Thanks to the insights of the experts; I learned a lot. I sincerely find this matter scalable. Insiders see the subtleties, while outsiders just watch the excitement. Let me add a few more points:
1) I saw a big influencer overseas using this to FUD Bitcoin, arguing that the current mining pool hash rate proportion shows that the top two mining pools, Foundry USA (33.6%) and AntPool (17.9%), have already combined to exceed 51%, leading to a crude conclusion that if the two major pools collude, Bitcoin is doomed. This is a typical case of an outsider misunderstanding the situation because they overlook two points:
1. The situation of two mining pools occupying 51% and one mining pool exceeding 51% are completely different, like heaven and earth;
2. The computational power of mining pools does not represent the complete bribing of miners' computational power. When a single mining pool's power is too high, miners will typically choose to disperse their power to avoid risks.
Thus, Satoshi Nakamoto's POW consensus has reached this point, integrating factors such as computational power, economics, and interests, achieving a delicate balance. In the short term, there is almost no possibility of this being broken, so those who FUD BTC based on this matter should take a break.
2) Acknowledging what teacher Zhang Ren said about Monero's problems does not equate to acknowledging POW's problems. Even if POW has security risks under extreme conditions, it does not mean POS is the optimal solution.
In fact, the issues with POW have made ASIC a preferred solution, avoiding the pitfalls of general CPU/GPU mining. Moreover, even in a purely CPU/GPU mined chain, attempting to attack using an already disclosed bribery method for miners faces various challenges, such as exchanges increasing confirmation numbers and miners adding checkpoints, all of which can reduce the probability of being attacked.
You see, discussing the issues of POW focuses on POW itself, and comparing across consensus can lead to misunderstandings. In fact, there are security risk boundaries for each consensus that need to be acknowledged, and the methods of countering those risks are also different; one cannot favor one over the other.
3) I saw teacher 0xTodd retweeting my post discussing the concept of 'selfish mining,' which simply means that after a miner finds a block, they should immediately broadcast it, but selfish miners will secretly hide the mined block, forming a 'private chain.' When honest miners publish new blocks, selfish miners suddenly release their hidden longer chain, rendering the honest miners' work completely void.
This is actually a very rogue approach and the main attack method of Qubic. In reality, its computational power has not truly reached 51%; it may control around 30%. Can it briefly achieve a theoretical 'double-spending attack'? Because, using 30% of miners for selfish mining, a shadow chain is formed. When honest miners find new blocks, Qubic suddenly releases its hidden longer chain, rendering a large number of real miners’ blocks void, which theoretically can cause damage as if exceeding 51% of the hash rate. Further, if the distribution of Qubic-controlled mining pool nodes is wide enough, it can leverage network latency and other factors to further reduce the proportion of computational power, thus achieving the same effect of controlling the entire network's hash rate.
Therefore, Qubic's attack this time has a significant degree of randomness and concealment, which also means that once this method is made public, the threshold for using it again will be higher.
4) However, I had a discussion with security expert n33k about a possibility that Qubic may not use the same old tricks, but instead operate through a 'boiling frog' style attack, further bribing miners to increase their mining pool size, and then allowing some miners to purposely mine empty blocks, creating chaos for the normal operation of the Monero network.
If this continues, it will lead to more and more Monero miners fleeing since profits will decrease, and the experience will be terrible. In this way, the computational power controlled by Qubic will gradually increase, surpassing 50%. By then, it will be game over, and everyone will have nothing to play with. This type of chronic attack is indeed quite scary.
While there is no reason to prove that Qubic needs to do this, the possibility of this 'parasitic' chronic attack indeed exists. In the early stages, Qubic does not have to worry about some miners mining empty blocks in Monero; they will still receive $XMR rewards and can engage in AI training. Later, if Monero's profitability declines, they might attack other chains like Grin and Beam. Throughout the process, Qubic can consistently stick to its core AI training line, making its logic reasonable.
Because when the demand for AI computing power increases exponentially, and mining is no longer the only destination for computing power, the rules of the game have changed. The cost of attacking the network used to be 'purely burning money'; now there is an 'additional payer' in the form of AI training to cover the costs—attack costs are offset by AI profits.
This is my biggest concern about that article: AI demand is breaking the fundamental assumption of general CPU/GPU POW mining - 'miners rely on mining rewards, so they will maintain the network.' When computational power finds more profitable opportunities, this assumption no longer holds. Although this process will be slow, there is always the possibility.
