[1]http://online.wsj.com/news/articles/SB1000142405270230402680...
For instance, I think house construction is overpriced because I think "what's so hard about putting up walls around some ground?". The answer is everything is hard about it. But that it is hard is incidental. By that I mean, it was harder before, and now it is easier, and it should get even easier (there are 3D printers printing houses now) and therefore cheaper.
A non-programmer thinks the same way about software as I do about houses: "what's so hard about telling the computer to do grab this data and transform it?". Well, all of it is hard, but again, it's incidental. It would have been more expensive in the past because it was harder, and it will be easier in the future (Docker, NixOS, mathematically proven programs that do not require maintenance, etc).
It's nice being a programmer today but we won't ride this forever. As we automate ourselves out of the market, it will be a net benefit to everyone (as no one will have to program as hard or as much as today).
On the other hand, something that hasn't changed in 40 years is that end users are not able to precisely explain what they want. That is why we will always have jobs. There will always be some sort of exceptional rule.
Complexity expands so as to fill our current mental capacity.
Each level of abstraction divides the cost per feature and so multiplies the sum of total complexity.
Unions are obviously one solution but can be wrought with other issues. Besides that, it seems there really is no solution to wage stagnation outside of major failures or security breaches which hit the business' pockets directly.
I don't just mean in IT but it does seem to be an area where salary growth doesn't match up _at all_ with the amount of value it adds to a business, especially in the case of IT within a non-IT company.