✴Limit of IPFS?

[Ucy]

How sustainable is the IPFS concept?

I do not understand how Personal Computers around the World can hold so much IPFS data without limit. Where will the remaining data go if people begin to have their computers filled up with data and they have no incentive to increase their harddrives.

Google data centers for instance can hold so much data from YouTube, GDrive, G+, Gmail, etc because Google has incentive to hold them and keep expanding the data centers.
What incentives are there for participants in the IPFS system to keep holding more & expanding their harddrives?

[1713466876]

1713466876

[1713466876]

1713466876

[1713466876]

1713466876

[Kallisteiros]

What incentives are there for participants in the IPFS system to keep holding more & expanding their harddrives?

There's none. IPFS nodes only store the information they themselves are interested in.

IPFS team is currently developing a project called Filecoin, which would provide such incentives. They got $257 mil, but sadly, there has been almost zero news since their ICO. So nobody knows when they're actually gonna launch.

[Gabrics]

There is https://storj.io/ already working.
They don't have their own coins developed, but they have Ethereum based token.

[Ucy]

Interesting answers. Thanks brothers!

@ETFbitcoin, I researched Freenet after someone mentioned it here few days ago. I thought IPFS was a hybrid of Freenet. Guess the devs of IPFS got some idea from there... There isn't really much difference!

[d5000]

I thought IPFS was a hybrid of Freenet. Guess the devs of IPFS got some idea from there... There isn't really much difference!

The main advantage of IPFS to Freenet seems to be versioning. If you address a content item via IPFS, you would be directed - as a default - to the newest version, and you should also be able to select the version you want to fetch.

On the other hand, Freenet has additional layers for privacy (a bit similar to Tor) which makes it slower, but more difficult to censor. The drawback is that you can't select which content you store permanently - so even if it probably has no legal consequences (because you never know what you're storing because of encryption) you have to accept to store/serve all kind of nasty illegal or morally questionable content, from Nazism and Jihadism up to child pornography. For me, this is a reason why I never got enthusiastic about Freenet. Blockchain-based storage services like Steem, however, have the same problem.

In contrast, IPFS is more like torrenting - you seed what you like, and your own files, and these other people pay for.

Filecoin, in my opinion, is not really necessary. We have Sia and the already mentioned Storj, which can both be used to create a decentralized market for the storage of pinned IPFS content.

[spiker777]

On IPFS files will be gradually lost from the network as fewer and fewer nodes store the content of those files. Most protocols have an incentivization system in place to ensure that old files are retained, usually by giving these master nodes a greater cut from the block reward for ensuring they don't delete their files. But it's definitely something that needs to be resolved before IPFS can take off.

[just_Alice]

How sustainable is the IPFS concept?

I do not understand how Personal Computers around the World can hold so much IPFS data without limit. Where will the remaining data go if people begin to have their computers filled up with data and they have no incentive to increase their harddrives.

Google data centers for instance can hold so much data from YouTube, GDrive, G+, Gmail, etc because Google has incentive to hold them and keep expanding the data centers.
What incentives are there for participants in the IPFS system to keep holding more & expanding their harddrives?

IPFS is decentralized and works as a content based addressing, more like BitTorrent works. Thus, they face two main problems - free space and around-the-clock accessibility to files. That's why the IPFS team is working on Filecoin. It's like a blockchain on top of IPFS and it will incentivize people to keep files on their harddrives, because they'll get a reward for it. This will not only be a solution to a space problem, but also it will make nodes go online all the time, so that people will have access to files 24/7.

[d5000]

On IPFS files will be gradually lost from the network as fewer and fewer nodes store the content of those files. Most protocols have an incentivization system in place to ensure that old files are retained, usually by giving these master nodes a greater cut from the block reward for ensuring they don't delete their files.

I think all solutions where the currency (master)nodes also are expected to share the whole file system are not scalable and will not find acceptance.

Pinning is actually the way to go. The publisher that wants others to share their content, and guarantee to share it permanently, shall also pay them. You can organize this blockchain-based, like Sia/Storj does (and Filecoin probably will do, too). I'll probably soon test Sia as a client (renting space) so I get an impression of reliability, performance etc..

In theory, what would be ideal is that people needed not to pay in a specific "Filecoin" or "Siacoin", but to have a marketplace covering all cryptocurrencies, including Bitcoin - the most popular. Surely, all these ICO-based projects won't like that, but I'm sure it will happen eventually (so I would not recommend to invest in a coin whose only purpose is to serve as a payment of this kind of marketplace).

There can be additional mechanisms for readers to share content (e.g. like ZeroNet does). This is great for blogs with a "fan" base. But for example a video sharing site like YouTube would probably need professional hosting services which can pin contents on machines in well-connected datacenters.

[cellard]

I thought IPFS was a hybrid of Freenet. Guess the devs of IPFS got some idea from there... There isn't really much difference!

The main advantage of IPFS to Freenet seems to be versioning. If you address a content item via IPFS, you would be directed - as a default - to the newest version, and you should also be able to select the version you want to fetch.

On the other hand, Freenet has additional layers for privacy (a bit similar to Tor) which makes it slower, but more difficult to censor. The drawback is that you can't select which content you store permanently - so even if it probably has no legal consequences (because you never know what you're storing because of encryption) you have to accept to store/serve all kind of nasty illegal or morally questionable content, from Nazism and Jihadism up to child pornography. For me, this is a reason why I never got enthusiastic about Freenet. Blockchain-based storage services like Steem, however, have the same problem.

In contrast, IPFS is more like torrenting - you seed what you like, and your own files, and these other people pay for.

Filecoin, in my opinion, is not really necessary. We have Sia and the already mentioned Storj, which can both be used to create a decentralized market for the storage of pinned IPFS content.

Hearing freenet makes me feel very old. I remember that ancient software from years ago. I was curious about it and tested it. Eventually I learned how it worked, and I also realized how people stored all kinds of illegal data on it, including illegal pornography and other deranged shit. I decided to not actually install the program and test it myself because I don't want none of that sitting on my drive, even if it's all encrypted. What's the point? Sorry but I would not risk that. This is the problem with file-shared models such as freenet and co.

On the other hand with Bitcoin there's no risk involved, it's just a database with numbers. I just don't want to have to store illegal encrypted data in order to use a software, I don't feel right about that. What if the encryption is broken some day or something.

[d5000]

Hearing freenet [...] I decided to not actually install the program and test it myself because I don't want none of that sitting on my drive, even if it's all encrypted.[...]
On the other hand with Bitcoin there's no risk involved, it's just a database with numbers. I just don't want to have to store illegal encrypted data in order to use a software, I don't feel right about that. What if the encryption is broken some day or something.

Well, both Freenet and Bitcoin actually are "a database with numbers". However, I see a difference: Bitcoin has a clear goal to be used as a financial instrument, and data storage is only a "residual" use case. Freenet, in contrast, is clearly made with data storage in mind.

In Germany at least, I've read that Freenet users are not in danger to be sued because of the encrypted stuff they could serve if they don't share it actively, because there is no way they could know about it. Even in countries where there could be a problem using Freenet, I guess Bitcoin full nodes are even less in danger, because you can assume that you're using it as a financial instrument and not with the intention to share illegal stuff. In most legal systems a "good intention" is something that counts if a case goes to court (although it doesn't save you from punishment for every crime).

As I already wrote, for data I prefer IPFS and/or torrenting - I myself can choose what to share. (It would be interesting if projects like Steem(it) had legal problems already because of that issue.)

I've tried Storj, but not really satisfied if you need to store big files. Besides, there's no guarantee that your data is accessible 24/7.

I still prefer using regular cloud storage and encrypt my data before upload/sync if necessary or use open-source storage cloud and deploy it on my own VPS.

Obviously everybody is free to use whatever technology (s)he wants to store data, also to "pin" content on IPFS. But blockchain-based markets like Sia/Story are definitively interesting as they are quite transparent and middlemen-less and thus should tend to offer very cheap prices. For stuff which is not totally important it could be an interesting market niche, combined with IPFS. As I wrote, I need to test it still.