The DAG KNIGHT Protocol — Elevating Kaspa

Sir Knights

On December 18th, 2022, Kaspa held a Community Crowdfund to raise 70 million KAS to finance the construction and implementation of the DAG KNIGHT protocol. After 12 days, on December 30th, 2022, the crowdfund met its goal entirely at the community’s request. So, what is the DAG KNIGHT protocol, and why did the Kaspa community want it so badly?

The DAG KNIGHT Protocol, written by Michael Sutton and Yonatan Sompolinsky, is a three-year work of genius and consensus upgrade for Kaspa’s PHANTOM GHOSTDAG Protocol. KNIGHT was set in motion with the goal of resisting 50% attacks without knowing the network latency ahead of time.

Typically, blockchain protocols assume a parameter of network delay for best and/or worst-case scenarios without the ability to adapt to real-time latency. This assumed latency parameter keeps blockchains restrained to slower confirmation times to avoid security risks and instability.

A 50% or 51% attack occurs when a group or entity owns more than 50% of the nodes on the network, granting the controlling parties the ability to alter the blockchain. However, with network delay left unaddressed, if latency spikes, the percentage of ownership needed for an attack decreases. Subsequently leaving the chain more susceptible to an attack. To fix this, blockchains sacrifice block confirmation time to account for the ailing network health by setting an estimated/assumed network delay parameter.

The White (Green?) DAG KNIGHT

This is where the DAG KNIGHT in shining armor comes into play. The KNIGHT protocol actively evaluates and responds to the network conditions and adjusts block confirmation times accordingly.

Once the Kaspa rewrite completes, DAG KNIGHT will be developed on a network that is already able to push 32 blocks per second!

Not only is that blazing block speed impressive, but when applying KNIGHT to our advancing internet infrastructure, the protocol really starts to shine. As the internet improves and inevitably becomes faster, Kaspa won’t be stuck with an aging 10-minute or even 1-second block time. Instead, it will be parameterless and fully adaptable.

So now that we know with DAG KNIGHT, this latency cannot be defined until network conditions are evaluated by the protocol(parameterless). Let’s dive under the surface and see how the protocol evaluates the network.

Cluster Around

The protocol starts by analyzing groups of blocks sampled from the DAG called k-clusters. On examining these k-clusters, the protocol selects a k-cluster with the most network delay that also covers 50% of the network. Then, with this valid k-cluster, the protocol adjusts the block confirmations to a sufficient speed to counter an attack. By doing so, KNIGHT is constantly adjusting for the lowest latency valid k-cluster with 50% coverage.

Since KNIGHT performs according to latency and the protocol is high-speed while the network is healthy. When the network is slow, additional blocks are added. Through this process, DAG KNIGHT finds valid k-clusters and references honest nodes to secure chain ordering, adding another layer of security to GHOSTDAG. Also, transactions may not be confirmed if GHOSTDAG suffers excessive delays and the latency bound is violated (due to an anomaly). KNIGHT, however, will remain intact at the cost of speed and is self-stabilizing.