I Sold Out for $20 a Month and All I Got Was This Perfectly Generated Terraform
29 points by adamcstephens
29 points by adamcstephens
If someone offers me a flying car that runs on the literal blood of third-world orphans, I'm not interested in how many miles it gets per gallon or how comfortable the heated seats are; I'm not accepting it because the existence of this conveyance is obviously unethical.
Your friend doesn't care about the second-order consequences of his slapdash software falling apart or failing, and the associated human costs, because he has decided that dollars in his bank account are the only measurement of his work that matters. This philosophy can be successful, in an internally consistent way, but it is deeply nihilistic and selfish. We should strive to have positive impacts on our communities, our industries, and the next generations in ways that are not measured in dollars.
If you start by arguing against something on the basis of its negative externalities- the costs to society, to the planet, the erosion of social contracts- and then you're tempted into trying it anyway on the basis of personal convenience, efficiency, or career advancement, your ethical objections were not sincere to begin with. The normalization of this selfish behavior you are choosing to broadcast is even worse than the 20 dollars a month you're choosing to shove into Anthropic's money-furnace.
The only way to win here is to live off the land, have no income, have no health issues, etc, etc. You’re eventually going to need to participate (unwillingly) in the system “built by AI.”
I am not particularly happy about where we’re going, but it’s happening whether I get in the van or not. Becoming an actual janitor isn’t even going to fully protect me.
We can offer alternatives, but Capitalism isn’t going to accept 20x slower at 40x the cost. So, we need to become 40x faster … with existing tools? I dunno what the plan for that is…
one of my biggest questions atm : what will happen if the financial incentives of one or more hyperscalers change? e.g. openai folding, anthropic changing business model from "utility" to trying to capture more of the software value chain. I mean, they already say the quiet part very loud that they want to put us all out of a job.
Ed Zitron is very vocal about the financials here and how they can’t possibly work long term. I think what ends up happening, then, is there’s a race to private GPUs / compute to run the “open” models, locally. That’s already happening to some degree, though, the cost is enormous right now given the hyper scalers purchasing power… The models, themselves, are comparable to the last generation frontiers, which were already capable.
If you start to bundle that capability with the additional techniques that are coming out to write working code (mostly around validation efforts), and probably 1000s more techniques currently being developed, I think we’re going to see this stuff have longevity outside of the hyper scalers which fail to sustain it as a product.
What I’d really like to see is an effort to do “Lemon Extract” — basically, advance the state of the art tools for programming and improve our productivity by 10x, 20x, 30x somehow, without the need for the models. Short term, this isn’t going away, and it’s relatively cheap. Take advantage of that to build a world where we don’t need to rely on it to get enormous productivity gains in the future — that is the reason this stuff is so enticing, right?
a flying car that runs on the literal blood of third-world orphans
Please realize that rhetorical flourishes such as this one help nobody but your upvote count.
When the author talked about their friend who doesn't care about code quality and always makes the deadline, I recalled this quote: "Consistent mediocrity, delivered on a large scale, is much more profitable than anything on a small scale, no matter how efficient it might be."
Then he said, “You know what the difference is between you and me? I know I’m a mercenary. You thought you were an artist. We're both guys who type for money.”
This his home as I haven’t really used my art degree with the market demand warm bodies that could write code. I found paradigms & others that treated it more like a “craft”, but the craftsmen aren’t the ones getting the last laugh here… especially if mostly we were building CRUD apps that didn’t need so much detail put into them.
It can be condensed down to “because most things on the internet originally existed to find pornography and/or pirate movies, stealing all content on the internet is actually fine because programmers don’t care about copyright”.
You also can’t have it both ways. OpenAI can’t decide to enforce NDAs and trademarks and then also declare law is meaningless. If I don’t get to launch a webmail service named Gmail+ then Google doesn’t get to steal all the books in human existence.
Except that copyright and trademark are separate laws, serving different legal purposes. And that selective law enforcement is a thing that has existed since law enforcement itself. This all-or-nothing logic does not apply, we see laws being ignored and laws being enforced overzealously every day.
because we all pirated music in 2004, intellectual property is a fiction when it stands in the way of technology. By this logic I shoplifted a Snickers bar when I was 12 so property rights don't exist and I should be allowed to live in your house.
Please, no. That is a totally false equivalency. The snickers bar had to be produced and is missing in someone's inventory after the theft. The music is copied and you don't notice anything, except for a potential lost sale which might have never materialized in the first place.
It's unfortunate that these points make the post a lot more inaccurate and unappealing than it has to be.
Leaving everything else aside, I had it as kind of a cached fact in my head that these things would be bad at Terraform and similar after reading this article (which makes that claim), so it is useful to know that this is not so (at least according to this other article).
Well, this very thread right here is a microcosm of every other thread on this site (and that other one): person A reports "this is the second coming, it cured my cancer, and cleaned my bathroom" and then person B reports they did not have that experience
So, I'm pretty sure the only way you can know for sure if any random tool will do what you expect is to try it, and then, tomorrow morning when tool version++ comes out, try it again to see if it got better or worse, goto 1
I'm pretty sure the only way you can know for sure if any random tool will do what you expect is to try it
This was ultimately already true for any other kind of software. People often get spun out looking at synthetic benchmarks as a proxy for just trying your actual workload, when that proxy often over- or undersells the actual end result you're going to get.
tomorrow morning when tool version++ comes out, try it again to see if it got better or worse
Unfortunately if the fundamental unethical nature of these systems is the problem, it seems unlikely that's going to get fixed in a patch any morning soon.