Aptos Incentivized Testnet 2 | Highlights and Updates
We recently concluded a very successful AIT2 that brought the Aptos community closer together. Thank you for participating and making AIT2 even more successful than AIT1! We have taken the time since the conclusion of AIT2 to evaluate the outcomes and would like to share updates around the details of our incentivized testnet and our Mainnet launch this autumn.
Here are a few key updates, before diving into the AIT2 retrospective:
- Aptos Labs remains focused on launching the Aptos Mainnet to bring the community the safest, fastest and most reliable foundation for building within Web3.
- Previous AITs taught us that incentivized testnets require time from preparation and registration, to execution, and then finally analysis. To continue to facilitate a healthy path to launch and maintain our expectations for AITs, AIT3 will be our last incentivized testnet before Mainnet launch.
- AIT4 and additional AITs will happen after Mainnet. Specifically, we will continue to host incentivized testnets for each major upgrade of the software.
- Our roadmaps are driven by community feedback and experience. As a result, some features from AIT3 and AIT4 have been deferred for post Mainnet.
- Our selection pool size for AIT3 will be similar to that of AIT2 as we focus on improving the network leading up to Mainnet launch.
- The team is planning a code freeze for Mainnet in the coming month, and shortly thereafter, we will publish a long lived Testnet for better community collaboration and preparation for Mainnet.
- AIT3 registration — August 19
- Code freeze — August 25
- AIT3 launch — August 30
- AIT3 concludes — September 9
- Mainnet Launch — Autumn
- AIT4 — Winter
AIT2 Highlights and Rewards
AIT2 has been another successful test driven by the enthusiasm and dedication of the community. There were over 225 community nodes, 26 of which were run by Aptos and Aptos partners. The nodes were spread across 44 countries and 110 cities around the world.
First the rewards! The team recognized the efforts made by so many and modified the rewards ranges to be more inclusive.
The requirements were:
- Perform all the required objectives on time including joining the validator set within 24 hours and exiting the validator set upon the conclusion of the network.
- Meet node liveness as defined by metrics push data ≥ 95%. An overwhelming majority of participants met this goal!
- Participation as defined by one consensus proposal per hour ≥ 95%. Though given that we put increased load and data refresh lag, we reduced this to 90%.
- Vote in at least 25% of consensus rounds.
The rewards were as follows:
- Full reward: 500 Aptos tokens, for those that were able meet all the criteria above
- Partial reward: 300 Aptos tokens, for those that were able to achieve at least 50% on both liveness and participation along with the other criteria
- Bonus: 200 Aptos tokens for each of the following criteria:
– Top 10% and upgraded at least once.
– Operators that kept their nodes in remote regions, while sustaining all the requirements above. As they helped us better understand the nuances of how decentralized node distribution affects performance.
– Operators who went above and beyond in helping us identify critical issues and significant contributions.
Note: Aptos tokens can only be given to non-US participants, due to regulatory considerations. Any rewards in connection with the incentivized testnet will be provided only following successful mainnet launch and any tokens provided will be subject to a minimum one-year lockup period. More details are provided in our terms and conditions.
- The community was heard! The heroes within the community were recognized and helped us improve our process and guide other operators through this process. The collaborative nature of the community truly reflects the Aptos ethos.
- Incentivized testnets remain a great way to validate software stacks. The team along with operators were able to identify glitches and as well as opportunities for performance wins. The Aptos Labs team and the operators coordinated successfully across four software upgrades with zero downtime on the network.The most notable of these was an upgrade that reduced bandwidth consumption by over 50%.
- Operators are attentive and help us better understand the nuances of deploying complex, decentralized software. In this phase, we explored a voting-based rewards algorithm. Some participants migrated to regions that ensured they were more likely to have their vote included. Those that didn’t still helped us validate that being placed in remote regions can be challenging and require better understanding of latency and bandwidth limitations.
- Rewards based upon voting disincentivized geographical distribution. In our consensus protocol, proposers only wait for the first ⅔ of stake-weighted votes, limiting the opportunity for geographically remote nodes from effectively participating. This leads to a bias towards clustering of nodes within a region as operators try to maximize their voting performance. The team is actively working on a better rewards design aiming to reduce this bias. The diverse ways in which the community set-up their nodes helped us understand how different configurations impacted the protocol.
- This experience allowed us to identify opportunities to improve the operator experience across tooling and deployment considerations to significantly reduce cost and operations. Some operators were impacted by network changes and we had yet to add network address updating to our toolset. Similarly operators helped us notice high usage of bandwidth that resulted in us offering software upgrades to reduce the overhead and invest more time into this space in the time leading up to Mainnet.
Coming Up Next
Thank you for your continued support as we move towards Mainnet and for the success and learnings that came from AIT2. In the coming weeks, we’ll share a more in-depth update about AIT3. We also welcome questions and comments on our newly launched forum. Until then, continue to follow us on Twitter and star us on GitHub.