Sustainability NNS Proposal

Thanks for this Jesse. I just want to throw this related idea into the mix: Zero-carbon subnets


Thank you, @icpjesse

This is brilliant. I’ll be following the discussion with interest and voting as appropriate.


Cool idea @icpjesse, I would vote in favor of this proposal.


Feedback on the three bullet points:

Firstly, point 2,

I don’t think this is necessary. Beliefs fall under the same category as Opinions, and some people’s opinions will never align with the majority. Focus on the facts.

Your first bullet point:

This is a clear, “Yes.” In terms of type of assessment and internal, or external; here is a paper produced by Tezos, though it was privately commissioned it uses the review process, which is absolutely essential: . All this stuff will take time, commissioning a private study like this would seem the quickest way to get the ball rolling.

Regarding, “…where that electricity is sourced…” IPFS is working on a progressive approach. They intend to issue Bakers using renewable energy with certificates and IPFS users will be able to choose these Bakers as custodians of their data Filecoin Green

Transparency is key, it therefore feels appropriate to suggest Dfinity works with several Universities with leading climate and crypto departments from different parts of the world to carry out a range of studies. It may be that other blockchains are more efficient, but that is not the point (IMO). The first objective is to design a provably robust reporting mechanism so people know IC emission data is true, the second objective is to establish that the IC is less impactful than traditional/legacy systems.

Not all truths will be palatable, but without the information appropriate decisions cannot be made.
More broadly, using universities could enable comparison with compute and storage on AWS, comparison of streaming on YouTube and Spotify, comparison of payment transactions with Credit and Debit Cards. This kind of data may assist Developers with their architectural choices (IDK, but it’s a starting point).

This is also a, “yes.” However, it probably follows some of the above.

Accepting that there will be a climate impact is necessary. Nothing is entirely “Carbon Neutral” and one has to be mindful that the public isn’t stupid. Therefore, Carbon Offsetting is a tool that needs to be used carefully (We’re living in a golden age of greenwash | Greenpeace UK).

Finally, what IF… the Community Fund included a pool of money Universities studying climate impact could apply to for (say) $10k grants to top up nearly funded studies?


@NickM Just wanted to drop a note to say thank you for such considered and nuanced feedback. I’m about to post another comment specifically on bullet point two, and I’m in agreement on updating it. I’d welcome specific language suggestions for how to rewrite it for the same resulting impacting decision making prioritizing sustainable solutions, but without the belief part. Or would you say simply cut it altogether?


Pulling in this conversation between myself, Coolpineapple18 and @wpb on twitter which I think has excellent feedback (edited to remove usernames for readability)

threaded from this tweet:

:pineapple:I personally believe that the climate and biodiversity crisis is the biggest threat we face and like the idea of making the #IC the greenest blockchain. So I generally support the concrete steps to monitor consumption, have a dashboard and target reductions…

:pineapple:…However I’m uncomfortable with the NNS ratifying a belief. IMHO global warming is a fact supported by a huge amount of scientific data but I don’t think the NNS should be in the business of making statements. The point of blockchains is credible neutrality.

:pineapple: If we open the door of making non technical statements of belief then we have become then things get very political very fast. As individuals we may stand with Ukrainian, believe in global warming, support BLM… ect but a technical system should not be in the business of belief…

:pineapple:we should be a neutral platform for the world with certain technical properties - including energy efficiency - not a DAO with beliefs that takes a stance on issues. If we do that we lose people countries and cultures who disagree as potential users.

Then @wpb chimed in:
I can agree with your concern about ratifying a “belief” and wouldn’t mind seeing that language changed in the proposal to something that would be perceived to be more diplomatic. However, there is a strong marketing opportunity here that makes the proposal worth pursuing.

:pineapple:I would just edit the middle bullet point to make environmentally sustainability an ongoing objective. (Similar in framing to long term R&D proposals on other topics Motion Proposals on Long Term R&D Plans)

:pineapple: BTW: There is some tension with performance and security eg replication factor but we can frame this as an objective “how do we improve environmental sustainability without sacrificing security and performance” and I think that question could lead to interesting designs.

1 Like

Cut it.

And I say this with an itchy typing finger as I am a sucker for a debate! But let’s not create a drain on headspace for countering individual opinions.

FYI: I am a fan of the Trivium, which structures arguments into three parts:

Grammar = identify the object of the discussion. A “Sustainability NNS Proposal” will provide transparency of the Internet Computers energy usage. Metrics can include, txts, storage, etc. Energy usage of the InternetComputer is currently unknown. A “Sustainability NNS Proposal” addresses the absence of such data.

Logic = examples of the grammar in action. Other chains are reporting energy usage of txts, storage, etc (examples) and are delivering reports (examples). This enables comparison of energy usage (examples). Those blockchains reporting Energy usage are attracting attention from energy conscious developers and users.

Rhetoric = personal opinion. This always come last! IMO transparent reporting of the InternetComputer’s Energy Usage MAY provide dev insight into more efficient tech/smart contract solutions and MAY help society address the Climate Crisis. BUT we will only know this once we have access to trustworthy data.

Trivium - Wikipedia.

1 Like

Hi All! Thanks to everyone who has provided feedback on here or on twitter, especially @NickM @diegop @alexa.smith @wpb @Kyle_Langham and CoolPineapple on twitter - your time and effort to engage in this project has been super helpful, and as always reminds me what makes the IC community and the NNS unique.

My edits are minor and simple: I’m eliminating the second bullet point about belief and climate change - I’d love to include it, but would rather separate it out into a separate proposal down the road with more action attached to it for long term sustainability planning. This proposal works best as a first step to gather data and set a broad direction.

My updated bullet points are below - please feel free to submit more feedback and we will move this to the NNS shortly for voting! Because the motivation section is unchanged, I’m omitting it from this part, but feedback there is also welcome.


  • Jesse

What about those of us that just want to see the world burn…


a good idea is a good idea…


1 Like

Me: I have a lot of work to get to today.
Also Me: ok one more meme but this one has @Kyle_Langham



Will follow updates with interest and look forward to participating in votes!


This is Luis from DFINITY.
My team was responsible for supporting node providers with building the 1st generation of data centers founding the infrastructure of the IC.

The infrastructure of the IC runs in professional data centers. The majority of the datacenter providers are already carbon neutral. This is mainly because this market is an oligopoly which doesn’t allow many differences between competitors. This of course means that we have to trust their emission reports.
We thought about making carbon neutral data centers mandatory, but there were different reasons why we decided not to do so:

  • The community wouldn’t be able to verify that and as I said above on the end everybody including the node provider would need to trust the data center provider or the entities that certified them.
  • This requirement would exclude smaller data centers providers that are important for the decentralisation of the network.
  • The footprint of the IC isn’t only defined by the power consumption of the nodes and the cooling. A lot of resources and energy was needed to build the hardware. The intent of the IC is that every node that was added to network will be used as long as there enough nodes to create a subnet. Because they are running in a optimal environment and seven nodes are enough to create a subnet the IC nodes can run for much longer than in legacy cloud platforms.
  • We don’t want to support greenwashing. The sustainability of the IC isn’t lonely defined by the power consumption.

Not sure if this needs an NNS proposal. I would even say that ppl that don’t believe in global warming and its consequences would hardly understand what the IC is.
Speaking from three years at DFINITY I can confirm that environmental sustainability is an objective in almost everything we do. From avoiding traveling and commuting to refurbishing workstations to avoiding plastics in the office.
The interest of the IC is the sustainable use of all resources involved to run a canister. Beginning with hardware and going up to the protocol efficiency. And there is a lot of potential for improvements but different levels of control. The community can improve the protocol efficiency but we can’t control the hardware efficiency. I personally would like to see hardware vendors optimizing the environmental footprint of the hardware. Actively recycling resources and using them to build new hardware for example. Or using reusable packaging for their hardware.

That’s technically hard to achieve. Estimating the power consumptions doesn’t make sense. We would need to enforce smart PDUs in the data centers. We did that with our lab racks. These PDUs are expensive and data centers bill you for installing and running custom PDUs. Some data centers don’t allow them because they need that data themselves to decide about cooling needs based on the power consumption.
Even if we would be able to get that data from every rack running IC nodes it would only show the power consumption of the nodes. The cooling power isn’t measured. Some data centers are highly efficient. Hot aisle and cold aisle are separated and controlled strictly. They have recuperators that are getting back the energy from the hot aisle. Other data centers care much less about the cooling efficiency which can make almost the half of the total power consumption.


Hi Luis -

Thanks for your detailed notes, as well as your work at DFINITY!

I wanted to provide a few questions and follow up to your notes:

  • We don’t want to support greenwashing. The sustainability of the IC isn’t lonely defined by the power consumption.

I picking this out for a stand-in for all of your notes from my first bullet point, but they are all good. Please note this section asks for an audit and report, not a requirement. I also fully support your statement here about not embracing greenwashing - I guess my question is how to do get more info to work with and market IC as a green blockchain without greenwashing? Right now I have almost nothing to work with regarding IC’s carbon footprint. I embrace you wanting a holistic picture, but I also feel like I want more reporting and info than we have now. I’m open to rewording this section asking for a report if you have specific suggestions on how to do it.

Not sure if this needs an NNS proposal.

I have removed the second bulletpoint from my proposal per earlier feedback

That’s technically hard to achieve.

Re: my request on a reporting panel - I want greater visibility and data. This was the solution I proposed, I welcome a rewrite that moves it to a place you and your team can support.

Thank you again for taking the time to read and provide an insider’s view of my proposal - I’d love actionable suggestions on how to improve it to make it something we can implement and act on when it (hopefully) passes.

  • Jesse

I hate seeing someone fall for the climate change meme so I would like to invite you to re-open your mind and play with the idea you have been had. A lot of countries already force their citizens to pay a climate/carbon tax so in the eyes of our rulers who mind you only allow us to use their “approved” power sources the “problem” is long $olved, don’t forget to tip your taxman either.

If the programming is buried too deep within you mind and you cannot let go of their incepted ideas then rest assured you will feel mighty happy when carbon credits get brought in.

1 Like

Since conversation has died down, I will move this to being a proposal on an accelerated timeline. Please leave additional feedback or notes!

1 Like

Thank you for acknowledging this is a complex problem and for sharing concerns regarding gathering data.

Similar concerns are acknowledged by the authors of “The Energy Footprint of Blockchain Consensus Mechanisms Beyond Proof-of-Work” published on April 4th 2022. This study has good provenance, being a collaboration between University College London, Imperial College, Kings College and University of Bayreuth. The concerns, including data integrity and hardware configuration, are approached with appropriate scientific best practice, “We approach this research gap by formalizing a basic consumption model for PoS blockchains.” The abstract is quite short and is definitely worth a speedy read.

The above study compares Ethereum, Algorand, Cardano, Polkadot, Tezos and Hedera.

This Jan 2022 report by Crypto Carbon Rating Institute looks at Total electricity consumption [kWh/year], Electricity per node [kWh/year], Electricity per transaction [Wh/Tx] and Total carbon emissions [tCO2e/year] for Cardano, Polkadot, Solana, Tezos, Avalanche and Algorand. No single blockchain out performs all the others across all criteria, which isn’t surprising

The conversation around energy consumption across all aspects of IT is only going to grow. Therefore, finding ways for the Internet Computer to participate in these discussion is imperative, or once again it will be notable only by its absence.

1 Like

@icpjesse Hold up there I don’t think we solved this whole part of what sustainability data DFINITY thinks it can feasibly get. I have a few questions/potential solutions for @Luis

I understand that it’s hard to understand the full impact, but hopefully we can start with a few basic metrics that are based off two things.

  1. Power consumption from the Boundary Nodes (which I understand are 100% DFINITY controlled atm)
  2. Emissions reports from the data centers that are running nodes on the IC.

This way, DFINITY can present boundary node power consumption estimates with near ~100% certainty, and can provide 3rd party metrics as well. Those 3rd party metrics may be inaccurate, but the goal would be to start by trying to provide at least one or two metrics that are accurate to within an order of magnitude.

I would imagine DFINITY would want to use this information to market the IC, and the more accurate the data from data centers becomes the better. It might even make sense to incentivize the data centers that provide accurate power consumption with additional rewards for the time being.

1 Like

Just FYI I am moving ahead with getting the proposal submitted to the NNS. My reasoning for doing so is two fold:

  1. All of our action items are “Ask DFINITY…” so they are nonbinding and the answers to these questions will need to be solved by DFINITY no matter what. I’m optimistic it passing will help management decide to allocate further resources to help answer the questions more fully.

  2. It’s very likely that the cost of having an NNS proposal fail will dramatically rise in the near future. The fate of this good-faith proposal is unknown, but given the controversial nature of it, failing is of course a real possibility. Given how much discussion here has slowed, I’m motivated to get it submitted. I expect that once live on the NNS this thread will become significantly more active.

By design, this proposal leaves everything in the hands of DFINITY for execution - we are just trying to set a direction and voice this need for another marketing tool (plus save the planet, etc).

  • Jesse

Sorry @icpjesse for the delayed response. We had a long easter holiday in Europe.

I tried to explain that DFINITY can’t provide such details because we don’t have them. Technically the node providers won’t be able to provide that details without installing additional hardware. Even if they would do that these very precise metrics wouldn’t allow us to draw any conclusions about the sustainability of the IC.

You could do an estimation of the current total power consumption of the IC with details that are already publicly visible: The number of nodes per data center and node provider you can get from the ICA’s public dashboard. This gives you a number of active nodes per DC. There you can also find the DC provider/owner and look for their carbon footprint. On the ICA website you can find the specs for the current dual-socket AMD Rome nodes. With these details you can calculate a total power consumption of the IC under full load and how much of the power is provided by DCs that claim to use renewable power sources. It still doesn’t say anything about the sustainability of IC.

Drawing conclusions to sustainability based on such energy source and consumption details, nevertheless how precise they are, is what I would call greenwashing.
Wasting power stays a wastage nevertheless where the power is coming from and wastage is never sustainable. What the power is used for is therefore relevant for the sustainability in the very first place?

Let’s assume we need to run something that is sustainable by nature. Let’s say a car pooling or a second hand exchange portal. The sustainability could then be reduced to the protocol efficiency and its energy source. A comparison between a legacy LAMP setup and the IC would probably already show a higher efficiency if you take the redundancy and geological distribution of a subnet into account. But that’s hard to prove and we are improving the protocol efficiency constantly.
But now let’s assume we need to run something that can only work on-chain. Let’s say a voting machine providing the first democratic elections in an autocracy or an NFT portal giving the exploitation of the rights back to the artists. Here you would need to compare the IC with other blockchain projects where the technical edge of IC would most likely show a higher protocol efficiency.

1 Like