Module 38: Crypto

Blockchain, Bitcoin, DeFi, and the promises and problems of decentralised finance

Part A · the problem crypto solves — trust without intermediaries
Why this matters
Crypto is a coordination technology, not magic money
The core idea is stark: can strangers maintain one shared ledger without a bank, state, or platform owner deciding what is true? Bitcoin answered with proof-of-work. Ethereum added programmable contracts. DeFi, stablecoins, NFTs, DAOs, and the regulatory fight are all consequences of that first design move.
Mental model
A database with economic armour
A blockchain is not efficient in the ordinary software sense. It repeats work across thousands of machines on purpose, because redundancy is the price of not trusting one operator. The question is always whether censorship resistance, open settlement, and self-custody are worth that cost in a given use case.
Cryptography and trust — the foundations
Select a concept above.
The common misconception is that Bitcoin made copying digital files impossible. It did not. It made it economically and socially very difficult to rewrite a particular shared history of ownership — and that distinction matters enormously.
Satoshi's Bitcoin whitepaper — what it actually proposed
The nine-page "Bitcoin: A Peer-to-Peer Electronic Cash System" was posted to a cryptography mailing list on 31 October 2008. It proposed a chain of digitally signed transactions, timestamped by a distributed network using proof-of-work, where the longest chain represents the consensus truth. The whitepaper never used the word "blockchain" — that term came later. Satoshi Nakamoto mined the genesis block on 3 January 2009, embedding a newspaper headline: "Chancellor on brink of second bailout for banks." This was both a timestamp and a statement of intent. Satoshi's true identity remains unknown; they disappeared from public communication in 2011, holding an estimated 1.1 million BTC — coins they have never moved.
Part B · how blockchain works
Blockchain anatomy — a visual diagram
Block #1 (Genesis) Prev Hash: 0000000... Nonce: 2,083,236,893 Timestamp: Jan 3 2009 Merkle Root: a3e... Hash: 000000000019d6... Block #2 Prev Hash: 000000000019d6... Nonce: 1,639,830,024 Timestamp: Jan 9 2009 Merkle Root: f4b... Hash: 00000000839a8... Block #3 Prev Hash: 00000000839a8... Nonce: 877,262 Timestamp: Jan 9 2009 Merkle Root: 9b0... Hash: 000000006a625... Each block's "Prev Hash" must match the previous block's hash exactly — altering any block breaks every block after it
Immutability is a consequence of the data structure, not a rule. To alter Block #2, an attacker must also redo the proof-of-work for Block #3, #4, and every block since, faster than the honest network is adding new ones. With Bitcoin's ~600 exahash/s of global hash rate, this is computationally impossible for any realistic attacker.
Consensus mechanisms — how a decentralised network agrees
Select a mechanism above.
Mining and validators
Bitcoin miners do not verify transactions out of altruism. They compete for a block reward — currently 3.125 BTC per block after the April 2024 halving, plus transaction fees. This game-theoretic incentive makes honest behaviour more profitable than attacking the network: a miner controlling 51% of hash power would earn more by mining honestly than by attempting a reorg that would likely crash the price of the coins they hold. In Ethereum's proof-of-stake model, validators lock up 32 ETH as collateral. Dishonest validators are "slashed" — a portion of their stake is destroyed. The capital-at-risk creates skin in the game without burning energy.
Wallets, keys, and addresses
A Bitcoin wallet does not hold coins. It holds a private key — a 256-bit random number — from which a public key is derived using elliptic curve cryptography, and from which a public address is derived by hashing. Coins are on the blockchain, associated with an address; the private key proves ownership and authorises spending. Lose the private key, lose access permanently. The 12-or-24-word "seed phrase" is simply a human-readable encoding of the private key. "Not your keys, not your coins" became a mantra after the FTX collapse — about $8 billion of customer coins vanished because users had entrusted their keys to Sam Bankman-Fried's exchange rather than holding them directly.
The blockchain trilemma — and where each chain lands
Decentralised / secureScalable / fast
Bitcoin
~7 TPS, maximally decentralised PoW
Ethereum L1
~15 TPS, programmable settlement
Optimistic rollup
High TPS, 7-day challenge period
ZK rollup
Proof-based scaling, complex systems
Solana
~3k real TPS, high hardware demands
Centralised exchange
Fast because users trust the operator
The trilemma states that any blockchain can achieve at most two of: security, decentralisation, scalability. Layer-2 rollups sidestep the trilemma by offloading execution while inheriting Ethereum's security for settlement. A centralised exchange achieves maximum throughput by abandoning both decentralisation and the blockchain entirely for its internal database.
Part C · Bitcoin — the original and what it is (and isn't)
The store of value thesis
Bitcoin's maximum supply is hardcoded at 21 million coins, of which approximately 19.7 million have already been mined. New coins are issued roughly every 10 minutes as a block reward, but that reward halves every 210,000 blocks (roughly four years). The 2024 halving cut it to 3.125 BTC; around 2140, new issuance will stop entirely. The inflation hedge argument holds that because central banks can expand fiat money supply at will (the Fed doubled M2 between 2020 and 2022), a provably scarce asset should hold value over long timeframes. The counterargument: Bitcoin's 30-day volatility has frequently exceeded 80% annualised, making it a poor store of value in any short or medium timeframe — it correlates more with risk appetite than with inflation.
Bitcoin as a payment system
Bitcoin was originally titled "A Peer-to-Peer Electronic Cash System" — but it has largely failed at the cash part. Bitcoin's base layer processes about 7 transactions per second, compared to Visa's 24,000. Block confirmation takes 10 minutes on average, and fees spiked to $60 per transaction during the 2021 bull market. The Lightning Network is a layer-2 payment protocol that routes payments through off-chain state channels, settling only occasionally on-chain. El Salvador made Bitcoin legal tender in 2021, but adoption was limited, and the IMF pressured them to scale back the mandate as a condition of a $1.4 billion loan in 2024. The main practical use case for Bitcoin today is as a speculative asset and cross-border settlement rail, not everyday spending.
Bitcoin price cycles — major bull and bear phases
Each cycle has been driven by a different catalyst: 2013 by Silk Road notoriety and the Cypriot banking crisis; 2017 by ICO mania and retail discovery; 2021 by institutional adoption (Tesla, MicroStrategy) and COVID-era stimulus money; 2024 by the approval of US spot Bitcoin ETFs in January, which attracted over $30 billion of net inflows in their first year.
Block subsidy per halving era — Bitcoin's declining issuance schedule
The subsidy halves every 210,000 blocks. In 2009, miners earned 50 BTC per block; by 2140, the reward rounds to 0 satoshis. From that point, transaction fees must be sufficient to compensate miners — this is the long-run security debate. The chart uses a proportional scale, so the 3.125 BTC era (2024) appears vanishingly small against the original 50 BTC era.
Bitcoin halving supply model — explore the cumulative schedule

Drag to see how much BTC will have been mined by each halving event.

Halving #1 (2012)Halving #32 (~2140, last)
...
...
Part D · Ethereum and smart contracts
What Ethereum added
Vitalik Buterin published the Ethereum whitepaper in 2013, at age 19. His insight: Bitcoin's scripting language was deliberately limited to simple payment logic. What if the blockchain were a general-purpose computer? Ethereum added the Ethereum Virtual Machine (EVM) — a sandboxed computing environment that runs on every node simultaneously. Smart contracts are programs stored on this computer. Once deployed, they run exactly as coded, without any possibility of downtime, censorship, or third-party interference. Gas fees are the payment for computation: every operation costs a specific amount of gas, preventing infinite loops from stalling the network. The Merge in September 2022 switched Ethereum from proof-of-work to proof-of-stake, reducing its energy consumption by an estimated 99.95%.
Tokens, NFTs, and the EVM ecosystem
The ERC-20 token standard, introduced in 2015, enabled anyone to issue a fungible token on Ethereum in minutes. This spawned the 2017 ICO boom (and subsequent crash), the 2020 DeFi summer, and thousands of essentially worthless tokens. ERC-721 is the non-fungible token standard: each token is unique and not interchangeable with any other. The 2021 NFT boom peaked when Beeple's digital artwork "Everydays: The First 5000 Days" sold at Christie's for $69 million. The critical misconception: owning an NFT does not mean owning the underlying image file — it means owning a blockchain record pointing to a URL, which may or may not still exist. Most NFT projects lost 90-99% of their peak value by 2023.
Major blockchain platforms — explore and compare
Transaction throughput comparison (transactions per second)
Theoretical vs real-world throughput differ dramatically. Solana's 65,000 TPS figure is a theoretical maximum; real throughput under load has been around 2,000-4,000 TPS, and the network has experienced multiple complete outages. Visa's 24,000 TPS is a network capacity figure; average daily throughput is closer to 2,000. The comparison also ignores finality time: Bitcoin transactions are considered final after 6 confirmations (~60 minutes); Solana reaches finality in ~400ms.
Part E · DeFi — decentralised finance
What DeFi is — replacing banks with code
Traditional finance requires permission: to open a bank account, to get a loan, to trade an asset. DeFi protocols are open to anyone with an Ethereum address. At its 2021 peak, DeFi held over $180 billion in total value locked (TVL). By 2023, that had collapsed to around $40 billion — primarily because DeFi's yields depend on token inflation and speculative activity that cannot be sustained. The "composability" of DeFi — the fact that protocols can be stacked like building blocks — is real and powerful. It is also a risk amplifier: one failing protocol can trigger cascading liquidations across everything built on top of it.
Uniswap and AMMs — prices from pools
Uniswap's x*y=k formula means the product of the two token quantities in a pool must remain constant. If a pool holds 100 ETH and 300,000 USDC, the implied price is ~$3,000 per ETH. A buyer pushing the ETH balance down to 95 must push USDC up to 315,789 — paying ~$3,158 per ETH. Arbitrageurs keep pools near external market prices, which means liquidity providers are effectively selling what rises and buying what falls. This is automated price discovery without an order book, but it introduces impermanent loss for liquidity providers.
Lending protocols — collateral first
Aave and Compound do not lend like banks. They require overcollateralised loans: a borrower might deposit $10,000 of ETH to borrow $6,000 of USDC. If the collateral value falls below a liquidation threshold, automated liquidators repay the debt and seize the collateral at a discount. There is no personal credit scoring and no relationship with a lender — the code enforces the rules. This makes DeFi lending capital-inefficient but also means it cannot easily fail the way a bank can through fractional reserves.
Stablecoins — crypto's dollar rail
Stablecoins are the workhorse of DeFi because they remove volatility while keeping blockchain settlement. USDT and USDC are fiat-backed promises; DAI is crypto-collateralised and overcollateralised. TerraUSD was algorithmic and relied on an arbitrage relationship with LUNA to maintain its peg — a mechanism that is only as stable as confidence in the system itself. In May 2022, confidence broke, redemptions accelerated, LUNA supply hyperinflated, and $60 billion of market cap vanished in 72 hours. The lesson: a stablecoin is only as good as its collateral, redemption mechanism, and behaviour under stress.
DeFi total value locked (TVL) — rise and fall
TVL is denominated in USD and is doubly volatile: it falls when users withdraw funds and when underlying token prices fall. The 2022 collapse was a sequence of cascading failures: Terra/Luna (May), Three Arrows Capital insolvency (June), Celsius Network freeze (June), and finally FTX (November). Each failure removed liquidity and confidence simultaneously.
The word "yield" is dangerous in DeFi. Sometimes it means real fees paid by traders or borrowers. Sometimes it means token emissions that dilute existing holders. Sometimes it means hidden leverage that will unwind violently. The right question is never "what APY?" It is: who is paying me, why, in what asset, and what can break?
Impermanent loss calculator — 50/50 liquidity pool

Impermanent loss is the cost of providing liquidity when the price ratio of the two pooled tokens changes. Enter the price change of one asset to see how much worse off you are compared to simply holding — before accounting for any trading fees earned.

Enter values above and click Calculate.
Formula: pool value relative to hold = 2√r / (1+r), where r is the new price ratio. This ignores gas costs, time-varying fee volume, price path dependence, and smart-contract risk. Impermanent loss is only "impermanent" if the price returns to the entry ratio — if it does not, the loss is realised on exit.
DeFi failure modes — where the money actually goes (breakdown by category, 2020-2024)
Key theft (often through social engineering rather than brute force) and custody fraud account for over half of all losses. Smart contract exploits get the most press but are third. Bridge exploits are disproportionately large in dollar terms because bridges are single points of failure concentrating billions of dollars. Even audited contracts are not safe: the Ronin Bridge hack ($625 million, 2022) exploited compromised validator keys, not the contract code itself.
Part F · NFTs, DAOs, and Web3
NFT market cycle — trading volume over time
The NFT market collapsed roughly 97% from peak to trough. Legitimate remaining use cases include music royalty tokenisation (artists receive automatic payments when their NFT is resold), gaming items with verifiable scarcity, event ticketing (preventing scalping by tying tickets to identity), and digital certificates of provenance for physical artworks. The core misconception was that "ownership" of an NFT conferred rights over the underlying content — in almost every case it does not.
DAOs — decentralised autonomous organisations
A DAO is an organisation governed by on-chain votes, with rules encoded in smart contracts. Token holders vote on proposals — treasury spending, protocol changes, or personnel decisions. MakerDAO governs the DAI stablecoin system; Uniswap DAO controls a $3 billion treasury. The reality check: voter turnout in most DAOs is below 5%, and governance token ownership is heavily concentrated — the top 10 addresses in many DAOs control over 50% of voting power. The 2016 "The DAO" hack remains instructive: a vulnerability allowed an attacker to drain $60 million of Ether, and the Ethereum community controversially hard-forked the blockchain to reverse the theft, demonstrating that "code is law" has limits when the code contains bugs.
Web3 — the vision and the critique
Web1 (1990s) was read-only: static pages. Web2 (2004-present) is read-write but centralised: your content lives on platform servers under platform terms. Web3 promises read-write-own: digital assets, identity, and data controlled by users through cryptographic keys. The critique, articulated by Moxie Marlinspike (Signal founder): users do not want to run servers. In practice, "decentralised" Web3 apps still rely on centralised infrastructure — Alchemy and Infura provide the API nodes most DApps use; a 2022 Cloudflare outage took down large portions of the DeFi ecosystem. The decentralisation is often a frontend illusion over centralised backends.
Use case maturity — where crypto applications actually stand
Proven demandSpeculative / narrative-driven
Stablecoin settlement
Clear demand where dollar rails are slow or blocked
Bitcoin self-custody
Censorship-resistant savings with real track record
DeFi trading
Real usage, but cyclical and risk-heavy
DAOs
Useful treasury tooling, difficult governance
NFT art
Real niche, heavily distorted by speculation
Metaverse land
Narrative-first with fragile, thin demand
Part G · regulation, collapses, and the bigger picture
Regulatory stances by region — explore
Select a region above.
Major collapses and failures — a timeline
The pattern across every major collapse is consistent: (1) an unsustainable yield or price appreciation attracts capital; (2) leverage amplifies the system; (3) a small disruption triggers liquidations; (4) liquidity dries up faster than any regulator can respond. FTX was not a DeFi failure — it was an old-fashioned fraud where customer funds were used to prop up Alameda Research. The crypto framing was largely irrelevant; it was a custody failure with a crypto wrapper.
Bitcoin energy consumption in context
Bitcoin's ~150 TWh annual consumption is frequently cited as its environmental cost, but context matters. Roughly 50-60% of Bitcoin mining used renewable energy as of 2023 (Bitcoin Mining Council voluntary surveys). Proof-of-stake Ethereum uses ~0.01 TWh/year after The Merge. Gold mining uses an estimated 130 TWh/year and produces significant physical waste. The global banking system consumes an estimated 260 TWh/year. Whether Bitcoin's energy use is "worth it" depends entirely on whether you accept its value proposition.
Dollar-cost averaging calculator — Bitcoin accumulation
Enter values above and click Calculate.
This uses a simple compound growth model and does not account for taxes, exchange fees, or Bitcoin's extreme volatility. Historical Bitcoin CAGR from 2015 to 2024 was approximately 50% — but includes multiple 80%+ drawdowns. A 20% annual return assumption is conservative relative to history but aggressive relative to traditional assets. Past performance is not predictive.
Crypto market composition — what is the market actually made of?
Market cap figures as of early 2025. Bitcoin dominance has rebounded from 38% at the 2021 alt-coin peak to ~55% in 2025, as retail speculative interest shifted to memecoins and institutional money concentrated in Bitcoin ETFs. "Other" includes thousands of tokens, most of which have negligible real liquidity.
What crypto has genuinely delivered
A permissionless global settlement layer for value. Remittances at a fraction of Western Union's 8% average fee. A credible alternative to inflation-ravaged local currencies (Argentina, Turkey). Programmable money enabling financial instruments without counterparty credit risk. Ethereum's smart contract model inspired a genuine computer science innovation. A stress-test of monetary theory that forced central banks to think seriously about digital currencies. Stablecoin settlement, Bitcoin custody, and Ethereum smart contracts are not imaginary achievements.
What crypto has failed to deliver (so far)
Everyday payments: crypto is rarely used to buy things. Banking the unbanked: most DeFi requires significant capital and technical literacy — and can become "selling volatile risk to people with fewer protections." Truly decentralised governance: most protocols are controlled by a small group of insiders and VCs. Privacy: Bitcoin is pseudonymous, not anonymous — blockchain analysis firms routinely trace transactions. Web3 applications that non-technical people want to use at scale, beyond speculation.
Ask who bears loss
Every crypto design has a loss allocation rule. In a bank failure, deposit insurance, regulators, shareholders, creditors, and courts fight over loss over months or years. In crypto, loss may fall instantly and irrevocably on key holders, liquidity providers, token buyers, bridge users, or governance voters. The absence of an intermediary does not mean the absence of trust. It means trust moved — from institutions to code, to auditors, to token distributions, to anonymous developers. Understanding where trust actually lives is the most important question before using any crypto product.
Part H · questions a smart, sceptical reader would ask
Is Bitcoin actually scarce if anyone can create a new cryptocurrency?
Bitcoin's 21-million cap is enforced by the rules of the network, but nothing stops anyone from copying Bitcoin's code and creating "Bitcoin2" with the same supply cap. Thousands have done so. The distinction is network effect and security: Bitcoin's value is partly a function of its hash rate (making it the most secure proof-of-work chain by an enormous margin), its liquidity, its regulatory clarity, and its 15-year track record. A copy starts with none of these. The analogy is not "gold vs fake gold" but "original Levi's 501s vs a manufacturer making identical jeans" — technically identical, but the brand and history matter to buyers.
Could a government just ban Bitcoin?
China effectively banned it in 2021 — and succeeded in moving the majority of Bitcoin mining out of China within months. But Bitcoin's global node network continued running without interruption. Banning Bitcoin in one country primarily harms citizens of that country, rather than shutting down the network itself. A coordinated global ban would require unprecedented international cooperation. The US approach post-2024 has moved in the opposite direction: Bitcoin ETFs are now listed on NYSE and Nasdaq, and the current administration has explicitly embraced crypto. A ban in the US would face serious constitutional challenges around property rights.
If blockchain is immutable, how was the 2016 Ethereum hack reversed?
The Ethereum community voted to execute a hard fork — a change to the protocol rules that rewrote the chain's history to return the stolen funds. The majority of miners adopted the new chain; a minority refused on principle and continued the original chain, now called Ethereum Classic (ETC). This episode revealed that "immutability" in practice means "immutable unless a sufficient majority of stakeholders agree to change it" — much closer to how traditional institutions work than crypto advocates typically acknowledge. It remains the most honest example of the limits of the "code is law" ideology.
Is crypto primarily used for crime?
Chainalysis estimated that in 2023, approximately 0.34% of all cryptocurrency transaction volume was associated with illicit activity. In absolute terms that was about $24 billion, which sounds large until you note that the UN estimates $800 billion to $2 trillion is laundered through traditional financial channels annually. Cash is far more useful for crime: it leaves no audit trail. Bitcoin's blockchain is a permanent, public record — every law enforcement agency now has blockchain analytics tools. The persistent "crypto is for criminals" framing reflects the early Silk Road era and has not kept up with the data.
What would it actually take for Bitcoin to replace the US dollar?
The US dollar functions as a global reserve currency because the US has the world's largest economy and deepest capital markets, US Treasuries are the world's primary risk-free asset, and the dollar is stable enough to serve as a unit of account. Bitcoin satisfies none of these conditions currently. To serve as a unit of account, volatility would need to drop to low single-digit percentages annually — comparable to gold's. That would likely require a market cap 50-100x today's. Even ardent Bitcoin proponents like Michael Saylor generally argue for Bitcoin as a reserve asset ("digital gold"), not as a transaction currency replacing the dollar — those are very different claims.
Who are the real winners from the crypto industry so far?
Early Bitcoin miners and holders who sold near peaks (2013, 2017, 2021). Venture capital funds that invested in infrastructure rather than tokens — Andreessen Horowitz's a16z Crypto raised over $7 billion across its funds. Exchanges: Coinbase generated $3.1 billion in revenue in 2021. Layer-1 blockchain foundations and their early-employee token holders. On the other side: retail investors who bought near peaks — the 2021 peak saw over $1 trillion in aggregate paper wealth destroyed in the subsequent bear market, with losses concentrated among retail participants. The distribution of outcomes has been steeply unequal, with sophisticated early entrants extracting value from subsequent retail waves.
If blockchains are so inefficient, why use them at all?
Use them when the inefficiency buys something valuable: censorship resistance, shared settlement between parties that do not trust one operator, self-custody, or public auditability. A normal database is better for a payroll app, a private inventory system, or a social network run by one company. A blockchain starts making sense when the operator itself is the risk. That is why the strongest use cases cluster around money, settlement, and assets that need neutral infrastructure — not around applications where a trusted company would do the job better, faster, and cheaper.
Is Bitcoin anonymous?
Bitcoin is pseudonymous, not anonymous. Addresses are alphanumeric strings, but every transaction is public forever, so blockchain analytics firms can cluster behaviour and connect addresses to exchanges, sanctions lists, hacks, and real identities. Privacy can improve with careful coin control or dedicated privacy tools (like Wasabi Wallet's CoinJoin), but ordinary users leak information constantly through exchange KYC, on-chain patterns, and IP addresses. Cash is often more private in daily life than transparent blockchains — one of the more counterintuitive facts about Bitcoin's design.
Did Ethereum's Merge make transactions cheaper?
No. The Merge changed Ethereum's consensus mechanism from proof-of-work to proof-of-stake, cutting energy use by about 99.95%. It did not increase execution capacity enough to significantly reduce gas fees. The cheaper-transaction strategy is rollups and data availability improvements — specifically the Dencun upgrade in March 2024, which introduced "blobs" for L2 data storage and cut L2 fees by roughly 90%. The Merge and the fee reduction are often conflated, but they are entirely separate technical changes with different timelines.
Are stablecoins safer than volatile crypto?
They are safer on price volatility, but they introduce a different set of risks: issuer solvency, reserve quality, legal jurisdiction, and redemption mechanics under stress. A fiat-backed stablecoin depends entirely on the quality and accessibility of its reserves — Tether (USDT) has been criticised for years over reserve transparency. A crypto-collateralised stablecoin like DAI depends on its liquidation system functioning during market crashes. An algorithmic stablecoin that relies mainly on confidence can fail with brutal speed, as TerraUSD demonstrated in May 2022. "Stable" describes the price target, not the underlying risk profile.
What is the best single question to ask before buying any token?
Ask: what gives this token durable demand that is not just the hope that someone later pays more? For Bitcoin, the answer is fixed-supply settlement and monetary premium built over 15 years. For ETH, it is payment for blockspace, staking yield, and collateral in a large application ecosystem. For many tokens, the honest answer is weak: governance rights that nobody uses, token emissions that dilute holders, insider unlocks that create constant sell pressure, or simply a ticker symbol and a compelling narrative. The inability to give a clear answer to this question is itself an answer.