we’re all in on it and doing our best. it’s a white/yellow/black (research/specs/code) effort happening in parallel with various relationships

👀

There was a question brought up in a Hebrew TG group regarding other pow techs and how they compare to Kaspa. Sharing the Gemini-translated answer in case it’s useful to others. ——— Fundamentally, protocol Z (forget the name; the answer does not depend on it) is not an “interesting” protocol when compared to Kaspa. I’ll try to explain why. Broadly speaking, within the mining paradigm itself, Kaspa, in a sense, already encompasses every pow technology imaginable. Why? Because a block-DAG, in its general sense, allows for the simulation of any other protocol beneath it. This is because, at the information-sharing level, the block-DAG, as constructed in Kaspa’s mining protocol, maximizes possible knowledge sharing. When a miner mines a new block that points to all the tips they see in their local DAG, they recursively share the entire history as reflected by these tips. To illustrate this, consider a btc miner who receives three blocks mined concurrently (a rare event due to the ten-minute block time, but probabilistically possible). Since it’s a block-chain, the miner selects a particular block in some manner—IIRC, in the default implementation, it’s based on the first one received. What happened here? The moment they chose, and assuming they indeed mined the next block, they censored information they possessed at the time of mining. At the time of mining, they knew about the existence of three tips, but the rest of the network doesn’t know they were aware of them. In a DAG, however, the block would point to all three tips, and all information would be shared. On top of this general block-DAG structure, one could, for example, also run the longest-chain ordering protocol. Meaning, DAG mining maximizes shared information, and therefore, any other protocol can be run using it or on top of it. In my view/parlance, a block-DAG isn’t just another neat structure; it’s a canonical, ultimate structure. There isn’t, and mathematically, there won’t be a better structure. Note that we haven’t even discussed the ordering protocol yet, i.e., Ghostdag. So why isn’t everyone using block-DAGs then? First, briefly: a. It’s difficult to efficiently implement a general block-DAG. Maintaining its structure is hard because it’s not ordered or predictable; the structure can be completely amorphous. It’s even harder to maintain the structure in a way that allows for efficient topological queries (like reachability, for those familiar). b. A protocol is needed that can maximize the use of all this information. Elaborating: a. One of Kaspa’s hidden gems (from both research and engineering perspectives) is how we efficiently maintain the DAG structure. This is a non-trivial problem, to the extent that it wasn’t clear if an efficient solution existed for a dynamic, constantly evolving DAG. This clean and efficient implementation is one of the factors enabling speeds of ten blocks per second (BPS) or more. It’s far easier to maintain a simple chain structure, or a “braid” of chains with a pre-defined, orderly structure. b. Ghostdag is a protocol that explicitly uses all the information it sees to deduce the set of blocks representing “honest” miners. I won’t delve into this, but a note for the advanced readers: I conjecture that the k-cluster structure underlying Ghostdag is a canonical structure for achieving consensus under communication-bound models (both synchronous and partially synchronous). This is why the Dagknight protocol also uses a k-cluster, only it does so more intelligently. In short, this is getting lengthy. So, why is Kaspa fundamentally superior? Because its canonical and ultimate DAG structure allows the most efficient and fastest protocols to run on top of it. The practical implication is faster transaction confirmation times, approaching the a priori internet latency limits (with Ghostdag) or actual realtime limits (with Dagknight, Kaspa’s future protocol).