As the news of Ethereum merge is heating up the cryptoverse, the interest in both L1 and L2 is exploding. The intention of this article is to take a look at the ETH transfers on Optimism from a network analysis angle.
For this analysis we are going to query data from FlipSideCrypto using their ShroomDK. We’ll be only interested in ETH transfers on Optimism. On Flipside data tables we can find data from 15th of June and we’ll be looking at all the ETH transfers happened on Optimism until 22nd of July. We will be treating ETH transfers as a directed graph as transferring ETH from one account to another indicates a direction. A node will be represented by a wallet/contract address while an edge is represented by a transfer between these nodes, as this is a directed graph ETH that comes into a particular address will be considered as an inward edge and a transfer going out of an address will be considered as an outward edge. Since the main intention of this graph is to find out which wallets are connected we don’t weight the connections based on the number of times they send ETH to each other. Please refer the Appendix for tools, methodology and SQL queries used for this.
There are 218186 nodes in other words, wallets/contracts transferred within each other during that time period making 489539 connections (These connections are not number of transfers, if A send B, 100 times and also C send D 1 time both considered as one connect, a connection between A and B in the next scenario a connection between C and D). Now, this doesn’t look a lot of connections for the given amount of nodes, which means we have a density of 0.00001028 (where 1 means all the nodes are connected to all the other nodes while 0 means no connections between any nodes) and a transitivity of 0.000055 (This measures the tendency of nodes to cluster together, the higher the value the denser the networks communities internally) Since we see that transitivity is higher than density we could speculate that when they connect they tend to make smaller, denser communities?.
So, first what is a degree? Well, not the one you are offered by a university, but in our context a degree means how many connections a node has.
Unsurprisingly we can see Hop protocol has the most number of connections, which means it’s the wallet which most number of wallets have send ETH to (Since 15th of June of course) which makes perfect sense as it’s a bridge that help people to send ETH from mainnet or other L2 to Optimism. Out of the labelled wallets we can see most of these top 20 highly connected wallets are dexes or dapps which provide services to the end users.
Now these connections have two directions so let’s look at how many connections these top wallets have in terms of inward and outward directions.
Now we can clearly see that Hop protocol has more number of users who has sent ETH than received ETH from Hop protocol. Interestingly Socket.tech registry has only 9 users that’s connected by outward transactions similar pattern can be seen on Li Fi and couple of unlabelled wallets.
We know that this network is not densely connected this shows that 99% of the wallets has 12 or less connections, in other words their are many ‘lonely’ users who have no interactions in terms of ETH transfers with other users, we can see 28.1% has just 1 degree (Has connected with only one other node in the network) those are the ones probably bridged ETH from a bridge like Hop and just HODL?
Also this shows that the ETH transfers on Optimism is highly active among a small number of users which includes but not limited to dapps, and dexes.
Let’s first understand centrality in simple terms. This is like a rank that a node gets based on the location they are positioned. Degree centrality is based on the number of other nodes it has connected to so, higher degree centrality means it has higher number of connection. However, influence may not just mean having lots of connections but having lots of connections which are also important and influential this is something Eigenvector centrality calculates. (A person who knows the top politicians in a country would be more influential than someone who has just high school friends).
Now in our graph we see most of the highly connected are also highly influential too. However if we take Li Fi or Socket Registry they seems to have very high quality connections as the Eigenvector centrality is quite higher than their degree centrality.
Alright, let’s look at some the communities inside this bigger network…
Here the names represents the most influential nodes in that particular community
These were just only 5 rather larger and interesting communities out of 1000s, based on Modularity, modularity looks for the nodes that are more densely connected together than to the rest of the network.
Interestingly, we can see some well known labels in all the communities except the green community which also has other interesting structure. Let’s quantify that a bit.
Note: The size of the marker represents the transitivity
We looked at the density and transitivity of the whole network, the communities are we saw in the pictures are densely connected than that of the whole network while Red Community is mostly densely populated among them. As we saw, the Green Community has their own internal sub communities which is evident with the higher transitivity. Blue community seems to be the Hop protocol’s community and undoubtedly it doesn’t seem to have many other internal communities, the transitivity is even lower than that of the entire network but denser than the main network. Which could mean the users who made transfers with Hop may be a less active crows in terms of connecting with others (Introverts :D). While the green group is actively making connections within them (extroverts :D)
The Green community seems to be a very interesting community, yes it has a high influential ‘figure’ but compared to other communities and also the entire network as a whole the influence is lower, which suggest that compared to the other communities it’s more resilient. Blue community has a very highly influential wallet, Hop protocol, and the pattern is similar across other 4 communities.
The ETH transfers on Optimism is an interesting network with lot of beautiful pictures :). However, the network seems to have a very high reliance on a handful of wallets. For example, Hop protocol, Uniswap Router etc. This introduces risks, as a failure or an attack on these points may send some huge ripples across the network. On the other hand we can see some interesting smaller communities being formed inside, some are growing around known entities like 1Inch protocol, Orbiter etc. The interesting community was the Green Community, which is rather resilient. It will be interesting to know who are they and what are they doing :)
As a fun thing, you can try out this interactive network which has the top 1000 most important nodes here in this link.
ShroomDK was used to query the data from FlipSide database.
Networkx on Python was used for this analysis.
SELECT time, ifnull(f.address_name, eth_from_address) as eth_from_address, ifnull(t.address_name, eth_to_address) as eth_to_address, amount from ( select min(block_timestamp) as time, eth_from_address, eth_to_address, sum(AMOUNT) as amount from optimism.core.ez_eth_transfers group by eth_from_address, eth_to_address order by time) left join optimism.core.dim_labels f on f.address=eth_from_address left join optimism.core.dim_labels t on t.address=eth_to_address
Searched on Github for few of the labels on top wallets which were missing in FlipSide tables.
AZTRO369 in Red community is actually a user on OpenSea :)