Dissecting D.A.G.G.E.R. & The Bee Analogy


In this article, we dissect GenesysGo's Directed Acyclic Gossip Graph Enabling Replication Protocol (D.A.G.G.E.R.) through the lens of "The Bee Analogy." As always, for a deeper dive into technical specifics, please refer to the D.A.G.G.E.R. Litepaper. If you'd like to experience D.A.G.G.E.R., try the Hammer Testnet Game.

The Bee Analogy

In the D.A.G.G.E.R. Bee Analogy, operators are diligent bees, and events are flowers. Each flower (event) has unique pollen (metadata) collected by bees (operators). When the bees are scattered across the meadow (decentralized), each is capable of performing key activities (modules).

Communications Module

Bees (operators) communicate with each other about the existence and location of flowers (events). This is akin to how operators exchange information about events with other operators.

Verifier (sub-Module)

In the Verifier, bees (operators) verify the quality and uniqueness of the pollen (transactions) they've collected. This is similar to operators verifying the authenticity and uniqueness of incoming transactions.

Forester (sub-Module)

In the Forester, Bees (operators) organize pollen (transactions) into honeycombs (Merkle trees) for efficient storage. This represents operators packing transactions into Merkle trees for efficient storage and verification.

Graph Module

In the Graph Module, bees (operators) collectively determine the order in which flowers (events) are visited based on the pollen (metadata) collected. This mirrors how operators analyze the graph to derive the consensus ordering of events.

Bringing it to Finalization "The Honey"

Additionally, when a bee (operator) wants to be sure it has found the correct pollen (event E can strongly see event X), it traces back its path to the specific flower (event) through the pollen trail (parent and self-parent relationships). Suppose the bee (operator) finishes its search, finds the specific pollen (metadata), and verifies the paths from the event to the specific pollen, and more than two-thirds of the visited flowers contain it. In that case, the bee concludes the pollen is strongly present or "strongly seen."

For the hive (the system) to reach a consensus vote, it requires a super-majority of bees (operators) to agree on the sequence of flower (event) visits.

This bee-like process is asynchronous. Each bee (operator) independently searches for specific pollen (metadata), contributing to the beehive's (D.A.G.G.E.R. system's) consensus on the order of flower visits. The collective knowledge of the beehive on the sequence of flower visits and the quality of pollen is like a verifiable ledger, shared and known across all bees - the honey, in this case, would represent the consensus-ordered list of transactions.