I’ve spent some time now following Ethereum and studying the system. Ethereum is a general-purpose open-access globally-distributed computer that can’t be shutdown. It includes an economic system where computer operations are paid for with “gas” which is a derivative of the network’s native and crowd-funded currency token, Ether. Here are some of my thoughts on the technology in outline form.
- Ethereum is not a DevOps play
- It’s slow and currently equivalent to the CPU in a mobile phone from 1999, making it cumbersome and unnecessary as a backend for most applications
- If the network were to accommodate an app like Facebook, the inferred computing power requirements of each node would result in a highly centralized topology, so why not use Amazon RDS in the first place?
- What gains can a decentralized computing network of unknown and untrusted peers realistically achieve through sharding and parallelization? Given the ambigious performance characteristics of each peer, what is the data distribution model?
2. Unclear what the killer app is
- Resources in the real-world can already be allotted on an individual basis by any number of other payment mechanisms.
- It’s trivial to handle digital payments to open the door to an AirBNB or unlock the passenger door to a self-driving car
- The maid oracle snitching on the AirBNB tenant based on the sensor levels on the smoke detector sounds kind of fantastical
- The self-regulating-renting-and-optimizing car or condo contract still has human owners, how is liability handled in the event of an accident?
- It’s a general purpose programming language so it can ‘do anything’ which creates low signal:noise (do I really need it or am I just using it because I can?)
3. The blockchain-enabled Internet of Things is a market that approaches near perfect informational symmetry
- Near frictionless machine to machine payments and transparent public markets means there is very little ‘edge’ or market disconnects to capitalize on
- If you aren’t taking advantage of disconnects in the market the marginal revenue approaches the marginal cost
- At equilibrium would result in limited or fleeting opportunities for most participants
4. It’s not clear a peer-to-peer Uber running as an autonomous-corporation on Ethereum would be an improvement over the existing service
- Enhanced privacy for user/provider but at what sacrifice?
- Cut out the middle man but you enable a transparent market (bid/offer) which will likely lead to lower prices and less profit for the driver, and then what if you get in an accident?
- Ultimately the driver ceases to exist and the age of Minority Report is upon us, so who owns the self-driving car and who holds the liability in the event of an accident or bodily harm?
- Not clear if the surface-level cost savings and transparency of the open market outweigh the ambiguous risk-profile
5. Contract law is a function of government, not everyone is a perfect coder, and not everything is legally disrupt-able
- The trend is towards consolidation of legal power
- Businesses rely on courts to interpret law which is outpaced by innovation
- Individual contracts aren’t directly modifiable once deployed, but if you try to build a modular architecture you change the trust model
- The barriers to auditing contract behavior will be prohibitive for most people, so the trustlessness for users of the system decreases at scale
- Impossible to test for every edge case in code (it’s what you think you know that just ain’t so)
- The ‘sharing economy’ already fights an uphill battle against the establishment providers (taxis, banks, etc), and decentralized applications will likely be far less funded while still keeping harpoon-able maintainers and contributors
6. Ethereum is really good for running lightweight programs where you need extreme redundancy and security, or for the transfer of value between two unknown parties
- perfect for peer to peer wagers and gambling (trivial to write a rock, paper, scissors game with money involved and distributed between strangers)
- crowd-funding the release of information
- building your own custom wallet protocols (programmatically guarantee that only X amount of Ether can be moved from account Y per given interval)
7. Ethereum’s programmability does not inherently improve p2p finance over existing technology
- if I loan you money you need access to it in order to earn me a rate of return
- if the money is tied up in an Ethereum contract you won’t be able to take advantage of it
- if the money leaves the Ethereum contract, there’s no recourse no matter how smart the contract may be
- it’s hard to imagine the disbursement of significant amounts of value being triggered by oracle-scraped external events, and multi-sig is not a groundbreaking technology