f311a

This does not account for anti fingerprinting. Also, are there any new tricks that are not in fingerprintjs?

Prior to LLMs, I've rarely seen such headings when reading blogs, but now they are everywhere.

From the article:

> A kiosk that kept dying

> “It’s my device”

> Reality television

> The grind

> The relief pitcher

This is one of the giveaways for me.

How is the general knowledge of Qwen 3.6? Do you need to explain things outside of algorithms to it? Since the size is so small, I guess you need more explanations to it. General knowledge helps with coding when your don't specify a lot of details and ask for big changes.

Why is it so small, but expensive?

Open Router

Input /M $0.45

Output /M $3.20

Cache read /M $0.05

Throughput 27 tps

It would be a very nice model at 200-300 tps and if it was dirt cheap. What's the limiting factor of optimizing speed and price for inference providers?

Opencode said they are working on matching the old prices using their own inference.

Right now, they give 4100 credits for Luna and 63 000 for Deepseek on their prepaid plan (both are 2x)

It's just model size and heavy RL, sometimes they overfit on specific tasks. RL can get you very far, prior models did not have such a focus on RL for agentic setups.

Look at deepseek, they improved it just by doing a lot of RL and you can see it from how it behaves. You provide very little information about a task, but since they are trained on similar tasks, they come up with a lot of assumptions and details on their own, because they were trained with such an info during RL.

The only problem I have with LLMs in Go is that they also make a lot of concurrency mistakes, in the same way as people. It's easy to fix though, just by asking to double check the code

What’s everyone’s opinion on learning new tech things in this day and age? My opinion swings between positive and depressing vision of the future.

I still learn new stuff, but I’m afraid it won’t have any value in a year or so.

For example, I’m pretty good at optimizing low level stuff, but right now you can just ask LLMs to do so and they are pretty good at it. They will profile the code and suggest reasonable options like 90% of the time.

It takes years to test it thoroughly in real environments. You don't want your data to be silently corrupted.

What's the reason for it? Does not make a lot of sense to keep all the commits elsewhere

Regexes are pretty slow, though.

Unfortunately, simple URL parsing breaks on so many things. There is a reason on why every URL parsing library is at least a few thousand LOCs.

One common way to test it is just to pass ipv6 url: http://[f021:d981:b487:e57d:193e:550e::]/

You can do this using fn button, without moving your hand every time you need to press them.

It's very convenient, when your hand don't move at all and can reach every button.

Why do you need screenshot button for programming? How many times do you press it? You can just use fn button for that.

People are programming on a 36-key keyboards just fine and they don't miss any keys. It's easier to press fn + something than reaching out for screenshot button, which is usually far away. That way, you don't have to move your hands.

There are a lot of subtle bugs. I'm on macOS too, but perhaps my workflow is unusual? Every time I open it, I find small bugs. There are so many features that it's hard to maintain them, I guess.

For example, a lot of disable buttons do not actually disable stuff. They just hide it from the UI.

This does not sound fun. It's better to work on your side projects with manual coding. You will learn more.

Retyping things is inefficient for learning. It's like trying to retype calculus solutions — you don't learn from it. Even if there is an explanation of why the code is written in such a way, you did not come up with it, and you don't know alternative solutions. It is a practice for memorizing, not for building your intuition.

A better option is to write it yourself first and ask LLMs for better options. They are pretty good at it, especially when you need to optimize hot loops.

Of course they are. I'm not even a native speaker.

I read the whole thing and it was my impression.

It does not matter which prompt you use. Even a long article with a small input (of what to write) will include weird details and structure.

People tried to fool Pangram by using custom prompts and instant feedback (rewrite until it passes). After hundreds of rounds, the text was still detected.

Although it's harder to detect such text by humans, there will be clues if you want to write a whole book based on a few pages of input.

It will, there is no doubt.

But look at the first paragraph of the article:

    > The writing was on the wall, and now the writing is  everywhere. Literally. Artificial Intelligence is writing about as well as humans — at least well enough to cause confusion and speculation — and that’s creating an existential crisis among authors, confusion among readers, and legal action within the industry.

He talks about the current state.

That's true, but I can still tell if a big article was mostly written by AI.

The author talks about writing cheap stories, but right now, you can't do this. Cheap stories are pretty bad, because there is very little human effort to polish them. The more effort you put into AI written stories, the better they become, but it's not cheap at all. You have to rewrite and review them.

> People are going to eventually figure out how to polish and refine the usage and outputs so you cant even tell it is AI.

Personally, I'm gonna go offline after this and meet more people. One of the reasons I read comments is because it's human conversation. When everything will be ridden by bots — I'm out.

There is no point in consuming this.

The model is already up on Opencode, but they require a consent to use Chinese datacenters.

This is a well-known author, but how come he thinks AI is better at writing? AI writing makes me vomit.

Also, I think he's too enthusiastic about the future of readers. In my opinion, reading books will be a very rare hobby pretty soon. The decline started more than 10 years ago.

I do Python, Go and Rust. Go works the best, I would say Python is worse than Go.

Rust works perfectly fine, but when I use Rust, I usually pay closer attention to performance, so I have to guide it a bit, to improve cache locality, use simd, avoid unnecessary allocations and so on. Terra has the same issues with Rust. If you always prompt models to achieve the best performance, the code is usually no the one that I want, they optimize unnecessary/cold parts or blindly optimize stuff where compiler takes care of the optimizations already.

See my other comment for more information on how I work with it. In short, keep the changes under 1k lines, context under 120k (ask it to use subagents), drive the architecture yourself.

I've been driving flash model for 90% of my tasks. It's better than pro (for unknown reasons), very cheap and fast.

I try to keep changes under 1000 lines and drive architectural decisions myself, barely notice any difference compared to frontier models. The rest 10% is to spot bugs, security problems and to investigate better architecture, which flash can also do pretty well, I just cross check it.

Faster iterations are way better for me, I hate waiting for 5-10 minutes on small changes. I tried to use recent versions of Kimi and GLM, but they use too much thinking for no reason and are pretty slow because of it. I also often feed a lot of data to it, without worrying about hitting the limits: dependencies (to find bottlenecks in them), logs, performance dumps and so on.

Also, it will never complain about security guards, I've been using it to reverse engineer binaries.

Personal blogs are not a business, usually.

There is PR somewhere in the mosh repo to support block cursor. Just build it manually. Mosh is abandoned, unfortunately.

> no one will visit your website.

Yep, let's suppose an average HN user reads 30-100 articles a month on independent websites (personal domains) posted on HN. Out of 100, he probably subscribes to 0-2 at best (via RSS/e-mail). Now, imagine how many times you need to appear on HN to get your own audience. There is too much friction for a user to subscribe.

Substack is similar to a social platform. People open it to read new articles in a similar way to HN, and they can subscribe to anyone by pressing a single button. There is a recommendation feed as well.

Since Google search for tech content is dead, getting views on independent blogs is extremely hard.

Substack gives a really nice exposure though, unlike medium and other platforms.

It's very hard to gain readers now on independent websites. Google search is dead (very few people now search for tech content), social media platforms penalize you for posting links and so on.

You can get a few thousand subscribers on Substack just by being consistent and interacting with other people. Now try gaining 1000 subscribers to an e-mail newsletter or RSS from your personal blog. People who tried it know that 5,000 views on your blog can convert to a few subscribers, and you need to get those 5000 somehow first.

A lot of people just repost their articles to substack for a reason. It's not forbidden. You can repost them from your personal blog.

> less training competition

I think you meant less research and experiments in big labs because they don't get all the AI money.

Training is expensive, but they also have more than 10 000 of employees combined and they cost a lot of money.