A new way to look at vibe coding to point out that it isn’t coding.
It’s hiring a company. In the year 2012, you could email someone with a request for something you want made, and if you have money they’ll reply back over network the code you need.
AI companies just tightened that loop. You hired a company to write software for you. That doesn’t make you a developer or a programmer.
My team has been trying for years to get apps for internal use and we have waited for IT to either buy or make something. One day I just said, “fuck it” and started building the tools we need. I’m not anywhere near done, but the team has started using them, and it’s increasing productivity.
So someone thanked me for coding the tools, and I said, “I’m not really coding. I’m more acting as a product manager. I gathered the requirements from the users, brought it to the developer team, and act as the go between.” I think that’s the best analogy. And like a product manager, I have to take on faith that the devs are doing what I ask without being stupid.
People say shit like this all the time, and it always irritates me.
“We built a house down south.” No you didn’t, you hired a contractor to build a house for you.
AI has certainly made it worse, though. I keep having conversations with people where “I’m writing a novel” somehow turns into “I’m having ChatGPT write me a novel.”
But it’s also used by actual programmers with strong credentials who seem to actually use them intelligently (e.g Torvalds) which muddies the water a lot for the general public, and allows these people who are basically the programming version of a script kiddie (leveraging powerful tools they barely understand to accomplish things that look impressive on a surface level, and then misrepresenting the results).
Could I ask what model you use locally? I’ve got a bunch of computers and vijia cards and would love to experiment with a local model. I’ve never used the subscription models, but I’d love to play around with something secure and private!
But yeah generally, Qwen 3 / Qwen 3.5 type models, will work reasonably well and reasonably fast, at 4B or 2B. 9B works if you close basically everything other than LM Studio lol, or, this latest DeepSeek v4 Flash variant is considerably more performant than just standard Qwen 3.5 9B.
But yeah the way Bazzite does LM Studio via BoldBrew, it is at least theoretically reasonably containerized and isolated… not even gonna pretend I am competent enough to give any guarantees there, but also, I don’t think anyone is capable of saying anything LLM related is 100% certainly correctly understood.
Oh uh, yeah, you’re gonna wanna be docked/plugged in to a power source, if you want more than …idk, 0.5 to 1.5 hrs of battery life while the LLM is actively working on something.
Thank you so much for your detailed response! My user tag I don’t remember setting for you proves correct again!
I never ended up picking up a steam deck cuz I don’t leave the house, so I’ll be using a desktop with maybe a 1070 or 2070s maybe? I’ve never tried Bazzite but I’ve got a bunch of different distros on various machines at home. I’ll throw Bazzite on my Medicat drive. Thank you so much again!
YMMV with other hardware, I genuinely have no clue rofl!
Maybe start with 2B… try to work up from there?
Maybe Nvidia cards are just… better generally, due to generally better CUDA support for LLMs than ROCm?
I have no idea!
The Deck uses an AMD APU, RDNA 2, shared LPDDR5 RAM… so… yeah i have no idea how to translate that to a more standard PC set up, in terms of performance expectations. Tooooo many variables.
(Unless they mean the new one, in which case I’d say, don’t be a pussy and play the old version, getting soft locked in the water temple builds character)
A new way to look at vibe coding to point out that it isn’t coding.
It’s hiring a company. In the year 2012, you could email someone with a request for something you want made, and if you have money they’ll reply back over network the code you need.
AI companies just tightened that loop. You hired a company to write software for you. That doesn’t make you a developer or a programmer.
I’ve worked with overseas coders since 2012 - Claude is much easier to communicate with, faster and more responsive.
Yeah it’s more like offshoring.
My team has been trying for years to get apps for internal use and we have waited for IT to either buy or make something. One day I just said, “fuck it” and started building the tools we need. I’m not anywhere near done, but the team has started using them, and it’s increasing productivity.
So someone thanked me for coding the tools, and I said, “I’m not really coding. I’m more acting as a product manager. I gathered the requirements from the users, brought it to the developer team, and act as the go between.” I think that’s the best analogy. And like a product manager, I have to take on faith that the devs are doing what I ask without being stupid.
People say shit like this all the time, and it always irritates me.
“We built a house down south.” No you didn’t, you hired a contractor to build a house for you.
AI has certainly made it worse, though. I keep having conversations with people where “I’m writing a novel” somehow turns into “I’m having ChatGPT write me a novel.”
But it’s also used by actual programmers with strong credentials who seem to actually use them intelligently (e.g Torvalds) which muddies the water a lot for the general public, and allows these people who are basically the programming version of a script kiddie (leveraging powerful tools they barely understand to accomplish things that look impressive on a surface level, and then misrepresenting the results).
I have an entirely free LLM I regularly run locally on my SteamDeck. I didnt sign a contract for anything, I just downloaded it.
I could power the Deck running the LLM indefinately with maybe $300 worth of solar panels and a camping grade power station.
Now it ain’t nearly as fast as hooking in to an LLM thats being run on a data center, but that’s tool use, not contracting an outside firm.
We can democratize this technology if we want to, we do not have to rely on environmentally destructive, oligarch enriching cloud computing.
That sounds like a cool setup, what kinda stuff do you use it for?
See my reply to Rai’s comment!
=D
Could I ask what model you use locally? I’ve got a bunch of computers and vijia cards and would love to experiment with a local model. I’ve never used the subscription models, but I’d love to play around with something secure and private!
Bazzite -> LM Studio (installed via BoldBrew)
Uh what… model am i on now lol… hold on…
Qwen 3.5 9B DeepSeek v4 Flash by ‘Jackrong’
But yeah generally, Qwen 3 / Qwen 3.5 type models, will work reasonably well and reasonably fast, at 4B or 2B. 9B works if you close basically everything other than LM Studio lol, or, this latest DeepSeek v4 Flash variant is considerably more performant than just standard Qwen 3.5 9B.
But yeah the way Bazzite does LM Studio via BoldBrew, it is at least theoretically reasonably containerized and isolated… not even gonna pretend I am competent enough to give any guarantees there, but also, I don’t think anyone is capable of saying anything LLM related is 100% certainly correctly understood.
Oh uh, yeah, you’re gonna wanna be docked/plugged in to a power source, if you want more than …idk, 0.5 to 1.5 hrs of battery life while the LLM is actively working on something.
Thank you so much for your detailed response! My user tag I don’t remember setting for you proves correct again!
I never ended up picking up a steam deck cuz I don’t leave the house, so I’ll be using a desktop with maybe a 1070 or 2070s maybe? I’ve never tried Bazzite but I’ve got a bunch of different distros on various machines at home. I’ll throw Bazzite on my Medicat drive. Thank you so much again!
YMMV with other hardware, I genuinely have no clue rofl!
Maybe start with 2B… try to work up from there?
Maybe Nvidia cards are just… better generally, due to generally better CUDA support for LLMs than ROCm?
I have no idea!
The Deck uses an AMD APU, RDNA 2, shared LPDDR5 RAM… so… yeah i have no idea how to translate that to a more standard PC set up, in terms of performance expectations. Tooooo many variables.
Slippity sloppity
It’s the difference between: wow this 14 year old developed a pc port for ocarina of time
Vs.
Wow this 14 year old spend money to develop a pc port for ocarina of time
Has he not heard of an emulator?
You can just play it on PC, no port needed
(Unless they mean the new one, in which case I’d say, don’t be a pussy and play the old version, getting soft locked in the water temple builds character)