Content
@
0 reply
0 recast
0 reaction
Garrett
@garrett
There's way too many people that disregard or write off Solana without even exploring it for themselves
10 replies
7 recasts
31 reactions
Cassie Heart
@cassie
I did explore it. I've even historically participated in their main discord, reported security issues, tried to reason through their design choices, and ultimately, I just don't like it. I feel like there's things they could improve, and then there's things that are too extreme of efforts to change that are sorely needed.
1 reply
0 recast
8 reactions
Garrett
@garrett
Would love to hear more of your thoughts here! What issues do you have with it in terms of design, needed improvements/changes, etc. One of the biggest things I see is that Solana might be the first chain to truly get to human scale and perform well enough for the average consumer to use daily
2 replies
0 recast
0 reaction
Cassie Heart
@cassie
proof of history necessarily makes validators do way more work than is required. They call it a verifiable delay function but definitionally it fails on this due to the lack of succinctness (all hashes must be validated, albeit can be in paralllel). The bandwidth required for validators is somewhat of a centralizing force, but further the stake barrier involved has lead to a superminority that controls sufficient stake to take over the network: out of 1375 total validators, 19 have enough stake to control.
1 reply
0 recast
4 reactions
Garrett
@garrett
Don’t fully follow the implications on the first point On the second point, do you think this centralization of validators changes as the networks scales and becomes more valuable? Maybe the validator set gets more diversified as there’s a larger incentive with TVL on the network
1 reply
0 recast
0 reaction
Cassie Heart
@cassie
The first is partly what leads to the second. Because proof of history is so beefy, it makes validators necessarily top heavy machines, increasing centralization. Seems doubtful validator sets will increase unless they get all of the network into zk
1 reply
0 recast
0 reaction
Garrett
@garrett
Ahh got it. Do you have a rough idea of what it might cost to run one of these top heavy machines? Any further color on why you think it’s unlikely for this validator set to increase?
1 reply
0 recast
0 reaction
Cassie Heart
@cassie
Definitely not cheap: https://docs.solanalabs.com/operations/requirements#hardware-recommendations I don't believe this is going to decrease in reqs. The other factor is limits to networking scale — until we're regularly in the land of 10GbE or higher, intra link latency between peers sending shreds is a problem for the network. They have no intention of sharding, so unless bandwidth increases while preserving latency or we get faster than light communications, there's always going to be an inherent upper bound.
0 reply
0 recast
1 reaction
