Varun Srinivasan
@v
A longer thread with problem statement and reason. Eth charges fees to protect against ddos and lots of similar precedent in tradfi. So eth never "goes down" or is "too hard to sync" in practice. Fees are more palatable because there is expectation of profit per tx. For social media, there is (usually) no expectation or profit or precedent for a paywalled app that succeeded. It seem inevitable that we need at least some upsell for users to get in and contribute before we can charge them. But we can't also let the network go down or get congested, because then no one can build apps.
3 replies
10 recasts
80 reactions
Varun Srinivasan
@v
The problem of "bots will" overrun the network is mostly solved. As long as there is some reasonable limit (captcha, sub-cent fee) this is preventable. The main goal is to reduce costs to the point where apps can underwrite users. We have a few paths to doing this. 1. increase network capacity (snapchain) 2. reduce storage unit sizes (and by extension demand on network) 3. change the network itself (use da layer or central server) (1) is something we are doing and we can charge less per unit (2) is what we are trying to push on here to get costs to a point where apps can sponsor users. (3) will be the most scalable, but unfavorably advantages warpcast over other builders. also makes it hard to build apps since the "true" state of the network cannot be fetched easily.
2 replies
4 recasts
36 reactions
Varun Srinivasan
@v
An interesting left field idea is app-based storage proposed by @vrypan.eth and @deodad a while ago. In this model storage is a good that you can purchase and divide up amongst arbitrary other accounts. So Warpcast can pre-pay for 10,000 units of storage and use some heuristics (captchas, phone number etc) to give it to some users for free. The main benefits are: 1. utilization is much more efficient, a lot of users who pay for storage leave around unused bytes, which apps can give those to new users. 2. costs for storage are lower and apps can arbitrarily decide what storage users get, so you have more freedom in how you want to onboard users.
2 replies
1 recast
2 reactions
vrypan |--o--|
@vrypan.eth
A DA-ish approach was suggested here. Which indeed, has the problem of making it harder to fetch the whole state. https://blog.vrypan.net/2024/06/26/farcaster-l2/
0 reply
0 recast
1 reaction