I went to query these same blocks on StarkWare database. The number of blocks are a bit more that probably is due to the database update and the fact that some of these blocks are not yet confirmed on L1. But since we are working with average numbers this shouldn't cause any problem.

Blocks contain a variable number of transactions of different type inside them and as you see there were 334k transactions inside 2903 blocks which makes it a 115 average transaction per block. This relates to the complexity of computation and also storage needed for transaction.

The fee mechanism of StarkWare is complex but as the Documents points:

>StarkWare’s sequencer will only charge the fee required to cover the proof cost (potentially less than the max fee). Presently, the sequencer only takes into account L1 costs involving proof submission. link

So, as not technical as I am, it seems that the proof submission on L1 is the biggest factor in determining the fees paid in each transaction (the more capacity your transaction occupies in a block, the more share of L1 fee is charged for it). On this basis, it makes sense that StarkWare users pay way less fees than direct ethereum users.

I am going to test this by doing a simple query and math. I showed you that there are 115 transactions on average in each StarkWare block sent to L1. So, it makes sense to check latest 115 transaction on Ethereum to see how much fee was paid for them.

As table below shows, this amount is 7.26 for last 115 transactions on Ethereum while it was only 0.0173 ETH for each block on starkWare.

Conclusion

• Fees on StarkWare are much cheaper than Ethereum L1.
• Things you can do on StarkWare is limited and even the amount of ETH that runs around the chain is nothing compare to ETH. So it needs more time and test to be able to be judged fairly.