Topic timeline
Cardano Tech & Research over time
3,662 entries across all AMAs, in chronological order, showing the first 150.
12:41 The third generation is defined by three problems that come from success: scalability, interoperability, and sustainability.
15:07 Peer review is masochistic, with papers rejected and authors yelled at, but it gives real assurance that the claims may be right.
15:48 The plan was to build cryptocurrencies like a mathematician would, in Haskell and with formal methods.
16:48 Formal verification tackles the semantic gap between an ideal paper and the real code an engineer writes.
17:52 A formal specification plus a proof by simulation shows the code matches the spec, so it has no bugs as long as the paper is correct.
18:33 Most software skips this because it is horrendously expensive, done in tools like Coq or the Isabelle that IOHK prefers.
19:16 The seL4 verified microkernel took over five years and 180,000 lines of Isabelle to produce 7,000 lines of C.
20:40 Scalability should mean that as you gain users you gain resources, a property BitTorrent already has.
21:42 Cryptocurrencies are harder than BitTorrent because they add transaction processing and data storage on top of moving data.
22:24 Generation one assumes everyone holds a full copy of the chain with a reliable connection, assumptions that break at scale.
23:24 Before scaling, you need bedrock and peer review, so the first question in 2015 was what a secure ledger even is.
23:44 The GKL15 paper defines a secure ledger, and proves that Nakamoto consensus forms one, so Satoshi did something magical.
25:07 Using that foundation, the team proved that proof of stake can reach a security level equivalent to proof of work under certain assumptions.
25:28 The Ouroboros proof of stake protocol took six revisions, rejected from Oakland before getting into the Crypto conference in Santa Barbara.
26:29 Beyond consensus you need a network protocol, and Cardano chose RINA, John Day's recursive internetwork architecture.
28:53 Runtime Verification and Grigore Rosu used the K framework to write formal semantics of the Ethereum VM, the KEVM.
29:54 From that they built Yella, a register based virtual machine that looks like LLVM, making it easier to build compilers.
30:55 With semantic space compilation you write a language's K semantics and get a compiler for free, already done for Java, C, JavaScript, Solidity, and Plutus.
32:57 Interoperability between crypto systems is a cryptographic problem: prove that incoming coins exist and were not double spent.
34:00 The non interactive proofs of proof of work paper builds tiny bridge proofs between chains, and is being translated to proof of stake.
39:52 On backward compatibility, Phil Wadler warned that accepting Solidity means accepting all its sins, like the Parity wallet exploit.
40:33 Cardano splits into a settlement layer for accounting and a computation layer, which can run both an EVM clone and Yella.
40:53 Developers choose between safety and familiar tools, and the strong security guarantees live only on the encapsulated Yella side.
57:31 The peak TPS is unknown, and the harder question is propagating messages through the gossip protocol, studied with Delta Q and PNSol.
58:12 Ouroboros runs in epochs, electing slot leaders via cryptographic sortition using a source of randomness and follow the Satoshi.
59:13 The next generation of the protocol runs epochs in parallel for n times the throughput, adjusting like mining difficulty.
59:54 Moving and storing data stays the hard part, needing RINA and sharding as chains grow to exabyte scale.
1:02:42 For financial contracts, a domain specific language may beat a turing complete one, as Project ACTUS shows with 30 composable primitive contracts.
1:03:43 Marlowe, built at Kent with Simon Thompson, is a turing incomplete paint by numbers language for financial contracts.
1:04:43 For robust smart contracts there is Plutus, built with Phil Wadler and Darryl McAdams as the revenge of Haskell.
1:05:48 Mobile stays a walled garden controlled by Apple and Google, and if they get too draconian the answer is for the industry to build its own hardware.
6:54 Zero-knowledge proofs and zk-SNARKs from first principles, covering verified computation, Pinocchio and privacy use cases like Zcash.
10:58 Using zero-knowledge proofs to validate sidechain transactions without holding a copy of the whole source chain.
12:02 Details IOHK's zero-knowledge research streams including the million dollar Edinburgh lab led by Markulf Kohlweiss.
15:01 Building the Mantis client from scratch in Scala with the Grothendieck team as the leanest functional Ethereum client.
18:56 Corrects the misconception that Cardano smart contracts can only be written in Haskell.
19:22 Lays out the accounting ledger computation model built with Marlowe and Plutus to out scale Bitcoin's transaction types.
21:22 The Ethereum style stateful computation model and its current scaling limits pushing projects off chain.
22:54 Details the Runtime Verification partnership, the K-EVM formal semantics work and the new IELE (yellow) virtual machine.
23:49 Semantic space compilation as a universal compilation target letting any language target Cardano smart contracts.
26:35 Multi-party computation as an off chain alternative for games like poker and decentralized exchanges.
29:53 Names the MPC toolkit effort Lebowski and describes the Kaleidoscope and Royale poker protocols that preserve card privacy.
52:52 Lists IOHK contributions back to Bitcoin like the Dandelion paper and lessons from Blockstream's Simplicity language.
54:50 There is no minimum stake to register a stake pool and correct the delegated proof of stake misnomer for Ouroboros.
56:07 Describes why delegation matters for cold storage holders who cannot stay online to run their elected slot.
57:00 The game theoretic model for sizing the ideal number of stake pools between too few and too many.
1:00:30 Staking stays possible even at 0.1 percent of the network, letting anyone participate or delegate.
1:01:01 Predicts stake pools will compete on fees and services like Oracles and payment channels, lowering costs over time.
1:01:47 The chain quality metrics used to tune Ouroboros parameters and detect an unhealthy network.
1:03:21 Previews Ouroboros Hydra as the next generation sharding research to scale transaction throughput via amplification.
1:04:27 Credits Byzantine fault tolerance researcher Matthias Fitzi for leading the Hydra scaling research.
1:32:20 Points to the rigorous UTXO wallet formal specification as proof the research translates into real engineering.
1:43:03 References IOHK's published mesh networking paper by Bernardo David for incentivized off-grid connectivity in Africa.
1:45:20 Proposes an open source 200 to 500 dollar ATM design researched at the Edinburgh trusted hardware lab with Huawei co-funding.
6:12 Previews upcoming Ouroboros improvements including faster blockchain download, accelerated transactions and sharding.
8:39 Updates on Ledger and Trezor hardware wallet support, waiting on their firmware while a new shorter address format gets finalized.
12:41 Explains why the Linux wallet is delayed, the current file storage approach needs to shrink from 1.3 million files to about 40,000 before it ships in 1.4.
20:59 Teases new Mantis wallet features, Ethereum and ERC20 support, and non-interactive proof of work research for lightweight side chains.
31:39 DAG based protocols trade strict block ordering for throughput, and why that tradeoff is not a magic scalability fix.
35:31 Confirms Cardano is researching similar structures for Ouroboros Hydra once Genesis wraps up, with a first paper expected by the end of 2018.
51:17 Announces a new staking incentives paper submitted to WINE 2018, projecting the system will converge toward roughly a thousand stake pools.
53:36 Envisions stake pools evolving into layer two service providers running relay nodes, oracle feeds and a lightning style network as the protocol matures.
57:37 Weighs extending the UTXO model against a separate ledger for smart contracts, crediting the language design team he has assembled.
1:00:12 Explores running workshops so the community can read and write formal specifications, pointing to Runtime Verification's ERC20 checking service as a model.
1:02:56 Compares formal methods to a medical specialty most engineers never encounter, admitting existing tools and textbooks are too academic and not developer friendly.
1:05:07 Argues verification and specification tooling should be blockchain agnostic rather than locking developers into one platform.
1:19:32 Defends the academic peer review pipeline behind Cardano's research, walking through how conference submission, rebuttal and acceptance cycles work.
1:28:54 Recounts a CT-RSA panel alongside Turing Award winners debating proof of stake, with Ron Rivest in the audience.
1:31:23 Details sidechain validation research where a slot leader maintains several ledgers concurrently as an early form of sharding.
1:34:29 Introduces RINA as a long term networking research project, a policy driven universal network layer meant to make future protocols easier to build and optimize.
2:49 Cardano is a best guess at a third generation cryptocurrency, defined by scalability, interoperability and sustainability.
3:16 Real scaling means gaining resources as you gain users, like BitTorrent, while Bitcoin and Ethereum are replicated systems where new participants only shrink everyone's slice of the pie.
5:06 With over a thousand coins in use, interoperability is about standards for moving value and information between ledgers, via Interledger and sidechain research on both proof of stake and proof of work.
9:15 Ouroboros and the sidechain design were built by Aggelos Kiayias, Peter Gazi and Bernardo David, and Lancaster work on liquid democracy supplies the governance model.
12:28 A sidechain connecting a public ledger to a private exchange ledger enables proof of solvency, real time continuous auditing, front-running prevention and tokens moving in and out of Cardano.
13:23 Pushes as much as possible off chain with verified and multi-party computation, and let people write contracts in Java, C++ or JavaScript alongside dedicated high assurance DSLs.
13:50 Asked whether Cardano's smart contract model is Turing complete and uses gas, the answer starts with separating accounting from computation, which lowers regulatory risk and stops a bug bricking the whole chain.
14:48 The settlement layer CSL uses an extended UTXO model with a Turing incomplete language to start, opened up carefully via Template Haskell and Plutus.
15:42 LLVM was chosen over WebAssembly after consulting UIUC, Grigore Rosu and LLVM co-creator Vikram Adve, producing K semantics and the lightweight IELE virtual machine.
16:40 Runtime Verification's 19 person team handles gas and memory management, and the K framework behind it has 15 years of theory and users like NASA and Boeing, with Marlowe and Plutus as the high assurance DSLs.
17:34 Developers should pick whatever language they like and register its semantics on chain, and an EVM clone shipping alongside IELE will run Solidity and Viper, with a model like EOS attachable later.
18:28 Asked whether a proof of stake economy is sustainable, punts to Aggelos Kiayias and Peter Gazi, who invented the first provably secure proof of stake protocol and speak next.
20:49 State and payment channels, trusted hardware and Bernardo David's MPC protocols can run poker, blackjack and even decentralized exchange off chain, scaling to millions of concurrent users.
16:54 Ouroboros is about three years into a research agenda now bearing fruit toward finality and sharding through Hydra.
22:16 Breaks down what the ADA token actually does beyond being a legacy style store of value and means of exchange.
23:08 Proof of stake block elections through the Follow the Satoshi protocol, where your odds scale with how much ADA you hold.
27:36 Computational fuel, Ethereum style gas costs for running programs, as a further dimension of ADA's value.
45:21 An emerging third generation of blockchain and teases an enterprise product built around a blockchain less blockchain for data.
58:32 Waves off a coin burn question by pointing out ADA has no dedicated reserve of coins to burn.
58:54 Recaps choosing Haskell for the core, then spinning up a Rust project called Prometheus that became the Icarus wallet.
1:00:39 Reflects on Haskell being powerful but temperamental compared to Rust's easy portability across mobile and web.
1:01:30 Tackles whether exchanges holding user ADA in custody can be kept out of the staking process.
1:02:42 Proposes a visually distinct exchange address type that is automatically excluded from stake weight.
1:03:54 Sketches a dual key design letting exchanges reverse a stolen transaction within a roughly 12 hour settlement window.
1:05:08 Addresses segregation needs industry wide adoption or game theoretic assumptions about staking behavior break down.
1:05:59 Brainstorms human readable addresses through named lookup tables and a friends with wallets web of trust model.
1:07:09 Rules out BIP39 style word based addresses as impractical since matching normal entropy needs 30 to 50 words.
1:08:03 Prioritizes shorter, sequential index HD wallet addresses as the near term fix, to be backported into Daedalus.
1:08:56 Draws the line between financial contract languages like Plutus and Marlowe and general execution engines like KEVM and Yellow.
1:10:22 Marlowe and Plutus exist to write rich, verifiable financial relationships, echoing why Satoshi limited Bitcoin Script.
1:11:16 The client, server, smart contract triangle developers face when deciding what logic actually needs to run on chain.
1:12:48 Explains why building Yellow on K framework semantics lets any programming language plug in without writing a new compiler.
1:13:57 The two layer ledger design means Cardano can host multiple virtual machines like KEVM and Yellow side by side.
1:18:29 Considers whether Cardano will add Monero style ring signatures so funds can be spent more privately.
1:19:45 Breaks privacy into layers: whether an address can be linked to a person, whether network use itself is visible, and whether transaction amounts are hidden.
1:21:34 Warns that OS and hardware backdoors, like a CPU frequency side channel attack that stole a PGP key, limit real world anonymity.
1:26:02 Lists the privacy toolkit on the table: bulletproofs, ring signatures, dandelion routing, and plausible deniability wallets.
1:37:56 Rejects giving any protocol level committee the power to reverse transactions, favoring market based insurance and custodial options instead.
1:46:23 Opens a question on where AI and machine learning fit into crypto, starting with complex event processing in modern finance.
1:48:11 Argues consolidated, transparent on chain data feeds could prevent the kind of hidden risk buildup that caused 2008.
1:49:23 Blockchain gives AI and machine learning models a more holistic, unified data feed than markets have ever had before.
11:07 Outlines plans to build Cardano's own off chain micro payment channels that settle on chain after Shelley, leveraging stake pools.
12:38 Points to a new RapidChain sharding paper and older research on block propagation limits that bound how large blocks can practically be.
23:14 Interest in zero knowledge based compliance systems that let people control their own data instead of banks.
26:40 Confirms early research into a dynamic transaction fee system that responds to ADA's price rather than a fixed fee.
32:42 Confirms native tokens will live on the settlement layer as first class citizens, paying fees in the token itself rather than ADA.
35:45 A blockchain land registry could store title history and prevent the same property being recorded twice.
41:08 Cardano's technical role as anchoring off chain land records by hashing them into the public blockchain, similar to Factom.
45:05 The double spend problem with offline transactions and how trusted hardware enclaves can prove a key was erased and unique.
47:48 Expects trusted hardware enclaves to be ubiquitous in phones and laptops by the mid 2020s, and floats open source enclave alternatives.
49:44 Confirms developers can bring their own currency and consensus mechanism through the control layer while staying interoperable with Cardano.
52:05 Contrasts proof of work mining as a puzzle contest with proof of stake leader election, stressing that stakers still only get paid for doing the work.
54:15 Floats a future reputation factor that could bias staking rewards toward good network citizenship rather than stake alone.
56:16 Points to the incentives paper with Elias Koutsoupias projecting over a thousand stake pools and roughly five percent annual inflation.
57:50 Compares the K framework to a set of skis built for programming language semantics rather than any one blockchain.
1:01:07 Semantics based compilation could let Cardano support many programming languages for smart contracts with little upgrade friction.
1:12:52 Open a question on complex event processing with a smoke and fire analogy for how humans piece together unrelated signals.
1:14:57 Uses a hacked news Twitter account and fake White House attack tweet to show how algorithmic trading reacts instantly to events.
1:16:26 Uses a credit card fraud detection travel pattern example to illustrate complex event processing.
1:18:00 Concludes Cardano should not natively support complex event processing, leaving it as a business opportunity for DApp and oracle providers.
1:34:40 Confirms stablecoin research is underway though nothing is ready to announce publicly yet.
2:06 The Ouroboros BFT paper is now published on ePrint and describes a newly designed, top down network stack.
2:48 Details progress on VRF and KES implementations for Ouroboros Praos and early research into XMSS signatures.
6:54 Points to a newly opened multi party computation repo from the Kaleidoscope research line and hopes it connects to Yoroi.
21:36 Answers a consensus question on why an honest longest chain stays denser than an attacker's, tied to the no prior 51 percent attack assumption.
22:00 Targets roughly a thousand stake pools, tuned by the Oxford incentives paper, aiming for far more decentralization than Bitcoin or EOS.
29:22 Frames smart contracts for web developers as removing trust and censorship risk from existing user experiences rather than inventing new ones.
31:38 The trust model shift from a centralized server saying trust me to decentralized computation for things like gambling and finance.
32:20 Stresses avoiding vendor lock-in so developers can keep familiar tools like React or Angular while swapping blockchain backends freely.
34:01 Decentralization doesn't fit everything, like an MMORPG or an internal company payroll server.
40:24 There are no plans to use Elixir, despite liking the language.
41:42 Previews the upcoming parallel chains paper converging with the Hydra side chains research toward Cardano's sharding approach.
42:59 Explains why governments like the US are standardizing post quantum cryptography now, anticipating quantum computers within 10 to 20 years.
44:35 Cardano is looking at XMSS style hash based signatures for quantum resistance, likely limited to delegation rather than everyday transactions due to size.
45:30 Expects next generation post quantum signature schemes within five to ten years, moving delegation and ledger maintenance to quantum resistant crypto sooner than transactions.
46:41 Flags understanding a quantum adversary's full capabilities, beyond just breaking signatures, as an understudied research topic for later.
48:39 Mass address generation can't threaten security because the key space is astronomically large, like finding one painted grain of sand on a Hawaiian beach.
52:06 Walks through core cryptography concepts like confidentiality, authentication and zero knowledge proofs, and how adversary models shape protocol design.
54:12 Admires how Ouroboros research builds rigorously from formal blockchain definitions like the GKL model up through networking, calling it elegant provable work.
57:56 Plutus likely won't get full dependent types like Idris or Agda, favoring refinement types via Liquid Haskell instead.