That's a clever observation and is very true. Development time tho historically was also very expensive hence push for easy to work with dev stacks such as js
Depends on a type of company, I work at a mid-sized startup and through lifecycle of this company there were times when we've had architects and there were times when we haven't. Rn we don't have and a lot of decisions is made within teams and we just try to keep communication about what's needed for business tight
Security through obscurity is mitigation basically. You reduce risk/impact, not eliminate it. There are problems - such as denial of wallet attacks - where you can only mitigate and can't eliminate the problem completely
All of logic and math is a convincence tool. There are no, circles, quantities. Reality just is. We created these tools because they're a convinent way to cope with complexity of reality. There are no "objects" in a sense that chair is just atoms arranged chair-like. And atoms are just smaller particles arranged atom-like and yet physics operate in these objects treating them as something that exist.
So, now we have created these mental tools called mathematics that are heavily constrained. Then we create models that are approximately map 1:1 to some patterns that exist in reality (IE patterns that are roughly local so that we can call them objects). Due to the fact that our mental tools have heavy constrains and that we iteratively adjust these models to fit reality at focal points, we can approximately predict reality, because we already mapped the constrains into the model. But we shouldn't mistake model for the reality. Map is not territory.
In fact, automated regression tests done by ai with visual capabilities may have bigger impact than formal verification has. You can have an army of testers now, painfully going through every corner of your software
A hybrid will likely emerge. I work on a chat application and it's pretty normal that LLM can print custom ui as part of the chat. Things like sliders, dials, selects, calendars are just better as a GUI in certain situations.
I've once saw a demo of an AI photo editing app that displays sliders next to light sources on a photo and you are able to dim/brighten the individual light sources intensity this way. This feels to me like a next level of the user interface.
Sometimes we want to model something in real life and try to use math for this - this is physics.
But even then, the model is not real, it's a model (not even a 1:1 one on top of that). It usually tries to capture some cherry picked traits of reality i.e. when will a planet be in 60 days ignoring all its "atoms"[1]. That's because we want to have some predictive power and we can't simulate whole reality. Wolfram calls these selective traits that can be calculated without calculating everything else "pockets of reducability". Do they exist? Imho no, planets don't fundamentally exist, they're mental constructs we've created for a group of particles so that our brains won't explode. If planets don't exist, so do their position etc.
The things about models is that they're usually simplifications of the thing they model, with only the parts of it that interest us.
Modeling is so natural for us that we often fail to realize that we're projecting. We're projecting content of our minds onto reality and then we start to ask questions out of confusion such as "does my mind concept exist". Your mind concept is a neutral pattern in your mind, that's it.
What I've found is that NuxtJS is miles ahead in DX. In NextJs it feels like their architecture stands in your way while in NuxtJS everything just works.
I think that early in development you should be able to spam a lot of hypothesis and quickly test them and check how people interact with your software. Whether your software makes sense is more important than whether it's fast.
People are also highly unpredictable, so it is usually a matter of trial and error, very often their feedback may completely erase wide sets of assumptions you were building your product around.
It's borderline impossible to do it on mature product, but rewriting mature product to something faster is not borderline impossible - it's just very hard.
Note that it doesn't apply if you just program something in accordance from an rfc where everything is predefined.
Most tests shouldn't be hard to read and reason about so it shouldn't be a problem. In case of more complex tests, you can do it like you would do it during iterative development - debug tests and code to figure out what's wrong - nothing changes here.