Selfish Mining Fallacy - Debate II

"Bitcoin system, we characterize how the threshold varies as a function of message propagation speed in the network.”
· There has been no analysis of the message propagation speed or format within this paper. The authors state they have done this and yet clearly model a system that has no relationship to the inter-connectivity within bitcoin nodes.

· They extend to comments such as the following that do not even act as conclusions to the fallacious models and data that have been presented later in this paper:

o “This implies that, if less than 100% of the miners are honest, the system may not be incentive compatible: The first selfish miner will earn proportionally higher revenues than its honest counterparts, and the revenue of the selfish mining pool will increase superlinearly with pool size.”

o The network gamma in bitcoin has been incorrectly modelled. The authors simply pull figures out of the hat based on a presumption of Sybils and their affect within a power law graph. Unfortunately these are completely unrelated to bitcoin and the network model expressed within that network.

o Even taking this into account, it is crucial to note that under one third of network power even expressed in the selfish mining model is not only of the lower profitability then honest mining but delivers a lower revenue and increased cost.

o Figure 2 in their paper demonstrates that any mining pool with less than 30% of the hash rate can never hope to get even the revenue of a standard mining strategy. The result of this is at a greater cost. The selfish miner loses guaranteed blocks in an attempt to gain more in a Martingale strategy.

o This itself is a cost. In addition to this, running Sybil nodes requires additional virtual machines all of which could either deliver a profit in other uses or not be hired in the first place. The systems need to be managed and maintained. This is a further cost that is not taken into account. In this only addresses the revenue model.

o We will also detail later in the system the flaws associated with treating the selfish mining strategy as a discrete time process. All poison processes of this type are based on continuous time analyses. The theory used here is known as queueing theory. When applied correctly, the authors assumption of the same number of blocks being mined is shown to be false. We detail this later in this paper.

· A strategy that requires higher cost investment for a lower expected revenue is not one that would attract rational investors. Rational investors are driven towards profit. This seems to be a common misapprehension held by many people within the bitcoin community. This inability to understand the difference between revenue and profit has had many dire consequences and negative effects already.

“The blockchain is maintained by a network of miners, which are compensated for their effort in Bitcoins.”
· This demonstrates a subtle error associated with the nature of money and wealth. A mining reward is allocated, however the nature of the compensation to each miner is oversimplified here.

· Miners are not compensated, they compete. This is a key fundamental difference that the authors of the selfish mining paper have not fully understood. They do this at cost to themselves in the hope that they will profit from the difference in the cost associated with gaining revenue and the amount that they can sell or trade bitcoin for on the market.

· All money is traded. Money is not a consumption good. We either trade it for other forms of currency author goods and services that we seek to consume. As such, a part of this compensation that they claim to be allocated is the profit that each miner individually manages to create through their own efficiencies.

"Yet, since the current protocol has no guaranteed lower bound on this threshold, it cannot automatically protect against selfish miners."

It is called profit.

Like many academics who have not worked in the real world, they fail to understand this.

Revenue is what matetrs, it is what you can claw away from others in grants and the like and it does not require that you manage it wisely. Hence one of the many flaws in this paper.

At 49% the presented model provides 100% of mining rewards to the Selfish Miner.

There is a 3% chance that a Honest miner will start with 5 blocks in a row. There are many other instances where the selfish miner at this level flip flops one block after another with the honest miner.

These scenarios all provide revenue returns to the honest miner, the SM is not reversing blocks.

Yet, somehow... they manage with 0 gamma (and 51% of the network = 51% gamma just to themselves in this wrong model).

They are saying, those blocks are not revenue when they clearly are....

"But a savvy pool operator can perform a sybil attack on honest miners by adding a significant number of zero-power miners to the Bitcoin miner network"
For an Erdos Renyi graph of the distance and radius of Bitcoin, you will need around 1 million Sybil nodes for each real node.

That comes to around 1 Billion virtual machines.

For on demand pricing (excluding costs to maintain and run) this comes to $0.0059USD an hour.

Only 141,600,000 USD a day at current costs...