Andy Yen’s values are incompatible with Proton’s mission. He advocates for poor security practices, supports fascists, and he verbally abuses his subordinates. As such, I can no longer in good conscience continue to financially support a company with such terrible leadership.
Unfortunately for me, my subscription just renewed last month on a multi year plan, but it does give me time to browse for and experiment with alternatives.
If you’re able, I encourage you to do the same and cite Andy Yen as the driving factor for your decision. We can’t let these tech bro fascists destroy what little semblance of privacy we have left.
Yeah, and I knew a carpenter with 40 years of experience too. He was a shit carpenter, but he always got work because he’d been a carpenter for so long.
I may only have 10 years of experience coding but I can still see how shitty the output is, and even listed an example. No one should be driving a car with the engine strapped to the hood, even if they’re good at driving.
If it’s some basic program like a simple video game mod that changes a value or two, sure, let the janky ass code do it.
Cryptographic high security stuff though? There’s no way in hell that thing isn’t going to introduce a security risk.
That said, it is at least useful for finding bugs and security flaws. But that’s because an LLM is a statistics machine, and it’s therefore better at reading data than it is at creating it.
I always like to say to people with a lot of experience that are still bad: you don’t have 40 years of experience, you have 40 times 1 year experience.
Then you should be putting more time into your tools. Nobody uses a blank of wood to hammer something, and then decides 'wow, hammers aren’t good tools," you’re just using a shit hammer.
And, to be clear, I work in a full-stack environment and regularly generate HTML and CSS without the issue you mentioned… you can’t just pull a frontier model without any skills/rules/mcps and expect it to just do the work. You need design docs and thorough rules for your environment. An LLM is a tool, not a developer.
That’s totally fair. They can only get ~75% of the way there on crypto stuff rn. Amazing at the other side, though (pen testing)
The example I cited was a frontier model set up with specific rules etc.
That’s exactly why I realized how shit LLMs are. This was Cursor set up with its VS Code fork and designed specifically for HTML and CSS. It still goes crazy on the amount of div classes it makes compared to what a human would do. The website it coded is inefficient compared to what a human would do.
An LLM is a tool but it’s no hammer - its a lump of iron pretending to be a hammer. And you’ll get just as messy results.