I see software literacy as a tool which will become increasingly relevant as the world starts wanting more diverse software. You will start seeing florists, dentists, architects, and foremen writing programs to solve issues which are not complex enough to warrant a professional software engineer, or which require so much domain knowledge that they couldn't be written by most software engineers.
That's really what this is about, in my opinion; at least personally I wouldn't be so misguided to think that the point of software literacy is to churn out a country full of specialist software professionals.
One could argue that if you're able to use the computer to catalogue and enhance data, turn data into information (like visualizations) -- then you're able to use a computer. A lot of this can be done in Excel (or another powerful spreadsheet program). But then most will also tell you that spreadsheets are visual programming tools. And that you might need to do (and automate) transforms on data that still need some form of "actual" programming -- be that VisualBasic, PowerShell, Awk/Sh -- or "proper" programming in a "proper" language.
Another side to "everyone should know how to program" (a side I find less interesting) is the growing need for software engineers, as we move to computerize more and more both public and private sector "systems". I think the lack of engineering talent is real, but that's not my main motivation for showing kids how to code -- just as I don't teach kids how to do math because I hope they'll work as mathematicians.
Math helps you think, helps you invent, and so can computers. If we want the next generation to be impatient for the kind of programs that we had, like Sketchpad. Or even an environment like Smalltalk -- I think teaching them to program is a good start. It's also a great way to educate them how crappy (and simplistic as opposed to simple) many of the "apps" and games are.
For more, see eg: "Doing with Images Makes Symbols" by Alan Kay
Meanwhile, the STEM skills that are utterly necessary in today's world - statistics, linear algebra, logic, computer programming, electrical engineering, chaos theory, ecology, molecular biology - are largely absent from the traditional curriculum.
But then, that's the problem with top-down planning. You end up with systems adapted to the prevailing conditions at their birth, and nobody willing to stick their neck out to change them.
Encouraging people to play with code and telling them they're going to help fulfill a shortage of software developers are two different things.
See eg: Dropbox: https://news.ycombinator.com/item?id=8863
For what it's worth, I still think BrandonM's comments are spot on. But I think it was also obvious that easily "dropping" files "in the cloud" was a great service to those that weren't inclined/could set it up for themselves.
You don't have to write a web server in Perl, you can write a script that saves people from editing some horrible text file in Word to remove extra lines.
Similarly, some apps are complex 3D games, and others are timers that let you know when your egg is cooked, or your break is over.
Start small.
The problem I see with the push for everyone to learn code is that it doesn't just expect people to have basic literacy with computers, but wants large numbers of them to choose programming as careers, if only they could be convinced how rewarding it is (financially or otherwise).
To use your examples, everyone should have basic literacy in finance, the law, mental health, and writing. But not everyone should become a lawyer, a finance professional, a psychiatrist, or a novelist. In fact, for many of the aforementioned professions, you could argue we have too many people practicing them already.
Similarly, you could argue we don't need more coders who only want to do it because someone told them it's financially rewarding or is the right thing to do.