> The mobile web as a whole has gotten faster due to network speeds+cpu improvements
Making something faster by throwing more hardware at it doesn't meaningfully count as making something faster IMO, you can make the most inefficient piece of software "fast" by throwing the biggest CPU and network you can find at it.
The real issue is why should an otherwise capable CPU from 5 years ago struggle to render the average website today, when it really shouldn't be that hard. Scrolling through someone's marketing website _should_ be a painless experience on even a low-end budget phone.
I don't know the direct answer, but I have a deduction in my head to work with...? I'll just put it out there: I think speed is deeply impacted by high-level frameworks that parse or compile at runtime. React Native is a framework on Android, which is a framework on Java, which compiles at runtime. IDK if anything in that chain compiles down to Assembly or machine code before you open an app made in React Native. That tied with the bloat of JIC background services sitting idle eats bandwidth. Garbage collection checks operate in a loop, checking again and again if all these unused but loaded processes still exist at their addresses. And when you pile those on each other, it seems relatively easy to see how modern CPUs don't seem much faster than chipsets from 5-6 years ago.
I stopped programming around 8 years ago because I hate the current MVC model most software is created and maintained with. What got me interested recently in dipping back in was a video on branchless programming. I love the idea of unit testing at the machine code level for efficiency, and then figuring out how to trick the compiler or runtime and the chipset into making quick, predictive outputs to reduce idling on branches or making 15 steps for something doable in as little as 4.
That feels like a completely opposing direction to take given the current priorities of engineers across almost all industries, even oldtime ones like Gaming.
> The real issue is why should an otherwise capable CPU from 5 years ago struggle to render the average website today, when it really shouldn't be that hard. Scrolling through someone's marketing website _should_ be a painless experience on even a low-end budget phone.
Because marketing/product/design decides to add bells and whistles. Optimization is also not zero-cost effort. The business has to pay for it. I would assume that most business don't think it is worth it.
>The real issue is why should an otherwise capable CPU from 5 years ago struggle to render the average website today, when it really shouldn't be that hard.
That depends on what is a capable CPU. I would say even iPhone dont have an Capable CPU 5 years ago today. ( iPhone 6, iPhone 6s not released yet )
And it is even worst on Android, and the current state of things aren't much better. Hopefully ARM will catch up in the next 5 years, but that means it will take another 5 years to filter down to market.
iPhone 6 came out in 2014; iPhone 6s on this day five years ago. Both are capable smartphones; the latter being capability enough to support the latest version of iOS.
Making something faster by throwing more hardware at it doesn't meaningfully count as making something faster IMO, you can make the most inefficient piece of software "fast" by throwing the biggest CPU and network you can find at it.
The real issue is why should an otherwise capable CPU from 5 years ago struggle to render the average website today, when it really shouldn't be that hard. Scrolling through someone's marketing website _should_ be a painless experience on even a low-end budget phone.