[allenu]

I completely agree. It's easy to pick apart legacy code and say what you would do differently, but it's starting from a place where we have learned a lot of knowledge about the product we're trying to build and the architecture we actually need. When you start writing code, that is when you have the least knowledge about what your target product is and the least knowledge about how best to solve it. Both of those things are only discovered over time, with multiple iterations, by which point your original design may be ill-equipped to handle the new world you find yourself in.

Of course, with experience, you get better at designing solutions to add some "optionality" in case things change, but it's a trade-off that you can't always make, nor should you always make.