Taiko alpha-2 testnet (a2) is live! Following our first testnet (a1), Snæfellsjökull, a2 continues the trend of Icelandic volcanos and is called Askja. Let’s scale this mountain on our road to scaling Ethereum securely.

Relative to alpha-1, the most prominent changes are:

To jump right in and get to testing, transacting, deploying, or proving, see the guide:

TaikoA decentralized, Ethereum-equivalent ZK-Rollup.taiko.xyz

To learn much more about this testnet, please read on.

https://taiko.xyz/docs/guides

About the Taiko Alpha-2 testnet (Askja)

Let’s home in on the most salient new piece of the test network: provers.

Provers

Validity proof generation has entered the network. As you may know, validity proofs (ZK proofs) are what proves the validity of the L2 EVM state transitions, and generally what makes a ZK-Rollup special.

Alpha-2 does not entail proving the full EVM execution yet, but rather proving the “public input” circuit. The public input circuit takes the data that a proposer pushes on to L1 and hashes it, and checks this hash value against the hash value that was automatically generated by the L1 smart contract. Effectively, it checks that the blob of data (the list of transactions) is valid, and that it matches what was committed to.

Oracle Prover 🔮

Since the ZKP coverage in a2 is still incomplete, Taiko holds a special address that acts as the oracle prover. Only this oracle prover can prove blocks as the very first prover (with empty proofs that won’t be verified onchain), all other actual provers must prove the block after the oracle prover with the same block headers.

Oracle prover is a way to ensure the blockchain does not enter an invalid state while ZKP coverage is not comprehensive. When ZKP coverage is indeed comprehensive, we will then remove this oracle prover entirely from the codebase.

Hardware requirements and benchmarks 🖥️

Generating validity proofs is a computationally-intensive process. Below is a sense of the minimum hardware required, and benchmarks if using this hardware.

Note: this is not representative of what a full ZK-EVM proof will eventually cost.

Rewards for provers 🫰🏽

To reward provers for their computing resources expended during proof generation, Taiko Labs is creating a reward campaign. It’s worth noting that under a ‘normal’ mainnet scenario, provers would be rewarded by the protocol for their work with a token that trades freely in the market (like ETH or a rollup token). In a testnet scenario we have to make the following modifications.

The Taiko protocol will account and administer proof rewards as it normally would, and mint test tokens (“TTKO”) to the prover of each block. These TTKO will then be redeemable for USDC at the end of the testnet reward period. The reward budget for provers on alpha-2 testnet will be up to 50,000 USDC, and provers will redeem their TTKO for their proportional share of the USDC. We expect the reward period to last for approximately 300,000 blocks, or 69 days. We use the assumptions of 20 second block times, and $0.15 average proof generation cost. Given rewards will be in USDC (representing a potential match to the currency costs are incurred in), we attempt to bake in a surplus of rewards of approximately 10% relative to provers’ costs. We will not reward provers after 300,000 blocks. Learning how the protocol reacts in the scenario where most provers stop proving blocks is part of this testnet’s desired outcomes.

Please note, profit considerations (if any) will heavily depend on a prover’s hardware cost, and certain assumptions we use may not reflect reality for the duration of the reward period. We reserve the right to adjust the program, with the ultimate goal of testing permissionless proving while providing a reasonable reward to provers on average.

What is not in this testnet

You may notice that this is in contrast to testnet alpha-1 which did allow anyone to become a proposer permissionlessly—and indeed, over 2000 of you showed up to act as a proposer!

The reason we have permissioned the proposer role in alpha-2 is for DOS-resistance. Specifically, given permissionless provers are now introduced onto the network, and given that proof generation consumes meaningful computing resources, we must limit who can propose blocks to be proven since proposing on a testnet consumes no real resources, thus creating a mismatch. That is, proposers can spam-propose blocks to L1 by just paying Sepolia ETH (not real ETH), requiring provers to expend real resources to prove them, potentially making the pending blocks queue interminable and halting the network (or requiring a much larger budget from Taiko to reward provers). By controlling the proposer for this testnet, Taiko can emulate a higher-fidelity environment to what real ETH costs for proposers would naturally lead to.

Get involved

Once more, here is the testnet hub where you can find documentation, guides, and links to all relevant apps and tools for testnet usage:

https://taiko.xyz/docs/guides

Just like the first time around, besides deploying, transacting, node running—and now proving—there are plenty of other ways you can get involved and help Taiko improve. Some of these include:

We will have ☎️ Community Call #2 ☎️ on Friday, March 24th at 15:00 UTC in our Discord server. You can find the ‘stage’ channel here. We will provide some general updates but will spend much more time answering questions. Please ask any questions you would like in the community-call-questions channel in Discord in advance. The questions can be testnet-related or general.

Thank you

Thank you all very much for being with us on this journey. It would not be possible without the community and broader contributors across Ethereum and other ZK or rollup efforts. We are thrilled in inching closer to a fully decentralized Ethereum-equivalent ZK-EVM with you.

Join us 💗

Explore open positions on our job board.

Follow us 🥁

To stay updated on the latest from Taiko:

Contribute

Contribute to Taiko and earn a GitPOAP! You will also be featured as a contributor on our README. Get started with the contributing guide.