I used to be around the rally racing scene, and it's utterly routine there to weld tons of re-enforcements into the unibody. Most people with the money will straight up laminate the suspension mounts with plate. The safety cage makes the car incredibly rigid, as does replacing all bushings. The cars work just fine. The people doing this have relatively ordinary welding skills. Modern cars are not some sort of high tech magic that can't be modified or repaired with ordinary metalworking skills... it's just usually not cost effective to do so.
But you are on the money with the cost effectiveness bit, that's exactly why it is perfectly possible to do these modifications with confidence: you are not operating under a budget that gets counted out to the last 1/10th of a cent, so you can do it right, even if it costs a few bucks more. Something a car company operating at scale would never do. Once the cost constraint disappears there isn't a whole lot you can't do.
1) reinforcing the whole car for a rally (which brings the car up to 1965 overengineered standards, and well beyond); and
2) simply welding two halves of a car together.
If you were to weld two halves of a car together, and weld a ton of reinforcements into the unibody, it would very obviously work fine.
The repairs you are proclaiming to be impossible are perfectly possible to do with good results. These repairs are often performed often, just not in the high cost of labor parts of the world where HN lives (the low cost areas buy our wrecked cars to fix and sell to the point where sourcing parts for major repairs is often easier there than in the vehicle's country of origin). There's autobody industry specific tools and literature that help professionals in the industry perform the repairs properly. It's perfectly possible to take a car apart and reassemble (with good parts) the way an OEM would have if you're willing to invest in the tooling (mostly for welding in hard to reach areas). Car passenger cabins aren't supposed to flex (but in practice they do so within acceptable limits) and OEMs provide service literature telling you what you can and can't modify, much repair info can be gleaned by reading between the lines of such service literature. The welding issues you are describing are simply not an issue for skilled welders. Your opinion of engineering education is hard to reconcile with the observed performance of new engineers in any/every field.
The conclusion of what you're replying to: "You can fix up a $100 Huffy with any old welder. A $2000 steel frame with thin-wall double-butted tubes? You want a really skilled welder and close inspection to make sure it's not compromised."
Yes, it's possible, and yes, I know folks whom I'd trust to do it. The guy in my local auto shop? Definitely not. Random car shop in the developing world? I've seen the results. They'll drive, but they're far from save by Western standards.
What you wrote: "if you're willing to invest in the tooling" "The welding issues you are describing are simply not an issue for skilled welders"
I mean, seriously. You responded to a something entirely different from what I wrote, repeated roughly what I actually DID write, and mixed a few errors in (if car cabins weren't made to flex, they'd be using a very different type of steel; virtually everything made to be tough is made to flex).
Please find a place in my post where I "proclaiming [repairs] to be impossible"
1) How much force it takes to flex it -- the spring constant
2) How far it will flex before it breaks
Multiply the two together (or specifically, 1/2 kd^2 where d is how far it will bend before it breaks, k is the spring constant, and 1/2 kd is the average force you're applying), and you get how much energy it takes to break it.
*Everything* flexes. Rigid things just flex less for the same force.
Things which break when flexed just a little bit are called brittle.
Have you tried doing that to anything other than a 1960s convertible?
My 90s junk handles that just fine (and they're not particularly rigid cars). Matter of fact I did it yesterday afternoon to change rear pads.
Generally speaking you need to get a floppy vehicle really bound up (think 80s Jeep Cherokee or minivan off road) in order to have issues with doors and even then it's usually rear hatch only. OEMs specially try to avoid having bodies that flex enough to be noticeable to consumers, let alone to the point of doors not opening. Heck, I'd be surprised if you can even design a car to pass modern roof strength requirements without also making it stiff enough to have the doors open and close regardless of how bound up it is. (Pickup tailgates are another thing since the bed is open top and bolted to a frame designed to flex somewhat and even then being able to open the tailgate with two tires off the ground is not unheard of)
It's similar in microelectronics. I can do a ton of stuff with an arduino because it uses really basic old, tough eletronics. I can put too much current into a pin, ok the pin dies but the AVR is still fine. A modern ESP32? A little bit of current or voltage above the rating and the whole thing is toast.
It's just like people putting turbochargers on their 90s Hondas. Naysayers could claim it's a fool's errand because if Honda, with all their engineering resources, didn't feel they could add turbocharging to their engines in the 90s and have it be cost effective and reliable enough, surely someone in their garage can't do it? But what they overlook is that if 5% of one of Honda's mass market cars blew an engine that would be a disaster, whereas a 5% chance an individual blows his own engine, which he may be able to replace for $500 to $1500: that may be a risk an individual is willing to take that a carmaker cannot.
With a 2020 car? I wouldn't trust it unless the person performing the work was a pretty good engineer. As engineering is more optimized, you can't just do stuff like this without being very careful:
1) Welds do leave a heat-affected zone around the weld which is weaker. If you're overengineered (1956) and everything is thicker than necessary, this doesn't matter. If you're near-optimal (2020), and using the exact amount of material needed to avoid failure under expected driving conditions, you might weaken things enough to see failures.
2) If reinforced, you'll affect the overall systems design. Engineered objects flex. Making one part more rigid than expected will put additional strain on other parts.
Counterintuitive as it might sound, optimal engineering makes designs safer. You analyze failure modes, put in appropriate safety margins, and do things right. Engineering classes drill that over-engineering is bad engineering. The flip side, though, is that if you can't tweak optimized designs like you can over-engineered ones without being careful and thoughtful about it.
Same thing with bikes. You can fix up a $100 Huffy with any old welder. A $2000 steel frame with thin-wall double-butted tubes? You want a really skilled welder and close inspection to make sure it's not compromised.