Including the initial epochs (where difficulty was set unreasonably high for security) expands the range of values for block time in a way that it makes the chart much less useful than it could be.
Could it be considered to remove these outliers from when the network started?
https://explorer.nervos.org/charts/average-block-time