lost because they don't even remember which of the job I was a contact for.

I don't envy anyone looking for a job right now.

My problem is that, in the JS ecosystem, every single time you go through a CI/CD pipeline, you redownload everything. We should only download the first time and with the versions that are known to work. When we make a manual update to version, than only that should be downloaded once more.

I just checked one of our repos right now and it has a 981 packages. It's not even realistic to vet the packages or to know which one is compromised. 99% of them are dependencies of dependencies. Where do we even get started?

I've been steadily watching my Google traffic plummet as AI overviews continue summarizing my content. My search impressions have remain the same though.

However, RSS! People have literally emailed me to tell me that they are unsubscribing from my newsletter but "Don't worry, I follow you through RSS." My traffic is larger than ever thanks to RSS. I syndicate the full content, not just a summary, yet people still click and visit my tiny blog.

Google traffic, mainstream social media, all of them don't care about my blog. But from time to time, random people send me an email, and that makes up for it.

One thing to take into account is that PR reviews aren't there for just catching errors in the code. They also ensure that the business logic is correct. For example, you can have code that pass all tests, and look good, but they don't align with the business logic.

We need to reframe it. At this point, what we call "AI" is not a technology, but a subscription company.

A technology is a tool you can adopt in your toolchain to perform at task, even if in this case it's outsourcing cognitive load. For a subscription company, well, as long as the subscription is active, you get to outsource some of the cognitive load. When Anthropic's CEO says that white color jobs will disappear, he means that he is selling Enterprise subscriptions, and that companies will inevitably buy it.

This is what I currently use on my blog and it works particularly well:

  navigator.share({
    title: "hello",
    text: "World",
    url: window.location.href
  });

Plus if the page has images, the browser will automatically use it.

Atlassian added this new AI feature to create audio summaries of Confluence documents. It's really impressive. No one reads documentation. I know it because I'll send someone a documentation, then they'll ask me a question that is answered in the second paragraph, meaning they didn't read it.

So their feature is not just text to speech, but a reading of a summarized version of the articles. But here is the problem. The documentation has no fluff. You don't want a summary, you want the actual details. When you are reading the document that describes how the recovery fee is calculated, you want to know exactly how it is calculated.

I've ran it on multiple documents and it misses key information. An unsuspecting user might take it at face value. So this feature looks impressive, but it misses the entire point of documentation. Which is *preserving the details*.

I used a similar metaphor in the past referencing "They Machine Stops" [0] by E.M. Forster. Yes, in the near future, we will still be able to read code and figure out what it does. I work on legacy code all the time.

But in the long term, when experienced developers actually feel comfortable letting LLMs write large swats of code, or when the machine no longer needs to generate human readable code, then we will start forgetting how it works.

[0]: https://news.ycombinator.com/item?id=43909111

I always thought I was weird for not ever letting the microwave beep when it ends. I always open the door a second before it ends. Sometimes you'll see me running across the house just so I can turn it off before that annoying beep. But then I started working at a co-working space, and I see a sticker on the microwave that says "Be kind, open the door before it beeps".

You sure can: https://developer.mozilla.org/en-US/docs/Web/API/MediaSessio...

Not fiction at all: https://www.bbc.com/news/technology-44561838

That's a pretty good idea. In the morning, have to go through my slack before stand up to remember what the heck I did the day before. You just gave me super powers.