CrimsonCape

> his contributions included investigating why generative AI systems suck up to users

Why does it take research to figure this out? Possibly the greatest unspoken problem with big-coporate-AI is that we can't run prompts without the input already pre-poisoned by the house-prompt.

We can't lead the LLM into emergent territory when the chatbot is pre-engineered to be the human equivalent of a McDonalds order menu.

"No taxation without representation" is a perfectly reasonable stance.

You are right. I went with ultramarine linux and am perfectly satisfied.

Questionably useful at the cost of personal computer components doubling. Unquestionably shafting the personal computer market.

Dude it would be so great to have a little device and you text it SMS messages as prompts. Then it just sits there and thinks away probably getting really hot lol.

It's unfortunate that there is huge interest in prolonging lithium battery life because it's an indication of products with hard-to-replace batteries.

protected lithium cells are fairly save and easy to replace into devices designed for them. Lifepo4 cells similarly.

We need more product engineers out there fighting the good fight.

I was ready to install Linux. I installed a new 1TB ssd in my laptop. I shrunk the windows volume using Windows' Disk Management.

Then I started reading the Arch wiki on this task. It forced me to learn things like MBR vs GPT. Then it said Windows by default makes an EFI partition way too small so I have to re-create a new partition by temporarily mounting EFI, saving the files, deleting the EFI partition, and recreating a new one.

This seems like a horribly complex task and I can envision about a million unwritten things that can go wrong that the answer would be "well duh, that's obvious if you had any experience with linux disk partitioning. I myself bricked a dozen PCs."

Deleting the EFI partition, if it goes wrong, by definition my system would be bricked until I could figure things out.

Also, everything must be typed into terminal exactly with no error and one chance. (If the typo causes the command to error, phew. if the typo causes something else to happen, beware)

So yes, I have a lack of taste.

I can't help but think that in 1910, both the concept of "fear of rejection" and "high standards" would have made no sense to people at the time. Yet I would agree that they are valid concepts today. We have to explore why these two concepts exist and why they did not exist in 1910. It seems valid to call them side effects of something bigger, what the bigger is I don't know. I don't see how society can address these two issues without addressing the other issues that lead to the existence of these.

It is hilarious and i'm not clicking any link lol.

I'm confused too. The parts are high quality and scream "in-house" and surely not the product of contract engineers or contract machinist work. But I don't think DJI has machinists or factories in-house? I would assume they are just the product designers but I guess it's probably some kind of unique workflow due to Shenzen.

I just anecdotally see that in the USA, the required iterative design process is too cost-prohibitive for injection molding, and likely the same for every other trade. So multiply number of trades (designer, CAD drafter, machinist, electrical engineer, software engineer, injection molder, assembler, etc.) multiplied by the number of experimental iterative processes required to build an institutional knowledgebase... it's cost-prohibited.

After seeing your renders on github, a question I always have is why voxels are stereotypically 1m square... wouldn't the ideal voxel engine immediately set the scale to something more high-res? I'm just drooling at the thought of a voxel engine where a block size is equivalent to the size of a physical pixel.

If I understand you correctly, you are considering Adams to be "off the rails" crazy and therefore you are condemning him, for having opinions?

Just looking at the injection molded shell of my Mavic Mini makes me cringe when thinking about the startup cost. It's the plastic shell; not the motors, nor the circuitry, nor the optical parts...and to think you could build that in the USA is laughable. DJI releases 2-3 models every 2-3 years... if you could even find a company in the USA machining the steel molds at that frequency (i don't think it exists) how are you going to afford the bill?

I can put over 30 years of kernel dev experience on my resume since my AI subscriptions were trained on 30 years of kernel development code.

So how can I get control of one Sears store and through that one store, singlehandedly resurrect the brand? Surely they must be ready to field Skunkworks-like ideas from young talent willing to risk and win...

This hypothetical new Sears vision is to create an Ikea competitor, but an honest Ikea, where Sears squares off with their biggest showroom-based adversary.

Ikea is beautiful but hollow, as their showrooms are seductive dioramas, but the resulting money exchange of cash for particleboard is like buying sex instead of meeting you life partner.

Instead, Sears would be the interface between the customer and the American-made premise. Sears sells American made products and represents American makers. Fuck selling washers and dryers. The showroom is not there to stock imports in cardboard. It's there to show you a concept, what your life could be if you stop buying junk. It's there to demonstrate quality. It's not particleboard at the core; it's a chunk of USA.

It would shamelessly capitalize on nostalgia and quality. Instead of nordic spartan whites and blands, it would have dioramas of pine wood wall panelling, cast iron wood stoves, chrome magnificence, Tiffany glass. Want a new desk? $5k. Made of US oak. Want the pine walls? The boardfeet are in the back, milled 5 miles away, how much square feet you need?

Even better, each store is fucking exclusive. Find local makers and concentrate their products in the geographically local store. Make people rabid and giddy to travel to each store even if they need to go five states away. Visiting is a unforgettable experience, like a museum that you can buy stuff.

Stage 2 is to use the Sears branded products like Ycombinator for machine shops. Sears injects some startup capital and now Sears cabinet pulls are turned on a machine lathe in a startup machine shop in Ohio. Furniture ateliers are like franchise Chick-fil-as.

I am of the generation of CAD users that is entirely digital, having no experience with bluelines or hand drafted sets.

And even now the industry is still changing. The old (for me) standard was obtaining reprographics from a full service supplier, signing a contract, leasing large format plotters, monthly delivery of toner, on-call maintenance.

My favorite plotter was an Océ and I think that company died? My last office was reliant on an Océ plotter which had a Windows Forms GUI app running on a never-updated embedded Windows XP. A perfect use case for embedded Linux but I guess it either predates embedded Linux or corporate offices at the time preferred embedded Windows.

Now we don't even print drawings.

I enjoyed the article because it is written by a stereotypical draftsman. While we all write in paragraphs, draftsman have some self-taught logic which results in this confusing blog format with random quotations, italicized sentences, arbitrary mix of heading styles, self-invented acronyms. Every draftsman I ever met 1. knew better than everyone 2. Knew you were always wrong 3. eternally pissed off.

> But using C# required us to contemplate whether and which dotnet runtime our client supported. Or did we need to ship our own? Isn't this just a small launcher stub? This was just too much complexity outside of our wheelhouse to put between our product and the user. This is not to say that the C# approach isn't valid. It is just that our limited understanding of that ecosystem and its requirements counseled against shipping it as a primary entry point into our application.

You should be able to compile a relatively small, trimmed, standalone, AOT compiled library that uses native interop. (Correct me if i'm wrong, dotnet users). Then there would be no dependency on the framework.

I been using C# since .Net Framework 4.6, jumped to .Net 6, and have kept up to .Net 9 when starting new projects.

My advice is to download .Net 9 SDK because it's now super easy to create a file with extension .cs, put your terminal window next to your code, and call `dotnet run file.cs` and it just works. The compiler will teach you a lot because the errors are extremely easy to follow.

I never bought into books because too many language features have been added in the past years that improve on "old" ways.

The only complaint I have is that Google is still the best interface for navigating the docs. Spamming google over and over. Solving this in an easily greppable way is on my wishlist.

Especially because I see the MSDN docs are rolling out new "Did you want to generate a sample of this? Here's a prompt you can copy paste into Copilot to generate an example: xxxxxxx blah blah"

Can someone ELI5 from the experience of hosting a static site, what is the next step beyond a static CDN?

I am guessing the next step is paying for a bucket that that allows virtualized containers to run an OS? Or having a constantly-up node.js backend to generate server-side HTML? Is this essentially the generic description of self-hosting Wordpress on an own server?

Does integrating a payment processor like Shopify require something more than static hosting?

Thanks, this helps. Do VBOs map 1:1 to meshes? Is there some point where I'm looping though meshes and sending every mesh to a VBO?

I see lots of shader related videos but to me the worst part of GPU code is that the setup is archaic and hard to understand.

Does anyone have a good resource for the stages such as:

- What kind of data formats do I design to pipe into the GPU? Describe like i'm five texcels, arrays, buffers, etc.

- describe the difference between the data formats of traditional 3D workflow and more modern compute shader data formats

- Now that I supplied data, obviously I want to supply transformations as well. Transforms are not commutative, so it implies there is sequential state in which transforms are applied which seems to contradict this whole article

- The above point is more abstractly part of "supplying data into the GPU at a later stage". Am I crossing the CPU-GPU boundary multiple times before a frame is complete? If so describe the process and how/why.

- There is some kind of global variable system in GPUs. explain it. List every variable reachable from a shader fragment program

Looking at the web page is kinda ironic. A yacht builder proclaiming how much better the materials are for the environment. We could just build less yachts, for example. Nobody needs a yacht.

There has to be an interesting commentary here regarding the necessity of productive endeavours that pay taxes and fund local governments and drive investors portfolios into the black, all funded by useless largesse.

Yeah especially when the unsolicited contact starts sending PDF attachments of "evidence".... felt like a classic phishing attempt

Here he is unironically citing Ted Kaczynski:

https://xcancel.com/RoguesPhilo/status/1814993224148066772#m

"I took 20 Americans on the most dangerous horseback riding expedition in central asia" and there isn't one AK47 in any of the photos. What a poser.

Do you believe there's a distinction between giving the minority too much power versus protecting them from having too little? It seems like the same thing said two different ways.

Yes.

I found the reddit ManyBaggers recently and there is a cottage industry of high-end bags that seem incredibly made for the price that are in no way luxury products.

You missed the part where he said "make a device people want."

Just some quick facts from the article:

The new requirement is that the maximum weight of a clothed worker is 124.7kg or 274.9 lb.

One worker said he was at 118kg (260.1 lb), and his target weight is 110kg (242.5 lb).

Not to throw shade at anyone in particular, but using the USA NIH.gov BMI calculator, for a 5'-11" man, the "healthy" range of BMI is 18.5 to 24.9, in mass that is approximately 60.32kg (133 lb) to 80.7kg (178 lb).