There is an inflection point coming in 2014 where it will cost more to make transistors smaller, so the cost of transistors will likely flat line as manufacturers stay at the current size according to this report.
In previous years people could make software and in 2 years the same software would run faster on a new machine that cost the same as 2 years prior. So up until now software developers have taken for granted that they can write software that is not that efficient and it is more economical to buy faster computers in 2 years than take more time to make the software more efficient. This was Microsoft’s strategy with Windows and Office for a long time. They built more features rather than make everything super fast at the moment of launching the software.
I think after this inflection point it will be up to the people making software to make their software cheaper to run over time and get more efficient. Software developers will not be able to take Moore’s Law for granted. This will require more diligence on software developers who might just be trained data binders. Software developers who argue that they made inefficient software because it was the fastest thing for the programmer to do will lose their value in the marketplace.
Some developers currently justify slower less performant software because it was quicker for the programmer to code. That is pervasive at a lot of software shops now (at least from the ones I’ve observed) and that will become a more costly practice in the future. The costly practice was masked before because companies could keep upgrading their computers every couple of years for faster ones at the same cost they did years previous. Except, after the upgrade the same program magically runs twice as fast.
The mask will be lifted after 2014 though, if this report holds true. It will be an opportunity for some programmers to differentiate themselves from data binders. It will also be an opportunity to make new tools for data binders to be more efficient.