You are reading content from Scuttlebutt
@elavoie

Energy is an Achille's heel of current major tech companies

Major tech companies and some critical parts of the Internet are powered by coal and other fossil fuels [1]. This is still the most economical way of providing 24h/7d access to viewing cat videos on YouTube, stalking your friends on Facebook, or using your Office 360 apps.

In contrast, if we can tolerate intermittent access, it becomes possible to use significantly less energy. As I am writing this, Low-Tech Magazine is being served with 1.15W from its battery. I can also browse my local database on SSB for 15-25W on my Macbook Air 2011.

A corollary of all this, is that as production of fossil fuels stagnates and other sectors of society start competing with the tech sector and the Internet for access to energy, SSB and other decentralized approaches are going to become more and more competitive. We may even reach the point where access to major services will become sufficiently disturbed, or that energy costs will make their business models less profitable, that local applications based on protocols like SSB will provide better quality of service, cheaper.

So there is no need to fight the big firms, we only need to keep building alternatives and wait for the sun, both figuratively and literally.

[1] In its environmental report, Google mentions buying the equivalent amount of renewable energy. This is therefore an accounting trick for greenwashing. As long as we can access Google services instantly, anytime of the day, there is probably going to be a fossil fuel plant somewhere powering the services.

@Luandro
Voted this
@andrestaltz

I was thinking about this the other day, but didn't look for metrics on this. But I thought that cloud computing or the client-server model is probably significantly more expensive than networks that follow the end-to-end principle. Basically because you need 2 computers (client and server) whenever you activate any feature. Always online servers also incur some costs just to stay online (this might be a consequence of location-addressed URIs as opposed to content-addressed URIs).

But apart from that, a huge amount of computation and networking in these tech giants go to big data, i.e. mass analytics and data processing, as well as (the more recent) heavy computation costs for machine learning.

I always find it hypocritical how Apple brags about its environmental responsibility, when the most responsible thing to do is produce long-lasting phones, not the 18 month release cycle they keep pumping.

@xj9
Voted this
@kravietz_
Voted this
@elavoie

Basically because you need 2 computers (client and server) whenever you activate any feature.

Yes indeed, plus all the internet routers in between, plus georeplication between data centers, plus monitoring to centralize all the information to give a global picture to engineers and management. And you also need energy and resources to host all the developers with all the perks of a top position in a tech company to monitor and maintain the infrastructure.

In comparison, SSB clients can use the Internet routers only for the delta updates which happen only once per piece of data and many of our developers have much more frugal lifestyles. If we use propagation algorithms based on physical proximity between smartphones, we can do away with the routers entirely. I have not tried to do back-of-the-envelope calculations yet but I suspect we use several orders of magnitude less energy and there is still significant potential for improvements.

A whole system comparative analysis of energy flow could be an interesting paper for the future!

@Ladislas
Voted this
@Ladislas

While i am agree with you @elavoie about the symptoms, i am not agree with the strategy (wating for the sun and let the big ones burn the planet and peoples :) ).

Sorry, it is in french but i can share this document about the social and ecological cost of the "cloud" :

https://www.soc-etudes.cgt.fr/download/592/les-cahiers-du-numerique/7695/592ff0d856ddftz_portfolio_1496314077.pdf

It is a nice peace of work from CGT comrades ;)

Meanwhile, are you aware of some reseach about evaluating the cost of a p2p system if a large amount of the population subscribe ? Maybe 1.15W by "a lot of people" will be more than the contemporary centralized systems ? Let me play the devil's advocate here :)

@Antoine Cailly
{
  "type": "tag",
  "version": 1,
  "tagged": true,
  "message": "%06iTbQGgD76m5JUZM8pdjG36fBUy7x4P2G3Q07XXsYM=.sha256",
  "root": "%6TJHwANg7Y+qFIZgSP3GWx99VlmjnG8124lIoKDzukY=.sha256",
  "branch": [
    "%/gmIcMeu4FcRinSgbwaCLT4NvdBl/1ywKiDM4WDu/nI=.sha256"
  ]
}
@Antoine Cailly
Voted this
@elavoie

@Ladislas Of course we don't agree and I think it is a good thing, let's continue disagreeing ;-). Thanks for the link, will get to it once I revisit the topic!

I am not aware about the cost of a P2P systems, doing background research on this will be for future work. Average humans can output 100W continuously so a small electrical generator would be all that is needed to power a smartphone and a radio to communicate. So large-scale P2P with smartphones using only human power should be perfectly doable with reasonable environmental impact.

Now creating the devices is another thing. The Archdruid Report I read right now argues that computers, in a few decades, will be abandoned altogether because non-computer tech will do the jobs we use computers for more economically. He argues we won't be able to maintain the industrial infrastructure necessary to make todays' computers because it is too energy intensive for anything else than fossil fuels. If that indeed happens, today's computers will become toys for the rich of the future and the rest of society will do without. Maybe we will live to see if he is right!

@Andre Esteves
Voted this
@Andre Esteves

Bulleye, elavoie...

https://www.zdnet.com/article/google-1-6-million-solar-panels-will-power-these-new-data-centres/

@greg
Voted this
User has chosen not to be hosted publicly
@Hamishcampbell
Voted this
@elavoie

@Andre Esteves That article is basically a press release for Google... It also fails to use the most recent figures: 7.6 TWh total energy use in 2017 compared to 5.2 TWh in 2015, and 3.7 TWh in 2013, a 46% increase in two years and a doubling in 4! It obviously won't keep doubling every four years for long if total world-wide energy production stops increasing...

Also using the most recent data Google makes available:

Solar Power Plants (from Zdnet article): 2 x 150 MW = 300 MW
Total cumulative renewable energy contracts: 2,960 MW
Average power use by Google in 2017: 7,609,088 MWh / (8760 h in a year) = 868 MW

So Google has contracts for about 3 times their average power usage, probably because it needs to deal with variations in power use. Also, these new solar plants represent only 10% of the total power use of Google. And the solar power plant power figure is probably peak power and not average power over the year, so the effective contribution of that plant is certainly lower than 10%.

The crux of the issue is that a 150MW coal power plant can produce 150MWh of energy every hour day and night as long as it has coal available. It can also ramp-up and slow down consumption of coal according to demand. A solar power plant, on the other hand, only produces energy when the Sun is available and storing the energy to use later is really expensive at that scale, more than burning fossil fuels. So if Google's quality of services don't depend on the elements, than most probably they use fossil fuels to be always available.

And the crazy train they are stuck on is that if they were not and chose to degrade quality of service to be powered entirely by renewable, their competitors would steal their market shares by using fossil fuels. So with all their might and power, they are stuck in a crazy economic game in which we will all, including Google, eventually loose.

Also the wording of the reports makes it quite unclear to me what Google is doing with all that renewable energy it buys, since it also buys energy from fossil fuels plants. Maybe they are reselling it and add that to the revenues? If that is the case, it would be more honest to say that Google is a middlemen in a renewable energy business that happens to buy (and sell some) renewable energy equivalent to the amount of energy it buys from fossil fuels plants to power its core operations. But to me those are separate operations that are brought together only for the purpose of accounting. Actually being transparent would make for less glossy environmental reports...

@elavoie

According to Google again, data centres currently use ~2% of world-wide electricity.

Using BP 2018 figures, that means
25,551.3 TWh x 2% = 511.02 TWh/year
511.02 TWh / (~1 billion Internet users x 365 days ) ~= 1400 Wh per user/day

That is a lower bound, because usage is variable and it does not count energy usage for the network; the figure is likely to be 2-5 times higher. Let’s stick to it anyway.

If I use my MacBook Air 2011, 12 hours/day at an average of 25W, that gives 300 Wh. That is a factor of 4.5x less than the lower bound. If I had a healthier lifestyle and spent more time growing vegetables and building velomobiles rather than writing papers and reading anxiously about the future of industrial civilization, I would spend maybe 4h for a total of 100Wh. So if we were to migrate the entire Internet to SSB and people mostly consulted data locally 4h/day, we could drop the world-wide energy usage of computing infrastructure by a factor of at least 14x.

Interestingly, 14x brings us into renewable energy territory (World-wide electricity mix: 15% hydro + ~7% renewable). Moreover, if everyone ran SSB locally, it would be a lot easier to build renewable energy infrastructure to power it. The infrastructure would be affordable to many people and they could use it for other things when not powering their laptop/smartphone. The simplest solution is you pedal 1h for 100Wh and you get 4h of SSB use. If it is sunny you use your solar panels, if it is windy you use your wind turbine, etc.

So there we go, the future of green computing is not more efficient data centers with machine learning, it is SSB plus home-made power infrastructure. We are going to beat the crap out of any industrial solution to the problem.

(Note: I have not factored in the energy usage of communication. Let's assume a volunteer-operated world-wide amateur radio system would use another 100Wh/(day-user) and I think the conclusion would be the same.)

@elavoie
Re: %UbmyE84JJ

What is too weird for society?

I don't think what is holding alternatives up is being too weird. I personally believe that it is because making a living while growing these alternatives is still quite uncertain and reserved to a few with enough privilege to dedicate a significant portion of their time to it. Once enough of us solve that problem and can grow alternatives to a sufficient level of maturity, more people will adopt them.

And if as a community we foster alternative lifestyles that are better aligned with limited resources and diffuse energy sources, and these become a better deal for those left out by mainstream society, then people will not only start using alternatives more, they will become part of their communities and help grow them.

I personally don't think we should oppose or fight Google/Microsoft/Facebook/etc. because they are all moving to Cloud-first business models and large-scale data analytics, which require a tremendous amount of resources and fossil energy. Once energy availability starts contracting they will have to compete against the rest of society for access and they will end up on the losing side.

SSB and other decentralized alternatives are by design significantly more energy and resource efficient, so I believe we only have to focus on having alternatives ready once the process of energy contraction starts happening. In the meantime, I think it is perfectly fine to use mainstream products if it helps us build alternatives. We only need to be careful not to depend on them.

And over the last week or so, I have been puzzling the ethical aspect of working for these companies. And I think it would be better for the community to have a Google/Facebook/Microsoft/etc. developer contribute money or time to SSB-related projects than not doing it. And it may even be good for the community as a whole if some of these companies started using SSB internally and contributed back developer time or research attention in a way that benefit everyone. We just need to be careful that the value created by the community will not be enclosed by a single entity.

@runKleisli
Voted this
Join Scuttlebutt now