I mean, if we only had plain old HTML for websites, all these people writing web applications would just be writing desktop applications. The complexity of the problems being solved wouldn't change, and it would basically be the same hell of large projects being done by large teams with little in the way of planning.
The main difference would be that nothing would run on OS X or Linux, because it would be too costly to support. Microsoft would have 99.9% of the OS market, and it would cost $300 a month to have a license. (Remember when you couldn't do online banking on Linux because they all used ActiveX? I remember.)
Basically, the field of software engineering wouldn't be any different. There would be different winners and we probably wouldn't have phones. But you'd still be using slow, confusing, crappy UIs, because that is the human condition. The programming language / platform wouldn't change any of that.
> all these people writing web applications would just be writing desktop applications
Good.
> The complexity of the problems being solved wouldn't change, and it would basically be the same hell of large projects being done by large teams with little in the way of planning.
That's a non sequitur... The complexity largely hasn't changed, it's been hidden by layers upon layers of libraries that aren't fully understood by the people building with them, even within the same company who built the library, and increasing levels of bloat. If it weren't for the never-ending pace of "marketing says we need a new UI" bugs might actually get fixed once in a while.
Imagine if people went from "we need shared libraries for disk and memory efficiency" to "we need completely separate environments for every module of our program" in 20 years. Oh wait, that happened.
There are plenty of non-web desktop apps that also never release any features, and whose engineers use the time saved from not dealing with "npm has found 600 security vulnerabilities that you can't update" to migrate their build system from Makefiles to Bazel or whatever. It's all the same -- the human race as a whole doesn't know how to deliver features with high velocity beyond a certain scale.
The proprietary desktop apps I use basically get updated every few months with what seems like about a day's worth of work for one engineer. (I use Fusion 360, which might actually be a web app in desktop clothing, Datagrip, and Simplify3D. Fusion 360 adds features and fixes bugs occasionally, but mostly each update tightens the screws on what features are available on your current payment plan. Simplify3D seems totally dead, I'd be surprised if they have any employees. Datagrip has a few updates a year, and they are things like "we fixed a bug when talking to some database you've never heard of via two tin cans connected with a piece of string". If you don't like a piece of software as it is right now, definitely never buy it. I do like all of those things, but they have definitely not discovered some sort of toolchain-based productivity miracle.)
The main difference would be that nothing would run on OS X or Linux, because it would be too costly to support. Microsoft would have 99.9% of the OS market, and it would cost $300 a month to have a license. (Remember when you couldn't do online banking on Linux because they all used ActiveX? I remember.)
Basically, the field of software engineering wouldn't be any different. There would be different winners and we probably wouldn't have phones. But you'd still be using slow, confusing, crappy UIs, because that is the human condition. The programming language / platform wouldn't change any of that.