Try Quests to earn cryptoTerra Fees
This dashboard investigates the relationship between the number of transactions and the transaction fee.
The graph below shows the fee collected vs the number of transactions. Based on the graph below, the hypothesis made earlier is false, where the fees collected are seemingly unaffected by the number of transactions, which doesn't make sense. So further investigation was conducted.
The total fees generated do not correlate well to the number of transactions. However, if non-paying transactions are removed, the total fees generated strongly correlates to the number of transactions.
Further proof is shown when these 2 types of transactions are the only non-paying transactions that appeared after March 17. Combining all these facts, it can be said that the fees collected in USD before March 17 did not strongly correlate to the number of transactions before March 17 is because these 2 types of transactions are bloating the number of transactions, thus polluting the data.
So, the number of transactions by message type of KRT transactions are found out and plotted. Note that the transaction count is higher in this graph because some transactions contain more than one msg_type if multiple msgs are sent in a single transaction. From the graph below, we see that msg_type oracle/MsgAggregateExchangeRateVote and msg_type oracle/MsgAggregateExchangeRatePrevote are the types that use KRT as transaction fee the most, where its number went down significantly after March 17 while the fee generated in USD remained about the same, as shown in the Evolution of Fee Generated in USD graph. This shows that these 2 types of transactions are bloating the number of transactions before that date without paying much in transaction fees.
To find out what was bloating the number of transactions before March 17, the number of transactions by native coin used as transaction fees is found out. The graph below shows that KRT was being used the most before March 17 however it dropped off tremendously after March 17, where the first of the non-paying transactions started appearing.
So how does the hypothesis hold if the non-paying transactions are filtered out? The graph below shows that the number of transactions after March 17 almost outlines the total fee collected, showing a strong correlation between the two. However, the total fees collected still does not relate to the number of paying transactions before March 17.
Upon further investigation, it was found out that some transactions did not require a transaction fee starting March 17. The percentage of paying transactions out of the total transactions is plotted out below. From the graph, we can see that most of the time, paying transactions are less than 50% of the total transactions after March 17.
Since Terra allows transaction fee payment using any of its native coins, in order to find out the correlation between the number of transactions and the fees collected, the fees must be denominated by a single currency. This is done by summing up all the coins used in a transaction as transaction by their respective oracle prices.
Note: Some transactions use multiple coins as transaction fee.
Transactions in Terra have transaction fees just like any other blockchain, in other words, each transaction requires a fee in order to be processed. This leads to the hypothesis that the fees collected are very dependent on the number of transactions, such that the more number of transactions, the more fees collected.