VideoCoin Technical AMA part 3
Our Chief Technology Officer, Devadutta Ghat spends time every few weeks answering the VideoCoin communities’ questions live in Telegram. We have an extra Telegram group discussion just for these technical questions. Below is a compiled list of the technical questions we have answers to. Enjoy!
Opening remarks from Devadutta Ghat,
“We’ve made a lot of progress on our testnet, Wallet Software and of course TGE Smart Contracts these last few weeks. We also brought on some critical engineering folks on to the team.”
“We have expanded our recruiting team, we found some really good folks who come to us from companies like Cloudera and Apple. Our engineering recruitment is in full swing. We will be making a public careers page where you can apply to join our team, but as always, if you have folks who want to join us, send an email to [email protected].”
“We also have a brand new marketing team who are doing a great job with hundreds of different marketing opportunities and partnerships!”
Question 1: We have a question about storage miners. Will Client1 (whom requested original video encoding process) pay storage miner post encoding or does Client2 (viewer of video) pay for that video to be stored with the storage miner?Also,version three does the encoding miner pay the storage miner post encoding?
Answer 1: This will depend on the implementation, let me explain.
In general, Client 1 pays the storage miner on equal intervals for time over the period of the contract for holding the file. So over the life of the video client 1 pays storage miner. If Client 1 requested encoding, yes, Client 1 will pay the compute miner.
Client and Miner will establish a contract to store the file and the client validates the proof that was submitted.
Possible implementations:
The client needs to pay multiple times for the same video input to be encoded into different formats in order to match codecs, color space etc..
Question 2: Is there a scenario whereas the requester of video (me with netflix) pays the encoding and storage miners?
Answer 2: That’s also possible to implement, we intend to keep the APIs transparent. So you could create your own smart contract on the network and use that to charge from clients. We will be providing a standard Ethereum interface, so lets say you want to create an ERC20 token on the VideoCoin Network, it is possible.
Question 3: So lets say you want to watch a show that has been encoded once for a specific format and you need it encoded in a different format. Is there a way to re-truncate that video to source input and then re-encode it to match my device requirements?
Answer 3: Great question! Yes, the client would submit another contract for the next format to be encoded. Each would be individual and multiples will be available. It will be a programmable setup for the desired input/end user.
The input outputs are very flexible. We are also working on porting an LLVM based compiler so we can run DirectShow filters directly from FFMPEG. This is super hard to implement, so I don’t want to get your hopes high right now, but we are working on it.
Question 4: One challenge for someone like myself in the videocoin network would be managing different contracts based on different content I wanted to view/submit. Is there any work in place to help manage smart contracts for this type of network whereas a not technical person could see past/potential contracts to engage in?
Or simply put how would I manage all my different content?
Answer 4: So when you build a complex Video Infrastructure platform, say like a YouTube, as a developer you would maintain a a lot of code involving API calls to third party services. Similarly, when you build on top of VideoCoin, you can expect to maintain a lot of code.
We will be providing sample code and implementations. But, you can expect the complexity to be proportional to the project.
Question 5: Do you think you have enough time to still launch VideoCoin alpha network in January 2019?
Answer 5: O, yes. On track as of now! We want to show a testnet demo as soon as we are able to in the end of this year or early January.
Question 6: Question: Could you comment, in general, on the Retrieval & Distribution aspect, status and goals for the remainder of the year?
Answer 6: We will be starting out with a testnet that will use a simple retrieval and distribution, single origin server and single “centralized” distribution.
Then we will steadily make the CDN decentralized. As you know, a decentralized P2P CDN is pretty intense work and we will approach it in stages. Right now we are mainly focused on encoding for this first testnet launch.
We will do the proof of transcoding and show an end to end transaction go through the network for encoding.
Question 7: What I’m looking for specifically is how I can describe what could be made transparent and tradeable through methods similar to yours applied to the life cycle of our personal data, with us added as stakeholders?
Answer 7: This is a brilliant idea, there are some ways you can use the concept of Proof of Transcoding. While the proof is specific to video, you can generically extend it to many types of computation using the same challenge response authentication methods. For example, if you want to build NLP models on the data you’ve collected, there is a way to extend a general purpose compute proof like Proof of Transcoding.
Question 8: Another question is, will FPGAs be good at Proof of Transcoding with an adequate bitstream?
Answer 8: You don’t need FPGAs, you can use standard CPUs, most CPUs have a video fixed function and we will use that
Question 9: Yes I know, but FPGAs are much faster, i know AWS use those. So i wonder if it’s an advantage to use those for proof of transcoding in the mining part of videocoin?
Answer 9: Depends on the FPGA, but in general the GPUs and CPUs of today have a very impressive video fixed function.
Question 10: I listened to Kellam Ames on Coin Agenda and he mentioned that the encoding, distribution and so on won’t be on the blockchain, it’s just too slow. Can you explain a bit how you will arrange that part. Will your blockchain send workout to miners and do you develop software for miners to download and use.
And will you also develop a wallet just for videocoin when it will have you own blockchain? I also would like to hear some time prediction of when your road map will be more specific, remember you said you will update that. And also when we could begin to see some code on github?
Answer 10: Nice! Kellam’s right. The blockchain is mainly a database that is used to store transactions and proofs. All infrastructure blockchain projects do this. They do off blockchain work and consolidate proofs on the blockchain. Like Sia or Storj.
And yes, we have a wallet in the works.
Question 11: The VideoCoin network, by Halsey’s own statements, is going to be predominantly comprised of enterprise-level data centers. Individuals don’t have the bandwidth or latency or reliability to act as enterprise-level edge or origin servers. They’ll be mostly limited to transcoding — the least important, easiest job in the equation, which can easily be performed by the aforementioned data centers.
The owners of these data centers all have the ability to accept USD — they’re big businesses! So I don’t buy your argument here — if we are accepting your argument, the only thing blockchain buys us is the ability to let individuals all over the world perform the transcoding job.
That’s not compelling enough for me — most of the transcoding is gonna happen in data centers the same way most btc mining is happening in data centers.
Answer 11: Great question! This is a blog I am supposed to write, never got to doing. Maybe the time has come. 🙂
Firstly, Encoding/Transcoding is one of the most important work as it is compute intensive and algorithmically complex part of the network.
The reason we need a new native blockchain is to make sure the proof of work to establish a transcoding job can be done reliably.
So, when you say “Blockchain is just an incentive layer”, it sounds simple, but it take a LOT of engineering to make a video encoding and distribution job to run and make the incentives work.
To your point that this is a B2B play and this can be done without a blockchain, here is something the blockchain is disrupting that not many people realize: Legal contracts.
One of the main problems with existing B2B business models is that you need to get into a deal with every single data center and ISP. Imagine the legal bandwidth we would need to make something like that happen. We would need a legal team the size of huge multinational company.
Blockchain revolutionizes this. All a data center has to do is install our mining software and they will be part of the network. No deals, no negotiations.
Sure, the first few folks will have to be convinced, but later on it will be an organic growth. Imagine how many people mine BTC, if all these people had to establish legal constructs with the BTC foundation, we would never have had BTC.
There are many other reasons for a “Why we need a blockchain”, I will do a better job in the blog, but here are some points to ponder on until I write that:
Decentralized applications — There is no way to build a truly decentralized network without a blockchain like database. If you have a question on “Why decentral”, we can go on a philosophical debate there.
Privacy and participation
Ecosystem of distribution partners
I know what Halsey meant by his statements, and it is true. But, there are a lot of other edge and end user compute nodes that are needed for the network to properly perform distribution and edge computing functions. Imagine our mining software running on your cable TV set top box. That is totally possible, and would create a very powerful network of distribution.
Question 12: Thank you for your detailed response. I will admit that my intuitions about the relative importance of transcoding and the inability for individuals to act as edge/origin servers could be unfounded. Time will tell — however, I struggle to imagine a world where Hulu or HBO would allow something as critical as data streaming to happen through a personal Internet connection. But we’ll see.
However, I do not understand your point about legal contracts. If you were going around building contracts with data centers around the world, you would be a traditional cdn! That’s literally what a cdn company does — they get POP contracts with data centers around the world and resell the bandwidth for profit.
The whole point of VideoCoin is to eliminate all that — if you have a data center, you can hook up your machines to the network and monetize your resources as a cdn without hiring engineers and marketing and support staff, and content streamers can hire your resources through an open marketplace — no marketing necessary!
So, with or without blockchain, you’ve obviated the need for POP contracts — instead of cdn companies building POP contracts with data centers around the world and reselling the bandwidth to streaming companies, you now have an open marketplace to connect streamers with data centers.
The POP contracts are gone — with or without blockchain!
Answer 12: Very insightful question again, thanks!
Re: Netflix or Hulu using VideoCoin — In the beginning, it will be difficult for companies like NFLX or HULU to use VideoCoin exclusively for their infrastructure. I agree. But, we’ve introduced an all new concept of “Relay Miners” who basically act as a bridge for these companies to gradually migrate to decentralized networks. It is similar to the way CDN caching is used. If there is a cache miss, the CDN forwards a request to the origin server. Similarly, a client will try to pull from VideoCoin network first and if it fails, it falls back to a relay miner or falls back to a centralized cloud backup.
Re: POP Contracts: You’re right, we have obviated the need for them, the blockchain is needed for people to transparently prove that work was done and get compensated for it. In the example of a traditional CDN, the buyer and seller is pretty much set, the seller meters usage and charges the buyer. With VideoCoin (and other decentralized systems), buyers and sellers directly execute a “smart” contract that can prove actual work was done. So the blockchain is doing the metering work in our case.
The main difference however is that with the traditional model, you trusted the seller to meter usage properly, but in a decentral blockchain powered model, you do not need trust, you can verify the proofs.
Join us for discussion in our Telegram group: https://t.me/videocoin