deleted by creator
iirc ipfs libraries are decoupled from filecoin, you can use whatever implementation in whatever language, and never even know filecoin existed
also, why do you think fc is a bad idea?
deleted by creator
Ipfs has always had a strongly layered philosophy. The team believes in modular software architects. They will never be coupled. Just look at their stack, heck they even made a protocol for self decribing hashes. Also libp2p is quite nice.
deleted by creator
yeah the diffence is that layered protocols would be intrinsically difficult to inject coupling. Protocol labs help create protocols, but its out of their hands now. For example polkadot uses libp2p and also did most of the rust implementation. Also many Ethereum project use lib2p2 and IPFS (although eth use DEVp2p for it networking) If protocol labs wanted to add something to the protocol that was helped just filecoin that would go against libp2p’s largest consumers and they would just fork the protocol anyways. Protocol lab’s approach is to be as plurilistic as possible so they can get the largest mindshare of developers using and working on the protocols, then eventually the protocols becomes self sustaining through the projects that are built on them.
team with plans to couple it
that’s conserning ngl, do you know of any source?
deleted by creator
that’s just protocol labs offering their own hosting, there’s nothing preventing you from not using it, no?
deleted by creator
they are offering it for free in order to incentivise more developers to adopt it, more people using it only means they need to pay more for hosting
plus, the value of ipfs is largely dependent on its flexibility, locking everyone to use filecoin would probably be not only technically impossible (many libraries are maintained by people other than protocol labs), but also weird from their side, as there is a multitude of applications for ipfs other than redundant bulk storage
obviously, there’s no way to know what’s going to happen, but in my opinion this is extremely unlikely: watching all the interviews with the people behind ipfs I get a strong feeling that these people are actually trying to do something good, and this is not simply a cash-grab operation
the docs are actually pretty good but some of the concepts go over my head. i wish there was more content on youtube and blogs though. most content surrounding ipfs rn are talks about the ecosystem and existing projects.
iirc,
content is distributed. so if your neighbour read aljazeera this morning, your computer doesn’t have to fetch to from thousands of miles away. this reduces network traffic load, and load on aljazeera’s servers. it makes they internet more scalable.
content is hashed, so if it changes, it gets a new hash. so you can search for today’s version of the newspaper, or you can search for last week’s version, because they have different addresses/hashes. nothing us ever lost or deleted irretrevably.
i think that’s the important part, but there are other big improvements that are built in, things people didn’t think of in the 1960 when they developed the internet protocols, but we now realise are needed.
Has anyone done a good breakdown of torrents vs IPFS?
I think the major difference is that IPFS uses a single DHT whereas torrents have a DHT (or tracker) per torrent. This mean that if there are shared chunks between media, then they will be deduplicated. The downside to this is that you have a single massive DHT, so lookups can be slower (and were very, very slow early on, but the situation is improving).
The other difference is that ipfs is built with a very modular stack. Networking, Self describing hash, ect, which should allow for reuse else where. IPFS is also very general, you can use different hash functions and chunking methods when importing data. This allows for existing hashed to be imported into ipfs (like a git repo, or a blockchain like bitcoin) and shared a crossed the network. But that modularity and generality can also come at a price of performance, so there are trade-offs.
they can’t be the future of data storage until they make an easy way to configure $IPFS_PATH in go-ipfs before initialization…