A useful PoW without replacing Nakamoto Consensus

[n0nce]

energy problem.

CPU only algorithms are one of the best places to go

Have you compared the energy efficiency of CPUs against ASICs?
And have you compared the CO2 footprint of residential energy mix to the CO2 footprint of an ASIC farm's energy source?

12% of the electricity in the US is renewable; while Bitcoin ASIC mining operations use around 70% renewable.
https://www.eia.gov/energyexplained/us-energy-facts/
So suggesting that mining farms are an environmental problem makes no sense.

In December 2019, one report suggested that 73% of Bitcoin’s energy consumption was carbon neutral, largely due to the abundance of hydro power in major mining hubs such as Southwest China and Scandinavia. On the other hand, the CCAF estimated in September 2020 that the figure is closer to 39%. But even if the lower number is correct, that’s still almost twice as much as the U.S. grid, suggesting that looking at energy consumption alone is hardly a reliable method for determining Bitcoin’s carbon emissions.

Also do consider that CPU mining farms would fuck up the consumer PC market even more than the GPU mining of Ethereum did in the past. You could run a PC without GPU, but not without CPU.

[1686298218]

1686298218

[1686298218]

1686298218

[ir.hn]

Many good points n0nce.

Firstly, when you talk about energy efficiency of CPU vs ASIC, you are probably talking watts/hash.  This isn't a valid metric because the algorithm determines how many hashes is "alot".  A slow algorithm could provide just as much security at 100 hashes/second as a fast algorithm at 1 billion hashes/second.  Hashes/second does not matter, what matters is how difficult it is to achieve 51% of the network hashrate.

Mining uses a lot of power.  It has been discovered that the reason for this is (of course the value of the coin and the block subsidy) but also the low CAPEX/OPEX ratio of SHA256 hashing.  In other words, the equipment to hash a tiny SHA256 input is very basic.  So this framework (an asic chip) can be produced cheaply and fed (with data and power) easily so that dozens or hundreds can live on a single board.  They can run reliably at high temperatures and speeds.

This leads to a massive ability for people with modest capital and lots of cheap energy to scale almost indefinitly.

On the other hand with CPU mining, CPU's are complex and difficult to feed (you often won't find more than one on a motherboard) and the capital (CAPEX) versus power cost (OPEX) ratio is very high.  You simply can't easily scale up to 100 computers (the equivalent of 1 asic hashboard) very easily.  Especially when these large miners can't dominate the market, everyone with a laptop from angola to zimbabwe will be competing with them.

For an example I am CPU mining as I am responding to you, browsing the internet, watching videos, there is no noticable slowdown. Wheras with GPU mining you can't use the computer that is running the GPU mining. Yet this dual use of CPU mining and browsing the web is producing 1/3 the coins as an S9 miner would produce at a fraction of the power usage and heat production.  So I don't think it would hurt the consumer market, in fact it would benefit it by not diverting extra fab capacity to unnecessary GPU chips or ASIC chips.  Especially with the moore's law memory req scaling, consumers would always have, at the very least, last gen of CPU's on the secondary market to buy up.

A little off topic, but that was one of the problems with GPU paradigm was that the memory scaling for EThash was linear not exponential, so GPU miners could basically use all the cards that have been on the market for the last 10 years.  Not so if you double memory req every 2 years then miners would only be able to use the newest cards.

[NotATether]

On the other hand with CPU mining, CPU's are complex and difficult to feed (you often won't find more than one on a motherboard) and the capital (CAPEX) versus power cost (OPEX) ratio is very high.  You simply can't easily scale up to 100 computers (the equivalent of 1 asic hashboard) very easily.  Especially when these large miners can't dominate the market, everyone with a laptop from angola to zimbabwe will be competing with them.

So I was looking at the prices of some dual-socket and quad-socket servers that also got some good memory inside them, and if you put a few of these together, you're staring at the price of an ASIC, but nowhere near it's performance.

It's all in the circuits. I'm not a hardware engineer or anything like that but the more stuff you can push to transistors directly, the more performance you'll get in any application.

And if you can't afford the short-time cost of accelerating a formula like that (likely because the problem will be irrelevant soon), then FPGAs are a nice alternative that gives you some flexibility in wiring it without sacrificing too much performance.

[tromp]

This isn't a valid metric because the algorithm determines how many hashes is "alot".

There's more to mining than hashing [1].

Not so if you double memory req every 2 years then miners would only be able to use the newest cards.

Then you'd want to use a memory-hard PoW that can be instantly verified with no memory, or PoW verification
will be unbearably slow.

[1] https://web.archive.org/web/20210506232722/http://cryptorials.io/beyond-hashcash-proof-work-theres-mining-hashing/

[n0nce]

A slow algorithm could provide just as much security at 100 hashes/second as a fast algorithm at 1 billion hashes/second.  Hashes/second does not matter, what matters is how difficult it is to achieve 51% of the network hashrate.

How difficult == how much power or hardware do you need. Therefore, for achieving the same level of security, you need to invest the same amount of real-world capital in form of electricity and hardware.
Security is directly proportional to 'waste'.

On the other hand with CPU mining, CPU's are complex and difficult to feed (you often won't find more than one on a motherboard) and the capital (CAPEX) versus power cost (OPEX) ratio is very high.  You simply can't easily scale up to 100 computers (the equivalent of 1 asic hashboard) very easily.  Especially when these large miners can't dominate the market, everyone with a laptop from angola to zimbabwe will be competing with them.

I understand that ratio, that makes sense. However, your numbers are a bit off. If we assume a 3kW miner that costs roughly 3000$ (it was over 10,000$ at the top): The S19 Pro.
For $10k, you can buy 10 rigs consisting of a motherboard, 12900KF and a PSU. The chip pulls 250W, probably 300W at the wall; achieving a similar ratio as with the ASIC.

With current ASIC prices (same machine going for $3k), the ratio may be 3x better on CPU mining, but it's not orders of magnitude better.

That's assuming nobody will build mining motherboards with 4 chips each or something like that. You know... like nobody expected SHA256 ASICs at the start.

[thecodebear]

Would number of transactions change if a year from now mining was 1/10th what it is today, or 10x what it is today? Nope.

You have the causality all wrong.
Amount of mining depends on amount of transaction fees, not the other way around.

With no block subsidy, if the amount of txs drops 10x, then there will be 10x less rewards available for miners  (actually even less if you go from (nearly) full blocks to never full blocks) and so mining must decrease 10x to remain profitable.

It's worrisome that senior members can get such a trivial implication all wrong.

A block can contain 4,000 transactions, or it can only contain 1 (e.g., mined right after the previous). Furthermore, a block that took very little time to be mined likely used less energy than one that took longer, while both blocks can have the exact same number of transactions.

Of course I'm not talking about individual blocks, but about longer term trends.
I'm also (like the poster I replied to) talking about on-chain transactions only. With L2, the number of off-chain transactions becomes completely unrelated to everything, as you could do a trillion off-chain transactions for 2 on-chain ones.

Sorry but you've still got this all wrong. Again, number of transactions has no relation to the amount of energy used for mining. There is zero relationship there.

Blocks are always going to be filled up no matter how much energy is used to mine those blocks. You're confusing number of transactions with cost of fees. Fees have to do with the size of transactions and how many transactions are trying to get into the block, not how many transactions are actually put in the block, because that second number essentially stays the same. We can safely assume that Bitcoin blocks will continue to be filled up in the future as they have been for the past like 6 years or whatever. So there is no real change in the amount of transactions happening per block.

If the case you are trying to make is true then mining would simply stay static forever once the block reward is done and mining only gets transaction fees. Obviously that is not the case. Energy use will go up or down with how profitable Bitcoin mining is. That relies on how many people want to get into current blocks and what the price of Bitcoin is, and then of course on the miner side things like price of electricity, efficient of mining machines, etc. Nowhere in that equation is the number of transactions.

During periods of Bitcoin mania when everybody's making transactions, electricity that goes into mining will go up even though no more transactions will be getting processed on-chain, because fees will go up.  In slower periods the same number of transactions (or more exactly, amount of data per block) will be getting processed but if the mempool isn't bursting transaction fees will be low and therefore electricity used will likely go down.

Again this is why education is so important. If you can't get this right how do we expect the casual politicians and voters and investors to understand this stuff?!
This is why it's important to state over and over and over again that any calculation done about energy per transaction is wrong and simply a nonsensical calculation to make. Now and into the future.

[oryhp]

Sorry but you've still got this all wrong. Again, number of transactions has no relation to the amount of energy used for mining. There is zero relationship there.

Let's say a block can accept exactly 1000 transactions. Imagine we have two forks Bitcoin1 and Bitcoin2 both of which are valued at $10 per coin and have the same supply.

Bitcoin1 has blocks with a single transaction along with a coinbase output.
Bitcoin2 has blocks with 1000 transactions all paying the minimum fee to cover their transaction size.

Which one do you think secures more energy per block?

[tromp]

Amount of mining depends on amount of transaction fees, not the other way around.
With no block subsidy, if the amount of txs drops 10x, then there will be 10x less rewards available for miners  (actually even less if you go from (nearly) full blocks to never full blocks) and so mining must decrease 10x to remain profitable.

Sorry but you've still got this all wrong.

Sorry, but your long rambling failed to show anything wrong in either statement.

Blocks are always going to be filled up

Plainly false even as we speak...