I live near the Columbia River where Google, Facebook, Microsoft all have data centers that rely on hydroelectric power. The dams produce clean energy, provide waterway for barges, prevent flooding, allow for recreation, and can back up water at night when demand for electricity is low. As the salmon fisherman will tell you, the salmon have adapted to the fish ladders and are spawning in the Columbia’s tributaries. Yet there are still political forces that want to breach these dams. Nothing is ideal, but hydroelectric energy is about as close as you can get to ideal in the Pacific Northwest.
This is objectively false. Salmon populations are more often than not considered in decline. The fish haven't "figured it out". They marginally persist solely due to intensive management that plainly consists of atrocities for species less popular in the media [0][1], millions spent on fish ladder theatre, and hatcheries.
The dominant mechanism in support of the fisheries being hatcheries that are basically a strangely laundered welfare program for indigenous, sport and professional fishermen each to their licensed proportion.
If you live in the Dalles, and you don't know the Columbia basin damns are an ecological nightmare.. it's unconscionable.
Not to mention the displacement of indigenous peoples from important traditional regions, the complete loss of stochastic annual flows feeding nutrient cycles in the river... etc etc.
Im no eco-activist. I don't necessarily think removing the dams would improve anything for anyone at this point, but if we're going to advocate for doing things a better way in the future - let's be honest: dams are horrible. The dams on the Columbia will eventually result in no salmon, no matter how hard we try. It's inhospitable. They used to run all the way up the snake into Idaho. Into IDAHO. Never again.
It's ok to advocate for energy, and the things you believe in, but let's not be belligerent about it - be honest.
In the Astoria maritime museum (several years ago now) I remember reading a passage that was proud of "..taming the wild Columbia, and turning it into a beautiful series of lakes and streams to wonderfully facilitate shipping and recreation.."
Just in case anyone who isn't from the PNW is wondering at the energy and number of replies OP has gotten about this:
Water quality and the salmon-centered river/stream ecosystem is, like, the local environmental thing everyone gets slammed with in school. Your biology class has you go look for arthropods. Your chemistry class has you go test for pH and dissolved oxygen and fertilizer contamination. Way more species (including plants) rely on the salmon than you'd guess but oh boy do you learn about it in a PNW school system.
And on the other side, the history of the dams is the history of the growth of the Northwest -- "your power is turning our darkness to dawn, so roll on, Columbia, roll on" is in the Washington state folk song and it's not even an aberration re: the region's folk songs / culture. ("Skagit Valley, Skagit Valley, / They would turn you to a mud pond / To run the Coca Cola coolers in Seattle, U.S.A.")
I just thought this would be interesting to contribute because I've found people from other regions sometimes have an ambient awareness more on the level of "Um, I guess there are fish in the water? And runoff seems bad?" and that just isn't possible for locals here, so that's why it's all somewhat heightened.
That's an interesting song in its own right. It was written under commission for the federal government by Woody Guthrie, and it was really about two of his favorite topics: improving the lives of workers and stopping fascism.
"Now in Washington and Oregon you can hear the factories hum,
Making chrome and making manganese and light aluminum
There goes a flying fortress, to fight for Uncle Sam
Spawned on the great Columbia by the big Grand Coulee dam"
"[Woody Guthrie's] 30 days at [the Bonneville Power Administration (the relevant hydroelectric institution)] is considered one of the single most productive bursts in his fruitful songwriting career."
From South Louisiana, it's the same for us growing up only instead of a dam it's the offshore drilling. We learn about the ecology of the delta, then go cheer the Saints in their black and gold jerseys (Who dat!).
I don't always express myself clearly, but I meant that in a more holistic sense. If you were to construct a Venn diagram of remove vs stay in the case of many of the dams in the Columbia river basin, it's hard to objectively weigh the costs and benefits without a perspective. Within that perspective is an implicit number of assumptions that fashion a bias.
The problem is the Columbia is immensely powerful, and as such it's impact on our society or environment with or without the dam create such a incomprehensible web of dependencies that it predicting the outcome would be akin to seeing the future. You can't.
The Elwha is an amazing story, but that dam was in disuse long before it was dismantled. The case for removing it was one-sided.
Back in middle school we went on a trip and saw them doing the deconstruction of the elwha dam. They were around halfway done. We stood on the banks and saw the process of the reservoir turning back into a river. There was around 50 feet of sediment with old flooded trees sticking out. It is hard to overstate the effects these dams have on nature. Seeing the photos of the elwha coming back to life is always awesome :)
Another problem of dams is the accumulation of heavy metals and other toxic sludge behind the dam itself (usually washes downstream in minimal quantities) and the deoxigenation of the watershed due to lack of flow.
On that note (interestingly and anecdotally) many of these dams are inundated with ash and sediment that should have flushed downstream from Mt St Helens!
The impact of these dams is just incomprehensible.
Edit: My comment concerning salmon was poor. The fisherman I know who fish above Bonneville, The Dalles, and even above the John Day dams do catch salmon have good reports, but this is only local and anecdotal. We know there are places where salmon are spawning where they haven't in 80 years which gives hope, but doesn't provide a good overall indicator. The dams have been a tough on the Salmon.
Thanks for that. I hate to nit pick, but the issues surrounding the dams in the Columbia river basin are so complex and frustrating.. I honestly don't even know who or what is right at any given moment. I've torched friendships debating culling - not even sure I was right or they were. It's just horrible. Everything about it.
The dams are hugely beneficial to society. The produce energy and expedite inland commerce, and that makes us all better off. At the same time everything comes at a cost, and the costs are easily just as catastrophic.
I've seen rivers three times the size of the Columbia that have no dams. They are like the arteries of the Earth. They wend and breathe through existence itself. Heave and fall feet in hours. To see this, to be near it is to feel everything upstream. It's like looking into the night sky, but instead of making you feel small it connects you.
The Columbia is dead. I get itchy about it because we've lost that and no one even knows anymore. We long ago muted a voice that used to sing to us, but people will only consider the loss in economic terms - salmon fisheries. We can't even talk about things like humans anymore.
I appreciate your words. The passion and the way you write about nature is great. I connect with it. It does remind me that humans have become numb to the world around us. That we are just animals dependent on the ecosystem and we disturb it at our own peril.
And you haven't even mentioned the Hanford Nuclear Reservation yet.
I volunteered at Audubon's WetNet (Wetlands Conservation Network) for most of the 90s. Their mission was to save the salmon from going extinct. So doing stuff like saving habitat (waterways, wetlands, shorelines, etc), keeping water cooler, opposing new fisheries (the final nail in the coffin for wild runs), etc.
People just don't get how bad things have become.
When GWB was selected President, I just couldn't continue.
I agree there is a delicate balance here but the thing is, if you let global climate change run wild, the salmon, and a lot of other critters, including us, don't stand a chance, with a dam generating electricity and a fish ladder, they stand a chance and so do we.
Hydo, Wind, Solar, and yes, Nuke, probably all are needed to overcome this challenge. Until some perfect thing comes along, fusion or whatever, we will ned to make decisions between the various imperfect options.
> Until some perfect thing comes along, fusion or whatever
Fusion won’t be perfect. It will also require resources, and generate waste.
The truth is that you cannot have something from nothing and any energy extraction system will need materials (metals, rare earths, concrete) and change land use patterns.
Salmon in the Pacific Northwest on the whole is dying. The entire salmon fishery will go extinct. Don't know where you get your information from, but Salmon in the Pacific Northwest is going the way of the cod fisheries in the North American Eastern seaboard.
Don't believe what your "salmon fisherman" will tell you - their whole purpose in life, their raison d'etre is to fish the salmon until every single last one is gone. Then they will move to another place to fish.
The fish have not adapted to the fish ladders and no salmon fisherman will tell you that. The dams do a lot but helping fish or the ecology is not on that list.
Its great that these huge corporations use hydro power for greenwashing, but there is a limit to how much hydro any river can generate and it would be great to use these sources for regular people and/or when there is no solar/wind.
Don't forget, E. Washington has a lot of wind farms going, it's pretty windy, there is a reason the Columbia River Gorge is so popular for wind surfing.
Also, placing server farms in northern latitudes is smart from a cooling perspective, you can use the Economizers on the CRAC units more often when it is cool outside.
Unfortunately, lately there is a push for lower network latency, and in response to that, there are now a lot of Edge Cloud clusters being built near urban areas where various fossil fuel power plants to meet peak loads or load leveling are often used.
Putting these edge cloud server farms in those areas will cause the carbon producing facilities to be activated more often. That low latency comes at a price to the environment more so than the centralized ones positioned in places like E. WA.
That's a strange use of "adapted", there. The logic here seems to be:
"Sure, salmon and trout numbers have gone down -- but not all runs have been completely wiped out. Therefore, they have "adapted" and there's really nothing to worry about. All so-called scientists talking about a 98 percent reduction in return counts since 1880, and warning that some runs are on the verge of extinction due to the combined effects of habitat reduction and climate change -- they have no idea what they're talking about, really."
Is the decline caused by the dams though? From where I'm standing (half a planet away), it's not so clear, they could go extinct and be perfectly adapted at the same time.
We don't need to it to scale to infinity. It's a weird objection. It's like saying human lifespans can't grow to infinity so we should stop trying to be healthy.
The truth is, humanity can continue to be more productive and efficient for millions more years. We can do more with less like we've been doing for centuries.
For example, the US economy is bigger than ever, yet we emit less carbon than we did 10 years ago, both total and on a per capita basis.
> We don't need to it to scale to infinity. It's a weird objection.
Is it? The proposition was that folks should have a growth mindset about consumption. Yet we're already reaching the limits of long term sustainably. I'm not advocating to stop trying to do better. Rather that we shouldn't assume we can indefinitely.
> The truth is, humanity can continue to be more productive and efficient for millions more years. We can do more with less like we've been doing for centuries.
Have we really been doing more with less for so long? Or have we just learned to extract more and more nonrenewable resources more quickly than ever before?
For example if your lifetime worth of waste had to remain in your own yard then how would it compare to those from past generations?
We should grow the good things and shrink the bad things.
That is very easy to say and hard to know how.
I do not think that our economic system depends on economic growth in reality. (It would be sick to depend on economic growth in PPP terms for economic justice) The entrenched power structure's path of least resistance for maintaining their position is economic growth. That is a different thing.
> The truth is, humanity can continue to be more productive and efficient for millions more years
It’s not truth. It’s a form of techno-libertarianism bordering on religion. I can’t find a single working scientist in any relevant field who believes this claim. Most scientists believe we are on our way out, along with the biosphere we’ve destroyed.
I know I'm not going to change your mind because that would require changing your whole worldview, but I would push back on the notion that this is a "techno-libertarian religion". It certainly has experts that endorse the view. A whole field of them. In fact, it has its own term in the discipline of economics.
You can be an expert and a techno-libertarian ideologist, though. Particularly in economics, which is a field rife with unverifiable dogma.
Besides, considering that we are living on a finite world, we don’t need to merely decouple economy from materials, because any use at all will at some point exhaust available resources. Sure, we can say “fuck everything, I’m going to keep going on as usual because in 100 years I’ll be dead and they’ll have that one wonderful technological panacea that will change everything (be it Mars colonisation, asteroid mining, or anything else)”. I don’t know how to call that attitude except for “techno-libertarian dogmatism”.
What I mean is, e.g., when the cost of energy drops people's energy bills in general do not go down, the have warmer homes. That has been known for a while. But is it still true if everybody can keep their houses at 21 degrees permanently? At thatr point does a reduction in energy costs reduce their bills or do they find something else to spend the energy on?
Billionaires tbf are a self-selecting sample of people who made it past $1m, $10m and $100m, and still decided to keep going.
As a % of the population, I don't know how many have those traits - 10%? 1%? - but with a population also numbered in billions, "enough to totally screw the planet" is a reasonable guess.
Really? Like in the next 200 years? That’s ridiculous. The scenarios from IPCC envision a future of prosperity, albeit at a reduced level than if we could somehow drastically cut carbon emissions.
> The current mindset in California is to cut down life style and eat bugs.
I live in CA, the valley specifically. I get why you have this image in your head, but I'm telling you it's not true. There are loads of normal people out here, with a select few very loud voices on certain topics. The problems those loud voices cause societally is likely more than their worth. This is true across the political ecosystem from what I've seen. It's not worth judging large groups of people by the loud and thorny voices among them.
I have friends telling me to not have kids. The entire woke narrative is about regressing life standards to that of a third-world nation, instead of aspiring to lift everyone up.
I'll be voting accordingly this next election, and defect for the very first time from the party that I used to vote for.
I was in a similar position as you. West coast liberalism is not great, I struggle to call it liberalism when I hear some people speak. That said, if you have liberal friends on the East coast chat with them. The culture there is loads better in my opinion.
Doesn't civilization itself induce demand for cheap clean energy given the environmental forcing of unclean energy? I don't see why the bitcoin case needs to be singled out any more than the yacht, private jet, gaming rig, or hell...even personal car cases.
> I don't see why the bitcoin case needs to be singled out any more than the yacht, private jet, gaming rig, or hell...even personal car cases.
Because Bitcoin literally is designed to waste electricity on proof of work. The Ethereum network has the processing power of a Raspberry Pi and wastes fantastic amounts of electricity to deliver that paltry power. Bitcoin alone uses more power than all of Google's operations.
I doubt that the total power used by the world's yachts and private jets per year is even half the power the Bitcoin network literally wastes.
Sorry, this is Poe's Law territory. I honestly can't tell if you're being sarcastic. A blockchain doesn't need some wasteful mechanism to be immutable. Boring old cryptographic signing does the same thing. But Bitcoin et al want their ludicrous "No one can be trusted! Except all the points of trust required to make the system works of course!" system.
> Doesn't civilization itself induce demand for cheap clean energy given the environmental forcing of unclean energy?
Now this is probably not a good comparison, but the first thing that i thought of was something along the lines of: "Burning tires is bad for the environment, so we should burn more tires until some factors force us to stop doing that."
Surely if there's a recognizably bad behavior, we should be able to stop doing it without being forced to, right? Where did the idea of being cautious in service of a potentially better future go? Why do we have such an exploitative mindset?
> I don't see why the bitcoin case needs to be singled out any more than the yacht, private jet, gaming rig, or hell...even personal car cases.
I don't have a yacht and i don't see why people should have them either, outside of sports or transporting many people (albeit there are more efficient ways).
I don't have a private jet and i don't see why people should have them either, maybe apart from transporting government officials in certain situations. Actually, planes in general are overused and people should rely more on trains, which are environmentally friendly.
My "gaming rig" draws around 100-200W in total under load, with a 65W CPU and a GPU that's power limited to 50% of its total power draw for longevity purposes. My homelab servers run off of 35W CPUs with iGPUs and both of them run with a total wall draw of 100W. Honestly, outside of media production, there is no reason for overpowered devices, plus the developers of games and much other software should be more concerned about doing more with less.
And i don't own a personal car, thanks to the public transportation infrastructure in my country not being crippled. I studied for both my Bachelor and Master degree in the city and never felt the need for a car due to being able to take the bus, tram or trolley bus wherever i needed to go. And even when you need a car (e.g. living in the countryside or needing to transport tools or other things), you shouldn't got for wasteful ones.
In summary, the human existence is one of excess, unnecessary luxury and waste. We absolutely should address that.
The opposite is true. People already have this unhealthy growth mindset in regards to energy production, when in fact the only possible way of averting the worse disasters of global warming at this point is to adopt a de-growth mindset. There is absolutely no way to replace all dirty energy with clean energy in the 10-20 years we have left to avoid terrible global waing (>2 degrees), and not even in the maybe 30-40 years left before catastrophic global warming (>3 degrees).
What previous resource intensive activities are these data centers replacing although? We used to ship far more paper around by parcel, and store them in temperature controlled rooms and shops in dispersed locations be it books, legal documents and more. Things that used to be air flights are now video calls. Videotapes, DVDs and video stores replaced by efficient streaming. Photographs stay on digital media instead of paper media. We fell trees and had 'paper' data centers in the form of pulp mills, which are also resource intensive industrial plants in remote locations.
Yes we still have paper and lots of it, but now it's significantly reduced.
When you look at the internet and the telephone compared to what we replaced it with, the internet comes out far, far ahead in environmentally friendliness.
> Yes we still have paper and lots of it, but now it's significantly reduced.
But a la Jevons Paradox [1] I'm sure the total amount of information being shuttled around now is far more than what was being sent around on paper -- bu orders of magnitude. So it's not at all clear that we still have a lower net ecological footprint.
> But a la Jevons Paradox [1] I'm sure the total amount of information being shuttled around now is far more than what was being sent around on paper -- bu orders of magnitude.
In general I'm glad for the easier access to information and being able to answer questions I have and learn things. But how much of that theoretically measurable amount is actually needed?
If we cut the amount of bits transmitted by half, how much would the quality of life decrease? If we doubled it, how much would people's lives improve?
That assumes the same amount of information is being passed around. The ease of the modern internet has led to more work being produced for servers to process. No video site 10 years ago had HD enabled by default, for instance.
I couldn't find a source that says when YouTube started showing videos in 720p by default, but I did find this circumstantial evidence that shows it was after October 2012, so at most 9 years 8 months ago: https://apple.stackexchange.com/questions/68318/view-youtube...
It's not really forever. Realistically it might be 30 years. And do you really think the ecological cost of storing 5 MB of data is more than manufacturing and developing a single photo? One hard disk can store a million photos. Will she ever get even close to 100k photos?
>> We used to print out a few dozen photos a year. Now my wife takes that many photos at lunch time
> Will she ever get even close to 100k photos?
Assuming a dozen photos per meal, 2 meals a day, 360 days a year... thats 8640 pictures per year. In 10 years, she'll be at 86k pictures of food.
For what it's worth, my own photo album has an average of ~7000 photos per year for the last 10 years (and that includes 2020 and 2021 being abnormal years), and I don't use social media. In less than 5 years, I'll cross 100k photos. Actually thats combined photos and videos. The videos take up more space. They used to be 1080p videos. They're now all 4k.
My Dropbox has 1.1M files in in, many of which are photos (and then stupid stuff like thumbnails of the same photos from iPhoto in 2008). However, my impression when they said all the compute and all the laptops is 2%, I'm like, "not so bad." We get a ton of value out of that stuff - music and books and movies and phones and pictures and even adult ed and stuff my parents did with physical stuff and don't have to travel so much because can video and stuff. If they can get right to repair and full-lifecycle closed loops going once (maybe?) the tech settles down a bit more, it'll be a net gain I'm pretty sure.
I am not sure paper usage is down. Overall, humanity is using more than ever.
As for video calls, they don't need massive data centres since it's supposed to be p2p for all sorts of reasons.
I am struggling to get where you come from. The actual content is so wrong, that it seems satirical, yet it's presented in a very serious, almost patronising way
Not everything can be measured with carbon emissions. For example, assuming the paper industry emits as much CO2 as the computing industry, does it mean they're equally polluting? Micro-electronics require hundreds of different chemicals/ore which means polluting many more extraction sites and using their local water supplies... then comes all of the ore refinement processes which itself requires considerable amounts of water pollution. Then of course, there's waste disposal which incurs even more pollution because to this day electronics aren't being recycled (because they're designed for performance/miniatureness not recyclability).
Overall, the IT industry has a much higher ecological impact.
It is, but it's worth making the trade-off explicit. In a day and age where 1/3 of the world population drinks water polluted with lead, and a recent study was published this week on HN claiming out of over 1000 water streams studied only 2 were unpolluted by medicine, the chemical treatments and quantity of water required for some production are key metrics to consider.
I mean it's not that i'm not worried about CO2 emissions. But tackling this problem is super easy and it's just no politician/businessman is interested in the answer (degrowth and low-tech). Forever-chemicals and polluted water sources on the other hand is a much harder and yet mostly-undiscussed problem beyond collapsologist circles.
+2°C is worrisome and will have significant damage on many species. Polluted water sources challenges the very idea of humanoid life in much of the world. It's interesting to note that both problems have the very same origin (industrial capitalism) and the same solution: don't produce shit we don't need, don't produce un-durable stuff, don't produce stuff from cheapest imported/synthetic materials.
Overall humanity is much larger and richer than it was 40 years ago. Paper use includes packaging, plates, toilet paper, etc so it’s hard to get actual numbers as it relates to IT. Especially when recycling initiatives have resulted in new uses for lower grade paper products.
That said, personally I am way down due to IT not just replacing forms but also physical books, newspapers, and even bills.
There are many billions of people on the world that have a lot of room to grow consumption to even come close a fraction of typical developed world consumption.
And some of the typical developed world consumption is ridiculous and should be considered an affront to human decency rather than an ideal for the world to reach.
Of course, that is why the only solution I think will work is making fossil fuels 10x more costly. Without $20/gallon gas/petrol at the pumps and $10k flights, I do not think we see any meaningful impact to environmental problems.
But until then, the music will keep playing and everyone is will keep dancing.
This is only true due to bad protocols and network infrastructure. As long as ISPs have incentives to make Netflix go fast and the connection to your actual neighbors go slow, P2P can't take off. Also multicast networks could easily outperform cloud infrastructure in many setups, but something something business incentives.
I'm not saying otherwise, although i don't have any experience with multicast (i just know people who were playing with that some years back). Feel free to point to some more detailed/critical resources if you have links at hand.
But even on unicast networks Bittorrent (and other P2P protocols) largely outperforms any cloud provider in terms of bandwidth, especially in parts of the world where your "cloud" company has no infrastructure.
What i was denouncing is that many ISPs (at least in France) make sure if you have 10Mbit/s upload it will work fine on a route to Google, but you won't ever get the full 10Mbit/s to your literal neighbor who's using a competing ISP... You'll be happy if you even get 1Mbit/s.
There's no technical reason for that, it's all political/economical considerations (the same that make us have asymmetric bandwidth, no static IP, no configurable reverse DNS, etc).
> Things that used to be air flights are now video calls.
I'm skeptical. What used to be a phone call is now a video chat, which arguably consumes more resources (more bandwidth = more infrastructure).
> Videotapes, DVDs and video stores replaced by efficient streaming.
This is entirely wrong. Efficient streaming uses peer-to-peer connections so if your neighbor has the video you're looking for you won't have to cross half your country/continent to reach a Netflix caching server. The cloud is the opposite of eco-friendly IT which is best represented by low-consumption services reusing existing devices/infrastructure.
>Today, the electricity utilized by data centers accounts for 0.3 percent of overall carbon emissions
So The Cloud, arguably the peak of human technology, uses contributes a tiny fraction of our total carbon emissions. Is curbing this technology in in any major way a productive way to reduce our emissions?
These emissions are all electricity usage converted to carbon emissions. Would it not be more productive to focus on transitioning all power generation to solar, wind, hydro, nuclear, etc?
Same way I feel when someone here tells me I should dither my PNGs or something to save the planet. Not eating a single steak would offset my entire life of using the internet.
> Not eating a single steak would offset my entire life of using the internet.
This appears to be hyperbole - from what I could find, one steak has a carbon footprint of 15.5kg [0] and one Google search has a carbon footprint of 7g [1]. That makes one steak equal to 2214 Google searches. Over a lifetime, 2214 searches seems low. In my browser history, I found evidence of 776 searches in the last 3 months.
Moving away from searches, apparently 1GB of data transfer produces 3kg of CO2 [2]. That makes one steak equal to 5.16GB of data transfer over the internet - or less than one hour of 4K content streamed from Netflix [3].
Those numbers don't make sense to me. I just watched this video[1] in 4K 60fps on Youtube. I think it's ~5GB. I watched it at double speed so it took 15 minutes to watch. Your second link says it takes 5.12 kWh to transfer 1GB, so it took 25 kWH to watch that video. That's about $5 at California power prices. Did it really cost $5 to watch that free Youtube video?
My ISP caps me at 1.2TB/month. That would be $1200/month, but I don't pay anywhere near that.
25kWh in 15 minutes is 100kW. That's 150x the max amount of power my overpowered desktop uses.
On HN people claim AWS overcharges on egress pricing. They charge between $0.05/GB and $0.09/GB[2]. Their "overcharging" price is much lower
(5% - 11%) than the cost of electricity if your numbers are correct, which doesn't make sense. Their ingress price is free, so they're not making up for lost money there.
The source for power per GB links to a paper published in 2012 [0] that describes how they arrived at 5.12kWh per GB. Maybe things have gotten more efficient in the last decade?
Note that the power figure includes all of the networking equipment between your desktop and the server, and the infrastructure required to run it like cooling equipment. Also note that watching a video at 2x speed may result in the server sending less data than watching a video at 1x speed.
I wouldn't put too much weight into that number. Studies of energy usage for data transfer vary by over 5 orders of magnitude depending on their assumptions. This paper [1], which is an analysis of 14 different studies, came up with 0.06 kWh/GB in 2015, with energy usage halving every 2 years. Assuming the trend continued, that would put it at less than 0.01 kWh/GB today.
The claim about CO2 emissions for a Google search was also contested at the time and later retracted by the original source [2][3]. Unfortunately, once on the internet, these things just keep getting repeated.
They are just dividing total consumption of data by total consumption of energy. While I would imagine both have risen substantially, I would imagine that consumption of data has risen by some multiple of consumption of energy.
So I bought Borderlands 3 the other day. It's about 130GB in size. At 5.12kWh per GB and UK energy prices of £0.28/KWh (keeping in mind the gas shortage), that implies downloading the game costs 5.121300.28 = £186. Per download. For a game that cost like £10 on sale.
From skimming the paper[1] they pulled that number from it seems... not the best metric. It looks like they are estimating the power consumption of the internet, then dividing that power usage by the amount of data transferred and claiming that is the cost per gigabyte.
But that's including a _lot_ energy cost that isn't related to _transferring_ the data, but storing/generating/running services to send that data etc
> But that's including a _lot_ energy cost that isn't related to _transferring_ the data, but storing/generating/running services to send that data etc
Well, doesn't that make it an even more accurate measurement of how much our data really impacts the environment? Data doesn't just exist at the moment its being transferred...
Transit providers accept traffic for delivery anywhere in the world at a tiny fraction of that figure, suggesting that the fully allocated cost per GB is not quite as high as they represent. I trust transit service is not in a venture capital funded bubble subject to collapse any day now.
The same analysis goes for storage and processing. The economic marginal cost of storing and generating data as such is quite low, probably lower than the transit cost, especially since most data stored is either soon discarded or transmitted a large number of times.
Both are among the most successful economies of scale ever invented, pretty much the opposite of what the article alleges. Measure this out as a percentage of GDP or transit/storage cost per hour of work or entertainment - it is certainly worth more than one airline industry in economic terms, advances we could hardly live without unless the plan is to roll back living standards to the 1970s or earlier.
Given that just recompressing images on the homepage of some sites I’ve worked on has cut page load size down by megabytes it sounds like it makes a difference.
Meh these "feel guilty" articles don't really convince me. We have nuclear power which solves basically all of this. I guess I'll keep streaming porn on my iphone 13 as the ship goes down although the guys in charge of the ship know what the solution is and how we can patch the hole. Embrace the absurdity.
We have some nuclear power stations built long ago, a regulatory quagmire, loss of production expertise, and Toshiba / Westinghouse atomic power division bankrupt [1].
I'm afraid that making new nuclear power viable will take quite some effort. I hope that this effort will eventually be exerted.
The world is dying, don't come at me with things that we can overcome with effort and ingenuity, when the stakes are "some high costs" vs. the end of human civilization. I get your point but I would hope you know that even the average supporter of nuclear power knows it won't be cheap or easy.
Yep. The amount of power that can be pulled to a building is a difficult thing to change. So the biggest cloud companies try to maximize the amount of power used for compute. There is a metric for this called Power Usage Effectiveness (PUE). It’s the ratio of totally energy usage to energy used for compute. Most modern cloud companies have PUEs <1.3.
The data centers in Ashburn, Virginia are mostly leased facilities and these kinds of facilities are more of the traditional raised floor air conditioning type facilities. However these facilities are also quite efficient with PUEs in the 1.6 range.
I can’t find where that stat comes from, however. As for everything else you said, it would be great to have sources. As far as I know, cooling is one of the major costs of running a data center. It would be illuminating to know why that’s now false.
The 70% thing came from a business development official in Loudoun County, a person whose job is to promulgate misleading boosterism. It doesn't make even a sliver of sense to someone who thinks it over for a moment.
Cooling is a major data center cost, but it's not 40% any more. All of the major datacenter operators claim a PUE of 1.11 or less. Facebook has been the most open and vocal about how they achieve this, for example at https://tech.fb.com/hyperefficient-data-centers/ (scroll down to "Systems").
Not sure what the call to action here is. Purely “ecologically”, I’m pretty sure staying home and binging YouTube is better for the environment than ordering a book online or even a short drive to a local library.
Sure you can document the ecological impact, but who's measuring the economic/cultural/social impact (positive and/or negative)? How do I measure an airline-sized carbon impact against a billion cat videos? Is there even a framework for that kind of accounting?
Until there is I think no amount of documentation is going to change people's behavior in the large. IMO the article is much better when citing specific impacts like noise pollution and water rights abuse.
I've often wondered if a "slow" datacenter could be created, that used low power processors. Amazon, for example, could offer a cheaper AWS tier that used the low power datacenter. Not everything needs to be fast. I'm perfectly happy with many jobs on my computer taking all night.
Because lower power machines can't run as many guests? I'm nothing like an expert in data center power consumption and relative efficiencies, but the last I had heard was that it was generally more efficient to cram a lot of users on high-end hardware rather than trying to run everybody on enough low end hardware to match the same overall throughput.
It's called graviton. Also from what I hear internally (at a subsidiary), price of service is mostly correlated to carbon footprint as power is a large factor in pricing. Yes there are profit percentages / etc but it's not a bad huristic.
I previously researched that USA does not allow private citizens to operate their own nuclear power plants. Why doesn't Amazon lobby against this law, and then build their own energy generation to offset their carbon footprint and reduce power cost? If it really is a big factor of their pricing model, it can be worthwhile, no?
I meant to include private companies too (my bad). These companies are running nuclear reactors for the state not for their own private use, which is what I meant.
Tangent: I've been trying to recall the author and title of a sci-fi short story that had slow computing as a central plot element. The protagonists had organic computers (eg DNA Turing machines or something). Massively parallel. Much slower than electrons. Used to chew thru intractable problems without deadlines. Like cracking encryption.
other than constant dissipative losses perhaps, I fail to see how speed of computation would effect efficiency for irreversible computers. you can use arbitrarily small amounts of energy to perform a computation with reversible computers.
can you explain why racing to finish with a laptop saves energy vs computing the same thing at a lower clock speed?
Google cloud has some data center choices which are indicated as having a cute green leaf. That could mean the building was constructed without using concrete or steel, and maybe techs ride their bicycles to the work compound. /s
In 2007 Google achieved carbon neutrality. I think that's the carbon offsets you mention.
In 2017 Google matched 100% of power usage with renewable energy. So that's not buying carbon offsets, that's buying renewable energy, although AIUI, due to temporal and spatial limitations, it can't be used 100% of the time and in certain situations gets resold as non-clean while the actual datacenter uses non-clean energy.
Google has committed to use actually clean power 100% of the time by 2030.
It seems like these things come up when someone has a axe to grind. Why do we use energy for anything? Because we want to do things.
How much energy do we use to farm sugar, junk oils, grapes for wine, hops for beer that only serve to make people unhealthier. How much energy do we expend to make junk fast fashion. How much energy does the US educational system consume in districts where the average kid can't read. How much energy does it take us to bomb civilians in the middle east and maintain the bases to maintain the capability to bomb civilians in the middle east.
And the answer is a very clear "we generally don't care" until someone has an axe grind.
I'm not justifying these things, just pointing out that I don't buy that these articles are just written out of a genuine interest in understanding and improving global energy consumption.
The problem doesn't seem like that people use a lot of cloud services. That's probably not going to change and probably shouldn't. Maybe we should stop shutting down all the nuclear plants and build new ones and get the problem dealt with if this is such a big problem.
What about calculating the ecological impact of the programming languages running on those 'cloud' servers?
Some of the most popular languages are the least performative (we all know which ones), but they make life easier for developers.
Thankfully, newer languages take performance seriously: Rust, Go, Nim, Crystal. A typed, compiled language gives you good performance for free and a smaller memory footprint (and the reduced computing resources that implies). When you consider the millions of servers in use, that additional language efficiency adds up to a substantial saving in electricity use.
In the computing field, we readily encourage throwing cheap hardware at a performance problem until a program runs fast. It's embarrassing that the "hardware is cheap" attitude is so widespread among developers. Imagine if a manufacturer said that they were going to make energy-guzzling fridges/washing machines or other appliances without regard to energy-efficiency.
In programming, everything is for the ease and comfort of the programmer and anything else takes a back seat - the user, energy use and ultimately the environment.
Aside - If you think programming language energy usage a ridiculous topic, consider the follow:
The creator of PHP, Rasmus Lerdorf, calculated potential CO2 savings due to speed improvements in PHP 7: fewer servers, smaller memory use and reduced CPU activity. (And remember this is an interpreted language.) You can watch a segment from his presentation below where he talks about the calculations he made of potential CO2 savings:
Here are two papers on this topic. In particular, the first is the shorter version (with a better/more journal friendly layout) and second is basically the technical report for the research. It does show some support for your hypothesis compiled languages are more energy efficient. C ranks in at #1 and Python is last at #15.
There is also a paper that I can’t find (I’m on my phone) that was done in conjunction with a Google researcher that reaches similar conclusions.
Pains me to say it over a link which resolves to MIT, but the 'staggering' in the headline is the sort of purple adjective which lowers the quality of discussion, and should be removed.
> In most data centers today, cooling accounts for greater than 40 percent of electricity usage.
Assuming the only power usage in a DC is actual compute and cooling, that would mean 100 kWh of power used would be 60 kWh compute and 40 kWh cooling, or a PUE of 100/60 = 1.67.
Hard to take the article seriously after making claims like that. (Others are also correctly pointing out that the alleged "power usage per hour of video" numbers that this and similar articles like to cite simply don't make any sense.)
" Today, the electricity utilized by data centers accounts for 0.3 percent of overall carbon emissions, and if we extend our accounting to include networked devices like laptops, smartphones, and tablets, the total shifts to 2 percent of global carbon emissions."
Proximity to HQ doesn’t inform the CIAs ability to wiretap. They can just use the Internet and their ability to influence lawmakers. CALEA says every telco has to wiretap their own infrastructure.
The CIA built a giant data center in Utah in the previous decade.
Should we take the claim that seventy percent of the world's Internet traffic both start and terminates in Virginia data centers seriously? Or is the claim that 70% of the world's Internet traffic either starts or terminates in Virginia data centers? If so, that is certainly not the site of the traffic. Traffic exists where it is in transit, not exclusively at the endpoints.
that's ridiculous. you might equally say too much internet infrastructure is optimized for adjacency to neuroscience. there's no other good reason for so much near HHMI Janelia
Ah, right, I was thinking of the NSA datacenter. Still, the fact remains, the CIA is not geographically constrained to Virginia, or even the east coast of the United States.
What? It's not like people are cancelling their flights so they could stay at home and watch a cat video?
I think more often a carbon consumptive habit (like watching a cat video) is replacing something non-consumptive like sitting on the toilet alone with your thoughts instead of obsessively checking how many likes your Instagram photo got.
“We wanna be free! We wanna be free to do what we wanna do. We wanna be free to compute! We wanna be free to run our virtual machines without being hassled by The Man. And we wanna get bitcoins. And we wanna have a good time. And that's what we're gonna do. We are gonna have a good time. We are gonna have a LAN party.”
What we may need instead are artificial lake-tiver-dams that are amenable to -pumped hydro- so as to solve the circadian rhythms of solar/wind. By pumping in the day and letting it run downstream in the night we can solve the capacity issue with a natural battery.
You need a large reservoir at the top, another at the bottom, and as much vertical separation as possible. There aren't all that many places in the world that offer the geography necessary. It can be a good solution in places that do have the correct geography, though.
I wonder how much lower it would be if data centers strictly ran on ARM cpus. Sure there are plenty other components and factors that come into play but it would be interesting to look at such a graph nonetheless.
> Sure there are plenty other components and factors that come into play but it would be interesting to look at such a graph nonetheless.
Not saying switching to ARM CPUs in servers would be bad but it's all of the other shit in a server that's using a significant amount of power. Even if an ARM CPU is more efficient per unit of computation a server still needs lots of fast RAM, various peripheral controllers, and storage. Replacing the CPU would likely reduce system power usage by several percent but not cut it in half.
Agreed. I wouldn't expect a 50% difference. Cloud services offer powerful GPUs, and other capabilities that can consume a lot more than a typical Intel xeon level cpu.
It would be interesting to see a breakdown of power consumption in current data centers.
> Steven Gonzales Monserrate is an anthropologist and a PhD candidate
Maybe this person should stick to studying skull sizes or whatever it is an anthropologist studies these days. I'm sorry to say, but what a load of horse shit.
Yes, data centers use energy. Essentially all of the big cloud providers have committed to carbon neutrality, most of them are pretty far done with getting there, and essentially all of them already source the vast majority of their power from renewables at this point. That's not just because it is greener, but also because it is cheaper.
Amazon, Google, Microsoft, etc. each maintain nice web pages on how far they are with these plans.
That's a trend with many other manufacturers as well. E.g. lots of automotive manufacturers. If there's a deep conspiracy where these companies are lying about what they are doing, that would be news worthy of course. But, I suspect that most of these companies are simply getting very good at managing absurdly high degrees of energy and water efficiency and cost.
Kind of counters the message of the article / the author's world view of course. So, instead this author is dragging in all sorts of things: waste problems, hypertension in people, "computational whir of data centers is not merely an annoyance, but a source of mental and physical harm". I mean, what the actual F** is he going on about here?! What's the point here? Modern technology is bad? Is there a point at all?
> If there's a deep conspiracy where these companies are lying about what they are doing, that would be news worthy of course. But, I suspect that most of these companies are simply getting very good at managing absurdly high degrees of energy and water efficiency and cost.
Believing in a deep conspiracy is probably violating Occam's razor, but I think believing in companies getting very good at gaming metrics and regulators not finding it in their interest to look too closely at positive results on metrics doesn't require too much conspiratorial thinking—everyone's happy in the short term.
* The Jevons paradox seems really relevant but there's no analysis done. The predecessors to the cloud seem important to look at if you're being honest about analysis of efficiency, but...
* "While in technical parlance the “Cloud” might refer to the pooling of computing resources over a network, in popular culture, “Cloud” has come to signify and encompass the full gamut of infrastructures that make online activity possible, everything from Instagram to Hulu to Google Drive." Okay, so we're using a definition of "the 'Cloud'" that extends to the services that run on it. Tricky to evaluate cost-benefit if you're doing that, but okay, sure. "Today, the electricity utilized by data centers accounts for 0.3 percent of overall carbon emissions, and if we extend our accounting to include networked devices like laptops, smartphones, and tablets, the total shifts to 2 percent of global carbon emissions." Sorry, what? So you don't actually mean cloud computing when you say "the Cloud", you mean everything connected to the Internet?
* All of the stuff about the wastefulness of protection against failure seems pretty sketchy to me, especially when considering that the preceding systems had all of the redundancy of every business that needed compute having its own physical servers scaled for peak. But again, there's no consideration of "what did we use to have" or "what would we need if we didn't have this"...
* Putting in a quote like "It festers in his mind, clawing at his thoughts, probing his sanity, poisoning him with a constant spell of dread and anxiety." to color the locals' objection to noise is pretty dramatic, so there must be good evidence for how awful it is, right? "Over the course of my fieldwork with the communities of Chandler and Printer’s Row, I learned that the “noise” of the Cloud uniquely eludes regulatory schemes. In many cases, the loudness of the data centers, as measured in decibels (dB), falls below the threshold of intolerance as prescribed by local ordinances. For this reason, when residents contacted the authorities to intervene, to attenuate or quiet their noise, no action was taken, because the data centers had not technically violated the law, and their properties were zoned for industrial purposes. However, upon closer interrogation of the sound, some residents reported that the monotonal drone, a frequency hovering within the range of human speech, is particularly disturbing, given the attuned sensitivity of human ears to discern such frequencies above others. Even so, there were days when the data centers, running diesel generators, vastly exceeded permissible decibel-thresholds for noise." .... So I'm supposed to read "uniquely eludes regulatory schemes" as... "isn't as loud as the limits for the relevant zoning." Cool. How elusive of them. What are the limits supposed to be? How often are they being exceeded? Why should I give credence to "some residents" on the importance of the frequency -- shouldn't there be studies about that? I live in an area with noise pollution so I am fully willing to take this issue seriously, and I can believe that the existing regulation isn't catching what it should, but this doesn't read like someone being serious about presenting the issue!
* "Historian Nathan Ensmenger writes that a single desktop computer requires 240 kilograms of fossil fuels, 22 kilograms of chemicals, and 1,500 kilograms of water to manufacture. The servers that fill the halls of data centers are dense, specialized assets, with some units valued in the tens of thousands of U.S. dollars." So the implication slithering between these sentences are that servers are worse than desktops vis-a-vis environmental impact of physical production, and desktops are really bad... but that's not actually shown by the fact that servers are expensive.
* "Even with these sustainability initiatives in place, environmental organizations like Greenpeace estimate that less than 16 percent of the tons of e-waste generated annually is recycled." Yes! This is a huge problem! Only... what percent of that is consumer-level and what percent industrial? Is cloud computing less efficient re: e-waste? The author seems to be saying that we should leave things in use until they break, but, well, in practice that's kind of all those standards codify anyway.
Ugh ugh ugh ugh ugh. I really care about the environmental impact of data centers and computing, and I want more people to care, but I want them to have better material to work with than this suggestive hand-wavey piece that isn't even compelling from an anthropological perspective.
One thought: does a server make better use of its environmental cost than a desktop?
Both now run multi-user operating systems, but due to both how OSes are sold and how desktops are managed, only one user will ever use it at a time. Over its operating life, the hardware of a desktop never realizes its potential.
I would have guessed that this would be the case, absolutely. It's not like the cloud makes everything better for e waste; for my home use SSDs are a giant pain because I need to monitor them for wear and replace them and etc etc , but it's so easy to let a cloud provider do it that I'm sure there are businesses cheerfully burning through SSDs who wouldn't be using them for on prem. But overall, for environmental analysis that isn't ideologically committed to going back to rocks and sticks, you have to balance what you get for the impact of a thing--and data centers squeeze a lot out of their stuff.
One thing to add: saying "as the cloud expands its impact expands" just might not be accurate when looking at businesses moving inefficiently run on-prem workloads onto it. I'm no economist, but what a thing displaces matters. The industry estimates I've read about say it's far more efficient than the way things were done, but I don't know if I trust those given their origin. I would love to read good critical content here (anyone have links?) but this isn't it.
The Australian Government circa 2010 introduced a carbon tax... it was one of the first governments in the world to do so. It causes prices to go up, miners to threaten to close mines as costs became too high, basically it was a shit show and the government went to election against a conservative party promising to remove the tax and make it impossible to bring it back, ever, if it won... of course, it won and has remained in power since then. Today, Australia is probably one of the worst offenders when it comes to carbon emissions per-capita, and its Government is hostile to carbon emission reductions except when it costs no jobs, no taxes and no worries to anyone , i.e. never.
I think we can see this scenario playing up again in quite a few other countries, unfortunately... for this reason, I don't think taxes will solve the problem.
The rationale of a carbon tax is to make prices account for the externality, so that people use carbon-emitting energy where it's still worth it, and stop where it's not. (And over time encourage everyone to find better ways to do things.)
You might see this as a quibble, but there's so much misunderstanding. E.g. the #1 objection is about the expense to poor people, but if all this tax revenue went straight back to a UBI it would not affect the above logic. The objection does have force if it's "funneled back into paying for its impact" instead.
>the #1 objection is about the expense to poor people
It amazes me when people who have no interest in the plight of the poor suddenly start spouting this.
Even more so when you consider that the poor aren't creating this problem, their carbon emissions are minimal. In many ways they stand to gain from changes to reduce emissions (improved provision for walking and cycling, better public transport, etc). The poor really don't give a shit if your flight across the Atlantic gets more expensive.
Yup, you're completely ignoring the argument about how to effectively focus everyone's efforts and ingenuity on fixing the problem of carbon emissions, instead substituting class-war logic, and assigning me to the other side of your war. Confirming the misunderstanding I brought up.
A Pigovian tax is not a redistribution scheme, it's a fix-misallocation scheme.
> The poor really don't give a shit if your flight across the Atlantic gets more expensive.
What does giving a shit have to do with what I wrote? Yes, a proper carbon tax raises the price of such a flight; wealthier people disproportionately take such flights, and therefore are paying; as UBI, wealthier people get almost none of that back, as there are few of them. There are probably fewer flights. Plane manufacturers (and competing transport systems) invest in finding more carbon-efficient designs. And the same pressure applied throughout the economy, without commissars decreeing who flies how much, etc., also throughout the economy. Do you want the problems solved?
