I realised I messed this analysis up by getting the current chunk size wrong; in current tests I think it’s 4mb max per chunk, not 32mb.
If this is the case, then with the other assumptions the same, the fee would cost not 30% of the total cost of storing data, but 76%. ($0.003986/($0.003986 + 0.00125)).
(0.00125 is the previously assumed ~6x SSD cost per mb, but as it’s 4mb vs 32mb, it’s $0.00125 rather than $0.01 per chunk.)
This means the fee would be costing over 3x what the nodes are getting paid, or the total cost to store a 4mb chunk would be about 4x what it’d be with the Native token.
If we needed to go to 2mb chunks for stability purposes, and each chunk required the same fee, which I’m assuming to be around:
… then the situation would be way worse, with storing data on Autonomi with ERC20 costing around 7x what native should cost.
Is this thinking correct?
Even with a 4mb chunk size, if the Arbitrum fee is going to be around 3x what we could reasonably expect to be the market rate for storage on Autonomi, then it’s a big reason to ensure the Native token is ready ASAP, as operating with ERC20 only will be a massive cost disadvantage for Autonomi & will not indicate the true potential of the network to the market.
Hope I’m missing something, but I fear I’m not… I’d like to see others make some estimates of reasonable pricing estimates for nodes to store data + the expected fee cost to see how big the fee is with other people’s assumptions in play.
I’d definitely like to hear the view of this team member:
… on their expectation of fee size as a proportion of total storage cost with ERC20 only payments.