The Cosmos team has launched Gaia-9000 in preparation for pulling the trigger on the Game of Stakes validator competition. Though they are not the same network, Gaia-9000 is critical to test parameters that will be used for the competition when it launches on November 11th.
The Cosmos team held a webinar on November 7th to give an update on GoS, and the conversation provided some juicy info on what validators should expect when the game launches.
It’s worth the listen if you plan to compete to win, but we’ve summarized the key discussion points below for quick reference on what the team is planning.
Important points from the webinar:
- A goal is to run GoS for 2 weeks once any critical flaws have been discovered and fixed, so the overall duration is expected to be 4 weeks. Cosmos Mainnet could be launched immediately following the conclusion of GoS — in early to mid December to deliver on the team’s current 2018 goal.
- What kinds of attacks could disqualify a player? Attacks that crash the network or nodes or attacks on the state machine. Those should come in through the bug bounty. The bug bounty will also be extended to SDK issues very soon.
- Flooding the network with governance requests is pointless at this time so slashing for governance vote misses will not be a part of GoS (or Mainnet it appears).
- Inflation in the GoS world is fast, so getting online and bonding your validator ASAP will give some advantage. The larger advantage will come when the Tendermint validators unbond. This will happen about 12 hours after launch. From there, the GoS players will run the network.
- The Tendermint valiators will “attempt” to remain online as full nodes to provide peering. Beware of changes in “altruistic” peering behaviors and don’t count on the seed nodes to know what’s going on.
- The 100 validator target for launching mainnet: Tendermint is much more performant than was originally envisioned. Currently running 2 second block with 80 validators across 3 continents. Zaki expects that Governance will expand the number of validators in 2019 as long as we stay in the 5–10 second blocktime range.
- More than 50% of participants in GoS hail from Asian countries.
- There will be transaction costs and gas fees and validators can set their minimum fees for block creation.
- Attempted winning through group delegation to 1 validator will be frowned on, and may cause disqualification. The focus should be on incentive compatibility level attacks.
- We might experiment with on-chain governance to disqualify teams behaving badly. There may be an additional challenge to get the community to upgrade to a non-Tendermint released version of the software. A 34% attack might halt the network and would be quite exciting to see.
- What is the key to winning GoS? The key to winning is to form a cartel. Restrict what blocks they vote on. What nodes they connect to, etc. Oligopoly competition is the path to winning. Purely defensive positions may do OK, but adversarial / collaborative cartels are the way to go. (send a note to email@example.com if you are DTC — down to collaborate :-)
- Misconfigured setups: Keep an eye on your peer connections. To low and you might miss blocks. Secure your host. Check out the Certus One knowledge base to get config best practices.
- Want to report some bad behavior during the game? Send an email to firstname.lastname@example.org.
- The prize pool: Atoms won during GoS will be part of the recommended mainnet allocation. Prizes are not delegation — they are property rights at Genesis. The goal in rewards is to incentive adversarial behavior as much as possible. The 300k Atoms will be distributed so that everyone who “credibly demonstrates that they participated” gets some. But the Top 10 performers might share 50% of the overall rewards. How much stake you have at the end of the game should drive how many Atoms you receive in the allocation. GoS performance will likely be a factor in ICF delegation decisions but most of those decisions will determined in 2019.