Hacker News Reader

Comment by nonbel | Hacker News Reader

Its amazing to see how much people waste in general. I did an analysis the other day and came up with ~ 3 TWh per year wasted just because reddit wants to use "fancy" (annoying to me) layouts:

>"So about 3 TWh extra electricity is being used each year. This is about the total used each year by Papua New Guinea, which is ranked 133/219 in electricity consumption by country." https://news.ycombinator.com/item?id=17449349

View on HN

At 10c/kWh that seems to imply an annual energy bill of ~$300MM just for the extra bandwidth? There must be a math error in either your math or mine because that seems like an insane amount to spend just on electricity for hosting a mid-size site like Reddit. (It's ballpark 3x their estimated revenue)

I'm pretty sure that somehow everyone who visits the site is paying for it.

That is really interesting to think about, however isn't a bunch of that going to be cached? IE the JS doesn't get loaded every page view, just the content.

And now I need to go down a rabbit hole to find out the cost / GB currently. I would expect that it has gone down in the last 6 years.

Please post your analysis, I'd love to see a more in depth one.

Very inaccurate, 95% of the requests would have cached those js files firstly, and secondly that is so mislead-ed its not even funny. they took the average cost of a DC and equipment in and and then added all Known dcs together to get to a magical number and say the only purpose of those DC's are to serve the internet. That paper is so inaccurate. The power cost of delivering 3 MB extra is minuscule on 99% of the networks. Only a hand full of people who use satellite only connectivity would suffer from 3 MB of extra data.

Man I hate analysis like these. At least yours isn't as flawed as that guy on reddit who didn't understand the difference between a Watt and a Joule (or kWhr), but what you did is still incredibly basic and ineffective. You aren't accounting for reddit simply frontloading the first page view, compared to old reddit reloading pages each link you click, or countless other possible optimizations.

Just measuring single page-loads using a browser extension is far to simplistic of a metric.

>"Just measuring single page-loads using a browser extension is far to simplistic of a metric."

All I know is the site has become almost unusable on an old tablet I have, it sits on a loading screen for tens of seconds before loading some simple text content. So whatever they are doing is not very efficient.

I'd love to see better analyses along these lines though. How many TWh per year on pron, how many on netflix/streaming video, how many on drinking/partying. There's all sorts of unnecessary stuff we have no intuition for.

Reply on news.ycombinator.com