Why an End of Year Summary?
The early days of Chainlink were shrouded in mystery, with very little communication from the team, so the end of year updates from Sergey were a focus for the community. In Dec 2017, Sergey held an impromptu Q&A in the official Slack channel to a limited audience. For those that haven’t seen it, I captured some highlights which are still an insightful window into the roots and ethos of Chainlink.
Also at the end of 2017, we saw the much hyped blog post entitled Chainlink, an Overview and Our Focus. At the end of last year, I also posted some end of year statistics and charts for Chainlink on Twitter, and I wanted to do the same again, especially since it doesn’t appear a formal end of year post will be coming from the Chainlink team.
Posting on Twitter provides reach, but the format makes it difficult to go into depth. Hopefully this longer form allows for more context behind the statistics, and insight into the current state of the network for a wider audience.
If you’re new to Chainlink, community member David Miller assembled a draft Chainlink Fact Sheet that provides a great overview of the reasons why Chainlink is a unique project. The official blog is another fantastic resource to get a base understanding.
Much has changed in the network in the last year with Chainlink’s May 30th mainnet launch on Ethereum. Measuring Chainlink on-chain activity has become increasingly interesting towards the end of the year as the network utility increases. When looking back at network usage in 2019, it’s most useful to divide the calendar year into three periods:
Pre-mainnet to May 30 all on-chain token activity was user generated, meaning it was to/from exchanges and users moving LINK between addresses. There was also some Kyber and other trade/swap related activity, though it only accounted for a relatively small portion.
Mainnet to December 6 saw the introduction of the first reference price feed. As mainnet launched, a handful of pre-approved Chainlink nodes began the job of fetching and reporting the price of ETH-USD from various data sources every five minutes. During this period, additional nodes passed security audits and were added to the contract, increasing the decentralization of the aggregated price detail that is recorded on chain (and available to use for free). With flawless execution, this stage of the network etched on-chain proof as an example to the entire blockchain ecosystem that Chainlink was the real deal, and the network had arrived and could be trusted.
December 7 until EOY was the period when Chainlink added additional reference smart contracts. This meant that nodes on the network were fetching and reporting price details for more trading pairs. After initial testing periods, this was the beginning of visible usage on mainnet for DeFi projects like Synthetix, Ampleforth, Loopring, and AAVE who each had unique needs beyond the initial ETH-USD price data. Although AAVE haven't formally announced their Chainlink integration at the time of writing, they are widely believed in research circles to be the next project to go live.
Over the course of 2019, there were many questions as to what early network usage would look like. It’s now clear that Chainlink will initially focus on planting their flag as a cornerstone for DeFi, providing reliable price data to power the exciting wave of unstoppable finance.
This year saw the network grow past user generated token movements, and into incentivized utility.
This chart shows the cumulative growth of job runs (requests to fetch price details) grouped by the increasing amount of Chainlink nodes being added to the network. The chart is sorted so that the oldest nodes are at the bottom. By looking at a colour band near the bottom like LinkPool or Fiews, you can easily see the total run count increase over time for an individual node. When stacked in the chart, the trend is magnified as new nodes enter to the network.
To incentivize early network growth, the Chainlink team is paying participating nodes a fixed amount of LINK tokens each time it fetches and reports price data. With that in mind, one way to look at this chart is that it shows the increasing cost paid by the team to build the early network. Another way to think about it, is the chart shows the increasing dilution of the tokens into the network as the team spends their reserves, effectively increasing the circulating supply on an ongoing basis. To be clear, I believe this is a perfect way to bootstrap the network.
The chart above shows the grouping of the same network activity by daily total, instead of cumulative over time. In this version of the chart, you can more easily see when each node was added to the network. The indicator calls out Dec 7th, when nodes began fetching and reporting additional reference pairs, changing the predictable pattern. The total number of oracle job runs on each day is different now, due to a combination of the volatility of the underlying asset, and the amount of testing being performed on mainnet.
In December, it became clear that DeFi projects began to use custom Chainlink solutions to provide decentralized price data. The chart below shows when the various related contracts went online, and which projects I believe each contract is associated with.
The chart below shows the percentage of transactions on the network that are attributed to nodes, which has increased steadily since mainnet was launched. At the end of the year, 85% of all on-chain transactions were related to network usage/utility - simply a massive number. With each new Chainlink integration, this percentage is bound to increase, and is quickly becoming an example of what a true utility network should look like.
The yellow shaded area above highlights a dip related to the frenzy of on-chain related activity when trading was announced on Coinbase, and Oracle and Google announced they had their eyes on Chainlink. These pieces of news in quick succession created a massive trading spike which increased the amount of user generated transactions that were related to trading activity.
DeFi projects are able to pay for the level of decentralization they need by selecting how many nodes are included in each job request. This means that each active node doesn’t partake in each contract. Instead, the teams decide which nodes are included, and the end result is seen in the charts above. The charts show job counts and LINK rewards paid to each node operator on the last day of the year, as well as the total for the entire year. The light blue is the ETHUSD price, which was running for the majority of the year, so it makes up most of the total rewards. Looking at the last day of the year is a much better representation of the current state of the network.
The team saw overwhelming interest from teams and individuals wanting to operate Chainlink nodes. The formal review process to operate a node on the early reference contracts quickly became backlogged. The team initially set an initial target of 21 nodes, and ended the year with 24 unique node operators serving up decentralized price feeds to various DeFi projects. On average, a new node was added every 9 days.
The nodes selected to participate in the initial network fall into a few buckets:
Professional PoS validators providing staking services for multiple networks
Service providers from the Chainlink ecosystem
Data providers who also supply price feeds from their core product offerings
Much of the community was hoping to be operators in the early network of oracles, but until the protocol supports staking collateral to ensure honest behaviour, the team needs to select operators who have business reputation at risk. DeFi projects with hundreds of millions at risk must have complete confidence in the network, and Chainlink's measured approach to early node selection is a welcome approach.
The visualizations above and below illustrate the relationships and level of decentralization between the reference data pairs in the inner circle, and the nodes in the outer circle which are fetching the requested price data. The video below shows relationship highlighting in the native Gephi application. A full resolution pdf and png copy of the network chart are available.
There are nearly 100 additional nodes online, waiting to be part of the network. Technically these nodes are operational now, although they are not actively participating. This should change over the next few quarters as penalty and deposit (staking) contracts are expected to be incorporated into the protocol alongside service agreements. I anticipate that once staking is active in the network, there will be far fewer jobs than there are oracles, so it's important to keep expectations in check for the long awaited, and much hyped milestone. Staking with Chainlink isn't like flipping a switch as it is on other networks, where all participants can instantly draw passive rewards. Chainlink is different, and that's part of the allure.
There are many costs associated with operating a decentralized oracle network. As a node operator, costs include infrastructure, development and monitoring costs, as well as subscribing to premium data providers to fetch data via API. Node operators also need to pay for gas costs each time they report pricing data on-chain. The gas costs are expected to reduce significantly once Threshold Signatures are introduced in the network in 2020.
In December alone, Chainlink has spent over 495 ETH on gas, valued around $70K USD. Some see this as a badge of honour, but I imagine the team is looking forward to reducing gas costs as soon as they can.
Both the #2 and #6 contracts belong to Chainlink.
From the team’s reserves, Chainlink currently subsidizes the costs for node operators and for the initial DeFi users in the network.
Node operators are paid a premium in LINK tokens as a reward for participating in the early network. LinkPool made their calculator public, which breaks down the node rewards and gas costs for the initial ETH-USD reference pair (which has since changed in frequency).
Rewards paid to node operators vary by the needs of the DeFi project requiring the data. The frequency with which the price data is reported is currently based on two factors - a predetermined frequency or ‘heartbeat’ such as every hour or two, and/or whenever the underlying asset experiences price volatility equal to a certain percentage, like 1%.
The amount of subsidy provided by the team for each DeFi project using Chainlink is unknown. I suspect the subsidy will reduce fairly quickly as the team demonstrates their critical utility on mainnet. As more DeFi projects use Chainlink, it will be interesting to see if projects who are paying for price data put measures in place to obfuscate or encrypt the on-chain results to prevent freeloading.
I’m curious to see how the team continues to incentivize the network as it grows, experiences gas cost reduction, and staking is introduced. The economics for all network participants will be an ever changing dynamic, and it’s going to be fascinating to watch the strategies the team employs to drive growth.
Smart Contract Growth
Measuring the amount of unique smart contracts over time that have received LINK tokens is my favourite leading indicator of network growth. Naturally, the more smart contracts that developers are creating that use the LINK token, the more utility that is being created in the network. This chart shows the trend well - the network is going vertical in the most important way.
On or around the 15th of each month, I post a breakdown of the on-chain addresses that are holding LINK. The breakdown groups addresses using the community generated Marine Corps rankings based on how many LINK tokens are currently held by each address. These ranks started as a meme to encourage accumulation, but I've found them to be very helpful during data analysis. The groupings are as follows:
The snapshot shows the month over month growth and allocation of the network participants. Many hold their LINK on exchanges or split their holding across multiple addresses, but all things being equal, this metric is a decent way to regularly measure project awareness or perhaps the sentiment of the wider crypto community towards the perceived success of the project.
Taking the values from each of the monthly updates created the following chart to show the growth of each address grouping over the last year. I include addresses that hold at least 1 LINK (not dust).
There are a few standout stats from this chart. The chart is sorted so that the lowest level Ranks are at the top. Total active addresses increased 164% year over year, up from a 56% increase the previous year.
The chart above shows the same breakdown of address counts grouped by Rank, but as a 100% stacked area chart. This illustrates the relative percentage change of accounts across each Rank. In 2019, 96% of all address growth came from addresses holding less than 5000 LINK. This trend became more pronounced later in the year, as was to be expected when the LINK token price increase made it difficult for new speculators to easily build large positions. Active addresses that received LINK for the first time in 2019, hold an average of 1987 LINK, excluding exchange addresses.
At the end of 2018, there were 22726 addresses that held at least 1 LINK, and 16585 (73%) of them still held LINK at the end of 2019. Of the 16585 addresses that held LINK at the start and end of the year and still do, 10% increased, and 13% decreased their LINK position, with 77% of addresses having no change. To me this low churn rate is a good indication that LINK token holders are educated, and patiently waiting for the next stage of the network.
This year saw a huge increase in the amount of exchanges that listed the LINK token for trade. At the start of the year there were 7 unique LINK trading pairs across 16 exchanges. This increased to 19 unique trading pairs, from 65 exchanges. Most new exchanges didn’t make a dent in the dominance held by Binance, but there was one exception.
The biggest shift of the year came when Coinbase listed LINK at the end of June (highlighted yellow). Binance saw an immediate drain from their balance when Coinbase opened for deposits. Frustratingly, Coinbase doesn’t publicly identify their hot/cold addresses, or deposit addresses, making the tracking of their balance difficult. Without including Coinbase, on the surface it appears as though holders are taking their LINK off exchanges, with the total balance seen going down 29% in the chart below, dropping from 125M to 89M.
Since Coinbase opened, there has been curious activity on-chain that has been noticed by the community. There have been a number of addresses that systematically accumulated 3 or 4 million LINK. The cadence of the accumulation is an indication that these addresses could be related to Coinbase.
The third image above inserts the suspected Coinbase addresses into the running balance held on exchanges. If these are in fact related to Coinbase, it would indicate that the total balance of LINK tokens held on exchanges has increased year over year, from 125M to 148M. With Coinbase’s institutional appeal and insurance offerings, I suspect this is a likely scenario.
There are many different participants in the Chainlink network, and there was notable growth across all roles.
The team is rapidly expanding marketing efforts after a major tour throughout Asia. They are now launching a new initiative with a large group of global advocates to organize and educate through local meetups. By my count, there are already advocates spanning 15 countries and 12 states in the US. This is a neat initiative, and should provide Chainlink a continued buzz in crypto-calendars, like the recent event held at the Google offices in Brussels.
Chainlink saw a massive amount of announcements for users, partners, and ecosystem members, totaling over 100 now. TheLinkMarine tracks the ecosystem visually with this great graphic.
The ecosystem is fortunate to have third party service providers building products exclusively for the Chainlink network. LinkPool, Fiews, and CLCG all had successful years carving out niches in the space.
LinkPool launched a Market for finding nodes, adapters, and data sources, and are preparing infrastructure-related services in anticipation of staking in the network. Nearly 100 nodes are registered in the Market already. After holding an initial crowdsale in 2018, LinkPool distributes LINK rewards to holders of their LP token. The amount of LP holders increased 96%, ending the year with 481 total holders who have received a total of 22,349 LINK. The 186 original crowdsale contributors have already had an ETH ROI of 12% from the passive rewards.
Running an enterprise level Ethereum node is difficult, expensive, and needs constant maintenance when used in tandem with Chainlink nodes. Fiews offers subscriptions to Ethereum nodes that are tailor made for operators running Chainlink nodes. At the peak this year, Fiews services processed 12 million JSON-RPC messages in a single day - more than 138 every second. Fiews also saw fantastic growth on the user side, peaking with 90 concurrent connections earlier this year.
The CLC Group launched the honeycomb.market, their first of many planned products. Honeycomb focuses on an essential component of the network - bringing data sources to the network in an easy way. They offer 282 API endpoints listed from 21 different data providers, all of which are preconfigured and made available pay-per-call from multiple nodes. The team held a great hackathon event for Smart Contract developers to build Chainlink solutions using the Honeycomb Market, and showcase the results in their Dapp Store.
Although Secure Data Links didn’t release a product in 2019, it’s going to be interesting to see what niche they carve as a Chainlink focused service provider in 2020. The Fourswords team is also starting to make some noise too, and will be one to watch this year.
An outsider measuring the performance of an elite software development team is an impossible task, so I’ll keep this section short. The team is believed to have multiple private Github repos and a private Pivotal tracker, so it’s incredibly unfair to measure development based on public information. What we know for sure is that CryptoMiso ranks the Github commit history of Chainlink team as sixth best out of all crypto projects. This builds on the previous year, where even with a smaller team, Chainlink was still consistently ranked among the top teams measuring in GitHub activity.
Looking at the Pivotal Tracker, and extracting the points assigned to features owned by each team member results in the following chart. It’s important to note that chores and bugs aren’t assigned points, so this isn’t a great representation of effort across the team, though it does show activity related to features.
Although seemingly still not on the radar of the wider crypto community, Chainlink was one of the highest performing assets in an overwhelmingly bearish year.
One of the most popular charts I posted this year showed the USD ROI year over year for Chainlink against other assets. The updated 2019 chart above limits the assets to those with at least 100M market cap, and at least a year old according to Messari. This cherry picks the filters a bit, with assets like SNX not 'qualifying' because of a token rename but had 3300% USD ROI, and also SEELE and MOF with even higher ROI but have a market cap just under 100M.
The LINK token appreciated significantly relative to other crypto assets in 2019. The token rose against ETH and USD by multiples, and is up over 200% against Bitcoin which had a good calendar year. With the LINK token moving in the opposite direction of the majority of the market, it's no surprise that Binance identified LINK as the asset least correlated to Bitcoin in a recent Report.
The token is near all time highs on the ETH pair, and seemingly in decent position on the BTC and USD charts. After out performing in 2019, many eyes will be on Chainlink in 2020 to see if it can continue the aggressive trajectory, or go through another quiet period of development while other assets outperform.
Chainlink saw growth across all metrics of significance in 2019. By any measure, the team had incredible success going into the end of the year - pushing marketing efforts, shipping code, building an ecosystem, and seeing real world adoption. The upcoming year will see many new features and users added to the network, and I look forward to watching the team continue to execute.
Monitoring on-chain activity has been relatively easy until now, but everything is about to get a lot more complicated. When I first started downloading and making sense of LINK transactions, the csv from BigQuery was 42MB, but that same file today is 467MB, increasing by ~100MB in the last month. Not to say that these are _big_ files, but monitoring network utility in a rapidly maturing network is ultimately a task best left for professional data scientists, ideally those tied to third party Chainlink reputation services. By the looks of things, the network is about to get incredibly exciting to analyze.
As a Chainlink supporter, it's hard to not be excited about 2020. The team is firing on all cylinders, and it's become increasingly obvious that they are planning many quarters ahead - seemingly always timing their moves to dovetail with the development of the wider market. As promising as 2020 appears, I'll be striving to mirror Chainlink's mindset of thinking beyond twelve months. Reaching the next milestone is exciting, but positioning yourself ideally for 2021 and beyond will be the ultimate reward.