A Look at the Kadena Network Launch
Community Mining as a Bootstrapping Mechanism
Kadena is a project that strives to solve PoW scalability by using several parallel chains that work together to form a unified braided chain. Over the first week of November 2019, the Kadena team decided to quietly launch the project’s mainnet to the public in limited functionality mode. Limited functionality simply meant no transactions or smart contracts could be used on Mainnet yet. The team has followed an interesting strategy to bootstrap the network and this article evaluates the pros and cons of the Kadena launch strategy.
Bootstrapping Through Mining
Kadena , aka Chainweb, rose to popularity among crypto VCs after the team closed a $12M investment round from accredited investors in April 2018. However, the project wasn’t commonly known amongst mainstream crypto users for one simple reason: they didn’t perform an ICO before launch. The team has obviously elected to run the ICO through Coinlist shortly after the mainnet launch to avoid complications with the SEC. This decision has put the team in a difficult position. PoW chains gain security from the participation of the community through mining. However, this community of individuals with vested interests didn’t exist because of the lack of the ICO.
CPU Mining Phase
The Kadena team approached the problem with an interesting solution: bootstrap the network through community mining. The team decided to start with a client that doesn’t support GPU mining. The only way one could mine on the network was via CPUs. This decision piqued the interest of many individuals who joined the project’s Discord channel to get more information about this chain that can be mined with CPU. This decision was mainly to protect individual miners from the large GPU mining farms that can price everyone out immediately once they join the network.
To make it even easier for these individuals to start mining right away, the Chainweb miner client didn’t require the miner to run a full node. The user could point their miner to one of multiple bootstrapping nodes that the team launched and start mining. Early joiners to the CPU mining party got super excited because they were able to mine KDA coins for a cost around 2–3 cents per coin. This is 98% discount from the current token ICO price.
Personally, I was able to mine 30 KDA coins overnight using an old laptop. People were rushing to rent cloud computing instances on AWS, Azure, etc, to mine KDA coins as well before the arbitrage opportunity closed.
The Bottlenecks
With hundreds of excited individuals renting computational power and throwing all the CPU hash power they could get on the Kadena network, without caring to launch a full node, a number of issues started to emerge early on:
- The public bootstrapping nodes provided by the Kadena team were overwhelmed by the mining requests from the thousands of mining nodes joining the network, preventing new users from connecting to those nodes even to sync a full node.
- With a rapidly expanding network, issues with the full-node P2P design started to emerge. The node P2P networking design was under-optimized leading to difficulties for network nodes to communicate and sync. Eventually, this led to 4 competing major PoW forks! Each of the forks has a number of nodes building on a different chain without any communication between these isolated islands.
- Competing chains meant that some individuals were working on chains that would eventually collapse and hence they were wasting resources. In fact, the peak in KDA mining rewards on Nov 7th in the above chart was due to miners mining on the “winning fork” while everyone else was wasting resources on useless forks.
- Miners who wanted to launch full nodes to support the network and mine usually couldn’t get their nodes to sync. Additionally, if they managed to sync their full nodes, they received an overwhelming amount of inbound from other users’ mining nodes preventing the node owners themselves from effectively utilizing their own full-nodes.
To the Kadena team’s credit, team members responded in a timely manner to these issues starting with fixing the P2P communication issue. The fix, along with a little bit of community coordination, eventually led to the collapse of several competing chains and the emergence of a single chain. Recently, the team also added a whitelisting option allowing a full-node operator to select which miners could mine to their node.
GPU Mining Tipping Point
A new network launch is always an interesting arbitrage opportunity. Few people know about the project and even fewer are ready to join the party from day 1.
For Kadena, the project team intentionally decided to delay pushing out the official GPU mining software to allow new users to join the network and mine using their home computers. However, with the lack of an official GPU miner, experienced GPU developers put their skills to work to come up with a GPU miner as quickly as possible. These developers could either perform secret mining on the network and have better economics than anyone else, or publically announce the miner and earn a cut from the dev fee embedded in the miner.
And that is exactly what happened. The first publically announced miner for Nvidia GPUs came with a whopping 25% dev fee. As a response to the excessively high dev fee, two community members: Edmund Noble and Alex Konovalov, released mining software based on OpenCL with a 0% dev fee less than 30 mins after the announcement of the Nividia CUDA miner! Since that moment, the GPU mining game for Kadena became wide open.
The OpenCL miner devs tried to protect the hoppiest miners by limiting their mining software to run 1 GPU type per device. However, this didn’t work as intended and, currently, there are several available GPU mining software for the Kadena network. It is fairly easy to distinguish the date when GPUs started to join the network from the difficulty graph. Soon, FPGA miners will join the party pushing the difficulty even higher.
What Has Kadena Team Achieved? What Could Have Been Done Better?
In my opinion, the Kadena team succeeded in achieving several major goals in its mainnet launch strategy.
- The project successfully built an early fan base within the mining community that is essential for the success of the project. This small excited fan base has helped the project bootstrap and scale from a hashrate of few tens MH/s in the first few days to more than 1 TH/s just 10 days after launch.
- The network launch without GPU mining or mining pools has facilitated widespread token distribution that is more effective than airdrops.
- The launch of a limited-functionality mainnet before the public token sale aligns with an important SEC guideline. As the network is already in operation, this reduces the risk of classifying the KDA token as a security.
However, there are a few things that could have been executed better — primarily around communication, documentation, and data management. For instance, the GitHub repo of the project didn’t provide updated information about the mainnet bootstrapping nodes and referenced the testnet nodes for few days after the mainnet launch. There were virtually no tutorials for new users to guide them through the process. Such information was only available through the discord channel which, to the team’s credit, was pretty active and provided immediate support. Lastly, the team could have more thoroughly stress-tested the node P2P networking protocol to avoid the network splitting, which occurred during the first week of the pre-mainnet.
Ultimately, this project was interesting to us because its launch deviated from the traditional token playbook. We wanted to illustrate the challenges that arose when the network was bootstrapped and tradeoffs that were made along the way.
Thanks to Emily Pillmore from the Kadena team for her Feeback on the article.
