Recap: Avalanche AMA on CryptoDiffer

CryptoDiffer hosted an Avalanche “Ask Me Anything” (AMA) event on Telegram on Friday, August 7 from 10 a.m. to 11 a.m. EDT.

The AMA was held in conjunction with Sperax, a financial public blockchain. For brevity, only the questions and answers pertaining to Avalanche are included in this recap. Some questions were edited slightly for clarity.


Nick (CryptoDiffer): Hello, everyone!😁

We are glad to meet here
Frida Cai (Sperax Global Strategic Partner) @FridaCai_Sperax
Joe Yu (Sperax Senior Researcher) @yuzhou87
Stephen Buttolph (Avalanche Software Engineer) @StephenTechSupport
Collin Cusce (Avalanche Senior Software Engineer) @hellmichigan
Jay Karahashi-Sofue (Avalanche VP of Marketing) @jayksofue

Can you introduce yourself to the community?

Jay Karahashi-Sofue: Avalanche is a highly-scalable, open-source platform for launching decentralized applications. Developers who build on Avalanche can easily create powerful, reliable, and secure applications and custom blockchain networks with complex rulesets or build on existing private or public subnets. Ava Labs is the company building Avalanche.

Many blockchain projects often build on their own chain and attract users, so why should people choose Avalanche to be their platform?

Jay Karahashi-Sofue: Avalanche will provide developers/projects looking to build decentralized applications with better performance and scalability; in the early stages of Avalanche, we’re engaging with mature ecosystem partners (e.g., dev tooling, dapps, etc.) to make our ecosystem more appealing to new devs in the future. Our immediate focus is decentralized finance.

We’re also minimizing friction for devs who are already in the decentralized space with our C-Chain. Currently, the C-Chain runs the Ethereum Virtual Machine so any dev who is comfortable with Ethereum (or Solidity, more specifically) can come over to Avalanche to try out the platform with minimal barriers to entry.

Hell Michigan: We are fully decentralized with the potential to achieve millions of participants, just like Nakamoto, but at TPS not-yet-seen in open permissionless consensus networks. In Nakamoto there’s a long wait for confirmations with Bitcoin taking up to an hour to receive enough confirmations to feel confident your transaction has made it to the cannonical chain. With Avalanche it’s immediate, requiring no confirmations. In fact, transaction average a finalization time in under 1 second. We had 4500+ TPS and in our earlier, lighter-weight implementations, we reached nearly 8k TPS without any hokey tricks to boost those numbers. Avalanche consensus is fully-decentralized consensus which runs at twice the transaction processing speed that Visa does on an average day. In one day, we can do 2x the transaction Bitcoin does in a year.

And we’re backed with the scientific rigor to prove it. Heck the fastest Classical consensus protocol, Hotstuff, was what Facebook’s Libra was going to use… yea that was authored by Ted Yin, co-founder of Ava Labs.

You can create your own VMs, so you’re not bound to just the EVM or just Move or just Pact. You’re able to create entirely new VMs of your own design, tailored to your needs. The consensus is completely divorced from the state machine!

We’re in an entirely other category and operate at an entirely new level. We’re not simply 10x over the previous work, we’re 100x.

That’s why people will choose Avalanche.

How does Avalanche Consensus work and what are its advantages compared to Classical and Nakamoto Consensus?

Stephen Buttolph: Collin Recently posted a nice article about this on medium (

However, the main idea is to utilize sampling to get statistical approximations of the preferences of the network. By performing repeated network polls, the nodes can adjust their preferences to be in line with the rest of the network.

This process is extremely powerful, because after just a handful of rounds, the network will have converged to a unanimous preference. Because Avalanche also has a way of tracking when this unanimous preference is established (through the same polling mechanism), Avalanche can be used to achieve consensus with an extremely small number of messages.

Communication complexity

Hell Michigan, in response to a Sperax answer: Avalanche consensus is O(1), fyi

Hell Michigan: here’s a high-level explanation of Avalanche for those interested:

In some leader-ful protocols such as Tendermint or HyperLedger Fabric, there is single choke-point through which all transactions have to flow. But Avalanche doesn’t have this, so could you explain to me why Avalanche protocol is absolutely leaderless?

Stephen Buttolph: This raises an interesting point. In a non-sharded consensus system, all nodes will need to receive all transactions. However, the process that must be performed to finalize the transactions is a little more complex. Protocols like Tendermint and Hotstuff introduce a choke point for every block. So, the leader will need to receive a message from every other node in the network for every block.

In Avalanche, there are no special roles between nodes. Every node only performs a small sample of the network. It is possible to introduce block producers into Avalanche, and still maintain this property. But right now there are no special roles in the Avalanche network. That’s why Avalanche is leaderless.

Consensus efficiency

Hell Michigan, in response to a Sperax answer: @FridaCai_Sperax “the current consensus that we are using is the most efficient among BFT-based consensus.” — you’re not using Avalanche so I disagree, but if you mean classical consensus, that’s HotStuff and the first author on that paper is our Co-founder Ted Yin.

While testing digital currencies in sandbox environments is extremely insufficient, what do you attribute the incredible increase of the DeFi market?

Jay Karahashi-Sofue: I think both approaches are reasonable. A sandbox environment can ensure steady success without getting in too much of a clash with regulators, while an open one without a sandbox can lead to incredible growth but also could lead to cases (in regards to DeFi) where retail gets burned due to lack of knowledge of the technology

Asynchronous network models

Hell Michigan, in response to a Sperax answer: Casper is partially asynchronous. Partially asynchronous is the most practical and the most analyzed model in academia. HotStuff? Tendermint? PBFT? All partially asynchronous.

You recently announced an integration with Chainlink. What benefits does this integration offer developers?

Jay Karahashi-Sofue: This integration allows for developers building on Avalanche to leverage the benefits of oracles; if you don’t know what oracles are, they enable real-time, outside data to interact with blockchain. Avalanche is continuously trying to grow its ecosystem to provide devs more tools and apps to use within their own projects.

Will Avalanche help Sperax in its exploration to DeFi?

Jay Karahashi-Sofue: We’re always open to collaborating with projects within the decentralized ecosystem and outside as well; in addition to building out our ecosystem with leading decentralized finance projects, we’re also engaging companies and institutions from the traditional finance and enterprise world.

Could you explain the efficiency of the Avalanche consensus protocol? Classical consensus requires every node to talk every other node. Nakamoto consensus requires running giant lotteries. How about Avalanche?

Jay Karahashi-Sofue: Collin (@hellmichigan) wrote a really great post that covers exactly this:

Many projects lie or do some really dumb tricks to report high TPS. What is your target for TPS that Avalanche expects?

Stephen Buttolph: On the Avalanche platform we’ve been focusing on implementing additional features lately. So I don’t really have a recent benchmark number to give you. Once things have stabilized we’ll focus back on TPS benchmarks. Our last benchmark on a large deployment (~1k nodes) was around 4500 TPS using single input two output transactions.

Avalanche is the 3rd generation of the main protocol. It will improve sideways by removing the missing aspects Classical and Nakamoto Consensus protocol, so do cryptocurrency ecosystem and Avalanche blockchain really change radically?

Hell Michigan: Ya I think it drastically will. Things like a decentralized Venmo were relegated to risky and often-broken Layer 2 solutions. No longer the case thanks to Avalanche and this can occur on Layer 1. There’s also the ad hoc creation of assets that Avalanche offers on a base transaction level which should be a huge improvement over ERC 20s. In addition, you can deal with these assets in any way you please. Thanks to custom VMs, you’re no longer stuck to one [way] of doing things like with Ethereum. You can use any VMs you want, and they all can pull from the same validator set that is used in the default network. The ability to float assets between chains and subnets is also going to change the way the world treats crypto.

Where can i buy Avalanche?

Jay Karahashi-Sofue: Avalanche is not available yet; once mainnet is launched (some time this month TBD), we will look to have AVAX available on exchanges. Until then, please follow our social channels for those updates.

Thank you for joining us today!

Nicolas Lemaitre: [Dabbing slide GIF]

Add Comments