199 karma · joined January 17, 2025
Don't just trust it. I've done websites as sample projects to work with LLMs to understand them better.... and while sometimes brilliant, they can miss really obvious things. One time it did user login but never actually verified the password set for a user login.
Treat it like an intern/assistant, not a magical thing that does work for you. While sometimes brilliant, they can miss obvious things, especially if you don't hold its hand. Verify, verify, verify.
> Give a man a fish, and you feed him for a day. Teach a man to fish, and you feed him for a lifetime.
Or in the context of AI:
> Give a man code, and you help him for a day. Teach a man to code, and you help him for a lifetime.
The physical demands are real, you will likely have back/neck problems from leaning into engine bays, looking up constantly, contorting your body to get at a bolt or connector.
You are constantly exposed to toxic fumes and chemicals in the air and on your skin. You will be burned from hot exhaust, cut open on dirty rusted metal, bruised from impacts or falling objects, hearing loss from air lines hissing, air tools and hammering on parts. Those are the lucky ones... unlucky ones get eye injuries (even with 'safety' glasses), severe burns or cuts and broken bones from crushing injuries or even death from a variety of sources.
The pay structure promotes cutting corners and speed. You are paid X hours for a task even if it takes you X*2 (oh no!) or X/2 (oh yeah!) hours. You don't get extra time and pay to clean dirt and mud from an area to prevent contamination or just do a good job in general.
Vehicles are generally designed to be easy to assemble, not repair. This is why a mechanic may spend 5 minutes cutting a hole in a plastic fender well to access a bolt and plug the hole afterwards. That bolt is easily accessible when the engine is removed, but dropping an engine (they come out the bottom easier) can take 2 hours. With a shop rate of $100/hr+, you'll want the quick hacky fix to replace that $10 gasket.
Often, there is no workaround like that, so you're racing the clock... hoping something else doesn't break along the way or you get hung up on a rusty bolt. And when something does break, you hope to any god that will answer you that you can get the part in a short amount of time.
$160k is not enough for some of those talented guys... they are like professional athletes in the sense that it's a young man's game and your body is getting used up every day you're in there wrenching.
Lead fuel was during a time when the cylinder heads were mostly cast-iron, and the valve seat was cut directly into the head. Cast iron is an interesting material, it's reasonably durable, but it corrodes/rusts very easily, especially when exposed to moisture. Gasoline and ethanol both have water as combustion byproducts, so when the engine is off and cools, some of that moisture condenses inside the engine.
Running straight ethanol in an engine without corrosion resistant materials causes much more wear over time because it tends to strip the protective/lubricative oil barriers away, causing iron to corrode when the engine isn't running. Modern engines are aluminum heads with valve seat inserts, stainless steel valves, better piston ring materials (high chromium I think? these were cast iron in the past).
Ethanol has a significant detergent/cleaning effect, even when at 5-10% concentration in gasoline. The valve stems also get some of their lubrication from the fuel, and gasoline is basically a thin oil, and provides protection to mechanical components, better yet with additives. Ethanol is also a difficult fuel in a cold start situation and requires good compression and a strong ignition system to kick it off.
I suspect the whole reason to want to keep lead was motivated by the bean counters involved. They saw a cost savings with lead in the fuel. Cheaper materials and no tooling changes. This means more profits.
IMO, the best way to "handle identity across Windows and Linux" is Microsoft's own tools. You can join Windows, Mac, and Linux machines into Entra now. For $8 a month you can get an F3 license for a user. This gets you the MS Office Suite (web only) plus Intune/Endpoint Management for 5 active devices, licensed Windows 11 Enterprise (good for machines without an included windows license), the ability to control Device Policy and Conditional Access Policy. The F1 license ($2.25) might work, but don't quote me on that (read-only office, no mobile apps, no Windows Hello for Business).
Mac and Linux machines aren't as robust as Windows for endpoint management. But the core features you'd want are mostly there. Apple business manager is needed and has to be paired with Entra, but it's not completely terrible. The Microsoft documentation is actually very helpful here.
Seriously, I completely agree with you.
Take away the internet. Except in a research/library scenario. Give them a limited time to complete tasks. This would promote a stronger work ethic, memory/recall and more realistic to time management skills. They need to learn to rely on themselves, not technology. The only effective way is to remove tech from the equation, otherwise the temptation to cheat to compete/complete is too strong.
I will say that as a Windows Administrator, the overall quality of the OS has fallen, and more attention is put into 'features' compared to performance, reliability, etc. I shouldn't have to wait for right click menus. 4gb ram should be a viable web browser and basic Office machine, even in 2025. The bigger and faster machines get, the lazier software companies get.
Generally, Windows is reliable for my Org. The OS is decent; yes it is annoying to have to remove the adware from an office machine. I struggle to understand why 16gb with a good SSD is the bare minimum to make the OS feel like it has a sense of purpose. I have 10-year-old hardware running Linux and mechanical drives that performs really well and only slows down when you're asking it to do heavy lifting.
Pen and paper only? Lock them in a faraday cage to keep RF from going in or out? RF jamming? In-person testing only. No devices, period. Supply hearing aids, glasses and even clothing if you have to.
AI systems can be fooled, people can be bribed, cameras and detectors can miss things. To me it's a risk vs reward equation. If the risk of getting caught is greater than cheating, people will generally not want to cheat. If there is no risk at cheating and only something to gain, cheating will be rampant. If you get caught cheating and there is no jail time, no financial risk, and you can take the test again... why not cheat? Make the barrier to cheat incredibly high and the cost of cheating even higher.
Negatives would include inconsistent KWh/performance between packs; Could you own a pack and charge at home or would you be restricted to renting one? Who is liable when it catches fire or you get in an accident and wreck a pack? A pack swap would probably cost more than just charging. Do you 'buy-in' to a pack separately or is it baked into the cost of the vehicle?
Then you have to get all the auto manufacturers to agree on a pack specification.
We have been solving the history problem with computing for decades now. My limited, imperfect human memory is a feature, not a bug. We get very good at the things we do regularly.
App security issues aside, my opinion of a more useful product would be about how can you make the person more efficient, not by doing something for them, but by informing them of how they can grow.
For example: "I noticed you... write many python command line apps; have you tried the python Fire library?"; you are struggling with some python basics; would you like me to schedule some time to practice Python with some of the libraries you use most?"; you do this task regularly; I think you can automate that like this.."
Hobbies are just jobs you don't get paid for. Focus on acquiring skills and enjoying it. When you stop enjoying it, it becomes mind numbing, tedious, work. I haven't "worked" in that sense, for years.
Using a service would be better for a final part in a better material for the sake of longevity, not just 3d printed either, but cnc'd aluminum for example. Plastics can only go so far, but for short durations in some stressful situations, like intake manifolds, they can give you information. Having a printer in house will let me prototype to verify functionality of some engine parts, and being able to print in resistant materials will let me make final parts of less stressed parts in areas exposed to fuels, oils, etc.
I can do what I need to do in CAD, design my own parts, etc. Other than the above, I'll use it for gifts, stuff around the home, rasPi and ESP32 electronic projects with home assistant, misc. enclosures, etc. I have a broad set of use cases but running production 24/7/365 isn't something I see myself doing unless I stumble upon a niche as I mentioned.
* how much downtime can be tolerated during a deploy or outage? load balancing and multi-region is more $$$.
* if you have a bunch of linux nerds and an efficient app -- a nginx webserver + your app + Postgres DB and Ansible to manage a single VM with Cloudflare in front of it might be a good option. Portainer in the VM is nice if you want to go with containers.
* if you have a bunch of desktop devs, containers and build pipelines with PaaS are a good option. many are resilient and have HTTPs built in.
* the smaller your infra/devops team, the more i would leverage team knowhow and encourage PaaS offerings.
* the smaller your budget, the more creative you need to be (ec2/storage accounts as part of hosting, singular monolithic VM has relatively flat costs, what free stuff do i have on my cloud provider, etc)
I think of a Dyson as more of a status symbol, like the latest Mac Book, a watch collection, or a Porsche. Dyson owners like telling people they have a Dyson. Most people just need things to work well for their use case, be reliable and affordable.
The saying goes... "Fast, reliable or cheap. Pick two."
Just `git clone someproject`, `uv run somescript.py`, then mic drop and walk away.