1/ Bitcoin has now captured the imagination as digital gold, but what happens when it becomes 1 trillion or 6 trillion dollars? What happens to the security and the electricity costs since 70% of the hash power is concentrated in low electricity cost places, like China/Iceland.
2/ In Nakamoto consensus, out of 100 nodes, there needs to be a minimum of 51 nodes that are bad actors to reverse the transaction history of Bitcoin. Also, there is a little known attack called a partitioning attack.
3/ A partitioning attack is an attack where state actors collude to seal off the internet so that geographically isolated nodes are only gossiping among themselves. Anything that’s geographically centralized is exposed to these attacks as it scales to trillions of dollars.
4/ Currently the cost of Bitcoin is mining is 20% from purchasing the machine and 80% from electricity and cooling. So 20% of the cost is focused on capital expenditures and 80% is focused on operational expenditures.
5/ It’s estimated that BTC uses .32% of the world’s electricity (cbeci.org) It would need to use 10% to reach a 6T market cap. How is this calculated, assumptions from linearly projecting current trends.
6/ It’s a linear projection with market cap. Today it uses 70-80 tWh, so multiply by 30x for 30x market cap. cbeci.org
7/ This paper talks about an alternative.
8/ In the 90s fiber optics made a breakthrough in transmitting data via light, but were bottlenecked via creating an integrated circuit from photonics.
9/ In the last 20 yrs on-chip low energy silicon photonic circuits have become commercial/ubiquitous in datacenters for handling data going in and out of fiber optic cables
10/ More recently this technology is being commercialized for analog deep learning processors. Specifically, they are projected to do low precision matrix-vector multiplication with 2-3 orders of magnitude better energy consumption.
11/ PoWx.org is proposing a new optical PoW algo, a variant of Bitcoin's PoW optimized for optical computing
12/ This new algorithm leads to a high capex (capital expense), low opex (operating expense) PoW paradigm based on low energy photonic chips that compute with light
13/ Miner’s main cost is the hardware so can be located anywhere doesn’t require access to cheap power
14/ Better for decentralization and stable hashrate: miners with photonic hardware have low operating expense so they are not sensitive to price swings, do not shut off their machines
15/ Hardware is based on silicon photonics, used commercially in telecom for data processing in and between datacenters, and being commercialized for AI compute
16/ Algorithm is modified hashcash, using new hash called HeavyHash that is a combination of traditional hash function like SHA and a weighting function that is optimized for photonics
17/ This will conceivably make it so that one can participate in the mining anywhere, even in major cities in the US, Europe, Latin America.
18/ This is a way to provide better geographic decentralization, lower electricity/heat censor ship risk, and protection against partitioning attacks.
19/ This preserves the best elements of Proof of Work without the best elements of proof of stake without the weakness of weak subjectivity with fork selection.
20/ Full disclosure, @DAGlabsOfficial is researching the possibility of using oPow.
