For the past few days, I've been trying to work out the plan, the coordinated plan, to move Threespeak into the new tech stack. I've written about this before, but it bears repeating: this thing is growing so many tentacles. But it's always like this. You begin changing one thing, and then a chain of changes must follow.
For the past few days, I've been trying to work out the plan, the coordinated plan, to move Threespeak into the new tech stack. I've written about this before, but it bears repeating: this thing is growing so many tentacles. But it's always like this. You begin changing one thing, and then a chain of changes must follow.
Although I've had a few conversations with Matt and Tibs, not everything is ironed out yet. For one, we are definitely not going to be storing videos forever, at least not the videos nobody watches. But how we are going to delete them, and when, is a question that's proving very hard to answer.
What I do know, and I say this with respect, is that most videos on video platforms are duds. They may get some plays in the very beginning, but they mostly go dormant. I certainly have plenty of those videos myself, and to be completely transparent, I'm glad they've survived, even if I sometimes forget they exist.
But, and this is the key part, there's a big budgetary difference between the Alphabet company and our little video platform venture. YouTube can store terabytes of video for pennies, and I'm sure if there's one thing they don't sweat, it's redundancy of their files. We, on the other hand, are battling sea monsters with handguns, and I submit we're doing pretty good too.
With all that said, the key thing, or perhaps the winning combination, is to give people, Threespeak users, a way to conserve their videos if they so wish. Give them a tool they can use to reprovide their videos to our infrastructure once the time we're willing to store them has expired. Again, I say this without actually having a set expiration date yet.
To be transparent, and to pull a chair up to the table, most videos, and I do mean most, go to sleep by the fourth day. In other words, most of their plays happen during the first and second days, but by the fourth, they're pretty much sleeping like a rock. This is understandable given the nature of Hive and its payout windows, but this also can't be how we dictate how long videos should exist.
I think we might end up hosting videos "for free" for up to a year. Meaning, if after a year nobody has backed up the video, nobody has become a reprovider of that content, and the video isn't one of those evergreen videos, it will get deleted.
Is this horrible? I ask. I don't really think so. But only, and this is key, if we also provide a tool for you, the content creator, the person who put effort into making that video, to keep said video alive for as long as you want.
A video that's older than a year and gets zero plays would eventually be cataloged as "community stored", which in turn would tell our platform that the video is very much in cold storage.
I can then imagine the Video Pulse service I'm working on expanding to seek out the IPFS CID out there, in the wild, so to speak. If it's still alive somewhere, Video Pulse could reprovide that CID to a hotnode for delivery.
Now, this means, and this part is crucial, that dormant videos hosted by our decentralized system, whether that's SPK Network or something else, would have a very slow process of waking up.
THAT is the compromise.
But in theory, this buys us years and years of running Threespeak within a reasonable budget, surviving even the beariest of bears.
That said, I need feedback. I want to know how everyone feels about this, and what you think would be a sensible holding period.
MenO