It Isn't Just Cryptocurrency Mining

Izabella Kaminska's Just because it's digital doesn't mean it's green reports on:

A new report by the carbon emission think-tank The Shift Project out this week highlights that not much has changed since [2014]. ICT still contributes to about 4 per cent of global greenhouse gas emissions, which is still twice that of civil aviation. What is worse, its contribution is growing more quickly than that of civil aviation.

Cryptocurrency mining is definitely a problem, but how big a part of the problem isn't clear. It could be quite big. Follow me below the fold for some surprising details.

The report warns that:

At the current pace of digital GHG emissions, the total over the same period of the additional digital emissions in comparison to 2018 will be about 2.1 GtCO2eq, which would cancel about 20% of the necessary effort made to reduce them.

Source

So the growth in IT energy consumption is a big problem. Where do the kWh go? The pie chart shows the breakdown. The slices marked P are for the production of various types of IT hardware; the others are for the use of the hardware once it has been sold.

45% of the total energy use by IT goes in manufacturing new hardware, 55% goes into running the installed base. Presumably there is some energy used in recycling, or more likely landfilling, the obsolete equipment but that isn't apparent from the chart.

Source

The Moore's-Law-driven 5-year life of IT equipment is thus a really big problem. And this is another area where cryptocurrency mining makes a significant contribution. Mining hardware has a much shorter service life than other types of IT hardware, typically 12 months. So if the split between production and use costs is the same 45/55% as for regular IT equipment, and the top 5 cryptocurrency's usage consumes as much electricity as the Netherlands (about 109 billion kWh/yr), producing their mining equipment with 5-year life would add more than another Belgium (about 90 billion kWh). Thus with a 1-year life it would add about 450 billon kWh, or a France. This is an over-estimate. Cryptocurrency mining hardware is run flat-out continuously for its service life, whereas other IT equipment has a duty cycle. So its usage proportion will be greater, and thus effect of the 1-year life somewhere between a Belgium and a France.

It is easy for manufacturers to design for a much longer service life, but so far it hasn't been economic. For an example, see Dave Anderson's presentation on Seagate's analysis of extended hard drive service life. Perhaps once people figure out that Moore's Law is dead, and Kryder's Law is dying they will insist on keeping their equipment longer. IT manufacturers, like US automakers in the 60s and 70s, have become hooked on planned obsolescence, so they are likely to respond by adding fins and chrome to their hardware rather than making it last longer.

Source

Overall, the improvement in GDP from each increased dollar of IT spending is dropping. The reasons are likely to be similar to those I discussed in Falling Research Productivity, because the underlying mechanisms aren't specific to research.

The 16% of the energy devote to running networks is likely to increase because traffic, primarily Internet video, is growing at 24% CAGR but technology isn't improving the energy efficiency of network gear that fast.

So don't think that by indulging in "Netflix and chill" rather than driving round to the pub you are helping solve the climate crisis. You need to curl up with a good Kindle book on a really old smartphone instead.