I feel the same way as you. But was unfortunately not surprised to see the replies you are getting here.

There are a ton of opportunities available right now to make new things. And make them better, more customizable, and more sovereign.

To the replies: be the change you want to see in the world, guys. That may be trite but focusing only on the negative will just make your own life shitty.

Software devs lost their pricing power due to LLMs but not exactly how most people think.

What's missed in understanding is 'how exactly does this functionality work for this specific case?' or 'can we implement this tiny one off feature in some legacy code base'. Both things are why you keep the guy that wrote it around. And you couldn't really replace him. Because digging into what he wrote was hard.

Now, LLMs can do that stuff better than the guy that wrote it.

Software devs were non-fungible. Now they're commodities. When things become commodities, they lose their value.

I'm not sure why I haven't heard people talk about this aspect. It's the biggest effect on jobs.

While this is true to an extent, oftentimes the important context is not in the code but in the head of the writer. The code is just the fence in the Chesterton’s Fence analogy. And that is still non-fungible and will (presumably) forever be.

> There exists in such a case a certain institution or law; let us say, for the sake of simplicity, a fence or gate erected across a road. The more modern type of reformer goes gaily up to it and says, “I don’t see the use of this; let us clear it away.” To which the more intelligent type of reformer will do well to answer: “If you don’t see the use of it, I certainly won’t let you clear it away. Go away and think. Then, when you can come back and tell me that you do see the use of it, I may allow you to destroy it.”

You're correct that it doesn't answer the why.

But it answers the what, how, and allows one-off features.

So the guy that wrote might (or might not) still have the edge with the why. But that's not the moat it used to be.

In big companies, the why is 80% of the work. I could swear actual dev work is less than 20% of a “developer’s” job at a standard large (non-SV/FAANG/tech-first) company. The rest is holding a lot of really weird organization-specific context in your head to make the right decision.

With my own tiny company, I used to answer questions about my code to support. Supporting the support. I remember doing that when working at big companies too.

Now, my support asks claude about the codebase to answer those sorts of questions. He's better than my memory.

Maybe we've had different experiences.

I love it. And no doubt, the smaller your company, the more profound the impact of “agentic” tooling can be.

But I am skeptical that people who have not seen a complex corporate environment understand how different it is.

> I could swear actual dev work is less than 20% of a “developer’s” job at a standard large (non-SV/FAANG/tech-first) company.

In banks its rather 10%, at least in mine is.

I am yet to see this pan out in the enterprise. Enterprises are full of mini kingdoms built by VP+ leaders with the tools they prefer or were sold on. And many of these tools are inherently and sometimes by design are cumbersome, expansive and not onboarding friendly. LLMs haven't breached this domain and this domain empirically is 80% of enterprise software. I am yet to see direct examples of llm agents replacing say 4 engineers out of an existing 8 person team.

Software devs were always fungible up to p99.9

So there are two competing narratives:

1. Mythos uniquely is able to find vulnerabilities that other LLMs cannot practically.

2. All LLMs could already do this but no one tried the way anthropic did.

The truth is one of these. And it comes down whether the comparison is apples to apples. Since we don't know the exact specifics of how either tests were performed, we lack a way of knowing absolutely.

So I guess, like so many things today, we can to pick the truth we find most comfortable personally.

People have found 0days assisted by LLMs for a while, and none of them wrote hype pieces to find an excuse not to release their 10x bigger model in the middle of a GPU shortage.

https://sean.heelan.io/2025/05/22/how-i-used-o3-to-find-cve-...

I really wanted to like anthropic. They seem the most moral, for real.

But at the core of anthropic seems to be the idea that they must protect humans from themselves.

They advocate government regulations of private open model use. They want to centralize the holding of this power and ban those that aren't in the club from use.

They, like most tech companies, seem to lack the idea that individual self-determination is important. Maybe the most important thing.

That is unequivocally true with some things. You don't want people exercising their "self-determination" to own private nukes.

LLMs aren't nukes.

They're more like printing presses or engines. A great potential for production and destruction.

At their invention, I'm sure some people wanted to ensure only their friends got that kind of power too.

I wonder the world we would live in if they got their way.

An LLM that can hack anything is not as harmless as a printing press. Please stop pretending it is.

The skill of shooting down ideas has never been more valuable, actually.

LLM's are an endless source of bad code ideas. Being able to sift through them and find the gems is the exhausting way to be productive.

I agree with the general premise that it is easy to shoot down ideas without thinking. But it's also easy to propose ideas without thinking.

Both are disrespectful if disproportionate to the effort of the other.

The core is not idea generation versus critique. It's the effort spent on each.

“I haven’t been hacked yet, my security is good enough.”

By the time they come for all of your internal data (the Sony hack over a decade ago!), it’s too late.

And does anybody recite the horror stories while making lousy corporate security decisions? Reading the headlines makes it seem like not.

They're trying for the vertical integration monopoly.

The times it works, it works well for the company at great cost to society.

Imagine the world we'd have if comcast got to control your web browsing experience.

If ISPs got started today, they'd sell the open web at API prices that no one can afford. Then sell the ISP's lock-in 'internet' for a low monthly fee.

My question is why people who don't want comcast's internet think other vertical integrated lock-in is fine.

Our markets game only works for the benefit of society if we have fair markets.

VC-backed loss-leader dumping to starve competition model breaks the game.

The stars are probably legitimate.

It's weird that most people in these comments are speculating fraud.

Why aren't companies with real money to gain from stars gaming the system to the same degree? Why do the other metrics - issues and pull requests - match up with its popularity? Why would the bots starring the repo mean that those same bots are not popular? Those bots are controlled by their users.

The project is extremely active because this is what everyone being able to customize their computing looks like. A mess.

But it's a good mess.

Github was the old code sharing model clearly not designed for this. I'm sure a new model for code sharing will come to fix the growing pains.

A ton of people who would have never been able to customize their computing experience are finally able to. And it is magical for them.

This means that those same people will finally value having access to source and use of open protocols.

It was always valuable to us because we had the power to make it matter. It never mattered to them because they did not. Now they do.

The last era of computing was defined by dumbing down computing for the masses. Less information, less customizable, and more metric driven. Control in the hands of the companies.

This new era will look more free/libre, more personal, and less enshitified. Control in the hands of the users.

This is a very positive development.

> Why aren't companies with real money to gain from stars gaming the system to the same degree?

How are you gaining money from stars? Why would Facebook bot stars for the react repo?

I think the implication here is that if we can’t find evidence of or motivation for companies paying to inflate their star counts, then we should cool it on the accusation that magically fraud has appeared in this case.

We also should remember that if this project had zero stars, we would hear crowing from these same people about how true and important that metric was. The idea that “open claw” paid for these stars somehow is mostly just reasoning backwards from the idea that no one would find this project interesting.

I don’t think most people are saying they are paid, I would guess that some people instructed their OpenClaw to leave some stars on the repo.

So the idea is some subset of OpenClaw users got a second GitHub account to vote twice?

Why?

To support their favorite project without having to do anything for it except writing a chat message? I’m assuming that OpenClaw can create its own GitHub account and give stars without a lot of human work.

So about 150 thousand people starred OpenClaw, then asked their bot to sign up for an account to star it again? I'm not trying to be obtuse, I'm just trying to get a sense of what we're talking about. Because if it is 1 person botting 300,000 stars (or 4x75k, etc), that costs real money. There needs to be a motive for that to be believable. If it is 150kx2, then that's a much wider (though still pretty unmotivated) phenomenon that someone would have blabbed about.

There are a bunch of open source projects that I want to see take off; I've never felt the urge to star one twice. I doubt that has to do with it being easier to say "Go star this project on github for me, with your own account" than it is to make a new account on github (which is not hard). I don't think that comes from any great moral fortitude, it's just...IMO hard for me to explain without an actual motive.

What's being alleged in this thread is widespread fraud via botting with no evidence of means or motive. As someone pointed out above, the argument for Facebook buying react stars is WAY stronger...and it is still really flimsy.

These sorts of tools will only be able to positively identify a subset of genAI content. But I suspect that people will use it to 'prove' something is not genAI.

In a sense, the identifier company can be an arbiter of the truth. Powerful.

Training people on a half-solution like this might do more harm than good.

It will just be an arms race if we try to prove "not genAI." Detectors will improve, genAI will improve without marking (opensource and state actors will have unmarked genAI even if we mandate it).

Marking real from lense through digital life is more practical. But then what do we do with all the existing hardware that doesn't mark real and media that preexisited this problem.

I agree. A mechanism to voluntarily attach a certificate metadata about the media record from the device seems like a better idea. That still can be spoofed, though.

In the end, society has always existed on human chains of trust. Community. As long as there are human societies, we need human reputation.

You could take a picture or video with your phone of a screen or projection of an altered media and thereby capture a watermarked "verified" image or video.

None of these schemes for validation of digital media will work. You need a web of trust, repeated trustworthy behavior by an actor demonstrating fidelity.

You need people and institutions you can trust, who have the capability of slogging through the ever more turbulent and murky sea of slop and using correlating evidence and scientific skepticism and all the cognitive tools available to get at reality. Such people and institutions exist. You can also successfully proxy validation of sources by identifying people or groups good at identifying primary sources.

When people and institutions defect, as many legacy media, platforms, talking heads, and others have, you need to ruthlessly cut them out of your information feed. When or if they correct their mistake, just follow tit for tat, and perhaps they can eventually earn back their place in the de-facto web of trust.

Google's stamp of approval means less than nothing to me; it's a countersignal, indicating I need to put even more effort than otherwise to confirm the truthfulness of any claims accompanied by their watermark.

The unix commandline tools being the most efficient way to use an LLM has been a surprise.

I wonder the reason.

Maybe 'do one thing well'? The piping? The fact that the tools have been around so long so there are so many examples in the training data? Simplicity? All of it?

The success of this project depends on the answer.

Even so, I suspect that something like this will be a far too leaky abstraction.

But Vercel must try because they see the writing on the wall.

No one needs expensive cloud platforms anymore.

Shell programming is high density inter-language glue. You simply have more options of implementations to call out to and so less to write.

I can trivially combine a tool written in rust with one written in js/java/C/whatever without writing bindings

> The unix commandline tools being the most efficient way to use an LLM has been a surprise. > I wonder the reason.

Because they are really, really well designed for humans.

Everyone is trying to reinvent the wheel and create "agent interfaces", but there is fundamentally no difference between what makes a text based interface easy for a human to use and what makes it easy for an agent to use.

If you want a better guess: It's because of the man pages for all the tools are likely duplicated across so many media for the LLM training that there's just an efficient pipeline. They go back to the 70s or whatever.

So, mostly re-enforcement along multiple vectors.

I'm not convinced. I don't want to rack servers and diagnose bad RAM like it's still the 90's, so I'm paying someone else for the privilege, especially to get POPs closer to customers than I want to drive or fly to setup, especially in foreign countries where I don't speak the language or know the culture. Fun for vacation but a recipe to waste time and money setting up a local corporate entity and a whole team when I can just pay GCP or AWS and have a server on the other side of the planet from me faster than I can book a plane flight and hotel reservation there.

There's also the maintenance of the server to be considered. Vercel or other PaaS/Lambda/GCP functions/etc serverless means there's just less crap for me to manage, because they're dealing with it, and yeah, they charge money for that service. Being able to tell Claude code, I setup ssh keys and sudo no password for you, go fix my shit; like, that works, but then the hard drive is full so I have to up size the VPS, and if you're stupid/brave, you can give Claude Code MCP access to Chrome so it can click the buttons in Hetzner to upsize for you, but that's time and tokens spent not working on the product so at the end of the day I think Vercel is gonna be fine. AI generating code means there are many many more people trying out making some sort of Internet company, but they'll only discover cheaper options only after paying for Vercel becomes painful.