How do you deal with DDoS attacks on your infra / sites? Especially ones from residential proxies or vpn networks? These are the hardest to defend against as they often are relentless and no amount of proof-of-work, captchas or otherwise seems to stop them from coming. I don't think it's something you can easily block at any firewall level, the sheer volume of IP(s) and Subnets is too much.
Why does it have to be called X? We already have x.com -- In fact what is it with this infactuation with calling everything "somethingX" or "X something"
> But something also got lost along the way, and it's that obligation to read slowly, to understand every line because there was no shortcut available
Not really sure what got "lost" here. But fundamentally nothing has really changed much, except for deeper abstractions, more powerful tools and the cost of "generating" (or producing) code approaching near zero (if you discount the cost of subscriptions to Claude/Codex, etc and the underlying costs of running those st scale).
You still have to understand, and read every line. Although now we just get tools to do so for us. But the understanding is still a requirement.
What happens one day when there is a significant outage of Claude or Codex and reliance on these tools as SaaS services is so great that it start to impact productivity and work? Will we just pack up our tools and go home?
> Therefore, there is no ... or «global stance on AI usage».
The fact that we're not starting to see alternative "code hosting" services because of the divide in our industry of the use of AI assisted or generated code is quite worrisome to me. Why? Because we shouldn't be in this socio-political situation in the first place, we should be getting along, with common goals, the love of creating, coding, whatever. Why do we now have to be so divided over new technology when we are the creators (or some of us anyway) of said technology?
Sure competition is good. Codeberg, Upd.dev, and many others. But can we somehow figure out a way to avoid being so polarizing?
I don't think so. Having build a custom agent with Anthrpic's SDK (in Go) and even running the agent on smaller quantized (non-frontier) models using Ollama, I seriously doubt you'd get good results. I read the same blog post, and I get the impression that the larger model with greater capabilities is able to infer more accurately without being constrained by large(r) system prompts, but I don't think this works so well for smaller less capable models.
Surely I don't need to explain how headless browsers work, bot proxies and distributed crawlers to you do I? If you haven't experienced this first-hand, I can understand. Maybe it warrants a blog post. But I can assure you, these don't work at scale, they might only stop the "less sophisticated" bots and maybe (just maybe) some unwanted spam.
Yeah no. Types don't make languages "more expressives". That just makes contracts (e.g: API(s) between callees and callers) more contractual and precise.
What does make a language more expressive (which may or may not be a good thing, that's a matter of taste) is powerful primitives that ladder up to more complex patterns and interesting ways to "express" ideas.
For example. One of the most expressive languages (arguably) is Lisp, or Io or other languages that have a property we call Homoiconicity -- meaning that the language's AST can be expressed and modified in the runtime of the language itself.
Macros are another feature that can make languages far more expressives too with similar effects.
Python's dunder methods, e.g: `__add__` are also another expressive feature (or anti-feature depending on how you use/abuase them) of Python.
In a work environment (at work), I often use Claude Code to fulfill work I would otherwise normally do with my steer and guidance. What I then do is go back to discussions, answering questions, reviewing others work, etc, until Claude has something for me to review.
This allows me to basically do "asynchronous" work and keep-up with the increasingly complex work environment where there are lots of things going on, sometimes too many things, all at once.
PoW schemes like Anubis don't work. Increasingly bots are using headless browsers and are basically able to solve captchas, proof-of-work(s) and basically bypass all any any attempts to block them. It's becoming impossible to stop bots from hammering your sites/services for unwanted traffic.
This tool doesn't seem to work very well. I tried to run it against my own harness that uses the Anthropic SDK(s) in Go and it turns up nothing. It doesn't seem to understand Go code?
Let's just put this argument to bed shall we? Let's consider another case. Leatherman pocket knives. It has a ~20% chance of cutting you, a 5% change of killing you if it pieces the wrong place in your body. I'm speaking from experience, where I sliced my thumb open (by accident of course!), but it was quite bad that I severed a nerve.
Needless to say, I don't feel compelled to take a moral or ethical stance on the use of such tools -- in this case; pocket knives and their inherent dangers.
Every time no matter how sophisticated or the no. of safeguards you put in place, can still inherently pose risks.
I'm not even arguing about food safety. Just think about the many restaurants that prepare and cook up the Japanese pufferfish. That's not even a tool problem right there, that's a skill and trianing problem. That shit™ will can can kill you if not done right.
Website: https://prologic.dev/ Twtxt: https://twtxt.net/~prologic Blog: https://prologic.blog/ Git: https://git.mills.io/prologic