v1rtl on Warpcast

Content pfp

https://opensea.io/collection/dev-21

0 reply

0 recast

2 reactions

greg pfp

Two opinions I’m starting to form: 1. In the future most data (80%) will be stored in one or more decentralized protocols 2. Those protocols will not use the EVM in any way Bullish crypto, bearish ETH

11 replies

2 recasts

21 reactions

Kyle Tut pfp

It will be IPFS. CIDs provide all the verification blockchains need while providing the most amount of flexibility related to persistence, decentralization, speed, etc.

3 replies

0 recast

1 reaction

Matthew Fox 🌐 pfp

Matthew Fox 🌐

Probably not Idealistically it could be but the reality is most people use it through a service provider - like you guys! Someone forgets to pay a subscription and everything gets garbage collected Always found other approaches to permanent decentralized storage more appealing for this reason , especially when you can pay once forever Your way closer to this then I am, so curious how you see the general landscape here I clocked out on this topic like 2 years ago so maybe I'm dated

2 replies

0 recast

0 reaction

Kyle Tut pfp

1/ Permanence is a meme. The vast majority of data should not exist for a long-time. Pruning state is necessary, deleting data is necessary. Filecoin and Arweave don't make fundamental sense. The idea of chopping up files into little bits and spreading them out across a global network is extremely inefficient. It causes significant issues. First, geography matters with files. If you want them fast, they need to be close to where you are. Filecoin and arweave are the anti-thesis of that. Second, filecoin and arweave can't run at the edge. This means they can't truly verify data on device, requiring additional systems or protocols to trust where the data came from. Third, they both require additional (expensive) consensus mechanisms. Files don't require their own consensus mechanisms. Filecoin and arweave will always be slow and expensive in comparison to IPFS. continued...

1 reply

1 recast

2 reactions

Kyle Tut pfp

2/ IPFS can optimize where files are stored to ensure speed, IPFS can run on any device, and IPFS can use the blockchain it is attached to for any consensus. Finally, the beauty of IPFS is its CID system. It allows it to be as centralized or decentralized as you want it to be based on the requirements. Neither filecoin or arweave provide the flexibility that IPFS does. / end

1 reply

0 recast

2 reactions

Matthew Fox 🌐 pfp

Matthew Fox 🌐

Great points appreciate the breakdown and perspective

1 reply

0 recast

0 reaction

Kyle Tut pfp

Final point, I don't think IPFS is ideal. The dht isn't great, the p2p isn't great. It has a lot of it's own issues. But, it's the closest thing we have today to where we need to go. While Pinata is obviously closely associated with IPFS, we are open to other options as long as it maintains content-addressable data with verifiable CIDs. Currently, the market supports it over anything else. If that changes, we will adjust. But, from our perspective, the answer is not Filecoin or Arweave.

1 reply

0 recast

1 reaction

v1rtl pfp

thoughts on EthSwarm?

1 reply

0 recast

0 reaction

Kyle Tut pfp

They do some interesting things similar to IPFS and Filecoin but have three main problems. 1) ETH specific. The advantage of IPFS is it is blockchain agnostic. So, it is interoperable across chains giving it a distribution advantage. 2) Like Filecoin and Arweave, it is built around solving persistence, specifically through incentives. This creates way too much technical overhead, significantly increasing cost and efficiency. 3) It can't run locally. If we are gonna have robots transacting crypto, the file storage needs to run all the way to the edge to ensure cryptographic verifiability. Tl;Dr: Mainly a variation of Filecoin/Arweave solving the wrong problem

2 replies

0 recast

1 reaction

v1rtl pfp

2 — incentives are not related to efficiency or performance. IPFS does not provide any storage guarantees, which is not ideal when you want to know if files are actually stored. Just having a clear incentive mechanism does not inherently mean it's more expensive. It means that you're safe from being rugged like with centralized pinning services and vendors can't dictate their rules. 3 — wdym can't run locally? you mean swarm's not embeddable?

2 replies

0 recast

0 reaction

v1rtl pfp

to add to 2, with IPFS when you buy storage from a provider, you buy a promise that that provider will keep your files for some time and instead of the price being calculated over a specific fixed formula (like in Swarm), providers set their own pricing plans which do not necessarily reflect the true cost of services. As an IPFS provider I can charge $10 per GB, with Swarm I can't do that

0 reply

0 recast

0 reaction

Kyle Tut pfp

The incentive mechanism isn't expensive, proving that the file is actually stored is, though. Because you don't have a known relationship to store the file, you have to do a significant amount of verification trying to prove that it's stored. That is expensive and inefficient. File storage isn't the same problem as blockchain so files can't be replicated on every node. Every instance of trying to decentralize a file across nodes will automatically make it more expensive and then, because they aren't on every node, you add computation to try to prove it. It's a fundamentally flawed approach to file storage.

1 reply

0 recast

0 reaction