Saturday, August 06, 2022

Is streaming really so bad for the environment?

There has been a plethora of articles recently about how bad streaming and subscription services are for the environment. Just one example is this one in The Guardian late last year, which makes the alarming claim that the carbon footprint of watching Netflix's top ten popular programmes is equivalent to driving a car out somewhere past Saturn.

Setting aside the fact that this calculation is, of course, for ALL of the mamy millions (billions?) of Netflix viewers, and so is an unnecessarily sensationalist claim (Guardian! Honestly!), it seems that these allegations, and those in many another similar article (for example, the claim repeated by several quite reputable sources that constant streaming of the 2017 hit Despacito on YouTube generated more carbon dioxide than the carbon footprints of 5 African nations added together, may not be as transparent and realistic as they seem. 

The whole process of streaming - from storage in huge datacentres, to transmission over WiFi or broadband, to watching or listening on various devices - uses electricitity, like pretty much everything else we do, and so it is perhaps no surprise that it generates a substantial amount of greenhouse gases.

Netflix itself releases estimates of its carbon footprint (as does YouTube, and a few other major streaming services), which appear relatively modest, not that I would necessarily trust Netflix's word over The Guardian's. But Europe's Carbon Trust has estimated that an hour of watching Netflix generates about 55g of carbon dioxide, the equivalent of about 300 metres of car driving. Given that people spent 6 billion hours watching the top 10 shows on Netflix (Netflix's own figures), that is where the figure of 1.8 billion kilometers in a car (roughly the distance to Saturn) came from.

But, hold on, 55g of CO2 is not actually THAT much, when you contsider that eating an 8oz steak generates over 2,000g of CO2, as does drying a single load of laundry, and an hour running on a treadmill can generate about 900g of CO2 (these figures are taken more or less at random from LiveScience).

The reputable organization CarbonBrief has taken it on itself to fact-check some of the more egregious claims by the popular press, particularly the widely-disseminated one that watching half an hour of Netflix is the carbon equivalent of driving 4 miles (6 km) in a car. It turns out that this figure (1.6 kg of CO2) itself came from a July 2019 report by a French think-tank called the Shift Project, which actually revised its own figures downwards 8-fold in June 2020 (having mixed up bits and bytes). I doubt The Guardian bothered to report that.

But CarbonBrief identified a bunch of other flawed assumptions in the original Shift Project calculations, including the average bitrate of streaming, the energy intensity of data centres and delivery networks, etc, concluding that Shift Project's original estimate of the carbon impact of half an hour's streaming (the source of that "driving four miles" headline, remember) was probably over-stated by around 80 times, or 90 times if we use a global average electricity production mix rather than an American one.

What we are left with, then, is the realization that streaming music and video entertainment is actually a pretty modest activity in terms of its carbon footprint, certainly compared to gaming, bicoin mining, etc. In fact, similar to boiling a lettle for tea halfway through. It's certainly not something need to be focussing our climate anxiety on.

My other conclusion? Be wary of extreme claims, even by relatively reputable sources. After all, extraordinary claims require extraordinary evidence (thanks for that, Carl Sagan).

No comments: