Blockchain Fetishism, Explained
What is it? And why has it endured?
Blockchains for real estate. Blockchains for climate change. Blockchains for dentists. There isn’t a problem out there that startup founders and crypto ‘builders’ haven't thought about solving with this distributed ledger technology. But does it make sense to use crypto’s favored database design for such fundamentally different business and public policy problems?
In this newsletter, I explain what public permissionless blockchains are and what tradeoffs they make against traditional database technologies. After stripping out the buzzwords, a simple conclusion follows. Public permissionless blockchains (or just blockchains, from now on) are a niche database design that is not well suited for most problems.
No particular news prompted me to write this. I just think the message is very important and want to convey it step-by-step as a reference for readers. Enjoy!
Just A Database
As I’ve written before, a public permissionless blockchain is little more than a database that’s distributed across thousands of computers. The Bitcoin blockchain, for instance, is just a shared public record of bitcoin transactions that include sender, recipient and amount.
Instead of recording transactions individually as you would with a new row in an Excel spreadsheet, multiple transactions are grouped and added together as ‘blocks.’ Each new block is then cryptographically ‘chained’ to the previous block, resulting in a block-chain structure.
The defining property of the database is that no single entity like the government controls it, since it’s decentralized across thousands of computers that have copies of the whole dataset. Why? Because Bitcoin’s governing computer code (the ‘protocol’) is designed so that it’s profitable for unrelated entities to independently secure and maintain the integrity of the ledger.
Database security and maintenance (a.k.a. ‘mining’) is incentivized with transaction fees and a reward of 6.25 new bitcoin (~$150,000) for each new block added to the chain, which happens every ~10 minutes via a lottery. Those incentives plus the way consensus over valid transactions is reached via a ‘proof-of-work’ algorithm makes attacks on the network extremely expensive. In practice, this secures the Bitcoin blockchain against manipulation and corruption.
Newer, more elaborate blockchains like Ethereum host more than records of transactions with sender, recipient and amount. On Ethereum, users can send thousands of different currencies, tokens, and importantly, upload and execute snippets of computer code called ‘smart contracts’ on the blockchain. Ethereum is also just a database, except it supports more data types and functionality than Bitcoin does.
Steep Tradeoffs
Blockchains pay a high price for their unusual decentralized architecture:
They sacrifice throughput (measured in transactions per second or TPS): Traditional databases that handle credit card transactions can handle millions of new entries per second, whereas Bitcoin does ~6 TPS and Ethereum does ~12 TPS.
They sacrifice settlement time: On Bitcoin, it takes an hour for transactions to reach final settlement. On Ethereum it takes five minutes1. On traditional databases, settlement is virtually instantaneous. (Wire transfers are slow due to regulation and company processes, not the underlying databases).
They sacrifice efficiency: Miners need to be paid a lot of fees and new cryptocurrency to maintain and validate shared ledgers. This results in transaction fees of $0.10- $100 or more during peak demand (for Ethereum). Fees and currency dilution to pay miners can total a staggering ~2% of the paper value of all crypto assets per year.
They sacrifice flexibility: Once a blockchain is live and deployed, changing the governing code to adapt to changing circumstances is uncertain and complicated (or impossible!). On a centralized database, changes are trivial.
They sacrifice reversibility: Once a transaction is added to most blockchains, settlement is essentially final. Unlike traditional databases, entries cannot ever be reversed or deleted, even if they are mistaken or fraudulent.
They sacrifice privacy: Once pseudonymous wallet addresses on most 2blockchains are linked to real people (through doxing, forensics or KYC at centralized exchanges), users’ entire transaction history and current balances become public. In contrast, traditional databases offer users privacy (if they aren’t hacked).
They sacrifice certain types of security: Private keys that grant exclusive access to crypto wallets can be lost or stolen. Once they’re gone, the assets in that wallet are irrecoverable, period.
What do blockchains gain in exchange for these steep concessions? As Jill Gunter wrote in a fantastic Coindesk Op-Ed: censorship resistance. Since the database is decentralized, no person, company or government can block transactions or activity on it. You can send bitcoin or ethereum to a known political dissident or terrorist organization, announce it publicly and nobody will be able to stop the transaction from going through or reverse it.
The censorship resistance isn’t perfect, obviously. Authorities can apply leverage where the crypto and the traditional financial system link, the so-called on and off-ramps like Coinbase. Or they can show up at your house with guns. But still, transactions cannot be meaningfully censored on the blockchain itself because authorities just don’t have enforcement tools.
THAT is the value proposition of blockchains. They’re a niche database architecture that achieves censorship resistance at an extremely high price. The price is that they’re very expensive to maintain and secure, don’t scale, limit privacy, and aren’t flexible or user-friendly compared to traditional databases.
Paying this price is only worth it if the activity is otherwise impossible on a good database. Examples include (1) buying synthetic US dollars (stablecoins) in Iran, Argentina, and Nigeria, where opening a U.S. bank account is illegal or impossible or (2) sending money across borders when transactions can be held up by sanctions or capital controls (e.g. into Cuba or out of Venezuela).
Technical Developments
People in crypto are working hard to make blockchains better while maintaining censorship resistance and decentralization. Layer-2 (L2) scaling solutions (like the Lightning Network for Bitcoin) are being built over base-layer (L1) blockchains to lower transaction fees and enhance privacy, and new blockchains with arguably better protocol designs (like Avalanche or perhaps Solana) have been deployed.
New algorithms to validate and secure networks (like proof-of-stake) have been developed to reduce the extraordinary energy costs of proof-of-work algorithms and privacy tools (like the Tornado mixer) have been developed to enhance on-chain privacy and more are in the pipeline.
But it’s an uphill battle. As I argued, the technical foundations of blockchain databases are fundamentally weak and even the smartest developers might fail to improve on them enough to compete with traditional database solutions.
Blockchain Fetishism
Will real estate improve if ownership records are put on an uncensorable but bad-in-most-other-ways database? No, it will get worse. Do tools to combat climate change need to be placed on a difficult-to-work-with but uncensorable ledger? No. Do dentists need a censorship resistant database to share information? No!
So why do people keep talking about these ridiculous use cases that don’t need censorship resistance? For many reasons, but partly because there’s a weird cultural fetish for poorly understood new technologies that are wrapped in get-rich-quick FOMO, buzzwords and technobabble.
To most people, blockchains are a magical frontier technology, not a bad database that sacrifices virtually all important properties to be uncensorable, which doesn’t matter for most problems anyways. Ignorance about the technology and the speculative nature of blockchain assets has produced an enduring cultural mania, even as the years go by without big use cases to justify crypto’s trillion dollar plus market cap.
What a lot of business and policy problems could benefit from is automation. Emerging market bonds and other securities are still traded by people over landline phones and instant messages. A lot of citizen-state relations are still conducted via mail or phone and in literal offices with employees in them. This could all be more efficient with computers. Does that mean you need a blockchain though? No.
Ninety-nine percent of the time, a regular database will work just fine.
(If you enjoyed this post, subscribe to the newsletter to get more posts straight to your inbox. Share the post on social media if you think your friends might enjoy it).
These are common heuristics. Transactions on PoW blockchains settle probabilistically, so after a certain number of blocks (six for Bitcoin), when the probability is sufficiently close to 100%, we say the transaction is “final”.
Notable exceptions include Zcash and Monero.