[NickM]

The Reddit analysis is pretty suspect - you can't assume that power usage scales anywhere close to linearly with data usage. If you look at all the devices connected to the Internet, I would bet that a huge majority of the power consumption stems from time that desktop computers/servers/switches/etc are either idle or underutilized.

[nonbel]

>"you can't assume that power usage scales anywhere close to linearly with data usage."

As noted in the original post, its based on 5 kWh per GB of data transferred as described here: https://aceee.org/files/proceedings/2012/data/papers/0193-00...

If you have a better method, I'd love to see it. This is back of the napkin stuff, it doesnt need to be perfect.

And using he same methods, HN is ~70 KB which is decent for a text-content site. The OP wired page is about 700 KB (but wired.com is 4.3 MB), slashdot is ~1.9 MB, and reddit now like 4.3 MB, etc. Pretty much the entire internet is over 90% unnecessary junk at this point even if you allow that the content has value.

[NickM]

This is back of the napkin stuff, it doesnt need to be perfect.

It doesn't need to be perfect, but if the result is off by an order of magnitude or more than it's pretty useless.

If you have a better method, I'd love to see it.

I'm not saying I have a better way of answering the same question, but that doesn't make your analysis more valid.

As noted in the original post, its based on 5 kWh per GB of data transferred

Yes, I believe that if you take the entire amount of data transferred by the Internet, then divide it by the total amount of power used by all internet-connected devices, then you get 5 kWh per GB (or at least you do if you base it off of the data available in 2012 when that paper was written). But that doesn't mean that if I download an extra GB of data that it took anywhere near 5 kWh to do that.

That paper you linked includes things like "total power consumed by all connected desktop computers" as part of the power budget, but this is obviously flawed reasoning. There are many things that computers do besides transferring data, and even if a computer is sitting idle it consumes a good deal of power as long as it's switched on. If I spend 1 kWh of electricity playing an offline computer game for an hour, then download a 1kB file, that doesn't mean that it costs me 1kWh/kB to download data.

[nonbel]

We can check from another perspective. Apparently the whole internet is around 70 TWh/year from datacenters alone in 2016 (link to actual study is broke): https://www.forbes.com/sites/christopherhelman/2016/06/28/ho...

Reddit is apparently the 3rd most popular website: https://old.reddit.com/r/technology/comments/8n959q/reddit_j...

No idea where this info comes from but it gives 1.8 TWh/yr for (the now fourth most popular site) facebook in 2016: https://www.statista.com/statistics/580087/energy-use-of-fac...

Does it make sense for total reddit usage to be ~3.5 TWh/year in 2018? It seems to be in the right ballpark, but who knows how circular these calculations are.