Crescendo is live. Kaspa moved from 1 BPS to 10 BPS in 2025.
Independent research resource
Kaspa, explained without flattening the roadmap into hype.
Kaspa is a Proof of Work blockDAG network. The serious frame is not "just a faster coin." It is a research-led attempt to keep Bitcoin-like monetary and security instincts while redesigning consensus around parallel block production and moving toward bounded, proof-oriented programmability.
10 BPSCrescendo live
PoWUTXO blockDAG
4Status lanes
Crypto awareness
Where Kaspa fits against Bitcoin, Ethereum, Solana, and the rest of crypto.
Use the comparison page when the question is not just what Kaspa is, but what lane it occupies: Proof of Work money, high-speed blockDAG settlement, and future programmability without pretending it already has Ethereum-level DeFi or Solana-level app adoption.
Toccata is the major hard-fork track: covenants, Silverscript, ZK foundations, sequencing commitments, and vProgs groundwork.
Igra and ecosystem work are practical application paths while deeper Kaspa-native programmability matures.
DAGKnight, RTD-style miner attestations, TangVM, and Proof of Useful Work must be described as roadmap or research unless shipped.
The clean frame
Kaspa is one answer in the crypto design space.
Bitcoin proved decentralized digital scarcity and Proof of Work money. Ethereum made the application layer programmable and pushed the industry to think about shared state, rollups, ZK systems, and composability. Solana changed user expectations around performance and app feel. Kaspa sits in a different lane: keep Proof of Work and UTXO instincts, replace the single-chain bottleneck with a blockDAG, and move toward bounded, verification-oriented programmability.
Why crypto at all
Money on the internet is hard.
The internet is good at copying information. Money is the opposite problem. It needs scarcity, clear control over spending, and a way to stop the same unit from being spent twice. Crypto matters because it tries to make ownership, settlement, and coordination possible on open networks without relying entirely on a central operator.
Why Kaspa exists
Bitcoin solved digital money, but not the whole performance problem.
In a longest-chain system, simultaneous honest blocks compete for one canonical history. As block production speeds up, more honest work collides. Kaspa comes from research that tries to use more of that parallel work instead of throwing most of it away. That is the starting point for understanding GHOSTDAG and the blockDAG thesis.
GHOST research
Yonatan Sompolinsky and Aviv Zohar publish work later cited in Ethereum's whitepaper.
Kaspa launch
Kaspa launches as a fair-launched Proof of Work network with no premine and no presale.
Crescendo
The network moves to 10 BPS, making the performance claim live rather than hypothetical.
Toccata track
Work focuses on covenants, ZK opcodes, sequencing commitments, Silverscript, and the road toward vProgs.
Status discipline
Live, near-term, roadmap, research.
Live / factual now
- Proof of Work blockDAG
- UTXO model
- GHOSTDAG consensus
- Crescendo 10 BPS era
- Pruning and UTXO commitments
- Wallets, explorers, visualizers, nodes, and mining ecosystem
Near-term / implementation track
- Toccata hard fork
- Covenants
- Silverscript
- ZK verification foundations
- Sequencing commitments
- vProgs groundwork
Architecture / roadmap
- vProgs as app-level verifiable programs
- Kaspa-native DeFi rails
- Solana-like cohesive developer experience
- Synchronous composability across programs
- Igra and ecosystem application paths
Research / speculative
- DAGKnight final form and activation timing
- 100 BPS with probabilistic predecessor selection
- RTD-style miner attestation and oracle designs
- TangVM-style reality-state ideas
- Proof of Useful Work via matrix multiplication
- Long-term post-quantum migration path
High-signal interview
Start with the 7-hour Yonatan Sompolinsky interview.
The Bitcoin Takeover S16 E41 interview is one of the best context sources because it covers the technical, cultural, and historical claims in one place: Kaspa as a generalization of Nakamoto consensus, fair-launch messiness, pruning and missing-history FUD, UTXOs, DAGKnight, vProgs, ZK opcodes, PoW vs PoS, social recovery, synchronous composability, and why the community needs more crypto context.
More from the Yonatan conversation
Insights worth preserving.
Generalized Nakamoto consensus
Kaspa is best introduced as a generalization of Bitcoin's consensus structure, not as an unrelated altcoin architecture.
DAG is not enough
A DAG is the causality/data structure. The innovation is the rule for ordering that DAG in a way that converges across honest nodes.
GHOSTDAG is live
GHOSTDAG is the current working system. DAGKnight belongs in the upgrade and research discussion unless activation is confirmed.
Selected chain is not longest chain
GHOSTDAG still uses a selected chain for ordering, but it is the center of the heaviest K-cluster rather than Bitcoin's longest-chain rule.
DAGKnight nuance
DAGKnight tries to remove the fixed K parameter, but it should not be described as live instant finality.
DAGKnight latency setting
The client-side latency assumption is about adversarial worst-case latency, not merely the smooth latency observed during normal network conditions.
Cyber-war resilience thesis
The strongest DAGKnight claim is resilience when network latency becomes unstable, not just a simple speed slogan.
100 BPS bottleneck
Higher block rates stress DAG processing because references and edges grow dense; probabilistic predecessor selection is part of the discussion.
RTD is majority sampling
The Mining the Internet framing is about using high-rate PoW blockDAG mining to sample an honest majority quickly for attestations, not just making payments feel faster.
RTD is also infrastructure
In the Staghunt framing, RTD is a metaproperty for infrastructure that gives coordination markets censorship resistance, permissionlessness, and fairness in real time.
Internet money flow is early
RTD-style oracle flows would need concrete data feeds, opt-in miner incentives, anti-MEV design, and careful implementation. Yonatan described the flow as far from ready.
Fair launch was messy
The launch story includes gamenet, unexpected hash-rate growth, a halt, and a second genesis pointing cryptographically to prior state.
Pruning is not missing money
Explorers may not archive every spent transaction, but consensus tracks the UTXO set. Explorer gaps are not missing UTXOs.
Pruning is not privacy
Pruning is a design tradeoff for throughput and storage, not a privacy system.
UTXO-set growth matters
UTXO compaction, fee mechanics, and storm-style UTXO pricing deserve more attention than they usually get.
Storm prices a real externality
The Storm fee idea is about charging for increasing the UTXO set, not simply making every transaction more expensive.
Native DeFi is not live
The serious path depends on Toccata, ZK verification, and vProgs as app-level verifiable programs.
vProgs are not ordinary rollups
The goal is app-level sovereignty without recreating isolated rollup islands that fragment liquidity and user experience.
One app per VM intuition
Yonatan framed vProgs as app-level mini ZKVMs that want to interoperate, rather than rollups trying to pull many apps into one island.
Solana-like is not Solana code
The target is Rust-friendly cohesive DevX; execution would run through sound ZKVMs such as ZK EVM, SP1, Risc Zero, Cairo-like systems, or similar choices.
L1 should stay thin
The intended design keeps L1 focused on verification, commitments, metadata, and sequencing rather than executing every app's logic.
Monolithic feel, not monolithic execution
The target is a cohesive developer and user experience without making L1 a global execution engine.
Synchronous composability
Async-only interoperability creates user risk and rollup advantages. Native synchronous composability is central to the vProg vision.
Stablecoins matter
Stablecoins are one of crypto's clearest product-market fits and would likely matter for Kaspa DeFi.
Security budget needs activity
The credible long-term answer is transaction activity, fees, DeFi, liquidity, and possibly future Proof of Useful Work research.
Tail emission is not the preferred answer
Yonatan leaned against tail emission for Kaspa, pointing instead toward activity, fees, and social expectations around ecosystem support.
Proof of Work has two layers
PoW matters both for monetary issuance cost and for consensus as an external clock/randomness source.
Leaderless sequencing
Fast PoW blockDAGs can have many miners competing within a round, which may improve censorship resistance and sequencing markets.
ASICs are desirable, with caveats
ASIC mining can improve security because specialized hardware is harder to summon on demand than rented general hardware.
Proof of Useful Work is not imminent
Matrix-multiplication Proof of Useful Work is an interesting research direction, not a short-term shipped feature.
Self-custody needs better defaults
Vaults, covenants, social recovery, and smart-wallet policies can make sovereignty practical without asking ordinary users to live like security experts.
Infrastructure should not be singular
Explorers, wallets, visualizers, SPV nodes, and full nodes should be diverse community infrastructure rather than one official interface everyone trusts.
Founder visibility is a tradeoff
Yonatan's continued presence is useful while major innovation is ongoing, but Kaspa should eventually outgrow reliance on one person.
Study broader crypto
The Kaspa community benefits from understanding Bitcoin, Ethereum, Solana, ZK rollups, crypto history, and earlier research.
Narrative correction
Better one-liners.
Bad: "Kaspa is Bitcoin but faster."
Better: Kaspa is a Proof of Work blockDAG that generalizes Nakamoto consensus so parallel honest blocks can contribute to ordering.
Bad: "Kaspa solved the trilemma."
Better: Kaspa makes a different set of tradeoffs: higher block rate and faster inclusion through blockDAG consensus, while accepting pruning, protocol complexity, and a still-developing app layer.
Bad: "DeFi is live."
Better: Native DeFi is a major design target. The serious path runs through Toccata, ZK verification, and vProgs as app-level verifiable programs.
Bad: "It is just another L2 path."
Better: The design aims to avoid rollup-island incentives by making app-level programs sovereign but synchronously composable and native-feeling.
Preferred sources
Primary source stack for future updates.
- hashd.ag and hashd.ag/raw for Hashdag / Yonatan Sompolinsky's writing archive.
- Michael Sutton on Toccata for covenants, ZK verification, sequencing commitments, Silverscript, and vProgs bridge context.
- rusty-kaspa GitHub and rusty-kaspa releases for what actually changed in code.
- Kaspa Research for KIPs, protocol design debates, fee mechanics, vProgs, covenants, and tradeoffs.
- Kaspa Q&A for detailed technical answers from developers and informed community members.
- KASmedia for weekly ecosystem context, interviews, and explainers.
- Bitcoin Takeover S16 E41 transcript for long-form context on what Yonatan actually said.
- Kaspa: Mining the Internet for the RTD, miner-attestation, and internet-money-layer framing.
What Kaspa is not
Honest limits matter.
- Kaspa is not a finished application ecosystem.
- Kaspa is not an Ethereum-style global shared VM on L1 today.
- vProgs and DAGKnight should not be written as already-live mainnet facts.
- Toccata should be described as targeted until it activates.
- X can help with primary-source discovery, but it is not a substitute for code, KIPs, releases, or full-context interviews.