Ethereum in 2019 — Through the Lens of a Block Explorer
The following blog post is an analysis of 2019 data with the goal of sharing actionable insights for Ethereum users. Sources of the data are the block explorer, website analytics, and lists compiled by Etherscan. Further analyses or counter-statements are welcome!
At Devcon V, we shared a view of Ethereum from the block explorer. With 2019 over it is a good time to update those numbers. To make sure the post is accessible to most, we’ve made an attempt at keeping it as simple as possible.
Summary:
- A curious spike in ETH transferred on 16 May coincided with a price increase that lasted until June, otherwise price was within the USD 100–200 range
- 2019 was more DeFi and less the “Year of the DAO” (or DEX)
- Addresses grew at a stable rate after the parabolic growth of early 2018
- USDT took over ERC20 volume after moving from Omni protocol to Ethereum
- Solc 0.5 outstripped Solc 0.4 as preferred compiler version of verified contracts
- Block gas limit spiked twice, in February (Constantinople fork) and September (Fairwin and USDT transactions)
ETH Price
As many will know, ETH bottomed in February and peaked in June before ending the year slightly below where it began, at USD 129.
An interesting view appears when you compare ETH price with the total ETH transferred. As would be expected, spikes in ETH transferred correlated with up/down swings in ETH price. One particular spike on 16 May led to a curious month-long rally ending at the June peak price, even as ETH transferred had dropped back to normal levels immediately.
Adding to the curiosity, the volumes on Coinmarketcap and Coingecko did not jump by a similar extent on 16 May, nor did it jump in the period leading up to the peak price. As a reminder, these websites track volumes transacted on exchanges but not over-the-counter (OTC) trades.
Is this normal? Looking further back, we can see that the two metrics (spike in ETH transferred vs up/down swings in ETH price) have generally been correlated from the end of 2017, at the peak of ICO mania. The only other exception was in November 2018, where there was no impact seen in either price or volume of ETH on CoinMarketCap or CoinGecko.
Additionally, when comparing price with the number of transactions, we see that there was no equivalent spike as there was for ETH transferred. This points to trades made by a small number of whales (large traders).
What does this mean?
One interpretation of the above data is that a small number of large traders made OTC trades that sparked a rally in ETH price. Alternatively, it could have been large whales offloading their ETH bags as price rose up.
DeFi, DEXs & DAOs
These 3 spaces are frequently referred to as primary use cases of Ethereum. Here we take a quick look at their progress in 2019.
There exists 200+ DeFi projects across 17 categories according to DeFiprime. To track growth, Aleth.io tracks some of these projects on three metrics: number of addresses, number of contract interactions and total ETH locked. For the latter metric, DeFi Pulse is commonly referenced. It shows that the market has gone up nearly 3x in 2019, from USD 200M+ to USD 600M+.
On the Etherscan loan tracker, we started tracking Collateral Debt Positions (CDP) on Maker in Dec 2017 (a recap of Maker, DAI and CDPs here). In the 2 years since, the numbers have skyrocketed.
It’s worth noting: though the number of loans (CDP) created shot up by 27x, the amounts of ETH supplied and DAI borrowed did not go up by more than 1x. One interpretation is that 2019 attracted ‘hobbyist’ users, who have started playing with CDPs without putting too much of their crypto wealth in.
As 2020 continues, we will look to add more loan data on Etherscan and explore the data in more depth.
The DEX (decentralized exchange) space is slightly more mature than DeFi (mature being relative!). Etherscan tracks the leading DEX by number of transactions on the DEX tracker. By this measure, IDEX retained its lead in 2019, with 44% of all transactions done there. At the same time, 3 new faces showed up in the top 5: Kyber, Uniswap and Oasis Dex.
A closer look at the monthly numbers in 2019 show a more nuanced story. Since September, transactions have dropped off on IDEX, potentially a belated impact from their implementation of KYC requirements in August. In contrast, Uniswap has gradually captured more of the market share. It ended 2019 at a close second in terms of transaction counts and looks set to lead the way in 2020.
Ether Delta showed a third interesting trend. A pioneer among decentralized exchanges, it has been mired with controversy, including hacks and scam allegations. Most actual trading has been done on ForkDelta, a community fork of the website (but using the same smart contract). Its numbers dropped significantly from 2018 and, while still able to hold a top 5 spot in 2019, by December it had dropped to a tiny fraction of total transactions.
Overall, there was a 30% drop in DEX transactions — from 6+ million in 2018 to 4+ million in 2019.
2019 was meant to be the Year of the DAO. It didn’t quite turn out that way, but there has still been progress made on this front. The most eye-catching was Moloch DAO, a fund-allocating DAO with ragequit mechanism that has inspired a community-wide meditation on Moloch and summoning of clones on Daohaus. Meanwhile, Aragon now stands at 1k+ DAOs created while DAOStack brought DAOs to the real world with events and hackathons. DigixDAO, one of the earliest projects in the space, discussed a dissolution mechanism which was eventually approved in January 2020.
What does this mean?
DeFi is showing strong adoption momentum, DEX are seeing a change in the players, while DAOs faced further experiments and realities.
Addresses
The number of unique addresses on Ethereum at 31 Dec 2019 had grown to more than 84 million, up 55% from the start of the year. The rate of address growth has remained stable after the parabolic growth of early 2018.
Out of this total, 76% (64 million) are normal addresses (ie. EOA/Externally Owned Accounts) and 24% (20 million) are smart contracts.
Interestingly, while normal addresses had slowed in growth compared to 2018, smart contracts actually grew faster by 4%.
Breaking down these addresses by ETH balance, we see that 99% of addresses have a balance of less than 1 ETH. Of those with at least 1 ETH, another 1% of those hold upwards of 1,000 ETH.
Some explanations for the high number of wallets with 0 balance are:
- Many addresses created are for one-off uses
- Many users hold balances with exchange and custodian services
- Many contracts with 0 ETH may have ERC20 or other token balances
- Many contracts are created to run code or are NFTs, not to hold ETH
A further look into these may be useful. Similarly, it’ll be useful to compare balance breakdowns between normal addresses and smart contracts as well as by labels.
What does this mean?
Higher rate of growth for smart contracts point to a focus on building in the ecosystem. In contrast, slower rate of growth for normal addresses point to a lower take up by retail users — user adoption remains a big challenge. Address balances show a consolidation of wealth in a small number of users. While this is not unique to Ethereum, it is worth more thought as the community strives to create a better financial system.
Contracts
From the 20 million contracts that exist on Ethereum, 78k or 0.4% have been verified on Etherscan. Verified contracts on Etherscan provides transparency for users interacting with these smart contracts and allow them to verify details related to a contract.
The number of new verified contracts slowed down, even as contracts as a whole had better growth. Breaking down by compiler versions, 2019 saw more developers gradually adopting Solc 0.5 (Solidity version 0.5) after the almost total dominance of Solc 0.4 in earlier years. All other compiler versions (including Vyper and Solc 0.6) made up a tiny fraction.
Note that many contracts are created for one-off use or as mainnet tests. Additionally, not all developers choose to verify contracts on Etherscan (we do encourage you to).
What does this mean?
Solc 0.4 and 0.5 remain the top compiler versions compared to others such as Vyper. New developers looking to build smart contracts are likely to focus on using the former. At the same time, there is a good opportunity for those looking to fill the gap of building supporting infrastructure for the latter.
ERC20 Tokens
As many would have predicted, USDT made up most of the token transactions on Ethereum in 2019. Despite moving to Ethereum from Omni only partway through the year, it makes up a whopping 25% of all ERC20 transactions in 2019. The next token by volume, DAI (including both SAI and DAI) is at 5% — a whole 20% lower. Faster confirmation times and lower fees have been suggested as reasons for the USDT shift to Ethereum.
ERC20 transaction counts support the narrative that DeFi is a major use case of the network, as stablecoins and exchange/liquidity tokens (eg. Bancor, Wrapped ETH) make up at least 30% and 8% of transactions respectively.
Next, we look at the distribution of ERC20 tokens by the number of token holders. Out of 6,000+ tokens with updated information on Etherscan, 68% do not have any token holders. The remaining 32% have an exponential distribution skewed extremely right.
A caveat — this measure does include tokens that did airdrops to a wide number of Ethereum wallets (readers will recognize these as names that appear uninvited on your Etherscan wallet balance).
What does this mean?
DeFi (specifically stablecoins and exchanges) make up a big portion of ERC20 transfers. Based on current trends we expect this to consolidate further in 2020. Separately, many tokens whose teams made the effort to update their information on Etherscan do not end up having users. This points to struggles for adoption even within Ethereum or a lack of compelling use cases for these tokens.
Geographical analysis
Out of the 5–9k nodes that were online daily in 2019, the US leads the way, with 3x as many nodes as second-placed Germany. China rounds up the top 3, who collectively make up more than 50% of all nodes. Meanwhile, the continental trio of North America, Europe and Asia hold at least 95% of all nodes.
A different analysis of geography is the number of unique visitors on Etherscan’s website. Overall, visitor numbers dropped by 50% compared to 2018.
Looking up the percentage of visitors by country, we see the same countries in the top 3 as with nodes. In this case, China tops the charts with almost the same amount as the US and Germany combined. The top 3 here make up only 37% of total users while North America, Europe and Asia make up 89% of users.
While the majority of countries had a drop in the number of website visitors, Germany and Ukraine broke the trend. Both countries grew to twice their 2018 numbers.
One caveat is that the use of VPN services likely affect both node and website numbers. We assume here that the numbers are broadly correct.
What does this mean?
With the vast majority of nodes and users concentrated in three continents, adoption isn’t just a matter of getting more users on to Ethereum. It is also about bringing them in from other parts of the world. Looking at the overall drop in website numbers, perhaps retention of Ethereum users or conversion — from speculative to more practical use cases — deserve a focus as well.
Gas Limit
Block gas limits had two spikes in 2019. To recap, gas is a measure of how much computation is needed for a transaction to go through and block gas limit is the max amount of gas that can be included in one block (more on these here).
The first followed the Constantinople hard fork on 28 Feb which deployed EIP 1234. It delayed the Ice Age difficulty bomb by 12 months, relaxing hash difficulty. As a result, block times dropped and consequently block and uncle counts surged as more orphan blocks were created. A good explanation by ETH Gas Station on uncles and their interplay with gas limits can be found here.
The second spike in block gas limits has largely been attributed to the infamous Fairwin smart contract and increasing USDT ERC20 usage. Fairwin, an alleged Ponzi scheme, racked up 350k transactions in September alone. Combined with USDT growing to a quarter of all Ethereum transactions, September saw congestion in network traffic. Miners then increased the block gas limits to reduce the congestion.
What does this mean?
The gas limit mechanics of Ethereum worked as intended. The first spike with Constantinople (delay to Ice Age) is a pre-planned mechanic to allow time for setting up of ETH 2.0. The second spike showed miners following their incentives (ie. raise gas limits as network gets congested and uncle rates drop).
Gas Price
Finally, an update on gas price. The daily average has decreased from 21 Gwei in 2018 to 16 Gwei in 2019. Combined with a lower ETH price, a standard ETH transfer spending 21k Gwei would cost you an average of 6 cents in 2019 compared to 25 cents in 2018. In comparison, the daily average cost to send BTC in one block (approx. 10 mins) was 92 cents (source: billfodl.com).
There were 2 outliers in daily average gas price, on 19 Feb (374 gwei) and 18 March (196 gwei). Excluding these, gas price generally stayed around 15 gwei other than during the network congestion in September.
What does this mean?
ETH 1.0 continued to improve its utility as a medium of exchange with lower gas prices compared to 2018 and its main competitor, BTC. It will be interesting to monitor how this changes in the future, with more applications being built on Ethereum and as the community starts shifting to ETH 2.0.
While the headline numbers of ETH price did not end the year on a bullish note, 2019 proved to be another eventful year. Importantly, these numbers alone do not capture the open, innovative spirit embraced by the Ethereum community. On Etherscan’s part, we hope to contribute by providing products, features and content that add value to all.
We hope these insights into the state of Ethereum in 2019 have been useful! Do let us know what other numbers you’d like us to look into.
To cap off this post, here are some amazing 2019 reviews by friends in this space:
