pragma_x

As someone who has never used MCP, or allowed an AI to directly talk to APIs, putting something between a model and a service just seems like common sense. If nothing else, it presents an opportunity to tightly control what the AI is allowed to do with external services, especially if the MCP code is outside of the AI's context. I would go as far as to say it's a necessary security measure.

What I find amusing about moving compute to a RAM bank is it _almost_ resembles where we were with ISA-based extended RAM back in the 1980's. Some cards featured a CPU that took over the whole system and/or functioned like an upgrade. Others were a "computer on a card" that provided other features. I think this goes to show how cyclic tech can be. So, something like Samsung's invention here might have gained traction, as overcoming the slow PC ISA bus would have been a huge accelerator, kind of like where we are now.

> Could anyone enlighten me as to why to use Obsidian in one's daily life ?

My handwriting has degraded from consistent keyboard use over the last 30 years. Obisidan has replaced my paper notebooks, mostly because of this. However, it is handy to be able to mesh those notes with hyperlinks, LLM output snippets, quotes from other sources, embedded images, etc. Doing all of those other things would be laborious to impossible with paper.

This still happens with full shadow-casts and everything. Look around, and look to bigger cities if you're living somewhere out of the way. If there isn't a troupe near you that does this regularly, you may have more luck closer to Halloween.

Yikes. I've tried accupuncture. It _can_ work wonders for pain. But as a cure-all? Total quackery. Sorry your friend was approached like that in a time of need.

What's old is new again. We'll be buying cartridges for AI capabilities on future architectures.

> A small drawback is that just isn’t readily available on developer’s machines (unlike bash and make)

Looking back, this was usually a substantial stumbling block for adoption of new tooling. Things have changed. All one has to do is look at what JS programmers will go through to scratch build entire toolchains on the back of npm. IMO, the bar has been moved up to allow _some_ installation friction rather than none.

IMO, if the tool is a solid improvement and installation is a one-liner, use it and promote it loudly so it continues to get support.

Same here. This part got me:

> The whale itself has never been sighted: it has only been heard via hydrophones, but its call has been detected since the late 1980s in a pattern that matches the migration of the blue whale and the fin whale. Those species call at 10 to 39 Hz and 20 Hz respectively.

So not only lonely, but sings in a falsetto by whale standards; a weirdo. A lonely fellow with a distinctly high-pitched voice.

I was wondering about this. I wonder if there are insurance products to close this gap? Or maybe some banks offer accounts with different kinds of purchase protection.

I'm with you. While I'm no fan of the risk involved with missing a CC payment, there's a mountain of difference between credit and debit when it comes to fraud. It's literally you trying to get your money back (debit) versus some giant corporation trying to get _its_ money back (credit).

Not GP, but for starters, the MSRP for that car in China is $32k.

https://www.carsdb.com/en/news/1476

I get that's an apples-to-oranges comparison, considering that labor and material costs (not to mention subsidies and regulations) are probably not the same. But even at $40k, it would be cost-competitive with other EVs if nothing else.

There may be another option.

Consider Japan's Kei car initiative, for example.

https://en.wikipedia.org/wiki/Kei_car#Description

https://en.wikipedia.org/wiki/Kei_car#Taxation_and_insurance

One route is to provide incentives, if not regulations, to force innovation on US automakers. The goal would be to yield products that are head-to-head competitive with imports.

> It seems to me though that the current strategy seems to be very short term thinking - trying to just hold back the tide.

I'm seeing that too, but from a different angle. The era of big trucks seems to be as much an effort to extract cash from the economy as it is taking advantage of a peculiar set of EPA and DOT regulations. Basically, "gettin' in while the gettin's good." It's not a long-term strategy because, at some point, people can't afford these behemoths and will go for the used car market next for cheaper goods. EVs may get caught up in that too, considering that they're aimed squarely at the sub-luxury tier and above. As we have no good cheap EV options at the moment, I think it's the same story.

This is huge, thank you.

Since you're using HTMX, I have to ask: do you have any tips or idioms for composing complex forms and UI without things getting out of hand? I love the approach, but I'm having a bad time figuring out where the ideal balance is between too few or too many HTMX-replaced areas in a page. Thanks.

> I used fixed point math which optimized well.

I feel like the idea of fixed-point is under-utilized and very under appreciated. There are loads of applications where this is a better choice, let alone more performant.

Just a guess: if you want to scale a sprite at anything less than a whole ratio (e.g. 1.5, 0.7, etc), you have to choose pixels to drop out and pixels to repeat , on some pattern that looks good. There are going to be scaling ratios that look like a hot mess, especially at a low resolution like 320x240.

Water consumption and localized atmospheric heating have been cited elsewhere as drawbacks. There have been some articles citing noise/vibration pollution (subsonic?) but I'm not completely convinced on that front. Personally, I would add electric grid load to the list.

In the worst case, if your local municipality sides with business over the little guy, that means potential brownouts and water shortages for you.

100% agree. Not only was VC++6 a stand-out product overall, but it was easily the better IDE out of the crop of options at the time.

Sadly, the product line got worse before VSCode came out. Things are much better now.

I completely get the concept and agree - this is great way to build this kind of durability in a workflow system.

That said, my gamer-brain wants to call this "Save-scumming at scale." Which is to say, a lot of people already know that this approach works, but maybe they haven't made the connection to abstract CS stuff.

Another strategy that can be used to build robustness is to build your workflow out of idempotent operations. That can be useful for situations where the workflow state is too large to back up. Instead, you just run the job from the top and it's a bunch of no-ops until you start making progress again.

I hear you on vendor lock-in. Everyone's freaked out about other companies getting the upper-hand with AI in the loop, so there's this charge to use the hell out of it at all costs. Meanwhile, we're quietly picking winners and losers on the service side of all this, and we'll have to live with that outcome for a long time.

At this point, I'm seriously considering what it would take to build a reasonable budget-AI box that's self-hosted. It wouldn't need to blow the doors off of Claude, just get me most of the way there. Maybe even build it out of used and/or last-gen GPUs and a beefy motherboard.

Interesting article. I appreciate the range of perspectives here, and the overall pitch to keep the most experienced in frame along side new-fangled advancements (AI).

The "speed" loop reminds me a lot of RAD. In fact, AI might be _the_ thing that helps us deliver on RAD's promises from decades ago.

https://www.geeksforgeeks.org/software-engineering/software-...

A vantage point from a very long time ago: The big social media services are pining for the days of CompuServe and Prodigy.

- https://en.wikipedia.org/wiki/CompuServe - https://en.wikipedia.org/wiki/Prodigy_(online_service)

Old, pre-internet AOL is also in the same category.

These are what I refer to as "walled garden" services, that existed up to and (for a short time) through the commericialization of the net in the early 1990's. They offered built-in private services for chat, news, forums, games, etc. As direct competitors, they had an interest in keeping their userbase coming back to just what they were offering, and how they offered it. They also fell by the wayside for cost-competitive (free) online services that offered broader and more interesting stuff.

Anyway, we're circling back to this. Big companies like Meta have a vested interest in locking folks in and keeping them blind to alternatives.

Bringing the fun back simply means offering something better by providing an unmet need. It worked before. Last time it was the humble web browser that broke their near-monopoly on computer-gazing eyeballs. Perhaps we need something new that's just as potent?

I've heard people describe this phenomenon in basically two ways:

- It's all relative to your teenage years, which splits up generations as a result.

- The past is another country.

The good news is: taste and disgust are also learned. So anyone can pick up and move to another worldview, if they're willing to do new things.

> Imagine being able to list your eBay items locally without having to have people needing to come to your house, or better yet, getting a cut of what you wanted up front since they're basically a pawn shop, and then they list it on eBay and turn a bit of a profit with a local pickup option available.

I kind of assumed there were already local businesses that already did this? Seems like a decent side-line for any drop-shippers out there. In any event, moving that activity into a local strip-mall would be super convenient for everyone.

It even comes pre-packaged with a theme song.

I came in here to comment the same. Our brains are wonderful pattern recognition engines and the reader would absolutely be able to more readily see the correlation between hex and character representations this way. It might even accelerate learning hex values in the process.

Moreover for the folks in the back row...

We may see Canonical or other commercial Linux vendors come forward with a government or enterprise-flavored solution for all this. But the important thing to keep in mind is that they're not selling Linux per-se. As the GPL prohibits this, these companies sell support for their Linux distro instead. That revenue goes into improving Linux and maintaining their distro (e.g. Ubuntu). But even with all that money changing hands, that they do not own Linux, the Linux kernel, or any other shred of GPL licensed stuff.

May as well link the full report too. IMO, this is a bit easier to read.

https://ehss.energy.gov/deprep/archive/documents/0308_caib_r...

In my short-lived stint as the same, I also had the same take.

> But not one of them ever came to me unprompted and said, “Let’s talk about your career growth.”

This quote absolutely floored me. The author had a lot of bad management.

> I ask straight questions and look for straight answers. One line at a time, one file at a time.

I've also taken to using the Socratic Method when interrogating an LLM. No loaded questions, squeaky clean session/context, no language that is easy to misinterpret. This has worked well for me. The information I need is in there, I just need to coax it back out.

I did exactly this for an exercise a while back. I wanted to learn Rust while coding a project and AI was invaluable for accelerating my learning. I needed to know completely off-the-wall things that involved translating idioms and practices from other languages. I also needed to know more about Rust idoms to solve specific problems and coding patterns. So I carefully asked these things, one at a time, rather than have it write the solution for me. I saved weeks if not months on that activity, and I'm at least dangerous at Rust now (still learning).

You are correct. You absolutely must fill the token space with unanbiguous requirements, or Claude will just get "creative". You don't want the AI to do creative things in the same way you don't want an intern to do the same.

That said, I have found that I can get a lot of economy from speaking in terms of jargon, computer science formalisms, well-documented patterns, and providing code snippets to guide the LLM. It's trained on all of that, and it greatly streamlines code generation and refactoring.

Amusingly, all of this turns the task of coding into (mostly) writing a robust requirements doc. And really, don't we all deserve one of those?