Following a significant rise in Zilliqa’s value, we recently had another look at this fairly new project.
Zilliqa’s team has kindly taken the time to clarify many aspects. We publish below in full without any modification their response:
Claim: “They’re using a DAG in the proof of work algorithm they say, so a node shouldn’t require much work at all.”
- DAG refers to Dagger Hashimoto, as in Ethash, which you can read more about here: https://eth.wiki/concepts/ethash/dag. This does not refer to Directed Acyclic Graphs.
- To join as a shard node on the Zilliqa network, you require 8.53 GH/s of hashpower (or ~470 GPUs).
- Proof-of-Work on the Zilliqa platform is only used for the purposes of security, in order to validate nodes looking to participate in the network for Sybil resistance; Proof-of-Work is not used for consensus. Zilliqa uses practical Byzantine Fault Tolerance (pBFT), which is explained here.
Claim: This was an ERC20 token. The transition to a full coin was made 106 days ago. Thus it looks like about 7 billion zil has been ‘printed’ in just 100 days. How? Well although this project is ‘sold’ as proof of work, it clearly isn’t. Not the sort of Proof of Work we know from bitcoin and ethereum anyway, with their ‘blockchain’ looking a lot more different:
- Token vesting release schedule from 2017 to 2020 was well documented by Messari and Binance Research articles.
- During our Token Generation Event, Zilliqa had a total of 12.6 billion ZIL tokens minted. The remaining 8.4 billion are mining emissions that began from Jan 2019 onwards when the mainnet launched.
- Zilliqa does not use Proof-of-Work for anything other than security as explained above.
Claim: You can see above the very different block rewards with the highest we saw being six zil. Then of course there’s the nearly 200k zil given per epoch. We’re told currently there are 16 epochs, meaning about 3 million is printed a day, or 3 billion in the past 100 days.
- Block rewards (of 197,244.577625571 ZILs) are minted every DS epoch (100 TX blocks) as documented here: https://github.com/Zilliqa/Zilliqa/wiki/Mining#reward-mechanism
- Everyday, there are ~16 DS epochs. You can track the live inflation tracker here: https://www.viewbase.com/coin/zilliqa
Claim: That raises the question of a missing 3 billion, or a potentially created one. That’s because as stated the ICO was for around 6.3 billion. Zilliqa itself currently has on multi-sig about 2.4 billion. So only about 4 billion was distributed. Around 3 billion was ‘printed’ through ‘mining’ as stated above, but total supply is 13.5 billion, so where is this 3.5 billion?
- Please refer to the respective links above to see how block rewards are divested and the inflation is tracked over time.
Claim: The first address is the funds yet to be distributed to miners. We tried to clarify how that distribution happens exactly, but without much success with that address saying there have been 859 transactions, yet it doesn’t say anything about what these transactions were. There’s no smart contract code here we can read, so just how this address is doing anything is unclear. Nor is it clear how this 200k given in the block we saw above is then given to mining nodes.
- Mining rewards are distributed via this coinbase address (0x000) like on Ethereum.
- As is the case on other blockchains, mining reward distributions are not regular transactions and are therefore not shown on the block explorer.
- The 859 transactions cited in the article are not related to mining reward distributions—these are mostly ZRC-2 tokens minting and burning transactions. An example of these tokens would be the stablecoin XSGD.
Claim: The second address was funded by Binance with it curiously corresponding to this ‘missing’ or ‘created’ 3 billion.
- This “second address” referred to here is the Binance cold wallet address.
Claim: Here the block finding seems to be fixed, by magic presumably, with it unclear who exactly is doing the validation.
- Block creation is done by the validators (miners). There are currently 2,400 miners forming consensus of new blocks.
Claim: There has been no community review of Zilliqa as far as we are aware, with the many usually quite busy blockchain experts seemingly not finding time so far to look at how all this actually works.
- There have been several research papers and community-created content that have evaluated Zilliqa, as well as an ETH Wiki entry with more information on sharding. You can find these links below:
- A Secure Sharding Protocol for Open Blockchains (October 2016; academic research published with the National University of Singapore): https://allquantor.at/blockchainbib/pdf/luu2016secure.pdf
- The Zilliqa Technical Whitepaper (August 2017): https://docs.zilliqa.com/whitepaper.pdf
- On Sharding Blockchains FAQs at Ethereum Wiki: https://eth.wiki/sharding/Sharding-FAQs
Claim: To us however on the surface it sounds like a fairly centralized system that currently has a very high inflationary rate with the security guarantees here appearing to be weak as in their own block explorer we can’t see how exactly the funds are being distributed to miners.
- Zilliqa has 2,400 nodes validating the transactions at any given time. Ethereum has 8,600. EOS has 21.
- In comparison, Zilliqa is fairly decentralised. Inflation is set at ~6.8% annually and the hashrate of the entire blockchain is 13.47 TH/s (7% of Ethereum’s), protecting roughly 1% of ETH market cap. Therefore, it is very secure.
Claim: In addition the use of a DAG makes the Proof of Work claim misleading as there’s hardly any real work, and thus hardly any real proof. Storage requirements here at scale are very considerable, hence perhaps why they’re ‘paying’ people to store stuff through ‘staking’ once phase 0 launches.
- DAG refers to Dagger Hashimoto (or Ethash) memory, not Directed Acyclic Graph.
- To understand the nature of “storage requirements” on Zilliqa, consider Moore’s Law. Storage size is dependent on unique addresses (externally owned/contract addresses), not transaction count. At this junction, with 50k unique smart contract addresses and 450k unique externally owned addresses, the state size is sitting at ~100mb. State size is not an issue even if we have 1 billion unique addresses (projected at roughly ~200GB). Your regular hardware now has 2TB worth of storage. Most problems people raise with state size is with regards to sync time, as with Geth taking a while to sync for Ethereum. In this issue, Zilliqa takes an innovative approach by capitalising on the deterministic finality that pBFT consensus provides, by doing checkpoints for sync every DS epoch. Therefore, syncing a node on Zilliqa takes merely seconds. (< 1 minute guaranteed)
Claim: You require 10 million zil ($300,000) to stake but in phase 0 even then you can’t, unless you’re Binance or KuCoin with whom this China based blockchain apparently has partnered at least for the staking.
- Zilliqa’s staking proposition requires that seed node operators must deposit a minimum of 10,000,000 $ZIL. For this, Zilliqa is looking at existing crypto exchanges, wallet and explorer providers, as well as operators from within the Zilliqa community. As these are already part of the Zilliqa ecosystem, they can leverage this ecosystem to engage with end-users.
- As per the Zilliqa FAQs on Staking, “A seed node operator that does not have the minimum number of $ZIL to stake may open up this service to other token holders who may not have the right expertise to run a seed node themselves. This is often referred to as delegated staking. In Phase 0, the way in which the custody of delegated tokens is handled is left to the seed node operators.”
- Zilliqa is not based in China. Zilliqa is a homegrown Singapore-based blockchain project.
So clearly there were some misunderstandings in regards to the workings of this somewhat complex blockchain, but in our defense if you are going to call something a DAG and don’t mean what everyone thinks it means, then it would be very useful to give the full name first rather than just the acronym.
In regards to the distribution we though Zilliqa kept 3 billion something, but it looks like they kept around 6.3 billion or 30%.
The team (5%) and ‘token treasury’ (10%) kept 17%, 3.5 billion. Then 13% at 2.7 billion was given to advisors (3%) and 10% was given in a ‘strategic sale.’
Around 3 billion has then been mined, but we thought that would have started once the ERC20 tokens were frozen 100 days ago. Instead above we’re told mining began in January last year so we made a calculation error in thinking it was 3 billion in the last 100 days when it’s 300 million with their cited stats page saying it’s 830 million a year.
That page also says there are 1.6 billion on exchanges, just 400 million in Binance, yet in their response above they said that 3.5 billion address is Binance’s cold wallet.
They’re in a different time zone so it’s usually difficult to find an appropriate time for full clarification, but in a proper blockchain like ethereum you can go to the addresses and see for yourself how much was minted in 2015, who it went to, and all the other relevant details.
That does start off at the 0x0000 address, with everything then shown and analyzable, giving the precise details of all relevant specifics.
Here however we can’t quite see how the coins are distributed to miners or nodes, with the latest address that got 197,000 zil, for example, saying there are no transactions.
So just how these funds are going to nodes or miners is not clear as this address alone is not meant to get all the funds. They are distributed according to some formula, which means we should be seeing how this 197,000 is moving and to where.
In something like bitcoin or ethereum we can of course see everything, especially where it concerns coin distribution. Here however we can see much else, but not the coin distribution.
Which makes this a different blockchain from what we’re used to, and by community review we of course meant independent blockchain experts who have some recognition and trust in the crypto space and who can read the code to see and show just what exactly it does, instead of what is claimed it does.
That’s not something for us to engage in depth, but the different consensus used here makes its sharding less relevant to something like bitcoin with Byzantine Fault Tolerant systems able to do it arguably far more efficiently than was done here without needing mining which from their description sounds irrelevant especially if you’re going to have stakers.