Reversible computing[1][2]. & I most highly recommend this fantastic lecture on the Physics of Computing:https://youtu.be/OEzuZ6BpC-g

https://en.wikipedia.org/wiki/Reversible_computing

https://youtu.be/qPgveUgK8P4

I always find it funny when projections decades into the future show straight lines. Even more so when the scale is logarithmic.

Obligatory URL: https://xkcd.com/605/

I agree with the general sentiment, but their point here (as I understand it) is to show that the current trend cannot continue like this. The source paper has a poor choice of word in an appendix, saying "when the energy required for computing will exceed the estimated world’s energy production" with "will" instead of "would", but it's clear from the text they don't believe this. The next sentence is: "Thus, radical improvement in the energy efficiency of computing is needed".

>Thus, radical improvement in the energy efficiency of computing is needed

Isn't that what all these new arm boards are?

In 1894, the Times of London estimated that by 1950 every street in the city would be buried nine feet deep in horse manure.

A human brain runs on 20W. I think we have a lot of work ahead of us

>So a typical adult human brain runs on around 12 watts

http://www.scientificamerican.com/article/thinking-hard-calo...

Why is the world's energy production a constant value from 2010 to 2040? Surely we produce more energy today than we did 30 years ago. Is this just a chart scale issue?

It's not constant, and the scale is logarithmic.

That said I think this prediction is really insane and goes against common sense.

What?! A constant in logarithmic scale is still a constant.

Parent said IT's NOT constant.

As pointed out by JosephRedfern, the production line is not exactly flat but has a very slight grade (http://i.imgur.com/oTUC2Nf.jpg). This means they postulate an increase in production.

>Why is the world's energy production a constant value from 2010 to 2040? Surely we produce more energy today than we did 30 years ago.

Not that much compared to the change in computers from 30 years ago.

We'd probably increase power production before that point, methinks

Or reduce power consumption.

Or reduce computers...

Or reduce the production since the market can be saturated by then...

The Joules/year on the graph is a flat line from now to 2040. It ought to read "does generate" instead of "can generate'.

It isn't a flat line: http://i.imgur.com/oTUC2Nf.jpg. The black bar I added is flat.

Fair enough, but that's still "projected to generate" not "can generate".