Imagine how much CO2 we could NOT emit into the atmosphere if we just yanked all the superfluous, useless and demonstrably harmful workloads from all the clouds and computers on this planet (instead, or in addition). No training your neural net for thousands of accumulated [CG]PU-years to make racial profiling, droning could-be-"combatants", or targeted advertising more effective. No high speed trading any more for ad real estate on the web, stocks, or natural resources.
I so, so wish that in many cases, people would perceive the actual value of simply NOT doing certain things. But we're spiraling into a lethal wet-bulb future on the back of so, so many things we not only could perfectly well do without, but that actually could never, in no conceivable scenario, have a net-positive effect on society in any way. What a darned pity.
This is because we don't price energy and raw materials as much as we value them. Putting a higher price on extracting natural resources will swiftly lead to less waste and less frivolous uses as they then become uneconomical. But as of now, compute is cheap and available, power is cheap and available; what else than wasteful use would you expect?
Hopefully not that corporation would somehow feel morally obliged to abide by some arbitrary standard and make less money along the way. But that is effectively what you are suggesting.
As an interesting aside to carbon taxes and such (which appear to be working to a degree):
Was shopping for ETFs a while back and came across the WealthSimple SRI portfolio which is based on their funds. They use a normal ESG index but it also de-lists any companies in the top 25% of emissions for their category.
This would seem to help avoid some collusion and complacency in industries like software that are just at a natural advantage in absolute terms compared with, say, coal plants.
Of course, it's a retail investor and consumer driven thing which leaves too much on the table for some, but I think there's something to the idea of getting companies to compete on the rate of improvement relative to each industry, in addition to internalizing some of the costs through taxes.
Interesting take on how to value something, even if it is a top performer but wasteful in how it uses resources, the lack of efficiency make it rank lower.
Thing is, most top of category performers in anything are wasteful in some resource. I think kinda by definition. Lets go way out into left field and say even top performing artists and athletes are wasteful, because the slope of the gain to expensive curve flattens the higher up you get.
I think for large orgs to be high efficiency it has to be baked into their culture. If you don't have high efficiency, the worst thing you can do is also get large.
I think the gamedev world could be akin to digital coal, toxic work env, low pay. Your entire labor force could walk/pivot to something else if better any dimension you compensate them on changes.
I sometimes wonder how much energy Windows Update consumes on a global scale. I have some old machines and updating can mean they are run at 100% for hours, multiply this by every Windows PC and it's a lot of energy.
I finally dropped Windows 10 from my HP Omen laptop a couple of months ago and installed Linux (had to have it to run Visual Studio until recently). The machine now runs cold when in the past it was distinctly warm all the time even when I had not been using it for hours.
I suspect that dropping Windows altogether has cut the consumption of his laptop by more than half.
I never do any graphics intensive stuff so I have no idea.
But perhaps that is what happened. But that really just makes it even plainer that Windows is very inefficient. Why would enabling the GPU cause high power usage when the machine has been sitting idle for hours with the screen off?
Not really doable, unless you're prepared to take it up with the largest consumer of coal, China, which consumes 51% of it. The US is distant third, after India. In fact as of now the US derives more energy from renewables than from coal (mostly hydro and wind - share of solar is abysmally low). That's the trouble with Western environmentalism. It's glorified NIMBY. Just because we destroyed on-shore manufacturing doesn't mean coal won't be burned elsewhere to make stuff people buy. If anything, more of it will be burned, since the environmental regulations aren't as strict there.
It’s something to take up with every nation emitting more than North Korea, given the scale of the problem.
In that regard, I simultaneously agree that bitcoin is a pointless waste we’d be better off without and also a rounding error compared to the single biggest and easiest wins we can start with and also that despite that it needs to be stopped (I am skipping a few of the steps for why that is still the case even if 100% of power was renewable).
Edit: I phrased that wrong. The maximum sustainable global emissions are approximately North Korea’s 2017 emissions. The sum of every country emitting less than NK is about 28 times what NK emits.
Typical american response, ignores population and history to point the finger at China.
I am not sure if all parties in US run this propaganda on all TV channels? or how do this people know to copy paste same bad statistics in the hope to delay doing something.
Even if your country is number 21 you should still do something proportional with your population and on how much damage you cause.
And for the historic part, if we could calculate how much country put CO2 and methane int he atmosphere they should proportionally put money in a fund for investments in poor countries to help them skip over the coal stage.
I think you have some issues with reading comprehension: I specifically blame the US in this case, for simply moving all these emissions to China instead of dealing with them for real and right here. We share the planet. Simply moving this crap around doesn't fix anything, and never did, if anything it exacerbates the atmospheric effects.
Also, China is not a "poor country" anymore. It's time it started setting money aside for "investment in poor countries", too, rather than build hundreds more coal power plants in the next five years.
Are you sure your text "US is only number 3" is not open to interpretation? Anyway I seen tons of similar bad arguments, that ignore the population , put China on number 1 and then say "let's do nothing until China cleans shit up and then until India cleans shit up and maybe after that we can find some other pretext not to do anything", same for plastics .
What each country could do is fix their local shit, as I said even if you have 1 coal plant, that smoke affects your local people and not someone on the other side of the world, fixing your local coal burning should still happen even if you are greenest of the green. Then if everyone would be logic and not blame the neighbors you get your clean air and less CO2.
To China's credit, they've just announced a ton of nuclear investment as well, something the United States no longer has the balls to do. This is (currently) _the only_ real solution to CO2 emissions that we have as a species, short of throwing billions of people back into poverty. The US prefers to pretend that solar is "it". But it's not, and it's never going to be, although its share will grow, of course.
All I'm saying is I don't think the world leaders perceive this as a problem at all. They perceive it as a money making opportunity, for themselves and their donors (maybe with the exception of Xi Jinping who doesn't need the money since he's not replaceable, and he can just take whatever he wants). Otherwise they wouldn't have flown to Glasgow in _four hundred_ private jets and wouldn't have driven around around in hundred car armored motorcades. A videoconference would more than suffice for what they "accomplished" there.
President of the EU, Ursula von der Leyen once took a plane to travel 31 miles [1]. That is not the behavior of a person who believes climate change to be a "problem".
> But it's not, and it's never going to be, although its share will grow, of course.
Unless we make fusion (or spacemagic which beats it), the future can’t be anything else. I’m expecting the world of 2121 to be 95% PV on an antipodal power grid. (It would take a while to build at that scale, but the sun never sets on such a grid).
Regardless of the progress on fusion and whatnot, China's move will force the US re-enter nuclear in the short term, if it decides to offer more than just platitudes to "climate crisis" goals. It's the only clean, abundant energy source supported by the sane and rational subset of both of the main parties.
> rather than build hundreds more coal power plants in the next five years
China's coal plants are a combination of two factors:
- Inertia/entrenched interests (i.e. it makes powerful people more wealthy). This is certainly a problem, but it's a hard problem to solve when even Xi Jinping's authority isn't enough to prevent it.
Yeah, and a lot of the transition is about pushing down prices of renewable energy solutions so that they become more competitive. It happens when those technologies gets adoption. So even a small country should be able to make a big impact if they adopt green technology early.
Prices of photovoltaics for example has gone down to a large degree thanks to early adopters and stimulus. Now it's quite competitive without such help. But it's not the only piece of the puzzle. Other technologies deserves to be pushed towards competitiveness as well.
Apparently, even all of computing is a small factor in energy use. (maybe bitcoin mining aside)
I too cringed hard when imgur and mp4/gifs started to pop... amounts of GB streamed just for short lived memes. It felt insane but all in all the value seems to be low.[0]
We got to focus on industries, globalization and heating apparently.
[0] I'm all for frugal minimal computing. I want to make a milliwat dumb terminal to be used as administration application client (low lag but low refresh TUI style). I'm sure 80% of tasks can be achieved through this. For the occasional 2D/3D user interaction one can still pop his 2020s smartphone
I have only read the abstract, so maybe I'm missing something, but how would moving workloads to different times really lower CO2 in the long term unless it's using surplus green power? They talk about "less carbon-intensive" supply, but if it's still 1% coal power at a particular time then using more power at that time will require burning coal to satisfy demands? It would seem to only help to place workloads at times where it could use up otherwise unused solar or wind power.
Also, leaving hardware idle is both having a CO2 cost, too (production of hardware that's not used), and, other people who are not CO2 conscious will probably move their workloads to those times where the hardware is under utilized and hence cheaper. So, the whole thing is a zero sum game, making one company's workloads appear greener at the cost of making other companies' workloads worse?
Co-author here :) Regarding your first point: You are talking about so-called marginal carbon intensity - and you are right, this would indeed be a better metric for incentivizing scheduling. Unfortunately it is extremely hard to compute (or better: to estimate), we discuss this in Section 3.4 of the paper.
Regarding the second point: This paper is more written from a data center operator or service provider perspective. They could, for example, (financially?) incentivize their users to give them some slack in when to schedule workloads.
Simply put: Computing should be expensive when energy is dirty. Governments could incentivize this via carbon pricing mechanisms (that are actually currently discussed at COP26)
In section 3.4 you end with "we consider marginal carbon intensity to be no practical signal for demand management at this point due to high uncertainties". But for the direct actual ecological value that would be the only number that counts, according to my understanding, so basically you're agreeing that right now temporal shifts have no ecological benefit? It appears that you hope that in the future, when marginal carbon intensity data might become available, and temporal shifting actually has value, this is a useful approach, and that as of today, preparing for that future is worthwhile. I can see that point.
I think this should be pointed out--this changes the proposal from something that looks like pseudo actionism (that I'm sure some companies' PR departments will still love) to an actual valid plan, even if it doesn't have immediate benefits.
PS. also, the marginal carbon intensity benefit still has to outweight the CO2 cost for buying more hardware (to afford running the jobs at those time points, and consequentially leaving the hardware unused at the corresponding counter times). Have you looked into this?
That was not was I was trying to say: Overall, we _should_ try to shift workload towards times where the average carbon intensity is low. Statements like "energy is usually cleaner at 1pm than at 7pm because of solar power" absolutely hold true. It's just that from the perspective of shifting a single compute job, marginal carbon intensity could be a better metric.
However, imho that's super speculative: The question which power plant will produce the "next" kWh of energy is very complex. It depends on a variety of factors like energy prices, weather (forecasts), expected future demand, contracts between energy providers and grid operators, etc. Additionally, one also needs information for all neighboring regions, as fluctuations in supply and demand are often regulated via cross-regional flows. That means, also in the future we will most likely not have a perfect methodology for estimating marginal carbon intensity.
Regarding the hardware: No, we did not look into this in this paper. The goal should not be to buy more hardware and data centers are barely ever used at full capacity. Also flexible workloads are not only important for carbon footprint (think of Spot VMs/Instances). However, real life resource and business constrains will definitely lessen the potential of savings.
Maybe, but maybe it should be what you should say? I'm consciously trying to nudge you into this direction.
> Statements like "energy is usually cleaner at 1pm than at 7pm because of solar power" absolutely hold true.
Sure. But is using more energy at those times also cleaner? I provided for a logic argument that it is only if there's surplus green energy at that time, i.e. if the need doesn't have to be satisfied by increasing use of non green energy sources during those times. After which you led me to the term of "marginal carbon intensity" which seems to express exactly this, which I'm happy with, so I said then that's the only number that matters, and if you don't have that number then currently you can't draw conclusion about the whole being greener through the shifting.
There's an argument to be made that one should move loads to 1pm since then the country could increase build out of solar energy, since there will be users of the surplus (instead of it needing storage to absorb, or left unused, or being used by more dubious consumers)! That would again be a good argument for the future, and I agree with it. But that's an argument I haven't seen you make--maybe you should.
> The goal should not be to buy more hardware and data centers are barely ever used at full capacity
Iff data centers are normally lesser used during times where the marginal carbon intensity is low, then yes, cool, no new hardware needs to be bought.
Outside of that, of course you could just ignore the need for more hardware and argue that you only shift so little load that it is insignificant compared to the reserve. But that would not be a plan for serious changes, just free riding? Maybe it's the same thinking that leads you (and others, you cite Google in the paper) used to argue that "at 1pm the mix is cleaner, hence use energy at that time": if the amount of load shift is very small, you are tempted to argue that the mix doesn't change, hence the shift is beneficial. But again, this is just ignoring an effect, since evidently if you shift a lot then that won't be true any more. I'm sure it's tempting to argue away the effect, but that's what greenwashing tries to do in general, and may be what makes a lot of people in this discussion angry. So better distinguish it from greenwashing as well as you can.
P.S.: just using up the reserve might of course be fine if moving away the shifted jobs again in case there's high non-shiftable demand coming up works quickly enough. (But it will be interesting to know how much that would be.) I don't really want to tear down your research, and some kind of simulation package to find out how much potential there is (and under what assumptions for the future etc.) would surely be cool. And I guess I've reached the point where I have to read the rest of the paper :)
First of all thanks for your extensive answers and interest in this topic :)
> There's an argument to be made that one should move loads to 1pm since then the country could increase build out of solar energy
That's a very good argument and I think you got the point of what I'm trying to convey: Marginal carbon intensity is a nice idea but extremely hard to compute in practice because you would need to anticipate decisions of many actors. Average carbon intensity, on the other hand, is a very good metric for shifting loads because you increase demand at times where there is many renewables feeding into the grid.
What would be even more interesting is actively exploiting the fact that sometimes energy is actually "for free" e.g. wind turbines are sometimes "shut off" if the grid cannot handle more energy because they are easier, faster, and hence cheaper to start/stop than fossil-fired power plants. I guess there is a high correlation between these events and low average carbon intensity, but I don't have any data on it.
> of course you could just ignore the need for more hardware
I completely see your point, but the available capacity could just be a hard constraint to load shifting efforts. An approach that is different and that I just came across recently is the Zero-carbon could (http://people.cs.uchicago.edu/~aachien/lssg/research/zccloud...). It's covered also in this talk: https://youtu.be/rX5H0oCxnfc. It's a very holistic view on the topic and somewhat overcomes the problem that we cannot reliably compute marginal carbon intensity for the entire power grid, by just optimizing for the stranded power in parts of the grid
I'd love for providers like DigitalOcean to show me the CO2e of any VM's + networking infrastructure I'm running. Optimally it'd include the entire LCA (so including manufacture).
If only we stopped up energy by making new GPUs and CPUs solving puzzles for random cryptocoin that is not a real currency used for trade and simply a gambling game of speculation. I wonder what would net energy saving be if crypto currency was deemed illegal - as in making it worthless by criminalising the act of converting *coin to any real currency.
Don't stop there. This is what I'd like to know by zipcode as well. What's the best time to charge my wife's Tesla, run large appliances, or GPU workloads in my garage. I bet there are a lot of consumer workloads that could easily be "shifted", and in some states that'd probably have non-trivial CO2 savings.
There's APIs that exist for this, including both current and 24-48h prediction.
Some EV chargers have had this built in for years, though it mostly works by time-of-use tariff. If you don't have one of those (or you want to be extra green) currently there's few end-to-end consumer solutions. The framework is there is most places though, probably just waiting for mass adoption).
At this rate let's just switch everything off. Get rid of everything and become hunter/gatherers again. This climate change fear mongering is getting really tiring. All these green changes ever do is make normal people's lives miserable.
CMV: claiming that computers produce CO2 makes no sense. Computers cannot produce CO2, they can only consume energy made by CO2-generating processes. Isn't the real way to stop producing CO2 to migrate to methods of producing energy that are better at mitigating this?
Furthermore, are computers even that power hungry? I cannot imagine they account for more than a tenth of our energy use.
What are we to make of, e.g. Google’s claims that their data centers run on 100% renewable energy? How can temporal workload shifting help reduce carbon emissions in the cloud if the energy source is already renewable?
Co-author of the paper here. Google currently only buys enough renewable energy to match their annual electricity use - their data centers do not run on 100% renewables all the time. You can read up on this here: https://www.gstatic.com/gumdrop/sustainability/24-7-explaine...
However, they are pushing towards operating on carbon-free energy 27/7 by 2030. For this to work, they will need to (and apparently already do) implement approaches exactly like the one mentioned in the paper.
I so, so wish that in many cases, people would perceive the actual value of simply NOT doing certain things. But we're spiraling into a lethal wet-bulb future on the back of so, so many things we not only could perfectly well do without, but that actually could never, in no conceivable scenario, have a net-positive effect on society in any way. What a darned pity.