Network managers without financial gains
DHT technology as used by Holochain is called agent-centric. Blockchain, e.g. Bitcoin, on the other hand are data centric. With Bitcoin, the network revolves around a single ledger that stores all the data, even though the data is stored in a distributed or replicated manner, with users interacting with it. In the TENDER economy, every user maintains their own data and interacts directly with other users.
(Figure 4: Transactions in data centric networks)
Example of an agent-centric cryptocurrency: When Alice transfers TENDER to Bob, both of them validate and countersign their common transaction. Bob will then check Alice’s transaction history to make sure the asset is not counterfeit. Then the transaction is published on the DHT, where all nodes that store it validate the transaction while ensuring network integrity at the same time.
(5: example of a agent-centric transaction system)
Holochain runs on user nodes. It is lightweight with every user acting as node, miner and staker at the same time. It can easily run on a mobile phone. This is possible because the node only stores its own transactions and a copy of some other users’ transactions. Another system that uses this approach is BitTorrent. There are no network managers that need to be incentivised just to provide upkeep. The incentive for users is to use the system.
However, there is no such thing as a free lunch. The network still needs upkeep in the form of bandwidth and file storage. Those costs are accrued at the point of each user, the node. Since a node doesn’t have to maintain the complete ledger like blockchains do, the cost of maintenance is a mere fraction of that of blockchains.
This system is even more elegant on closer inspection, as nearly all phones and desktop computers have idle capacity. Users with flat-rate bandwidth who charge their mobile phone daily will hardly ever have to pay anything extra for maintaining their node and thus the network.
TENDER’S answer to attack vectors
A main concern when designing a currency is to make it secure against a variety of attacks. The blockchain realm speaks mostly of the infamous 51% attack – the ledger of Bitcoin for instance can be manipulated by a user who controls 50% plus 1 node10. However, as we see below, there are design flaws that have already destroyed user trust in most cryptocurrencies without involving a direct attack. Let’s take a look at how the TENDER deals with some of the attack vectors and avoids design flaws:
Securing the system against 51% attacks
In contrast to single ledger blockchains, where one global consensus is achieved expensively and inefficiently by a group of middlemen (miners, stakeholders or other arbitrators), Pacio consensus is distributed, achieved asynchronously by the involved or interested parties – the users themselves.
Because the number of nodes grows with each user, it becomes increasingly difficult to achieve a majority. But even that would not necessarily compromise the network: if e.g. 51% of the nodes run a manipulated protocol, the worst that would happen would be the equivalent of a fork. The bad nodes would not be able to prove the validity of their transactions to the rest of the network. Each node can decide for itself what is right or wrong. Once it notices misbehaviour among its neighbours, it notifies the network about the breach and cuts connections to the malicious node (more on that in the technical specifications).
This would result in the network being segmented into two different parts by validation rules, i.e. a fork would occur. Both sets of nodes could recover missing data from other local nodes within their networks. Thus, all the work of organising an attack would not permit double spending, rendering an attack pointless.
Solving scale/spam attacks
The “cheapest” way to bring Bitcoin and Ether to a standstill is by spamming them with transactions. An attacker sends thousands of small transactions at the same time with high transaction fees. This guarantees that only these transactions are processed, bringing the network to a grinding halt. This approach is still costly, but way cheaper than a 51% attack. It is the DDOS attack of cryptocurrencies. This vector only exists because of a single ledger, with every node validating every transaction and the resulting abysmal transaction speed of blockchains. Lightning for Bitcoin and Ethereum 2.0 will make this attack more expensive and thus less likely, but the general problem remains unsolved.
Due to the architecture of Holochain, such scale attacks wouldn’t be successful since the network scales organically with its nodes. A spamming attack would only slow the transaction speed from the attacker to the recipient, but not the whole network.