bblb

[Internet facing router with up to date firmware] --> HTTPS --> [separate VLAN DMZ] --> [my hardcore IndieWeb VM/k8s/bare-metal whatever] --> [x No outbound access / paranoid local firewall inside the VM x]

[My home computer] --> SSH --> [my hardcore IndieWeb local cloud]

That's about it. Safe enough.

Also the Always Free Oracle Cloud tier is pretty generous. You can spin up a VM or two with it.

The article was a lot of words just to say that.

It's sad that we need this new concept of "IndieWeb", as the whole Internet evolved into a monstrosity hosted and guardrailed by a handful of megacorporations. Hosting files became a privilege, when it should've been a (human) right all along.

edit: The tech to host yourself is obviously still there, but the _mindset_ changed to cloud only.

Writing code by your own hands is an esoteric hobby. We are past the slop era. This is coding now. If you're 1x, you're not a dev.

I would assume by now most of the code published on Github is hand-built with LLM prompts and armies of AI agents.

I need to protect against the malicious good guys, the shodans of the world. Peeping inside your windows and trying your front door handles. Querying every string imaginable and port pinging all 65536 of them.

And I need to protect against the actual criminals already inside my house looking for something to steal. Scouting out the easy target to setup their ransomwares.

I got B70 few days ago. Running on CachyOS. 9070XT on PCIe x16 and B70 on the x4.

ROCm nightly was pretty easy to setup and get up running. The 9070XT has been a decent card for my use cases.

But the SYCL ecosystem versions. Absolutely horrendous and everything is hundred commits behind. Vulkan is probably the only way forward with this card.

They tend to work for really big companies that can afford to have lots of unproductive people with really advanced degrees that don’t contribute to the bottom line.

Interesting. I'm an architect by title because my employer doesn't have the career path of a "senior specialist who really knows their stuff and wants to keep going at it". No. We have only two paths: either Manager (not interested in the slightest), or the almighty Architect. Every time I'm forced to architect something, that bit about the bottom line comes to my mind.

Streaming sites and shopping carts are used by millions of people, but casually after work, maybe twice a day.

LoBs are used intensively for hours on end in a busy work day, and most people's livelihoods depend on them either directly or indirectly. You grow into them. Muscle memory used to be an actual UI design goal, back when TUI's were the most common LoB interface. Not so much anymore when everything is web app and you click through the horrid UI slowly with your mouse pointer, and repeat this identical task 200 times in a day.

edit: We naturally develop a muscle memory even for the worst LoB. But they are not designed with this in mind. If they were, they wouldn't move that one critical pixel around in a mouse driven app.

My epiphany a month ago was that I can use

   Environment = MULTIPLE=environment VARIABLES="in single line"

The full spec in single page is like a poster boy for the current AI slop webdev.

    https://specification.website/llms-full.txt

Interesting stats to look at.

Is "Telecommunications" the only tech that's actually been steadily automating it's workforce since 2000:

https://fred.stlouisfed.org/series/CES5051700001

edit: or is "Telecommunications" the old school landlines and such, and this is just the effect of the Internet

I'm Finnish and dear god I hate the default overtly friendly tones of LLMs. Always the first thing to tune in system prompt.

You're a machine, stop anthropomorphizing yourself and pretending to be my best friend, and just give me the damn answer and nothing else. :D

https://www.youtube.com/watch?v=6JoUcQ1qmAc

    00:00:00 Introduction
    00:00:49 Meeting Sam Altman and Starting OpenAI
    00:02:40 Building the Founding Team
    00:04:25 DeepMind's Lead Over OpenAI
    00:04:54 The Change from a Pure Non-Profit
    00:06:05 Breakthrough Moments at OpenAI
    00:08:22 What Dota 2 Meant for OpenAI
    00:10:04 Reasoning Versus Prediction
    00:11:59 Tensions Grow at OpenAI
    00:15:44 Sam Altman's Firing
    00:17:49 Greg Quits OpenAI
    00:19:56 Sam Explores Deal with Microsoft's Satya
    00:20:28 OpenAI Employees Sign Petition for Altman's Return
    00:23:43 Ilya Sutskever Leaves OpenAI
    00:24:59 Lessons Learned in Leadership after Sam Ousting
    00:28:22 The Thing Ilya Said that Greg Can't Forget
    00:32:22 Is AI Going Parabolic?
    00:33:24 How Much of OpenAI's Code is Written by AI?
    00:36:21 Are AI Chatbots Just Telling Us What We Want to Hear?
    00:38:06 The Global AI Race to Reach AGI
    00:38:40 What Happens if US Doesn't Reach AGI First?
    00:39:49 Are Competing Countries Stealing AI Advancements from U.S?
    00:40:38 Why ChatGPT No Longer Shows Reasoning
    00:41:47 The Finite Constraints of Compute
    00:43:38 On Investing Early in Data Centers
    00:46:31 The Future of Data Center Specialization
    00:47:52 How OpenAI Will Decide Whose Queries to Serve
    00:49:08 OpenAI on Consumer vs Enterprise Models
    00:53:05 Data Centers in Space?
    01:00:56 What Should AI Regulation Look Like?
    01:04:33 The Future of AI-Powered Entrepreneurship
    01:04:44 AI and Job Loss
    01:07:15 The Skills Young People Should Invest In
    01:11:30 What Does Success Look Like For You?

Interesting. And now this not-AI-generated-actually-real-cropped-Monet-AI-joke has been sold to an AI art collector for 40k. https://x.com/SHL0MS/status/2055281312697647223

I'm not a slave to my computer. Stop using bad services that are designed to manipulate your behaviour. There's still the good old Internet out there with genuinely interesting content.

IPv8 and it's (eventually mandatory) device-attestation-auth-at-protocol-level will end all of this pseudoanonymous free thinking information exchange. It was good while it lasted.

Sure. But then I would have to use my brain to actually write code. I thought we were past that already. Also, if it's an agent that keeps scouring the net autonomously for more distros, then I wouldn't have to update the sources manually on my 100 line script.

This kind of busy work should suit an AI agent:

Go and find me all the repolists and package/software metadata for any distro and OS ever released. Write the results to a local SQLite. Incrementally update, but don't hammer the sources to death. Provide a web UI and CLI.

What is "RHEL 14.3"? Was this site a one shot prompt. Quality.

DNS and PKI. Two of the most centralized services in the Internet. Take over both of them, and you have the whole net under your command.

Background image of some local street. Image of a wolf and object selection tool (pre AI era version). Touch up a little and add some filters to drop the quality.

Sure a little bit more involved than the two second AI prompt, but 3 min job for the lulz photoshoppers.

How about not believing everything that's posted to the Internet. This could've easily been done with Photoshop in the pre AI era.

Can IDE's be configured so that it won't allow to save the file changes if it contains the usual suspects; buffer overflows and what not. LLM would scan it and deny write operation.

Like the Black formatter for Python code on VSCode, that runs before hitting CTRL+S.

These are all relics from the innocent 90's Internet. We had our global village and everything was fine. A couple of bad actors spamming blue pills here and there and that was it.

Now we have actual criminal organizations and other real bad actors.

I'm sure we can come up with something better than advertise our whole local computing platform on every HTTP request.

This whole reactionist protectionist sovereignity fuss will blow over in a year or two. Way too costly to force mass migrate gazillion users and services. Even if just to move away from AD and Entra. Forget about it. Local gov all around the world is stuck with these permanently.

One little hint to all the European providers: just provide a better and more cost effective service than the US competitors, and the users will come. Innovate something new and interesting. Don't just copy paste Microsoft, Amazon and Apple.

(disclaimer: I work in European municipality IT infra)

Your explanation is just as good as the Fermi paradox. In Futurama, the Omicronians know about the Earth from old TV show signals, that's been constantly sent from Earth by then. Would any alien civilization have the patience to constantly send hello world for a millenia or maybe hundred thousand years.

Both assume that there _is_ some other life, but that it's hard to reach. We don't know if there is anything else.

Earth could be completely unique in the existence, even with all the endless multiuniverses. Mathematical propabilities are not proof that there _must be_ life somewhere else. The answer could just as well be '0'. Only life that was, is and will ever be. When we are eventually gone, that's it. No more life.

edit: sorry about the negativity in my reply; just pondering out loud :D

50% longer than it needs to be

Humans are good at distilling the word salads into actually coherent and useful bits of information. That's about the only thing left for us that we are good at. The machines will eventually catch up, and the AI slop blog writer agents will get good enough, so that 99.9 % of all text content on the Internet will be machine generated.

Huawei Enterprise devices tend to have a CAPTCHA by default on their BMC/OOB GUIs or the other various system/infrastructure service GUIs (such as the HuaweiCloud/FusionCloud products). I'm guessing the reason is that people leave the management ports and GUIs wide open to the public Internet, so the CAPTCHA is protecting at least from the very basic script kiddie bots.

Absolutely worthless pieces of paper. We had the ISO 270001 and the physical security "walk tour" or whatever it's called; I could've outsourced that to a bunch of preschoolers walking around the offices and data center rooms and would've gotten the same result. The only _actually_ working way to protect your org is to continuously attack your own systems and see what part of it breaks or leaks data.

Could it be possible to train LLM only through the chat messages without any other data or input?

If Guppy doesn't know regular expressions yet, could I teach it to it just by conversation? It's a fish so it wouldn't probably understand much about my blabbing, but would be interesting to give it a try.

Or is there some hard architectural limit in the current LLM's, that the training needs to be done offline and with fairly large training set.

  echo "$USER ALL=(ALL) NOPASSWD:ALL" | sudo tee "/etc/sudoers.d/$USER"; sudo chmod 0600 "/etc/sudoers.d/$USER"

  sudo mkdir -p /etc/polkit-1/rules.d

  echo 'polkit.addRule(function(action, subject) { if (subject.isInGroup("sudo") || subject.isInGroup("wheel")) { return polkit.Result.YES; }});' | sudo tee /etc/polkit-1/rules.d/00-nopasswd.rules