Computers ought to be our helpers which could understand our plans and implement them properly. Instead we are burying ourselves and those who will come later in what is essentially a deep math. And our software doesn’t even get faster or better, it just fits whatever performance:annoyance ratio is acceptable at the end of the day.
I think, and this is more of an intuitive rather than an informed guess, that the current practical languages are still too low level and that may be one of big reasons why software development is so complex. You probably feel it too, when you have to either A) go find, copy and fine-tune yet another boilerplate or a snippet, making it again too hairy to use as is next time, or B) use a standard solution which almost implemented another programming language on top of its configuration and isn’t really much easier to use than if it were just a from-scratch code. Shovels get better, but we are still digging careers with them.
And we had predecessors of a theoretical language which could be high-level enough to make programming great again. E.g. FoxPro with all its DOS restrictions was a platform where one could read, write, iterate, update data, never thinking how to map classes to tables or forms, or how to make it run. Or Delphi, in which most students could write client-server apps by dragging tables, queries, databases and forms right from a toolbar. The entry complexity skyrocketed since then or, more precisely, returned to “serious” levels of heavy C++ setups which only seasoned programmers by trade could manage without pulling hairs.
I don’t think that human[ity] will become any significantly smarter in the near future, and that, together with our all time high software demands, means that we will experience a disabling shortage of software developers until we fix our internalized tolerance to the increasing complexity.