Something that the article does not mention is that most of the large cloud providers are committed to 100% renewables when it comes to their energy usage. E.g. Google has been working on this for ages and most / all (?) of their data centers are running on clean energy.
Most of the non sustainable energy consumption happens on the consumer side where most of us are dependent on grids that are also shifting to more sustainable energy sources but still have some way to go.
Same for Microsoft, I think they intend to be carbon negative by 2025. And still there's people in the Netherlands complaining that datacentres use up too much of the green energy available in the Netherlands. Truly backwards way of thinking
> still there's people in the Netherlands complaining that datacentres use up too much of the green energy available in the Netherlands. Truly backwards way of thinking
Analyzing and critiquing the moves a massive monopoly -capitalist entity is ‘backwards‘? This sounds quite un-nuanced. Could you elaborate more what you mean? As it stands it sounds like a naive accelerationist perspective.
It is backwards because companies deciding to go CO2 neutral by using sustainable energy sources are contributing towards the Paris agreement. The same people who are complaining the loudest about the datacentres using green electricity, are the same people who would be complaining about the datacentres using fossils fuels, and the same people who use the services delivered by these datacentres.
Companies are facing backlash, yes, for using green energy. How can a company ever go green as long as this kind of thought patterns are mainstream
> How can a company ever go green as long as this kind of thought patterns are mainstream
It's fundamentally impossible. Capitalism is what brought this ecological madness down on it, because of its shortsighted profit incentives. The very same system cannot fix the problem, because nothing has changed.
Pointing out companies who used fossil fuels was intending to allow us as a society to realize we need to reduce consumption/production, not pretend we should shut up and live happily as long as we paint every single polluting thing with a green chemical glowing paint.
There is no such thing as green capitalism and there will never be. Consumerism, private property and competition is what destroyed this planet. Whether we go on and destroy new ones (hello Musk, Bezos and other psychopaths) or destroy the capitalist system depends on our understanding of those issues.
I fail to understand why you bring capitalism into this when it's the most efficient way of dealing with pollution, since it has a pricing mechanism for raw materials, pollution taxes, etc.
I've lived in the socialism on the USSR and seen what it looked like, it's day and night compared to how free countries deal with pollution. Guess what happens when raw materials are available for free, the economy is inefficient because why bother, and pollution doesn't matter because citizens can't vote anyone else into the office because there's only a single party. China happens. Rivers downstream from textile factories colored with dye-of-the-season, wire fences near cement factories opaque from dust, transport workers spilling diesel fuel into the ground for decades to hide slacking and sub-quota mileage, etc. I've seen it all with my own eyes.
When the first Western experts gained access after the collapse of the Soviet Union, they were utterly devastated at the scale of environmental pollution. They had never seen anything this bad. The solution to everything had been "dig a ditch and throw it in", and if anyone complained, arrest and beat them until they shut up, or send to a labor camp if they don't.
The main problem in the West is similar socialism for the polluters, albeit on a much smaller scale. They get to pollute for very cheap, mostly abroad in China. They should be taxed for what raw materials and their pollution actually costs instead of the current artifically low levels achieved through outsourcing to lawless shithole countries, and environmental damage from their actions would be greatly reduced.
> I've lived in the socialism on the USSR and seen what it looked like, it's day and night compared to how free countries deal with pollution.
It has more to do with free countries than free enterprise. Private businesses used to be notorious for stripping resources from the environment (e.g. strip mining, clear cutting) to keep the cost of resources low and they used to be notorious for dumping pollutants into the environment since they did not have to deal with the costs of pollution. It was public pressure that resulted in environmental regulation, and even then (as you mentioned) they try to circumvent the costs by outsourcing to jurisdictions that are more lenient.
I agree wholesale with you but would like to point out that in "capitalism" prices for raw materials are often so cheap as to be called free not only in lawless shithole countries (see the films "Tapped" or "Gasland" concerning shale and bottled waterwater industry), even fines for polluting are often a slap on the wrist for hugely rich corporations (who are too big to fail ...) Swiss billionaire Stephen Schmidheiny is a rare example of someone being held accountable...
But more to the point "The Corporation" makes the case that it's because a change in the law during the civil war let businesses have the same rights as citizens that such oversight occurs..they have more resources than individuals and often cause speculative havoc (ie buying up properties to make a Airbnb empire)
I understand your emotion and living in Romania I see the abandoned conglomerates and am glad that period is over...however if you discount the risk of chernobilian disaster (sea of Azov etc) that the lack of media and public accountability made possible I'd wager that capitalistic societies is powers of magnitude more polluting simply because people live behind paid layers (producer->wholesalers->distributors->retailers->consumer) paying for it by living in cities, working for buisnesses serving other buisness etc)
Manufactured boredom (from living in cities) ->Advertising->fashion/fads->brands-> excess packaging is also a thing and part of the problem...
Then there is the fact that businesses once created to solve a problem in the most efficient way often self perpetuate and bloat with govt closing their eyes to monopolistic behaviour in the name of customer pricing
In the city I live in, you can buy food from people from the countryside, rent/property is still cheap , so if you live simply you hardly have expenses and can avoid wasting your life working away while hardly seeing your kids grow up...
I understand you lived in a failed system but you shouldn't throw away the baby with the bathwater, authenticity counts, real food counts, generosity and food that is less processed or sprayed (because cost) counts
I feel having grown up in a WIERD (acronym) people there forget that the apparent sophistication is just a sheen and like the 300ml plastic bottles they constantly consume they are choking the rest of us on their garbage (ie the developed world using x times more than the rest)
I think you can see where I'm going with this...the problem with capitalism atm or the way government treats it is that it inherently needs growth, whereas the planet needs "decroissance" a french word for "ungrowth" a french politician, (Yves Cochet )I think, said something like our economy has to "ungrow" 10% per year , it's possible even without covid, but it certainly needs a whole new mindset or a drastic realisation that droughts are going to get much much worse...
If I'm not mistaken, the backlash wasn't because MS was using green energy, but rather the Dutch government had promised green energy to its citizens, and then when the wind farms were built, in a surprise move MS announced that it's building servers to use the energy (that was initially intended to go to the Dutch public). Basically the server was a "convenient" way for the Dutch govt to technically build new renewables as per the requirement (also see: the Urgenda Climate Case against the Dutch Government) while the new (and otherwise unplanned MS server) used up all the new energy.
It's in no way bad that the server uses clean energy, but the critique is that the government wriggled their way out, Microsoft got to boast about servers running on clean energy while the Dutch citizens were still stuck with the older CO2-intensive power source mix despite the govt's "commitment" to cleaner power.
The biggest problem is them using up billions of government subsidies which were supposed to be used to decarbonise existing energy needs, not attract new ones.
> most of the large cloud providers are committed to 100% renewables
I’m very skeptical of this. Until now this narrative has often turned out to be nothing more than corporate greenwashing, without tackling the root causes of the problem or helping shift the underlying systems. Take the example of Microsoft buying up renewable energy in Holland: https://youtube.com/watch?v=OiPoR9OvD0Y (Nederland als harde schijf | Zondag met [Arjen] Lubach)
I think it's probably important to point out that while Google may purchase 100% of their energy usage from renewable sources, this does not actually mean the datacenters run on renewable sources: https://www.google.com/about/datacenters/renewable/
I bring this up as renewables are currently supplemental and don't replace base load. Hydro storage isn't available/practical for all regions, nuclear has been deemed too expensive, so it's usually natural gas that slots in.
> this does not actually mean the datacenters run on renewable sources
Yes, this is a problem.
Let's say the grid uses 100 imaginary units of power per year. 50% is currently produced through solar and wind.
Company A buys 50 units and company B buys 50 units a year, until company A gets a crisis of conscience and decides to go green. They manage to get a deal whereby they pay for a certificate that says that their 50 units is produced through renewables. Great.
Company B, however, has customers who care less about green stuff so they just buy energy from the grid and "their" electricity is now just produced from fossil fuels.
In this case has anything really changed? It's probably a good thing that Google is buying renewable energy: I expect it raises demand for renewables in some way and it's good politically for companies to compete for green credentials. But it doesn't seem to do that much.
Isn't it worse than that? If both companies purchase 100% renewables from the energy supplier, do they say no to one?
I disappointingly found out that my carbon neutral tariff doesn't guarantee a purchase of renewables, it just guarantees that any energy I use that is over their wholesale bought amount of renewable energy will be carbon offset, which is complete crap.
Is the datacenter disconnected from the electric grid? Does it not have huge tanks of petrol to run backup generators? Was the datacenter built from locally-sourced materials? Those are the easy questions.
What "renewable" energy does it run and how was it produced? What's the carbox/fossil profile of those energy sources? How long do they need to run, and in what conditions, in order to offset this initial fossil energy invested? Those are the medium-difficulty questions. And as i much as i hate Michael Moore's rhetoric, "Planet of the humans" does a pretty good job of explaining why these questions are important.
What is the ecological impact of the cloud? Why do countryside folks have to have their data run across the country just to exchange messages with their neighbors, and what's the impact of that? What's the impact of a song that you downloaded from Youtube (replicated, dedicated centralized infra) instead of Bittorrent (decentralized, already-in-use hardware)? What's the impact of streaming this same song over and over? What's the impact of online advertisement and web trackers? How many people are working in concrete offices in a city in order to maintain that hellish infrastructure? Where does all the sand and minerals come from to build all that infrastructure as well as all the electronics you find in there? Those are the hard questions and when you answer them you'll never fantasize about green capitalism ever again.
In case you're interested in such topics, and what to do about it, but can't stand for Michael Moore, i strongly recommend the "End:Civ" documentary: https://youtube.com/watch?v=L4ccAJJrrjM
The transport of messages across the country or even the globe is, like shipping things around the world in a big boat, surprisingly cheap compared to the cost of the endpoints.
You can get a lot of containers on a boat and a lot of bits down an optical fibre. The cost of the crypto on the other hand, that adds up a bit.
Also BitTorrent uses a bunch of power on a bunch of endpoint servers which are often less efficient than custom silicon. Tricky to calculate.
I have wondered about this too. They say it is green, but for that to be true due to renewables being intermittent shouldn't there be at least sometimes when there just isn't enough green energy? And thus shouldn't they close or stop getting delivery of energy at these times?
Also, how green are their backup power generation? Outside batteries?
Until we have more renewable generation capacity than we need, any Watthour of renewable energy consumed at Google's datacenter is a Watthour of coal consumed somewhere else. All they do is very slightly subsidizing the buildout of more renewable generation capacity.
This article is very light on details. I'm also not sure what the difference between "traditional data centres"and "cloud data centres" is.
It also completely ignores (as does everyone else) the 700lb gorilla in the room - the embedded energy content of all of the computing equipment, from the video camera used to film content, to the networking equipment and data centre servers, to the final device used to consume that content. Manufacturing semiconductors is an incredibly energy-intensive (as well as very resource-intensive) process, and product lifecycles are ever-shortening.
Low tech magazine has a very definite agenda but I've found their writing really informative and I think they do a good job in raising how enormous an issue this really is.
> "traditional data centres"and "cloud data centres"
There is no difference hardware-wise. The difference is our cloud overlords have been promoting "on-demand scaling" as a proof "serverless hosting" is superior to traditional ways because it's greener and "we can shutdown machines when there's no demands". Which is arguably true, but so short-sighted it hurts to think journalists will fall into this trap.
> to the networking equipment and data centre servers
Sorry i don't have the data at hand, but some people estimated just that. Internet infrastructure as a whole was estimated to use several percent of the world's electricity. Those numbers did not count resources/energy for production of the hardware or the infrastructure, only recurring "costs".
The difference between sophisticated (Google, Amazon, Facebook) cloud data centers and traditional corporate data centers is Extremely Large™. Traditional ones are characterized by things like big central UPS systems, redundant AC/DC power supplies in each host, remote-control PDUs, overblown air conditioners, etc. Cloud data centers are typified by non-redundant power supplies with either distributed UPS systems, or no UPS, DC-to-point-of-load, and marginal ventilation where the cold aisles are hotter than the hot aisles of traditional data centers.
> Internet infrastructure as a whole was estimated to use several percent of the world's electricity
Well, if you define "several" as "almost 2%" then yes, but mostly in traditional data centers. Moving to clouds improves this figure. It is estimated that half of global IT power consumption is in cooling and power delivery, meaning the industry as a whole has a PUE of 2.0, whereas Google, Facebook, and Amazon have PUEs of about 1.1. In other words, moving workloads to the cloud can save almost half of the power.
Apple themselves claims that roughly 80% of the emissions of an iPhone comes from production; with the remainder being transit, use, and end of life recycling. As far as electronics are concerned, the best thing you can do is defer upgrading as long as possible.
Interestingly enough, the same is true for cars. It’s often better to just keep using an older car until it dies from a CO2 perspective (sometimes older cars emit other pollutants that makes this more complicated). Depending on the exact mix of electricity production in your area, it can sometimes be greener to use a used gasoline car than a new EV, although that is changing as energy production shifts towards renewable sources.
Agree eith everything you've said, but with one caveat - the resale market for cars and consumer electronics is completely different. Everyone I know has a drawer full of old laptops/chargers/phones that were purchased new and are only destined for landfill. Meanwhile, cars get reused long after their first owner gets rid of them even if they replace them every two years.
It’s certainly much easier to keep a used car running for a decade than it is to keep a used phone running, that’s for sure. And there’s a market specifically designed to support that too.
I think the most realistic choice here is to not upgrade phones every 1-2 years, but stretch it to 3. The recent trend of Apple to not include all the accessories in the box is probably a net good, given how many chargers almost consumers have already too.
4 is definitely doable. More if you're in the apple ecosystem. iOS 15 (not yet releases) still supports the iPhone 6s which was originally released in 2015. I bought one last year as a testing device for work, and it's still a great phone (and much cheaper - £120 and had recently had it's battery replaced). On the android side you have to be more lucky, but my Samsung S7 (releases 2017) is still going strong and recieved a security update last month.
I'm still using the iPhone 6s that I bought in 2015, and it's practically as good as new. It's still using the original battery, which doesn't seem to hold charge for as long anymore, but even so, it's doing remarkably well.
I use a iPhone 5s. Works fine and I have no reason to upgrade at all. I think I have it for like five years now. I did get the battery replaced last year though
My Nokia 6500 is starting to get a bit flaky, but it's done a good 12 years or so...
(It's hard to find a better phone with the same form factor.)
I get very attached to my devices, not sure I could deal with a 2-3 year turnover.
It's sad that 3 is considered a stretch. In Norway, you actually have 5 years of "reklamasjonsrett" (right to get it repaired or replaced if there's some fault the producer should have known about). And Samsung gives security updates for 4 years these days, so you don't even need to run unofficial Androids for the first four years (though really there should be some law about providing security updates, cf. the recent mybook incident and Internet of unpatched Things)
Could not agree more. If you halved the rate at which you replaced electronics (TV, laptop, phone, etc), you would probably do more good towards reducing CO2 production than buying an EV.
I see today’s internet as it is discussed in this article and in public discourse in general (with blockchain and all) as a massive black box dragon that has grown completely out of control and is now literally devouring the world. [1]
Another radically cooperative alternative is starting to emerge: fully agent centric networks, where modular and transparent app logic/functions validate and share the information that is important to a specific network of peers. Combine that with the exciting work of Mikorizal and Sensorica, which turn the model of Enterprise Resource Planning upside down (or as they describe it: “merging REA [the Resources Event Agent accounting method] with some ideas from Manufacturing Resource Planning (MRP) , Work Flow Management (WFM) and Constraint-Based Scheduling (CBS) systems” [2]), to create ‘Network Resource Planning’ systems and ‘Open Value Networks‘.
In other words: they are re-rooting economic networks for the digital age in a way that “enable[s the] internetworking among many different software projects for resource planning and accounting within fractal networks of people and groups.“ By focusing on protocols instead of platforms, a more balanced and horizontal web of trust and transparency can emerge.
The most exciting projects in the dWeb for me, that I mentioned above, are: http://valueflo.ws + holo-REA and holochain.
Government should start legislating that non-renewable energy companies must gradually raise their prices to a level such that demand is decreased to a sustainable level. That would be an easy thing for them to do and the outcome is a must if it’s for the survival of the planet. Does anyone know why this policy has not been implemented?
The government typically doesn't mandate companies set their prices to a certain level. Instead, they add taxes that then raise the price of the item. For example, there's already a gasoline tax.
I think it still eats up a ton (technical term) of unnecessary energy as we have moved from owning physical objects to digital copies of the same thing. For example music, video, etc. Before 2000 you only needed electricity to play content not store it. This question is : did the production of a vinyl record / cd use more electricity than a lifetime of storage of its modern day digital equivalent? Given the absence of scarcity of digital music I would hazard a guess that the number of people that can own a spotify track vs the physical track is so much greater that the energy requirements are exponentially larger.
> did the production of a vinyl record / cd use more electricity than a lifetime of storage of its modern day digital equivalent?
Replace “electricity” with “energy” and yes, the manufacturing and distribution of physical media likely does exceed that required for digital only over a lifetime. The details vary, storing and playing files on your phone is more efficient than doing so from a magnetic drive in a desktop computer and more than streaming it over the Internet on every play.
On a per song basis you are probably correct but on a total volume basis, more people have the opportunity to download and store their own copy of [insert album] vs the limited pressing of the actual physical copy of the album/cd. Per song the lifetime energy costs are lower but with a higher installed base , is it marginal?
> Even the predicted environmental impact of Bitcoin, which does require lots of computing firepower, has been considerably exaggerated by some researchers.
Exaggerated or not, what's aggravating is that Bitcoin is literally secured by the consumption of power through wasting it on hashing.
To process the same amount of transactions per second in a non-trustless (trustful?) system, you'd need, like, a few Raspberry Pis around the world. And the trustlessness is dodgy anyway, with a few big mining pools having over 51% of the mining, with everyone basically having to trust the developers, etc.
I'm not a fan of Bitcoin at all but that argument could be used for anything - "what's aggravating is that cars are literally moved by the consumption of power through wasting it on combustion in the engine", "what's aggravating is that computers are literally operated by the consumption of power to waste it on browsing HN", "what's aggravating is that TV images are literally sent by the consumption of power through wasting it on radio waves" and so on. The only case it seems like a waste is if you don't value the output. If someone values browsing HN, individual transport, watching TV, or distributed consensus then it's not wasted it was used.
Inefficient perhaps, but the energy use was intentional for a value to someone not discarded for no gain.
In those other processes the designers aim for efficiency - more useful product per entropy. In bitcoin it's exactly the opposite. That makes it unpalatable.
It is inefficient and unpalatable compared to traditional trust absolutely, even to some other crypto methods that try to achieve similar goals. Like I said I'm not really a fan of Bitcoin (or crypto to be honest).
That being said Bitcoin doesn't aim for less product per entropy. That would literally mean you get more reward for doing the same work in the same time but less efficiently which isn't true. That doesn't mean everyone always uses the most efficient hashing method, just like not every car uses the same engine because it has the best efficiency, it just means the aim isn't towards making less product per entropy. The product may not be aligned with what you want (e.g. "very efficient way to trust a transaction") but that's back to calling the energy usage all "just for waste" because you don't personally value the output the same way as someone else (which is true of every use of energy).
Bitcoin mining strives for as much efficiency as any other electronic process: ASIC chips are hitting the smallest process size available. Miners are extremely sensitive to electricity cost, so they must be efficient there. They do not compete with households for power.
If Bitcoin mining were aiming to consume the most energy possible, then the hashing would be done by hand on pencil and paper.
maybe I'm missing a semantic argument here, but don't computers always convert ~100% of their input power to waste heat? if my cpu consumes 140W, all of that gets dissipated through the heatspreader (perhaps minus a minuscule quantity of potential energy storing register/cache state).
If 100% of the input power was given to waste heat then your computer would just be a resistance heater and not a computer. Computing information is work which requires energy (which will also eventually end up as heat in the long run, as will all energy, just not waste heat) and there is a limit on how efficiently you can compute https://en.wikipedia.org/wiki/Landauer%27s_principle. I.e. it's not possible to compute something without doing work.
Waste heat in a computer is largely resistance, power conversion inefficiencies, and cooling. Heat generated by the actual computation is known as work https://en.wikipedia.org/wiki/Work_(thermodynamics) not waste heat https://en.wikipedia.org/wiki/Waste_heat. The ratio of the two describes the efficiency of a machine. Since some useful work must have been done for the computation to occur then it's impossible for 100% of the energy to have been used for waste heat.
I find myself dubious of the claim that after an hour of isolated number crunching, my GPU has done anything aside from generate heat (that wikipedia page doesn't appear to mention anything about computing information). If only 95% of it became waste heat, where is the 5%? Given that computing information with a computer is just switching transistors on and off in a clever way, it makes no sense that if we are unclever, then suddenly all of the heat becomes "waste".
> that wikipedia page doesn't appear to mention anything about computing information
From the opening of the Wikipedia page:
'Landauer's principle is a physical principle pertaining to the lower theoretical limit of energy consumption of computation. It holds that "any logically irreversible manipulation of information, such as the erasure of a bit or the merging of two computation paths, must be accompanied by a corresponding entropy increase in non-information-bearing degrees of freedom of the information-processing apparatus or its environment".'
Computation is in the first sentence, it's mentioned by name 8 times in the opening section - information 6 times. I'm not really sure how you came to the conclusion it doesn't mention anything about computing information. Even skimming it's hard to find a spot not mentioning it directly, by name.
> Given that computing information with a computer is just switching transistors on and off in a clever way, it makes no sense that if we are unclever, then suddenly all of the heat becomes "waste".
What you call "clever" and "unclever" are (essentially) ways of describing how entropy was changed. Another way of thinking about it is there is a very specific order to where energy needs to go and how it needs to be arranged to be "clever" and calculate something vs unclever and not have the answer to the number crunching problem.
Also note there is a difference between "energy that eventually ends up as heat" and "waste heat". All work will result in heat if you follow it long enough, some energy spent doing that will not go towards the amount of work done but will still end up as heat. The latter is known as waste heat. Waste heat is not just "heat I ended up with at the end".
> Waste heat in a computer is largely resistance, power conversion inefficiencies, and cooling. Heat generated by the actual computation is known as work https://en.wikipedia.org/wiki/Work_(thermodynamics)
According to ctrl-f, the word computation never appears on this page. The word information never appears in this context. Same with https://en.wikipedia.org/wiki/Waste_heat
> What you call "clever" and "unclever" are (essentially) ways of describing how entropy was changed. Another way of thinking about it is there is a very specific order to where energy needs to go and how it needs to be arranged to be "clever" and calculate something vs unclever and not have the answer to the number crunching problem.
None if this explains why the portion of heat in a clock cycle is different if we consider the clock cycle to be doing something clever. If some of the energy is not waste, it must have taken another form. What exactly is it? When we switch a transistor non-cleverly, presumably the energy never takes this form, and thus the actual switching of the transistor took more energy . Which makes no sense.
> All work will result in heat if you follow it long enough, some energy spent doing that will not go towards the amount of work done but will still end up as heat.
I'm not convinced this is true. It seems equivalent to the claim that all macroscopic objects moving through the universe will eventually stop, which I would ague is an assertion beyond science.
> Waste heat is not just "heat I ended up with at the end".
Fine, but we have to be able to point to where the other portion of energy ended up in the interim. Suppose I push black and white boulders up a hill. This leaves them with potential energy, which is somewhat less than the energy it took to get them there. We can agree that at some point in the distant future, the hill will decay, and the boulders end up back on the ground. Thus all the work eventually ended up as heat.
I can choose to arrange them randomly, or I can arrange them to create binary numbers. It seems as though you are claiming that if I create binary numbers with these boulders, the energy it took to do that is more than if I arrange them randomly. I want to understand how exactly it took more joules to do that, and where this extra energy is stored.
In computers the "work" is mostly not physically impactful apart from the creation of light from a monitor, or the application of magnetism to a disk. You could say all the energy that goes to heat instead is waste, but as with mechanical devices, there's no way to be perfectly efficient. Some amount of energy loss to heat is strictly necessary to process logic.
Yes, in the end, 100% of the energy that is consumed within a room, is not stored as potential energy, and doesn't escape as RF/light, ends up as heat.
The same argument can't be made for other goods. Bitcoin bakes the energy cost of transaction processing into the security model. Security and ownership transfer for cars and televisions relies on good old-fashioned physical locks and paper records. That has other inefficiencies, but not much energy cost. The introduction of the internal combustion engine to the world definitely involved some tradeoffs in terms of pollution and mineral extraction, but it was always possible to make cars more efficient without breaking the basis of the ownership record for cars.
The same argument can be made about other systems they just aren't as grossly inefficient as Bitcoin so it isn't (usually) worth mentioning. The energy cost of doing something is always baked in as it's impossible to do anything without using energy, the difference is simply the cost vs output ratio is shit (efficiency) not that Bitcoin has a large chunk of waste energy for a reason (decentralized consensus) nobody values.
I don't value it but obviously some do, same as some don't value individual transport over mass transit but I do. It doesn't mean a car is aggravating because it solely exists to waste energy in means a car is ridiculously inefficient. Same story with crytpocurrency.
I’m beginning to detect a pattern here. Exaggerating researchers are on the rise. It feels as though researchers might be incentivized to exaggerate. The news media help them with that. Maybe we need more research in this area.
Bitcoin miners use hashpower not only to find the next block but to secure the entire chain to that point. So each block secures over 650 million transactions right now, and that number only increases.
Your word "waste" implies a value judgement, and thus you have departed the realm of scientific inquiry.
The transactions-per-second energy argument is completely fallacious and often made in bad faith. It makes no sense whatsoever. Comparing Bitcoin (a settlement system) to some other credit-based payment system is not a genuine comparison.
Ignoring the Lightning Network in any discussion of Bitcoin shows a distinct lack of subject-matter awareness.
If you dig into the power grid system today, you'll find that a huge amount of power generated by the world's energy producers goes unused. The Bitcoin network uses far less energy than the amount lost during normal generation, yet I see no hand-wringing about this colossal "waste" boiling the oceans.
"Waste" is value judgement. For me, it's securing a network of decentralized transactions, so that power isn't wasted. Given my distaste for adtech and certain governments, I'd qualify their power and resource usage as a waste and that the world would be better off without them, even if the people who work in adtech or benefit from said governments would disagree.
The issue with Bitcoin is that there’s no cap on potential power consumption. It would eat up the sun if someone thought that would improve their returns.
Resistive heating is not efficient compared to heat pumps. The value obtained running the miners would have to be greater than the difference in cost between the two, and reasonably stable, for realistic people to be even started to consider it.
I don't think people generally think the internet as a network was wasting energy.
I tend to think it's mainly the mass amount of commoditised hardware that's the problem, especially if it's thrown away and not recycled, and recently hoarded for mining.
> From 2010 to 2018, the data workloads hosted by the cloud data centers increased 2,600... But energy consumption for all data centers rose less than 10 percent... What happened, the authors explain, was mainly a huge shift of workloads to the bigger, more efficient cloud data centers — and away from traditional computer centers, largely owned and run by non-tech companies... “The big cloud providers displaced vastly less efficient corporate data centers,” Mr. Koomey said.
This is fascinating, and the first I've heard -- I'm curious if anyone here knows (not just guessing) what the main factors are?
Is it economies of scale with power and cooling and construction? Is it massively higher CPU utilization because servers are on-demand? Is it custom-designed equipment? All of the above, or something else?
I'd guess that being sited in areas of cheap renewable energy would play a big part too, but if the shift described is solely "energy consumption" then I'm assuming it's treating renewables as the same amount of energy.
I know a lot of people here are anti-cloud and pro-hosting your own equipment in your own rack... but I'm now genuinely curious if there's a large environmental difference here.
News sites love interesting headlines, researchers and university PR departments love being in the news, often much more than they value truth.
The easiest way to get impressive research on power consumption is to use old per-unit consumption numbers to extrapolate power consumption based on today's usage (e.g. if transferring 1 GB of data took 1 kWh many years ago, just multiply "GB transferred today" with 1 kWh and claim that's what the Internet/zoom calls/... use)
"But Bitcoin, the scientists say, is something different — and a worry. The efficiency trends elsewhere in tech are blunted because Bitcoin’s specialized software churns through ever more computing cycles as more people try to create, buy and sell digital currency."
No, increased Bitcoin usage does not increase the amount of energy needed to process transactions.
More people are mining (and thus burning more power), because it is a profitable venture if you can get best prices on hardware and lowest costs of electricity.
The least expensive electricity in the world happens to be either stolen, stranded, or hydro / geothermal.
92 comments
[ 2.7 ms ] story [ 143 ms ] threadMost of the non sustainable energy consumption happens on the consumer side where most of us are dependent on grids that are also shifting to more sustainable energy sources but still have some way to go.
> still there's people in the Netherlands complaining that datacentres use up too much of the green energy available in the Netherlands. Truly backwards way of thinking
Analyzing and critiquing the moves a massive monopoly -capitalist entity is ‘backwards‘? This sounds quite un-nuanced. Could you elaborate more what you mean? As it stands it sounds like a naive accelerationist perspective.
Companies are facing backlash, yes, for using green energy. How can a company ever go green as long as this kind of thought patterns are mainstream
It's fundamentally impossible. Capitalism is what brought this ecological madness down on it, because of its shortsighted profit incentives. The very same system cannot fix the problem, because nothing has changed.
Pointing out companies who used fossil fuels was intending to allow us as a society to realize we need to reduce consumption/production, not pretend we should shut up and live happily as long as we paint every single polluting thing with a green chemical glowing paint.
There is no such thing as green capitalism and there will never be. Consumerism, private property and competition is what destroyed this planet. Whether we go on and destroy new ones (hello Musk, Bezos and other psychopaths) or destroy the capitalist system depends on our understanding of those issues.
I've lived in the socialism on the USSR and seen what it looked like, it's day and night compared to how free countries deal with pollution. Guess what happens when raw materials are available for free, the economy is inefficient because why bother, and pollution doesn't matter because citizens can't vote anyone else into the office because there's only a single party. China happens. Rivers downstream from textile factories colored with dye-of-the-season, wire fences near cement factories opaque from dust, transport workers spilling diesel fuel into the ground for decades to hide slacking and sub-quota mileage, etc. I've seen it all with my own eyes.
When the first Western experts gained access after the collapse of the Soviet Union, they were utterly devastated at the scale of environmental pollution. They had never seen anything this bad. The solution to everything had been "dig a ditch and throw it in", and if anyone complained, arrest and beat them until they shut up, or send to a labor camp if they don't.
The main problem in the West is similar socialism for the polluters, albeit on a much smaller scale. They get to pollute for very cheap, mostly abroad in China. They should be taxed for what raw materials and their pollution actually costs instead of the current artifically low levels achieved through outsourcing to lawless shithole countries, and environmental damage from their actions would be greatly reduced.
It has more to do with free countries than free enterprise. Private businesses used to be notorious for stripping resources from the environment (e.g. strip mining, clear cutting) to keep the cost of resources low and they used to be notorious for dumping pollutants into the environment since they did not have to deal with the costs of pollution. It was public pressure that resulted in environmental regulation, and even then (as you mentioned) they try to circumvent the costs by outsourcing to jurisdictions that are more lenient.
I understand your emotion and living in Romania I see the abandoned conglomerates and am glad that period is over...however if you discount the risk of chernobilian disaster (sea of Azov etc) that the lack of media and public accountability made possible I'd wager that capitalistic societies is powers of magnitude more polluting simply because people live behind paid layers (producer->wholesalers->distributors->retailers->consumer) paying for it by living in cities, working for buisnesses serving other buisness etc) Manufactured boredom (from living in cities) ->Advertising->fashion/fads->brands-> excess packaging is also a thing and part of the problem... Then there is the fact that businesses once created to solve a problem in the most efficient way often self perpetuate and bloat with govt closing their eyes to monopolistic behaviour in the name of customer pricing
In the city I live in, you can buy food from people from the countryside, rent/property is still cheap , so if you live simply you hardly have expenses and can avoid wasting your life working away while hardly seeing your kids grow up... I understand you lived in a failed system but you shouldn't throw away the baby with the bathwater, authenticity counts, real food counts, generosity and food that is less processed or sprayed (because cost) counts
I feel having grown up in a WIERD (acronym) people there forget that the apparent sophistication is just a sheen and like the 300ml plastic bottles they constantly consume they are choking the rest of us on their garbage (ie the developed world using x times more than the rest)
I think you can see where I'm going with this...the problem with capitalism atm or the way government treats it is that it inherently needs growth, whereas the planet needs "decroissance" a french word for "ungrowth" a french politician, (Yves Cochet )I think, said something like our economy has to "ungrow" 10% per year , it's possible even without covid, but it certainly needs a whole new mindset or a drastic realisation that droughts are going to get much much worse...
It's in no way bad that the server uses clean energy, but the critique is that the government wriggled their way out, Microsoft got to boast about servers running on clean energy while the Dutch citizens were still stuck with the older CO2-intensive power source mix despite the govt's "commitment" to cleaner power.
I’m very skeptical of this. Until now this narrative has often turned out to be nothing more than corporate greenwashing, without tackling the root causes of the problem or helping shift the underlying systems. Take the example of Microsoft buying up renewable energy in Holland: https://youtube.com/watch?v=OiPoR9OvD0Y (Nederland als harde schijf | Zondag met [Arjen] Lubach)
I bring this up as renewables are currently supplemental and don't replace base load. Hydro storage isn't available/practical for all regions, nuclear has been deemed too expensive, so it's usually natural gas that slots in.
Yes, this is a problem.
Let's say the grid uses 100 imaginary units of power per year. 50% is currently produced through solar and wind.
Company A buys 50 units and company B buys 50 units a year, until company A gets a crisis of conscience and decides to go green. They manage to get a deal whereby they pay for a certificate that says that their 50 units is produced through renewables. Great.
Company B, however, has customers who care less about green stuff so they just buy energy from the grid and "their" electricity is now just produced from fossil fuels.
In this case has anything really changed? It's probably a good thing that Google is buying renewable energy: I expect it raises demand for renewables in some way and it's good politically for companies to compete for green credentials. But it doesn't seem to do that much.
I disappointingly found out that my carbon neutral tariff doesn't guarantee a purchase of renewables, it just guarantees that any energy I use that is over their wholesale bought amount of renewable energy will be carbon offset, which is complete crap.
What "renewable" energy does it run and how was it produced? What's the carbox/fossil profile of those energy sources? How long do they need to run, and in what conditions, in order to offset this initial fossil energy invested? Those are the medium-difficulty questions. And as i much as i hate Michael Moore's rhetoric, "Planet of the humans" does a pretty good job of explaining why these questions are important.
What is the ecological impact of the cloud? Why do countryside folks have to have their data run across the country just to exchange messages with their neighbors, and what's the impact of that? What's the impact of a song that you downloaded from Youtube (replicated, dedicated centralized infra) instead of Bittorrent (decentralized, already-in-use hardware)? What's the impact of streaming this same song over and over? What's the impact of online advertisement and web trackers? How many people are working in concrete offices in a city in order to maintain that hellish infrastructure? Where does all the sand and minerals come from to build all that infrastructure as well as all the electronics you find in there? Those are the hard questions and when you answer them you'll never fantasize about green capitalism ever again.
In case you're interested in such topics, and what to do about it, but can't stand for Michael Moore, i strongly recommend the "End:Civ" documentary: https://youtube.com/watch?v=L4ccAJJrrjM
You can get a lot of containers on a boat and a lot of bits down an optical fibre. The cost of the crypto on the other hand, that adds up a bit.
Also BitTorrent uses a bunch of power on a bunch of endpoint servers which are often less efficient than custom silicon. Tricky to calculate.
Also, how green are their backup power generation? Outside batteries?
It also completely ignores (as does everyone else) the 700lb gorilla in the room - the embedded energy content of all of the computing equipment, from the video camera used to film content, to the networking equipment and data centre servers, to the final device used to consume that content. Manufacturing semiconductors is an incredibly energy-intensive (as well as very resource-intensive) process, and product lifecycles are ever-shortening.
Yes, I’d love to read more about that. i wonder if anyone here has a link to some research that looks at the full-er picture.
https://solar.lowtechmagazine.com/2009/06/embodied-energy-of...
Low tech magazine has a very definite agenda but I've found their writing really informative and I think they do a good job in raising how enormous an issue this really is.
There is no difference hardware-wise. The difference is our cloud overlords have been promoting "on-demand scaling" as a proof "serverless hosting" is superior to traditional ways because it's greener and "we can shutdown machines when there's no demands". Which is arguably true, but so short-sighted it hurts to think journalists will fall into this trap.
> to the networking equipment and data centre servers
Sorry i don't have the data at hand, but some people estimated just that. Internet infrastructure as a whole was estimated to use several percent of the world's electricity. Those numbers did not count resources/energy for production of the hardware or the infrastructure, only recurring "costs".
Although eu-west-1c ran out of nodes the other day so YMMV anyway.
The difference between sophisticated (Google, Amazon, Facebook) cloud data centers and traditional corporate data centers is Extremely Large™. Traditional ones are characterized by things like big central UPS systems, redundant AC/DC power supplies in each host, remote-control PDUs, overblown air conditioners, etc. Cloud data centers are typified by non-redundant power supplies with either distributed UPS systems, or no UPS, DC-to-point-of-load, and marginal ventilation where the cold aisles are hotter than the hot aisles of traditional data centers.
> Internet infrastructure as a whole was estimated to use several percent of the world's electricity
Well, if you define "several" as "almost 2%" then yes, but mostly in traditional data centers. Moving to clouds improves this figure. It is estimated that half of global IT power consumption is in cooling and power delivery, meaning the industry as a whole has a PUE of 2.0, whereas Google, Facebook, and Amazon have PUEs of about 1.1. In other words, moving workloads to the cloud can save almost half of the power.
Vendor lock-in is the only difference.
Interestingly enough, the same is true for cars. It’s often better to just keep using an older car until it dies from a CO2 perspective (sometimes older cars emit other pollutants that makes this more complicated). Depending on the exact mix of electricity production in your area, it can sometimes be greener to use a used gasoline car than a new EV, although that is changing as energy production shifts towards renewable sources.
I think the most realistic choice here is to not upgrade phones every 1-2 years, but stretch it to 3. The recent trend of Apple to not include all the accessories in the box is probably a net good, given how many chargers almost consumers have already too.
A quick google search about market share reveals: “AWS has 32% of the market, followed by Azure at 19%, Google at 7%, Alibaba Cloud close behind“
Why is Alibaba listed first in the list in the NYT article if American data centers together have 58% market share?
Another radically cooperative alternative is starting to emerge: fully agent centric networks, where modular and transparent app logic/functions validate and share the information that is important to a specific network of peers. Combine that with the exciting work of Mikorizal and Sensorica, which turn the model of Enterprise Resource Planning upside down (or as they describe it: “merging REA [the Resources Event Agent accounting method] with some ideas from Manufacturing Resource Planning (MRP) , Work Flow Management (WFM) and Constraint-Based Scheduling (CBS) systems” [2]), to create ‘Network Resource Planning’ systems and ‘Open Value Networks‘.
In other words: they are re-rooting economic networks for the digital age in a way that “enable[s the] internetworking among many different software projects for resource planning and accounting within fractal networks of people and groups.“ By focusing on protocols instead of platforms, a more balanced and horizontal web of trust and transparency can emerge.
The most exciting projects in the dWeb for me, that I mentioned above, are: http://valueflo.ws + holo-REA and holochain.
[1] https://www.kosmosjournal.org/article/seeing-wetiko-on-capit...
[2] http://www.jeffsutherland.org/oopsla97/haugen.html
Replace “electricity” with “energy” and yes, the manufacturing and distribution of physical media likely does exceed that required for digital only over a lifetime. The details vary, storing and playing files on your phone is more efficient than doing so from a magnetic drive in a desktop computer and more than streaming it over the Internet on every play.
Exaggerated or not, what's aggravating is that Bitcoin is literally secured by the consumption of power through wasting it on hashing.
To process the same amount of transactions per second in a non-trustless (trustful?) system, you'd need, like, a few Raspberry Pis around the world. And the trustlessness is dodgy anyway, with a few big mining pools having over 51% of the mining, with everyone basically having to trust the developers, etc.
The stability of bitcoin is the tyranny of the installed base. Even non-mining nodes will not relay invalid blocks that break the fundamental rules.
See also: Ethereum Classic, a case study in developer-as-dictator
Inefficient perhaps, but the energy use was intentional for a value to someone not discarded for no gain.
You can't have "all waste heat" as that would require doing no work which would result in... no waste heat.
That being said Bitcoin doesn't aim for less product per entropy. That would literally mean you get more reward for doing the same work in the same time but less efficiently which isn't true. That doesn't mean everyone always uses the most efficient hashing method, just like not every car uses the same engine because it has the best efficiency, it just means the aim isn't towards making less product per entropy. The product may not be aligned with what you want (e.g. "very efficient way to trust a transaction") but that's back to calling the energy usage all "just for waste" because you don't personally value the output the same way as someone else (which is true of every use of energy).
If Bitcoin mining were aiming to consume the most energy possible, then the hashing would be done by hand on pencil and paper.
Waste heat in a computer is largely resistance, power conversion inefficiencies, and cooling. Heat generated by the actual computation is known as work https://en.wikipedia.org/wiki/Work_(thermodynamics) not waste heat https://en.wikipedia.org/wiki/Waste_heat. The ratio of the two describes the efficiency of a machine. Since some useful work must have been done for the computation to occur then it's impossible for 100% of the energy to have been used for waste heat.
From the opening of the Wikipedia page:
'Landauer's principle is a physical principle pertaining to the lower theoretical limit of energy consumption of computation. It holds that "any logically irreversible manipulation of information, such as the erasure of a bit or the merging of two computation paths, must be accompanied by a corresponding entropy increase in non-information-bearing degrees of freedom of the information-processing apparatus or its environment".'
Computation is in the first sentence, it's mentioned by name 8 times in the opening section - information 6 times. I'm not really sure how you came to the conclusion it doesn't mention anything about computing information. Even skimming it's hard to find a spot not mentioning it directly, by name.
> Given that computing information with a computer is just switching transistors on and off in a clever way, it makes no sense that if we are unclever, then suddenly all of the heat becomes "waste".
What you call "clever" and "unclever" are (essentially) ways of describing how entropy was changed. Another way of thinking about it is there is a very specific order to where energy needs to go and how it needs to be arranged to be "clever" and calculate something vs unclever and not have the answer to the number crunching problem.
Also note there is a difference between "energy that eventually ends up as heat" and "waste heat". All work will result in heat if you follow it long enough, some energy spent doing that will not go towards the amount of work done but will still end up as heat. The latter is known as waste heat. Waste heat is not just "heat I ended up with at the end".
According to ctrl-f, the word computation never appears on this page. The word information never appears in this context. Same with https://en.wikipedia.org/wiki/Waste_heat
> What you call "clever" and "unclever" are (essentially) ways of describing how entropy was changed. Another way of thinking about it is there is a very specific order to where energy needs to go and how it needs to be arranged to be "clever" and calculate something vs unclever and not have the answer to the number crunching problem.
None if this explains why the portion of heat in a clock cycle is different if we consider the clock cycle to be doing something clever. If some of the energy is not waste, it must have taken another form. What exactly is it? When we switch a transistor non-cleverly, presumably the energy never takes this form, and thus the actual switching of the transistor took more energy . Which makes no sense.
> All work will result in heat if you follow it long enough, some energy spent doing that will not go towards the amount of work done but will still end up as heat.
I'm not convinced this is true. It seems equivalent to the claim that all macroscopic objects moving through the universe will eventually stop, which I would ague is an assertion beyond science.
> Waste heat is not just "heat I ended up with at the end".
Fine, but we have to be able to point to where the other portion of energy ended up in the interim. Suppose I push black and white boulders up a hill. This leaves them with potential energy, which is somewhat less than the energy it took to get them there. We can agree that at some point in the distant future, the hill will decay, and the boulders end up back on the ground. Thus all the work eventually ended up as heat.
I can choose to arrange them randomly, or I can arrange them to create binary numbers. It seems as though you are claiming that if I create binary numbers with these boulders, the energy it took to do that is more than if I arrange them randomly. I want to understand how exactly it took more joules to do that, and where this extra energy is stored.
I don't value it but obviously some do, same as some don't value individual transport over mass transit but I do. It doesn't mean a car is aggravating because it solely exists to waste energy in means a car is ridiculously inefficient. Same story with crytpocurrency.
I’m beginning to detect a pattern here. Exaggerating researchers are on the rise. It feels as though researchers might be incentivized to exaggerate. The news media help them with that. Maybe we need more research in this area.
Your word "waste" implies a value judgement, and thus you have departed the realm of scientific inquiry.
The transactions-per-second energy argument is completely fallacious and often made in bad faith. It makes no sense whatsoever. Comparing Bitcoin (a settlement system) to some other credit-based payment system is not a genuine comparison.
Ignoring the Lightning Network in any discussion of Bitcoin shows a distinct lack of subject-matter awareness.
If you dig into the power grid system today, you'll find that a huge amount of power generated by the world's energy producers goes unused. The Bitcoin network uses far less energy than the amount lost during normal generation, yet I see no hand-wringing about this colossal "waste" boiling the oceans.
It's secured by a mix of things, including expertise, investment and yes, a large energy component. Far from a new thing.
I tend to think it's mainly the mass amount of commoditised hardware that's the problem, especially if it's thrown away and not recycled, and recently hoarded for mining.
This is fascinating, and the first I've heard -- I'm curious if anyone here knows (not just guessing) what the main factors are?
Is it economies of scale with power and cooling and construction? Is it massively higher CPU utilization because servers are on-demand? Is it custom-designed equipment? All of the above, or something else?
I'd guess that being sited in areas of cheap renewable energy would play a big part too, but if the shift described is solely "energy consumption" then I'm assuming it's treating renewables as the same amount of energy.
I know a lot of people here are anti-cloud and pro-hosting your own equipment in your own rack... but I'm now genuinely curious if there's a large environmental difference here.
The easiest way to get impressive research on power consumption is to use old per-unit consumption numbers to extrapolate power consumption based on today's usage (e.g. if transferring 1 GB of data took 1 kWh many years ago, just multiply "GB transferred today" with 1 kWh and claim that's what the Internet/zoom calls/... use)
For example: Every GB of data uses 5 kWh [in 2021 according to https://www.aceee.org/files/proceedings/2012/data/papers/019...], so using 122 EB/month [https://en.wikipedia.org/wiki/Internet_traffic#Global_Intern...] = 1 464 000 000 000 GB/year = 1.5 PWh in 2017, which would be almost 7% of total power consumption.
That should say 2012.
Too late to edit now.
No, increased Bitcoin usage does not increase the amount of energy needed to process transactions.
More people are mining (and thus burning more power), because it is a profitable venture if you can get best prices on hardware and lowest costs of electricity.
The least expensive electricity in the world happens to be either stolen, stranded, or hydro / geothermal.