Test data versus Live data11 May 2023 14:43
Some key points from the PRN:
"QBT’s algorithm, known as Method B, has theoretically increased the rate of successful bitcoin mining by 2.6 times compared to standard bitcoin mining industry practices widely used over the same time period."
"The Company’s algorithm also theoretically reduced electricity consumption by 4.3%."
"Assuming continued successful progress with testing, the Company believes its proprietary software has the potential to significantly improve the financial performance of bitcoin mining once deployed at full-scale."
"Given the expense and accessibility of current bitcoin mining equipment, ASIC chipsets and mining strategies, these results are based on a bitcoin mining difficulty(2) lower than the difficulty currently in the market. However, the research results and the statistical methodology used to obtain them give the Company confidence that its algorithms should be scalable for commercial use. "
The commercial miners take transaction data from a pool, incorporate it into the Block and then attempt to achieve a Hash lower than the target Hash, by altering the Nonce and some other fields. Different miners can use different, or the same, transaction data from the pool to build their Block; Once the winning miner's Block is accepted by the network, everyone has to start again from scratch. Obviously the transaction data has an effect on the final Hash, so everything in the Block is relevant when finding a Nonce giving the desired result.
As pointed out by a previous poster, in the PRN QBT have used the term "theoretical" and "assuming" to describe some key points:
"Method B, has theoretically increased the rate of successful bitcoin mining by 2.6 times"
"theoretically reduced electricity consumption by 4.3%."
"Assuming continued successful progress with testing"
I'm curious what QBT used to fill the transaction portion of the Test Blocks, did they use historic transaction data, pseudorandom data, or something else? To test the machine learning algorithms fully, I'd expect they'd use something simulating real world transaction data. The biggest test will be using the algorithm with real world data and I'm not sure why they haven't done that already? If someone can set up a couple of rigs in a garage or shed and mine Bitcoin, why couldn't QBT have set up their own small mining farms and tested using live data at the current difficulty level?
Maybe the reason the price hasn't climbed higher than it has, is due to the use of the words "theoretical" and "assuming" in relation to key points.