My main desktop I use for full time software development and video editing is from 2014. I've run Windows on it forever. I put Arch on it ~8 months ago.

A full desktop environment I put together based on niri boots at 1.1 GB of memory. Every app I want is installed (tons of CLI tools, various GUI tools, Docker, video editors, browser, etc.) all together comes in at 10 GB of disk total for everything based on what `df -h` says for my root partition.

My machine does have 16 GB of memory so I'm not pressed for memory, but it's nice to know the machine runs on a system designed to not be bloated.

After switching from Windows 10 Pro, it felt like I got a hardware upgrade. Things open faster and generally speaking everything just feels snappier.

As an aside, it was super fun spending a few days setting everything up in a repeatable way which I've now replicated on a number of systems with https://github.com/nickjj/dotfriedrice.

I'm not a fan of either to be honest, PayPal once told me I was wrong with something related to taxes and a bunch of different reps told me what I was saying and reported was impossible. Their tax division specialists also replied by email with big bold red letters outlining how it's not possible and that I'm wrong multiple times. They were contacted through support cases I opened with other reps on the phone since they aren't directly accessible on phone.

Then I said I was canceling my account with them if this wasn't resolved since it would have resulted in me needing to pay $400 to have my taxes amended. Long story short, after being ghosted for 3 months they replied to me saying I was right and they indeed had the impossible problem, then fixed their tax forms a week before taxes were due.

It's really bad that a random person on the internet discovered a huge issue with one of their partners and their instinct was to require ~10 hours of back and forth phone calls, multiple emails, me giving them the likely problem and solution on day 1 only to be lead on and ignored for months until the very last second.

Believable is important because you have to internally 100% without a doubt believe that what you're doing is the right thing to be doing now.

As soon as the "what ifs" starts to creep in for the big picture items or goals, that can destroy everything. I'm not talking about running into technical implementation problems along the way (those can be fun), it's more like "did I pick the right language for this?" level of questions that sit in the back of your mind.

Personally when I find something to work on that I like and will have what I think is a favorable outcome, it's easy to put in 8-10 real 100% laser focused hours into a task every day, even if it spans weeks or months. I'd like to think most people can do this too, the hard part (for me at least) is having these things to work on.

I'm running Arch Linux and /etc/machine-id is present.

There's also an optional /etc/machine-info file that could exist. It's not a part of systemd and won't be created by default. It's more of an informal way to have details about the system in 1 spot. It was more popular when provisioning bare metal servers but still has value in the cloud. You can have key / value pairs on who to contact, where it's located, what type of machine it is, etc..

You can provide AI official sources to look at and dozens of prompts. I've lost track of the number of times where it didn't arrive at the right answer with tons of opportunities to correct itself based on feedback.

Just an endless of sea of "you're absolutely right to have brought that up, I didn't think about that" and other phrases it constantly uses when it fails to provide a solution. Fast forward 20 minutes later and it starts providing the same nonsense it did at the beginning because it forgot what it already said.

The code solutions it provides are so consistently bad but it's not limited to code. I recently tried a YouTube feature where it can generate AI thumbnails from your video. The results were really lackluster. It completely ignored my feedback like "use a real webcam photo of me that you see in the video", to which the AI recreated a completely different looking human that wasn't me. It even swapped out my real glasses with a rendering of glasses I don't have and kept on making incorrect assumptions about everything. After about 10 prompts and 20 minutes of waiting for thumbnails I gave up, it was really poor.

I don't think it's this because the outcome you get from AI isn't controllable. You can give it the best prompts and design suggestions and it'll still give you completely wrong or horribly written code.

If you were a manager and one of your reports kept producing completely wrong and horribly written code that other folks on the team keep bringing up as problematic in PR reviews or privately, that developer would eventually be fired for someone better.

But in the AI case, there is no replacement because all of the LLMs have severe problems.

I'm not sharing this to steal any thunder, but more to relate:

I've been making video programming courses since 2015 on Flask and Docker. Built up a whole business of doing contract work + courses. Released 500+ blog posts and YouTube videos for free to help share what I learned with no strings attached, but had the courses available for sale if anyone wanted to learn more or support my work. Organic search traffic and word of mouth was 99% of my traffic.

I sustained myself for a decade and I was living what I would consider my best life. I never made a ton of money but I got the flexibility to work on what I wanted and it was amazing hearing stories about the courses helping someone change their life for the better. Countless emails of people turning their life around by building things to help them and their families. Hundreds of success stories.

I also noticed an insane drop off around 2023 and by now in mid-2026, it's dead. I literally haven't sold 1 course in 40 days, not even 1. At this point I'm spending thousands a year just to host them out of pocket. Traffic to everything is down over 10x, etc..

I saw the writing on the wall a few years ago and I ended up doing full time work instead but I still post every week.

All that to say, I don't know what to do. I tried everything I could think of. Even did a 2 year weekly podcast in a related field and it yielded close to nothing, despite a number of folks emailing me saying it was their favorite podcast on DevOps / software development.

All I can say is, this really sucks. Our life's work has been gobbled up by trillion dollar companies and sold where we see nothing of it. I had to change my entire life around. I've gone back and forth on just deleting everything or at least putting it all behind a free sign up to stop AI but that doesn't feel right to the actual humans out there.

The real sad thing is internally, there's an endless number of courses I could make. This isn't anywhere near me wanting to stop. I have a ton more things to share. It's just impossible (for me at least) to move forward on it because these things take time and to be able to survive in this world you need money.

I think the issue with localizing it to a specific state is oftentimes contracts or working agreements have jurisdiction locations that align with the company. In other words, it might not be in NY. It could be anywhere, including different countries.

Yes, that was covered in the other thread. Presenting a backup as evidence has a lot less strength than the original.

If the other person doesn't have the email (they deleted it) and you don't have the original because Unattach deleted it to replace it with a different copy then we're only left with an offline backup on 1 person's machine. Who's to say that offline backup wasn't tampered with (modifying headers, etc.)?

THE GOOD:

Their FAQ says it uses OAuth to connect to your Google account and the emails never leave your browser.

It also costs 83 cents for 1 month so you can go nuts. Alternatively you can pay 1 cent per email if you have a handful.

I see another HN thread about it here from a few years ago: https://news.ycombinator.com/item?id=32462878

THE BAD:

I don't know if it can be used with total confidence because that HN thread makes it sound like they delete the original email and create a new email with the same meta data but some of the comments indicate that's not quite the same as the original. If you had to present this email as evidence, you could run into friction since the original no longer exists. Technically it looks like you can back the original up but as someone mentioned in a comment that seems like it would add a lot of friction in a legal case.

I have no intent on ever needing to use old emails with large attachments in a legal case but knowing this could be a problem makes me hesitant. Although on the bright side, it would be fine to use for anything you 100% know won't ever be used legally.

For example I'm always 1-2 GB away from my Google account being full. I've pruned Google Drive to the absolute bare minimum.

I've had my Google account for a really long time. There's tens of thousands of emails since day 1. However, there's many emails that have attachments.

For example my friends or someone might have sent me a bunch of images and there's a very long email thread going on with them. I want to delete the 300 MB of photos without deleting the email thread. I don't think Google has a way to do this. I'd easily be able to free up multiple gigs of space if this were possible.

I've already bit the bullet and deleted the biggest offenders but I have a ton of emails with 1-2 attachments (pdfs, zip files, some images, etc.) that might "only" be 15 MB but I definitely don't want to delete the email since it has a record of something. Not just the attachment but the corresponding email chain.

The drive might be too old to report back TBW. It's a Crucial 256 GB SATA 3.1 SSD.

It has LBAs written which is 138292009426. The sector size is 512 bytes.

I think you can calculate TBW from that?

(138292009426 * 512) / 1000000000000 = ~70.8 TB written?

Current machine has been running since 2014. SMART stats on the SSD shows 100,586 powered on hours which is 4,191 days.

I remember having around 280 days of uptime on Windows 7 when it went end of life. Having a UPS helps a lot to protect against short power outages or blips.

Nowadays I run Arch Linux, it's been 12 days since a reboot. Not trying to break records, I reboot to apply kernel updates when it's convenient. Since I use tmux and have terminal heavy workflows it takes 1 command and a few seconds to resurrect all of my sessions to get back to where I was at before.

The goal is after installing the official ISO, you can get a developer focused desktop environment up and running in ~10 minutes with 1 command. It's quite configurable so you can change around anything you want ahead of time or after you install it. I use it with vanilla Arch but I know of quite a few people using it successfully with CachyOS.

For CachyOS, the only manual adjustment you have to make before running the repo's install script is uninstalling `jack` since it conflicts with `pipewire-jack`. I could easily roll this into the script but so far no one has complained loud enough to automate this. I also wonder if CachyOS will eventually drop that AUR dependency in its official ISO.

If it matters to you, 99.9% of the code is hand coded. The only time I use AI is when I'm 100% stone walled on something and it's a last resort move. Even then it's copy / pasting small snippets into web based AIs where I fully review and refactor its output so you could say the 0.1% is still human vetted.

Moving away from Windows last year was so worth it. I've been wanting to switch since 2017 but always ran into hardware issues. The same hardware I had back then works beautifully with Arch today.

I've been fortunate to have a very limited amount of on-call events. At one place for 3.5 years there was 1 event. In another place there's been 2 in the last 9 months but on the bright side these events are taken seriously in the sense that dev time is immediately prioritized to hopefully prevent them from happening in the future.

All code being written gets reviewed by someone and there's an expectation tests are included. Of course that doesn't prevent all bugs, but there's an attempt at quality control by the teams producing the code.

I think part of this role (SRE / platform / DevOps / whatever you want to classify this as) is technical implementation but also coming up with systems and workflows to reduce downtime and risk when performing deployments. Not all management is open to change but IMO it should be brought up and taken seriously. There are companies out there who care about both providing value to customers while also keeping team morale high.

I do SRE / Platform type of work where I'm technically on-call 24/7/365 but as a salaried worker I don't receive over time or anything like that. If an on-call event happens where I end up putting in 2 hours on a Saturday or Thursday night, I'd use my discretion to leave early or start late another day.

In the roles where on-call was an expectation, it was focused to critical downtime events, not to answer a Slack message from someone working in a different time zone or non-standard schedule. I don't even have work Slack or email on my personal phone. If PagerDuty goes off from a critical alert I get called, that's the only way I get contacted outside of normal hours.