3 ms·

Huge achievement. A big, big stepping stone toward interplanetary communication. But while I love the idea of IPFS, it comes with a bunch of tradeoffs that I t

Huge achievement. A big, big stepping stone toward interplanetary communication. But while I love the idea of IPFS, it comes with a bunch of tradeoffs that I think make it very unlikely that it'll ever become mainstream. What I do think will happen in space, is similar to what already happens with PoPs around the world, no IPFS required. As an example: I believe that there will be a YouTube on the moon and YouTube on Mars, and YouTube servers caching most of the content on both, and it'll all work the same. Cross-planet communication will be high latency, but it won't matter.

toomuchtodo 3y ago
> What I do think will happen in space, though, is similar to what already happens with PoPs around the world. As an example: I believe that there will be a YouTube on the moon and YouTube on Mars, and YouTube servers caching most of the content on both, and it'll all work the same. Cross-planet communication will be high latency, but it won't matter. Importantly, this is enabled by content addressable storage in IPFS, where the address of the data is derived deterministically with cryptographic hashing. Agree it likely won't ever become super popular, but it is a core component of distributed storage systems without central orchestration. https://en.wikipedia.org/wiki/Content-addressable_storage https://en.wikipedia.org/wiki/Content-addressable_storage
- robertlagrant 3y ago
  > Importantly, this is enabled by content addressable storage in IPFS Can't it be enabled in all sorts of ways? E.g. edge caching, the way YouTube works today?
  - toomuchtodo 3y ago
    Coordination, namespace, and addressing challenge with traditional addressing and caching. It’s why BitTorrent is so good at distribution: it is uncoordinated edge distribution, IPFS is just a bit more formalized. Subresource integrity demonstrates why the crypto is needed.
    - robertlagrant 3y ago
      I thought BitTorrent's main strength (although that is one) is that it allows dynamic routing through lots of nodes, so you aren't pulling from a central cache, but from all sorts of clients. That was particularly useful in a world of asymmetric upload/download speeds; many people could upload to you at once, enabling your to saturate your downlink. That wouldn't be the case on the moon I would have thought, as you won't have a huge network of intermediate nodes to draw on. You probably just have one or two links, which you can push content through en masse. Tl;dr I don't see why what you're saying applies in this scenario.
m3kw9 3y ago
Is this a joke? If anything they would use a centralized service and not have to install a wallet to communicate.
- pocketarc 3y ago
  I think you might've misunderstood what I was saying. I absolutely agree with you. I don't think it will use IPFS. I'm saying in a hypothetical future in which we have colonised other planets, tech companies will just plop caching servers on Mars and the moon, completely obviating the need for IPFS or anything special. (edited original comment a little to clarify)
m3kw9 3y ago
And you gonna have 3 people watching YouTube in space after using millions to send them there?
- snvzz 3y ago
  For your example, sending one copy to space is better than sending three copies.
  - m3kw9 3y ago
    Not seeing how starlink can’t just put a hd and serve stuff.
    - snvzz 3y ago
      You'll need a way to find which starlink has the hdd that has the data you need to access. Content-based indexing (as in IPFS) handles that. Plus the data gets cached in the starlink you locally accessed, in your example. It might help to think about IPFS as a public CDN operating on an open specification protocol and an open-source implementation.

toomuchtodo 3y ago

> What I do think will happen in space, though, is similar to what already happens with PoPs around the world. As an example: I believe that there will be a YouTube on the moon and YouTube on Mars, and YouTube servers caching most of the content on both, and it'll all work the same. Cross-planet communication will be high latency, but it won't matter. Importantly, this is enabled by content addressable storage in IPFS, where the address of the data is derived deterministically with cryptographic hashing. Agree it likely won't ever become super popular, but it is a core component of distributed storage systems without central orchestration. https://en.wikipedia.org/wiki/Content-addressable_storage https://en.wikipedia.org/wiki/Content-addressable_storage

robertlagrant 3y ago
> Importantly, this is enabled by content addressable storage in IPFS Can't it be enabled in all sorts of ways? E.g. edge caching, the way YouTube works today?
- toomuchtodo 3y ago
  Coordination, namespace, and addressing challenge with traditional addressing and caching. It’s why BitTorrent is so good at distribution: it is uncoordinated edge distribution, IPFS is just a bit more formalized. Subresource integrity demonstrates why the crypto is needed.
  - robertlagrant 3y ago
    I thought BitTorrent's main strength (although that is one) is that it allows dynamic routing through lots of nodes, so you aren't pulling from a central cache, but from all sorts of clients. That was particularly useful in a world of asymmetric upload/download speeds; many people could upload to you at once, enabling your to saturate your downlink. That wouldn't be the case on the moon I would have thought, as you won't have a huge network of intermediate nodes to draw on. You probably just have one or two links, which you can push content through en masse. Tl;dr I don't see why what you're saying applies in this scenario.

m3kw9 3y ago

Is this a joke? If anything they would use a centralized service and not have to install a wallet to communicate.

pocketarc 3y ago
I think you might've misunderstood what I was saying. I absolutely agree with you. I don't think it will use IPFS. I'm saying in a hypothetical future in which we have colonised other planets, tech companies will just plop caching servers on Mars and the moon, completely obviating the need for IPFS or anything special. (edited original comment a little to clarify)

m3kw9 3y ago

And you gonna have 3 people watching YouTube in space after using millions to send them there?

snvzz 3y ago
For your example, sending one copy to space is better than sending three copies.
- m3kw9 3y ago
  Not seeing how starlink can’t just put a hd and serve stuff.
  - snvzz 3y ago
    You'll need a way to find which starlink has the hdd that has the data you need to access. Content-based indexing (as in IPFS) handles that. Plus the data gets cached in the starlink you locally accessed, in your example. It might help to think about IPFS as a public CDN operating on an open specification protocol and an open-source implementation.