It's only 5-24 terrabytes. That's not including tv shows, which would make it much bigger. But even those would still be small enough to fit it in a small box.
EDIT: Thinking about this more, it becomes almost like S3 for all web content. Each piece of content can be addressed with its SHA hash, disparate nodes can increase the number of replicas based on demand, etc.
IPFS, I'm pulling even harder for you now!
Subresource Integrity [1] includes the hash of the content/object in the link to the resource, as well as the URL/URI; this lets you do fun things like not fetch duplicate content if its already in the browser's cache from another site (think a specific version of jquery, for example).
Magnet links [2] are a great example of this in practice, but that's bittorrent-specific (although I would not be surprised if you started to see ideas from Magnet URI scheme work their way into Subresource Integrity):
"The Magnet URI scheme is a de facto standard defining a uniform resource identifier (URI) scheme for Magnet links, which mainly refer to resources available for download via peer-to-peer networks. Such a link typically identifies a file not by location, but by content—more precisely, by the content's cryptographic hash value.
Since it specifies a file based on content or metadata, rather than by location, a Magnet link can be considered a kind of Uniform Resource Name, rather than the more common Uniform Resource Locators. Although it could be used for other applications, it is particularly useful in a peer-to-peer context, because it allows resources to be referred to without the need for a continuously available host."
Github has a great explanation as well: http://githubengineering.com/subresource-integrity/
On that note, there as an article posted here on HN a while ago about Gopher. And at that point i found myself wondering if Gopher could have been modified or extended to handle such a scheme.
Exactly, although your browser would probably want to make an attempt at the URL specified before falling back to P2P/DHT if the server or URL wasn't available (you'd want to get objects like CSS, JS, or images FAST but other objects could probably lazy load in the background). The Magnet URI scheme already supports this behavior.
> On that note, there as an article posted here on HN a while ago about Gopher. And at that point i found myself wondering if Gopher could have been modified or extended to handle such a scheme.
"Everything old is new again."
Heh, that line pass through my mind every time i see a Slack headline...
Apparently a startup was attempting to offer a cloud PVR service, but then ran into the MPAA.
Before the encounter the company was planning to use de-duplication to keep storage expenses down. But that was nixed by the MPAA lawyers.
They insisted that each customer had his own, encrypted, storage area. Anything else would be considered a broadcast service, requiring a very expensive license agreement.
At that point i think the company founders basically threw they arms up in exasperation and shut the company down.
But this isn't the end of the matter for cloud-based 'cable operators': http://www.reuters.com/article/2015/07/17/us-filmon-copyrigh...
It seemed to have been a company getting ready to offer their service, and wanted to clear things with the MPAA (or the MPAA lawyers showed up unannounced). Only to find it impossible to both hit their planned price level and meet MPAA storage management demands.
This because the MPAA demands basically meant they had to massively inflate their storage capacity, and thus the hardware cost pr customer made them uncompetitive with existing PVR products.
Yet. We've got a long way to go before we hit physical limits in storage technology.
I'm just guessing there's similar factors in memory market where max theoretical memory isn't going to happen for business reasons if nothing else.
Back then you had various glossy magazines around, complete with a classifieds section in the back.
Apparently one use of those was to offer a like for like copying service. Meaning that you put up an ad and people would mail you floppies of various games and software, along with a list of same they hoped to get in return.
Apparently it eventually snowballed to the point where he had invested in a second floppy drive just to speed up the copying process, and ended up spending whole weekends switching floppies and stuffing envelopes.
Eventually he got so fed up he started returning envelopes unopened.
Secondly, i think Amazon accept Fedex-ed HDDs for their cloud services. In case the amount of data to be uploaded would take unreasonably long otherwise.