I think brewbuck's point is about doing something on those areas where one can intervene and that have a high impact. Assuming the numbers are right, I can surely see his point. And it's an interesting one. Any skepticism won't remove the fact it's a valid point.
Being fatalist about it, or comparing to other areas where we could do a higher impact, will do nothing. The whole point of Action is to do something, not sit around looking at what hasn't been done.
Yet, I do agree that to some extent that hoping software developers will optimize their algorithms is akin to hoping the web will start using less images. It's just not going to happen. Surely there's a desire to optimize code. But this ends when the code performs as required. And that is also affected by other external factors, like an increase in broadband or the average processor clock. Going beyond that requirement is often seen as unnecessary and even risky, considering the possibility of introducing bugs or making the code harder to maintain.
So we need to match our green conscious with what we have been taught during our academic studies and throughout our careers. They aren't just that much compatible. Green in this context, goes pretty much against everything we have bee told; Do not over-optimize, don't fix what is working, do not add unnecessary complexity, keep your code readable.
Ultimately you would need to point the finger at yourself for your use of the computer which is largely composed of moments doing nothing, or nothing deemed meaningful. Turn your argument on its side and you will be facing the same kind of responsibility concerning energy consumption in your house. In the end, there's so much software providers can do. The real use of the energy is what matters. And that is happening right now in front of you and as you read this.
So stop the Microsoft bantering. Makes you look like the rest of the useless bunch out there screaming for a greener planet, but refusing to let go of their comfort, their wool and cotton clothes, and their computers.