StarkNet is a composable zero-knowledge proof L2 rollup. It allows developers to build composable contracts using zero-knowledge proof technology.
A note on STARKS from ethhub:
ZK-STARKs (Zero-Knowledge Scalable Transparent ARguments of Knowledge) are a type of cryptographic proof technology that enables users to share validated data or perform computations with a third party without the data or computation being revealed to the third-party, also known as a zero-knowledge proof, in a way that is publicly verifiable. In simpler terms, a zero-knowledge proof can prove something is true without having to reveal what exactly it is proving. For example, ZK-STARKs would allow Alice to verify Bob's banking information using a zero-knowledge cryptographic proof instead of revealing the confidential information to Alice.
For a true on-chain world to emerge we require 3 things; cheap fees, cheap computation, composability. All these 3 things only currently exist on a ZK Rollup like StarkNet.
On-chain games will require many transactions, so these transactions need to be sufficiently cheap enough to allow gameplay to actually happen. This is impossible on mainnet, so it has to be on a Rollup.
Games need heavy computing power. We need to be able to run noise algorithms all on-chain and prove that they ran correctly. This is not possible on mainnet and it’s still going to be very expensive to run on an Optimistic rollup because of the way the EVM works.
ZK-Rollups shine in this scenario. A 1000 line function costs similar in transaction cost as a 100 line function (does not include storage). Only the proof needs to be validated on-chain, and the size of these proofs does not change with function compute needs. The greater the transactions, the cheaper the amortised fees.
‘Composability is to software as compounding interest is to finance’ - Chris Dixon
The composability of solidity contracts allowed developers to fully realise the potential of Ethereum. First, it was financial lego blocks (DeFi), now it’s on-chain metadata, next it’s games liquidity and item interoperability.
The Bibliotheca team is a huge fan of optimistic roll-ups, and we even deployed our Alpha protocol on Arbitrum. However, we do not see them as the ideal solution for a gaming protocol due to their cost and their 7-day withdrawal process. Bridge protocols have been able to reduce this with erc20 tokens, however, NFTs fundamentally cannot be bridged as easily as fungibles so they will be inherently locked into the OR.
Optimistic rollups will become cheaper in the future after call data compression and EIP-4488, but they are too expensive right now for anything but basic games to emerge.
We do not wish to have to maintain a specific app-chain when we can achieve the same outcome running on a L2 like StarkNet. With StarkNet, we get the speed and cost of an app-chain but with L1 security guarantees. We wish to build a gaming protocol and not a blockchain.
Sidechains have served their purpose of the time before rollups. Our thesis is that sidechains will decrease in dominance as rollups increase in dominance. This is a function of the market wishing for more security and more guaranteed uptimes. They do not share the same security guarantees as an L2 as they are running their own state machines, and if they go offline forever, your funds will be lost forever. This is counter to an L2 which post compressed data back onto L1, so in the event of an outage, you will always be able to get your funds returned.