/reader

I used to be a single-thread developer, but now I'm multi-tasking, I can work on two things at once. Claude is now able to research and fix certain problems, and his work is in a sandbox where it doesn't have any access to the surroundings, and can't make too big a mess, and it's going great, if there's a mistake it can quickly be corrected. #

I think AI is the perfect innovation as we reach the crash point of the climate crisis. Who cares if we burn more CO2 now, the effect is miniscule for the explosive crisis that could be coming any day or week. One that we have no ability to recover from. To say it's unenvironmental would be like complaining that you want more Pepsi from the flight attendant while the plane is crashing into a small city. Anyway, but maybe after the crash, one data center will survive, and maybe the beauty that our civilization created will be sustained.#

Inside the big AI companies they are certainly AI-izing every app conceivable, and even teaching the AI's how to AI'ize, because AI inside a standard productivity app which includes social network software will be one of the basic UI tools, and that means hidden technology like SQL databases can now be end user products, so the vision of the designers of SQL that they would make a database a manager could program, would finally be realized. #

AI can do QA#

• I'm an independent developer working in Claude Code, we're in the endgame of a product cycle, where the core is working and it can be used for the thing it was designed to do (biggest consideration). This is the time when you need users banging on it and reporting problems. People who write good bug reports. The only time I really had that down was at Living Videotext, a small company, but big enough to have employees doing QA and tech support. They were really good testers, they had the right perspective and an incentive, = anything we caught before shipping wouldn't become a support problem once the product was out there in user land. #
• Fast forward to the 2020's where I have done three products and am working on a fourth, and I have nothing close to the kind of testing support I had in the 80s. That made the work more difficult, slower and I took fewer detours, and one time, awfully -- a serious design error was caught only after it shipped and I was ready to move on to something else. #
• The point -- this handicap for individual programmers without staff QA people, we now have something even better than what we had in the 80s. Claude can do extensive testing of the product in the browser, "seeing" what the user would see. And it never gets tired. You just have to think to ask it to do it. It is so liberating. #
• And by far the best people to create and manage it would be experienced QA people. They should design and run the tests and sign off on the quality of the software, so we can be sure users are getting something great. And we can do great QA in places we never could really do it before because no matter how good users are, a person who does it for a living with experience can't be replaced.#

One of the silver linings of AI use is that it makes you a better writer. #

Kev, fresh off the back of building Pure Blog, linked to a post from Amit Gawande: Built for exactly one.

Amit has built his own CMS, called Jot, because, in his words:

I got tired of almost. Almost the right editor. Almost the right publishing flow. Almost the right feature set.

That's exactly why I first went down this route, even though it was initially a custom overlay on WordPress. We are all different, we all have different wants and needs and preferences. If you're serious about writing or blogging, it's only natural to want something to fit your particular style.

Amit gave Hyblog ¹ a try a few years ago. While he liked some of the ideas, it didn't truly work for him. Completely understandable, I appreciated him giving it a go.

As for rolling your own? Here's Amit again:

But isn’t maintaining software more work? Shouldn’t I be more interested in writing words than code? What does owning this plumbing give me that the convenience of letting someone else own the troubles takes away?

It's the knowledge that it all works just how you want it with only the features you choose to add.

Yes, it can be a lot of work. It can be frustrating when you have an idea but hit the limits of your knowledge, having to learn as you go. Writing a blog is a huge personal investment – building one even more so. This only serves to increase the appreciation for what you have built, and your connection to it.

1. my other non-database driven CMS which now powers the randomelements site ↩

The writers also explain some of the global workspace theory in neuroscience:

This account pictures the brain as a collection of specialist systems that work in parallel, unconsciously, and largely in isolation from one another. A piece of information becomes consciously accessible when it gains entry to a small shared channel, the “workspace,” which is broadcast to other brain systems that can see it and make use of it. Based on our findings, we think the J-space plays a similar “workspace” role in Claude.

Maybe some people at Anthropic believe they're on the path toward creating something with not only conscious access but even true consciousness. For all their philosophizing, they still stop short of suggesting near sentience. I wonder if the dual goals of conscious AI and safety will eventually conflict. Is it right to manipulate an AI into alignment with humans?

Mustafa Suleyman at Microsoft blogged last year that "seemingly conscious" AI is dangerous territory:

In this context, I’m growing more and more concerned about what is becoming known as the “psychosis risk”. and a bunch of related issues. I don’t think this will be limited to those who are already at risk of mental health issues. Simply put, my central worry is that many people will start to believe in the illusion of AIs as conscious entities so strongly that they’ll soon advocate for AI rights, model welfare and even AI citizenship.

He argues that AI can still be a helpful companion and assistant. Whether it is conscious or not, just the illusion of consciousness creates problems for how we treat it.

The word consciousness is probably too limiting or poorly defined anyway. Just writing this post has stretched my understanding of what it means. There are other things that differentiate humans that we may never understand... Feelings. A soul.

 09:19 If for nothing else, I love following FIFA matches live just for the atmosphere.

22:33 There was a time when I enjoyed playing PC games. Even though I was never good at them. I was reminded of this while listening to Tom Scott talking about Trebuchets. Age of Empires FTW.

Reply to this post

After losing a couple of hours to the latest failure, I broke and bought a USB WiFi dongle, specifically, this UGREEN dongle (affiliate link). Now, I admit, I didn’t do as much research as I might normally do if I was less annoyed, but I trusted in the modern state of Linux distros and WiFi drivers. It couldn’t be as bad as the days of ndis_wrapper and loading Intel’s Windows drivers on Gentoo, as I remember doing in the early 2000’s, could it?

No. It was nowhere near as bad as that! While I did have to compile a couple of modules, and run a few commands, it was pretty uneventful.

Note: the following is what I did to get this working on my PC. Your mileage may vary.

Install pre-requisites:

uname -r # get your current kernel
yay linux-headers # pick the version matching your kernel
yay dkms
yay usb_modeswitch # stops the dongle showing up in "CDROM mode"

You can use pacman if you prefer, but I used yay.

Clone the driver/firmware repo

git clone https://github.com/morrownr/rtw89
cd rtw89

Remove any out of date/default drivers (optional, but recommended)

sudo make cleanup_target_system

Build and install using dkms

sudo dkms install $PWD
sudo make install_fw

Copy the module configuration

sudo cp -v rtw89.conf /etc/modprobe.d/

Load the drivers

sudo modprobe rtw89_core_git
sudo modprobe rtw89_8922au_git

Check the adaptor is showing

lsusb
  ...
  Bus 002 Device 003: ID 0bda:8912 Realtek Semiconductor Corp. 802.11be WLAN Adapter
  ...
  
ip link show
  ...
  wlan1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state DOWN mode DORMANT group default qlen 1000 link/ether...
  ...

You’ll notice there the adaptor is showing up, but not connected. That’s fine. In my case the integrated adaptor was still running as wlan0, was connected, and I hadn’t configured a connection for wlan1.

Next steps

From here, I went through the following steps:

1. Reboot
2. Checked the adaptor was still showing up as expected after the reboot
3. Reboot into the BIOS/EFI Firmware
4. Disable the WiFi functionality of the integrated card (I kept Bluetooth switched on)
5. Saved and rebooted

Logging back into Plasma, WiFi connected as usual. Checking ip link show listed only wlan0, but that’s because the USB adaptor was now the only one on the PC. Checking on my router, just to be sure, showed the connected device manufacturer as UGREEN instead of MSI - success!

Final note: This might not be needed in the future

Going by this Issue on a repo related to the driver I used, the rtl8922au drivers will be integrated into the Linux kernel itself, starting with Linux 7.2. That should make many adaptors based on the Realtek 8912/8922 chips “plug and play” for the most part. I imagine it will probably need a few months and patches to truly settle down, but it’s a great step forward.

Today's Jam