Post
Topic
Board Altcoin Discussion
Re: AmericanPegasus initiates coverage on Sia. (Siacoin speculation topic)
by
Taek
on 10/03/2016, 22:19:12 UTC
Inoramus, don't you know that all these decentralized file storage systems (including Sia, Storj, Florincoin, etc) can't work because I explained that proof-of-storage (a.k.a. proof-of-retrievability) is nonsense because anyone can pretend to be running many nodes and keep it all stored on one node. Thus Sybil attacking the system and cheating its economics and data retention resiliency.


Same as for proof-of-stake deposits, this only has to be deposited once yet can be used to Sybil attack unbounded number of clients ongoing.

And shorting the coin can be more profitable than the deposit staked.

Reputation systems devolve to centralization winner-take-all paradigms due to the power vacuum created by the security hole (the potential to attack the coin and/or clients).

So, there are few problems here, the first is non-technical. I do ask that you don't smear insults throughout your posts because it makes it difficult to see your technical arguments and distracts from your core points. The second problem is that you are citing some pretty old information about how Sia works, and you are missing some important details that help bring the system together and secure it.

I tried to pull out your technical arguments in the quote above, not sure if I'm missing something.

The first technical argument you make is that "anyone can pretend to be running many nodes and keep it all stored on one node". I'm not sure if you are talking about a deduplication attack here, or just a standard Sybil attack. Sia addresses the deduplication attack by using different encryption keys for each portion of the redundancy. Because all of the pieces are using different keys, a host can't perform deduplication, they have no idea how the bytes map to eachother.

The Sybil attack is, in my qualified opinion, the weakest part of the Sia protocol. But we do have sufficient defense against it, and that comes in the form of proof-of-burn. By burning coins, you give yourself legitimacy, and then people can use you. People will prefer you in proportion to the number of coins you burn, perhaps favoring coins that have been burned more recently because there is more certainty about the value of the coins at the time of burning. Other than that, all burned coins are treated as equal. Host have persistent identity, which means you can identify them by their file contracts, and hosts are forced to burn (well, it goes to the siafunds holders) 3.9% all of their income in the file contracts. So, hosts are weighted based on burn, and hosts are continuing to burn more coins as time passes. To Sybil attack the network, you need to burn a significant number of coins, representing a significant financial investment. You suggested above that once the coins are burned, you have the reputation forever, but this is not true. If you start failing storage proofs, or otherwise demonstrating unreliability, clients will blacklist you.  Even if you maintain reliability, you will need to be continuing to burn coins to keep up with the rest of the network. Clients will continually be requesting out-of-band storage proofs on the data you are claiming to hold, and because of the siafund tax there is no cheap way for you to fake holding valuable data. If you are pretending to pay yourself, you are also losing coins to the built-in fee.

Sia has a second weakness here, which is that Nebulous owns most of the siafunds. So, it's expensive for non-Nebulous people to Sybil the network, but it's less expensive for Nebulous to Sybil the network (because we'd get back ~90% of all the coins we burned). Paranoid renters could require a second layer of burn which goes beyond the 3.9% siafund fee. We've also got a handful of other  techniques (such as 1 host per IP address) to defend against Sybil attacks as well. It's not perfect, but it's also going to be nontrivial to Sybil the Sia network. (I will establish that the proof-of-burn is not fully implemented, should be ready sometime around June).

Also you talk about shorting a coin and then using the money raised from shorting it to attack it. You would need to buy the short from somewhere, and you'd need enough money to both buy the shorts and then have enough to do the proof-of-burn. I'm doubting that someone would let you short a massive volume of some asset, if you are shorting that much they would probably get suspicious.

I do feel like I've done a poor job of explaining this. I'm happy to elaborate further, or address other points that you think I've missed, but please leave the insults at the door.