It's a fair point. It's hard to know where we are in the S curve. Considering how much better it's gotten in the past 12 months I don't get the sense that we're decelerating towards a plateau of capability, but it's certainly possible
Beautiful and witty prose to say "vibe coding sucks". He's not at all wrong about the state of AI coding in May of 2025. The 3 hours I just burned trying to get it to correct output bugs in a marimo notebook (which I started learning this week) is demonstrable evidence.
But it completely ignores the fact that AI generated code is getting better on a ~weekly basis. The author acknowledges that it is useful in some contexts for some uses, but doesn't acknowledge that the utility is constantly growing. We certainly could plateau sometime soon leaving us in the reckless intern zone, but I wouldn't bet on it.
This seems correct, and I suffer from the same issue as the author, but I'm curious if it's as simple as "you don't believe in the projects". Does Pieter Levels believe in all of his projects so much that they burn a hole in his brain, or does he just think "this seems like a good idea, I should do it and see if it works" and that's enough to get him to the finish line. I assume the latter, but it's just a guess.
I'm starting at Pacific Northwest National Lab on Monday in the data science and analytics group. I am coming from a 100% corporate background, and I am expecting a non-trivial culture shock to work overwhelmingly with academics and government folks. The reason I joined is I'm highly motivated to work on climate and energy projects, and it seems like an incredible place to do it.
We have to do everything. There's no either/or... we HAVE to reduce consumption, we HAVE to find lower-carbon alternatives for everything we continue to consume, and we HAVE to create technologies to remove carbon from the atmosphere.
It's not gambling or a delusion to say we must create new technologies to have a prayer of solving the problem. And of course there's no "magic pill" - we need to innovate in a thousand different ways, simultaneously.
The scale of the problem is hard to truly wrap ones head around, but it's crystal clear that voluntary sacrifice and curtailing consumption as the primary solution is a non-starter. That's not how humans work.
Also, what about developing countries? Are we going to tell them they have to consume less too? Seems unreasonable. So we're going to ask the developed world to consume less, and some of us might reduce a little bit... nowhere near the scale of emissions reductions that are necessary. Seems more like a rounding error than an actual impact.
I did a weekly "don't eat anything on Sunday and don't eat until Monday lunch" for about a year. Sunday afternoon/evening I'd be kinda groggy and grumpy. I'd wake up on Monday morning completely locked in - it felt like a high dose of adderol without ANY of the side effects. I'd get 2 days worth of work done in a morning. I'd always be bummed Monday afternoon because as soon as I ate the focus went away.
It's not about what's easier, and it's not an either/or proposal. Atmospheric CO2 is currently around 419ppm. Even if we cut our CO2 emissions to zero tomorrow, we still have 419ppm. The temperature might stop going up, but it's not going to go back down unless we get greenhouse gasses out of the atmosphere.
Most importantly, it's not certain that runaway warming can't happen even at our current temperature. Methane is being released from permafrost at our current levels of warming and is 20x as potent a greenhouse gas as CO2. Also, our natural carbon sinks (the ocean, macroalgae, forests, soil) are in decline whether or not we stop emitting CO2 - so CO2 may keep going up anyway as we lose biomass.
And let's be honest with ourselves here, we're not cutting emissions to 0 tomorrow. Or next year. Or by 2030. Or by 2050 most likely.
We have to reduce emissions, at a level that seems inconceivable. We ALSO have to pull CO2 back out, again, at a level that seems inconceivable. The ability to scale up CO2 removal to a planetary scale requires that we accelerate development right now.
I held my nose and jumped in feet first. Actually, I spent 3 months backpacking around Europe first, then went home and just didn't get a new job. It took 6 months to land a client, then 2 to land another, then 1 to land a third and then I was off to the races.
I would not do that approach again.
You've already got some consulting revenue, that's a great start. Do you have a solid 6 months of living expenses saved up? If not, do that. Do you have a portfolio of different projects you've done and a good way to communicate them? If not, do that.
In the small amount of free time you currently have, I'd look very selectively for a large contract. Once you land it, put in your two weeks and make the jump.
Once you're on your own, never stop prospecting. Even when you have too much to do to take on a new client, you have to keep the pipeline full - it's the only way to make it work because you'll invariably have some contracts end abruptly. I was real bad at that part of it, and that's why I stopped after ~4 years.
I did this in 2013. I not only quit my job, but my life. I packed up my entire apartment in a uhaul and dropped it off at goodwill. I packed a backpack and bounced around a handful of pacific countries for a few months (Costa Rica, Nicaragua, Indonesia, Thailand, Japan). Moved across the country to San Diego after I got back and started from scratch. Then after a year of working I took another 3 months and backpacked around Europe.
It's by far the best decision I've ever made. I had no objective and was completely fulfilled. My subsequent attempts at starting a location-independent business so that I could continue traveling indefinitely failed, or I might still be out there.
How would he look me up? He didn't know my name. Can lyft drivers see the full name of the people who they pick up, or just the first name? If they can see the full name, then MAYBE he looked up my friend, then looked at his friends list and found me. I find that extremely unlikely. I've also seen a lot of very, very strange "you may know" suggestions for people that I've only met in passing, and again without full names.
Oh, and also - this morning FB suggested I may know the Lyft driver from last night. I wasn't even the one who ordered the Lyft, I just rode in it. I don't have the FB app installed on my phone. WTF?
Recently my girlfriend discovered the depth of the program FB started to listen to the microphone, even if the app isn't on. She never installed the FB app on her new phone. Multiple times in the last month, some VERY specific topic of conversation IRL has shown up as an ad in her FB stream. Most recently, a coworker told her about a "Custom Martin" guitar he had just bought. Later that day there was an ad for a custom Martin guitar. She's never searched for that, liked anything related, there's absolutely no other reasonable explanation, and facebook has made it clear they were going to start listening. They were serious. We later discovered that new Samsung phones have FB installed as a "system app" that can't be uninstalled, and that app has permission to turn on the microphone whenever, without permission. Her phone listens to everything she says all day. Anyone with the FB app (whether they choose to install it or not in the case of Samsungs) is being listened to at all times. To me, this is worse than every single thing on RMS's list. It's kinda terrifying.
I worked for Hospira last year on the Plum A+ and 360 infusion pumps but not the PCA. I'm a little surprised such a blatant security hole wasn't caught, considering the magnitude of the regulatory environment we worked under. I'd never worked in that kind of environment before, and I left shortly after a successfully defended audit of our software development and tracking process and systems. (Although how successful is the development and tracking process if a year later this "bug" comes out?) My guess is that because we were beaten over the head day in and day out focusing on "if this software delivers the wrong medication the patient is probably going to die", and "if you make a mistake in the development process and in change tracking we can lose the ability to make these devices" that the idea of defending against malicious intent was de-emphasized/overlooked.