This article is attacking a straw man in which programming is equivalent to writing compilers.
The reality is that almost all of the sciences (and even some of the social sciences) have come to the point where you need to be able to program. Any data-driven field will require you to do data analysis. A decade ago you would need to know math, today you need to know math, and R. This is not the same as writing a compiler, but it is programming. And that's the best case scenario. In many fields, like genomics and ecology, the programming tasks become much more complex.
And that's just the sciences. The business world is also increasingly data-driven, and with increasing amounts of data there's increasing amounts of coding involved. Sure, not every field requires an understanding of programming, but not every field requires an understanding of history, and it's still taught year after year.
I want to up-vote your comment more than once. Just this week, my boss told me to aggregate and then analyze 90 spreadsheets worth of data along with associated metadata scattered around in various (structured) files. The task is one I'm going to have to do once every two weeks. It took my predecessor three days every time she had to do it, three days during which she had no time for any other experiments.
My solution? I wrote a Ruby script in 2 hours, then an R script in 1 hour (to run the stats). They're both flexible, reusable, and efficient. They get my boss her report the same day as my raw data come to me. And since my source data comes from another computer program, the process is flawless and does not suffer from copy errors or improperly paired metadata.
People should not have to do repetitive tasks like aggregation or mapping data to metadata by hand. We should be teaching every high schooler basic scripting and (more generally) that we can outsource the mundane to a computer.
Trouble is that there is a knee in the learning curve.
Everybody works with electrical equipement - it would be helpful if some of the 'non-technical' employees here understood about checking that it was plugged in, and not spilling drinks into it.
It's not clear that a little bit of PCB layout design would help that - just like knowing a bit of logo wouldn't help them with not remembering where they had saved the word file they expect me to find.
But isn't R is already a standard part in the curriculum of undergraduate social sciences degrees ?
I think that part of the issue is integrating programming not only in one course , but as a general useful problem solving technique, that one naturally thinks of using when faced with relevant problems.
The reality is that almost all of the sciences (and even some of the social sciences) have come to the point where you need to be able to program. Any data-driven field will require you to do data analysis. A decade ago you would need to know math, today you need to know math, and R. This is not the same as writing a compiler, but it is programming. And that's the best case scenario. In many fields, like genomics and ecology, the programming tasks become much more complex.
And that's just the sciences. The business world is also increasingly data-driven, and with increasing amounts of data there's increasing amounts of coding involved. Sure, not every field requires an understanding of programming, but not every field requires an understanding of history, and it's still taught year after year.