A “51% attack” constitutes of a circumstance in which a group of miners or a single person controls more than 50% of the networks mining hash-rate. Essentially, this means the entity would have the capability to disrupt the network.
What an entity CAN likely accomplish with a 51% attack:
-reverse transactions and thus double spend coins
-prevent confirmation of new transactions or target specific transactions, perhaps even halting payments entirely
What an entity can NOT likely accomplish with a 51% attack:
-Change the consensus of the protocol (add more coins, change the block size)
-Alter historical blocks that are deeply layered in the blockchain (potentially anything greater 6 confirmations)
It is important to note that maintaining a 51% attack comes at a continual cost to the entity. The output of the hashrate assumes that all of the available equipment to grant said entity their 51% share of the networks hashing power is actively working to disrupt the network. The longer the entity wishes to maintain said 51% attack, the greater the potential cost of said attack. Also, just because an entity controls 51% of the hashrate of a network, does not directly define a 51% attack, as they could potentially choose to continue to mine in accordance with standard network protocol.
Before we can understand why a 51% attack on the Bitcoin network is unlikely, we must first consider what proof of work consensus actually is. Proof of work is not a new concept. Proof of work has been around for as long as man has been able to harness the fruits of his labor and offer them as barter for the productivity of another man’s labor.
Consider gold mining.
Gold, which once was widely the soundest and most widespread monetary standard unit of trade (both because of its hardness and soundness as a money) requires a vast amount of effort and resources be contributed to it’s extraction from the earth. In many ways it is fairly inelastic to fluctuations in supply and demand, ie: a higher demand, which equates to a higher spot price of gold, does not necessarily mean more gold will be pulled out of the earth to meet that demand. The amount of gold produced remains fairly constant, which is what makes it a hard store of value.
Gold bars are essentially backed by a proof of work consensus. People understand that a certain amount of effort and resources went into extracting that gold from the earth and it now holds that value in the sense of a physical gold bar. Distribution of gold bars is, quite literally, a decentralized ledger tracking a claim on the effort exhausted to produce said gold.
Likewise, Bitcoin, which holds many of the same properties which make gold such a good money (however has all of the added benefits of digital distribution) has roughly always tracked in market value to it’s cost to produce. This “proof of work” consensus (which is technically artificially achieved with bitcoin) simply mirrors the proof of work consensus that has always existed as a axiomatic law of nature in human commerce.
Unlike gold however, Bitcoin potentially suffers from different vectors of attack. Gold, which can be seized in its physicality (by governments with the military and political resources to do so) and is often impractical in terms of its maneuverability and divisibility). While it would have been theoretically impossible to disrupt the gold network’s consensus, due to it’s large displacement of gold veins on the earth and near universal acceptance of gold as a store of value and thus a transaction medium, it is potentially possible that a single entity could control 51% of the mining of new gold, or attempt to clamp down on the trade and transaction of gold, or even seize gold outright.
Bitcoin is different in this sense, because as we mentioned earlier it’s value is derived not only from its monetary properties it has which are similar to gold, but from it’s network properties which make it transferable without regard to physical limitation of space and time, un-censorable, infinitely divisible, extreme hardness (compared to that of physical gold). It is this Bitcoin network which enable Bitcoin to function better as a monetary unit than Gold, in addition to it’s similar monetary properties which made gold valuable in the first place.
A 51% attack on the security of the Bitcoin network, would disrupt Bitcoin’s sole value creation proposition over gold as a monetary medium, by effectively rendering it’s value proposition useless. The sound money protocol of bitcoins are what allow participants to capture the entirety of the value the Bitcoin network creates.
A 51% attack could happen for, say a period long enough to disrupt it for a short term basis, but to sustain such an attack on a long term basis would be an ultimately ineffective strategy in regards to incentive. The entity controlling the hashrate, would have much more to benefit from the Bitcoin network by maintaining its value proposition and continuing to accumulate wealth from participating in the system, rather than continually disrupting the system and rendering its value useless.
The larger the supposedly “unsustainable” proof of work network grows, the greater the cost to disrupt said network becomes and the greater the loss in value of said potential entity incurs. Incentive to participate in Bitcoin’s network value is an offensive based defense Cost to produce Bitcoin via proof of work consensus is effectively it’s greatest strength. It harnesses one of the oldest axiomatic truths of tying a claim on productivity to wealth transfer.