[kkielhofner]

Except for the (currently) 5,505 Ethereum nodes[0] that are currently online...

If you (optimistically) assume that each of these nodes consumes 10w of power (that's an extremely low estimate - not every node is a Raspberry Pi) that's 55,0000 watts - or 55 kilowatts, all for a network that can process a maximum of 30 transactions per second and is used by a tiny portion of the population.

That's still absolutely abysmal efficiency. Nowhere near the estimated 150 terawatt hours of power consumed by PoW bitcoin but still completely unacceptable and ridiculous for a network where the ATH for daily active addresses was roughly 1.5m and the daily average hovers more around 500k.

481,000 kWh per year and again, looking at node stats the vast majority are amd64 so the real number is at least 5-10x this.

[0] - https://www.ethernodes.org/

[yokem55]

55 KW is a few racks in a datacenter. Figure 300w per u in a 42u rack, and 4 full racks will use the same amount of power.

[kkielhofner]

Precisely. How many users and traffic can be served by that infrastructure, and with what functionality (outside of blockchain)?

A lot more than 500k daily users interacting with smart contracts or updating a ledger at 30 tps...

The old version of the Netflix OpenConnect Appliance uses 650 watts of peak power to store 360TB of content and stream at 96 Gbps[0].

Assuming an average bitstream of 4 mbit/s (relatively high) that's 24,000 video streams on 650 watts. Over 2m people STREAMING VIDEO inside of 55 kW.

The new version is 190Gbps on 400 watts - at least 5-6m people streaming video inside of 55 kW.

[0] - https://openconnect.netflix.com/en/appliances/

[yokem55]

Those 30 transactions per second settle multiple billions in value per day. Even in a bear market folks are spending about a billion per year in transaction fees to have those transactions settled. And a good chunk of that power usage is just the value folks get out of being able to read the state of the chain without writing any transactions to it.

I get that you don't see the value, but plenty of other people do.

[kkielhofner]

I see and hear this a lot (and the data clearly backs up your numbers) but what I can't figure out is - billions of dollars in transactions per day for what realized real-world value, exactly? A billion dollars in fees for what utility?

With > 5 billion internet users as a TAM overall adoption for crypto other than trading on exchanges (which is almost completely off chain and essentially has nothing to do with blockchain, really) is miniscule. Every block explorer, DappRadar, MetaMask stats, etc - literally any data shows completely insignificant user adoption (active addresses, number of transactions, etc) when compared to the billions of users that could become blockchain users by installing a piece of software in a few clicks.

I know it sounds like I dismiss it completely but I've been active in this space for the better part of a decade and I still can't figure out where the use case is -or- better yet, the killer app that would lead to user adoption resembling anything even close to the web in the 90s. It's been 8-15 years (depending on your starting line). Where is it? Where are the millions (really should be billions by now) of Joe Everyone daily active users?

Absent anything approaching a semblance of widespread adoption (from the data I referenced) the cynic in me says most of the "value" you reference is overwhelmingly from wash trading and a lot of other unproductive activity with no value or utility other than whatever angle those actors have.

[yokem55]

If it all is wash trading, it is a lot of money to be spent on it. The fees assessed by dex LP's and gas fees would make anyone doing a substantial amount of wash trading lose money quickly and the activity would burn out.

The 'real world' stuff naturally happens in the real world, and not on chain. An example - last year I used a lending app to cover some of my taxes for a few months. I deposited asset X that I didn't want to sell at the time, borrowed their bespoke stable coin against it, swapped that for USDC, withdrew the USDC to coinbase, swapped it to USD and put it in my bank account to pay my taxes. When I could pay it back, I basically did the reverse. And at the end of the day because the lending application was offering 'borrow incentives' with their governance token, I ended up slightly ahead in after paying their repayment fee and gas fees. And all this was done with only substantial human counterparty risk when it came to interacting with coinbase. I found that useful and certianly helpful to me in a real world way. But if you only looked on chain it would not look like anything in the real world was happening at all and just lots of virtual bits moving around.

[kkielhofner]

The ingenuity of financial activity (especially with regard to fraud) never ceases to amaze me. While I can't come up with a wash trading practice that's net positive a lot of research indicates plenty of people do[0]. The provided reference talks about unregulated CEXs but one can only assume given the chains themselves have no regulation whatsoever I can only assume the practice is rampant there as well.

I just can't understand where the monetary volume you speak of comes from when a tiny number of merchants, platforms, etc accept cryptocurrency payments and (again) the number of transactions, addresses, etc just don't line up. As one example Venmo processes similar monetary volume with similar fees[1]. The difference is they have roughly 78 million users and every time I open the app I'm greeted with a running list of all of the people I know sharing their various (and copious) often daily transactions. The numbers for Cash App and others are similar. ~1B$ of daily volume for 500k - 1m users vs (assuming crypto address = user, which it does not) vs ~1B$ of daily volume for 78 million users... Yeah, something is going on with crypto.

Thank you for elaborating on your use case as an example of "shuffling bits around". Of course I'm not familiar with the situation but again (speaking to my lack of imagination/creativity) I can't imagine undergoing what ends up being a complex 12 step process involving multiple entities to obtain what is essentially a short term loan. Interesting and cool you pulled it off but not exactly something I see the masses being able to accomplish.

[0] - https://markets.businessinsider.com/news/currencies/what-is-...

[1] - https://www.businessofapps.com/data/venmo-statistics/

[aortega]

It's useless to reply to that guy.

[aortega]

>t still completely unacceptable and ridiculous

55 kilowatts is a single car. One car. How many datacenters has your bank? You have a very mistaken idea of computing power and efficiency if you think a Raspberry Pi is more efficient than a server CPU.

[kkielhofner]

What’s next? Comparing square footage of homes? Airline miles? I try to stick to things that are even remotely comparable.

My bank has 67 million customers and hundreds of products. The average US adult does 2.3 financial transactions a day. They also have branches, customer service reps, etc because real people in the real world need those things.

I’m sure they have much more compute because they’re handling many orders of magnitude more volume at transaction rates no blockchain could comprehend (let’s not start on Solana, please).

They also have the novel concept of complying with things like regulation, compliance, the law, and actually doing something about fraud. This also takes significant resources.

Yes, given that a Raspberry Pi is capable of running a full Ethereum node on an individual node basis at 10 watts all in it is the most efficient. All 5500 online, distributed, Ethereum nodes utilizing an aggregate of 55 kW is the most efficient power consumption scenario.

A server can’t keep the ECC RAM warm with 10 watts.