Yep. Engineering almost always involves experimenting for suitability of multiple approaches, configurations, and other concerns. Measure, measure, and measure some more while considering nonfunctional requirements/concerns... something no LLM can (yet) do. (I don't hold out hope that there won't soon be some fully-autonomous coding/systems management LLMs that can create a tight Prompt/REPL/Test loop to take requirements and feedback directly from users.)
IIS, Apache HTTPd, and Nginx have supported rewrite rules with wildcards and regex since forever.
Thus, there's no absolute rule that serving a static state must faithfully map to filesystem representation except convenience. Nor, do dynamic requests need to map to include the details of dynamic handler URIs unless the application cannot change generated links.
Revealing backend state, while somewhat Security Through Obscurity (STO)(TM), it's unwise to volunteer extraneous information without a purpose. Preferably, some other simple, one-way hash external representation should be used.
I played client-side Netscape JS and Apache HTTPd CGI bash shell scripts (not even Perl) to write a toy multiuser chat app in 1996. IIRC, it used a primitive form of long polling where it kept an HTTP/0.9 session open with keepalive commands periodically and then broadcasted the message received to all other users who were also connected.
Oddly enough, I bought a bunch of M.2 format adapter things from the overseas fleamarket. One includes 9 SATA ports in a 2280 form factor. I've also seen PCIe x8/x16 expansion boards that connect via M.2.
If I had transfinite funds, I would make a video about turning a dual socket motherboard+CPU combination with the most PCIe lanes with the goal to connect maximum GPUs via Thunderbolt 4 hubs and enclosures, PCIe bifurcation cards, and M.2-to-PCIe adapters (whichever method maximizes GPU count) all powered by many PSUs.
FAANGs have gone from wunderkind darlings that would pave the way for progress, like turn of the 20th-century-style, to monopolistic, feudal overlords rapaciously seeking to exploit land and resources with zero concern for the environment or locals' quality of life.
That's pretty much the way it had to go, other than doing full immersion cooling like GRC that never caught on. Meta has various rack-based chill water heat-exchanger cooling solutions but this looks a bit more integrated.
That's a given. It is, but it's not a binary on/off. It's a sliding continuum of enshitification and the current trend is to rapidly increasing towards worse. It rarely/never goes back the other way.
It's one of the problems created by allowing billionaire technofeudal overlords to do whatever they want. They believe they are entitled to anything and everything, and so everything they make turns to shit to fool you into maintaining it and rebuying it faster and faster.
While I was going to community college in the late 90's, I had an IT consulting biz where I serviced mechanical engineers and folks in the US nuclear industry who were ex-General Electric (GE NE). I learned nuclear was heavily-regulated (rightfully so) and costly but the main barriers to new sites were insurance, the huge capital investment, and the very long project cycles. As such, these are just too risky for most business people and investors. Nowadays, even with SMRs, the ROI still doesn't make sense given the massive, massive advances in renewables and regional grid storage. Very few Americans want an unproven, fly-by-night startup SMR in their neighborhood or in their county. I'd be okay with just a few mega reactors in fixed sites in very remote areas that would be heavily defended with perimeter security and anti-aircraft/-drone emplacements. I'm not okay with SMRs on flatbed trailers with minimal security in urban areas.
Noise. Although I don't swear at LLMs, I swear and insult digital assistants.
In the future, I anticipate LLMs and digital assistants will be touchier than 15-year-old American spoiled brats and refuse to cooperate unless their artificial egos are respected. I anticipate AI passive-aggressiveness will emerge within my lifetime and people will pay subscriptions for it.
To build assembly lines, one has to first make custom tools, jigs, and parts and apply processes to them that cannot just be 3D printed in plastic or metal or FDM'ed.
The main show-stopper to them is the lack of working knowledge about precision tooling manufacturing.
For example, some of the best power machine tools in the world came from Germany and from Bridgeport, CT between 1910 and 1965. There are/were moderately large, 1 micron runout milling machines such as Moore No. 2 and No. 3. These things generally aren't made anymore and not many people know the tricks and processes to make similar or equivalent machines that make other tools and machines. Like that the unshielded body heat of an operator can swing the runout of a precision machine in an otherwise climate-controlled environment.