text stringlengths 0 1.36k |
|---|
**Alex Sexton:** I think it's possible to set up your own servers... You may have to, like... |
**Mikeal Rogers:** Yeah, yeah... But again, we need to make this useable, right? And without setting up your own server, you can visit Roll Call and it will set up a point-to-point connection and never store any metadata or keep any kind of logs of that transaction anywhere else, and it's all encrypted and everything, ... |
These are issues that we need to start caring about as this becomes a bigger and bigger chunk of the internet. We need services that protect your privacy, and as soon as you wanna protect the user's privacy, the service provider loses access to read that data. So we have very different models. If you don't have a centr... |
Another part of this that we're not talking about that is a really big part of the decentralized web is building offline applications. One of the hardest parts about building offline applications is that you lose that central authority. When I go to Twitter and I get my timeline, it says that a bunch of people said thi... |
\[00:32:04.17\\\] So you run into a lot of the same problems with offline as you do with every other decentralized point. And once you solve these problems for offline, you naturally solve them for a lot of the other decentralized cases. |
**Alex Sexton:** Yeah, that's a good point. I think the solution to everything is homomorphic encryption, right? |
**Mikeal Rogers:** Why don't you tell us what that is? \[laughter\] |
**Alex Sexton:** This is one of my favorite things... Homomorphic encryption is whenever you can perform the same -- like, if you have a function, you can encrypt both the function and the input to the function, and the output is the encrypted version of the plain text actual output. If you had a function called Add 2,... |
The idea would be that you could have Gmail, and Gmail would be 100% encrypted on the backend, but one thing about an encrypted mail service is that you can't search in it anymore, and mail without search is effectively useless, in my opinion. So you'd either have to bring all the search locally, which means you'd have... |
Essentially, you can perform a search on encrypted data. It's kind of early days -- not that early, but it's gonna take a long time for that to exist and be good, but I think it solves a lot of the trust cases, while enabling user experience at end-to-end-to-end-to-end performance. |
**Mikeal Rogers:** One of the most fascinating things to me about all this is that we're not really waiting on a lot of web standards anymore; we're not really waiting on even these cryptography libraries to exist. There are encryption libraries right now that will work in the browser. What we're really lacking - peopl... |
**Alex Sexton:** Yeah. To be clear though, the homomorphic encryption still has a long way to go, and there's not just like an npm package that's gonna do it for you, or any packages can do anything that I said for you, except for very simple math. Anyways, not a podcast on homomorphic encryption, by any means. |
So we end up with a lot of demos in this space. We talk about NodeBots, we talk about VR, we talk about P2P, and all of it I think is massively interesting and is the future, but it keeps being the future, no matter how far into the future that we go... It keeps still being the future, so I'm wondering... |
**Mikeal Rogers:** I would push back a little bit on that. For instance, NodeBots were new and novel in 2012. I think I had talks at NodeConf on it and people were like "What?! Robots in JavaScript?" |
**Alex Sexton:** Yeah, for sure, but no one ships a NodeBot as their production robot. |
**Mikeal Rogers:** \[00:36:01.19\\\] Yes, they do... Yes they do. Skycatch is shipping basically a drone for commercial construction sites, that are now taking over all of Japanese construction, and building 3D maps of things... Everything on the bot is JavaScript and NodeBot, and in their cloud, in their frontend - it... |
**Alex Sexton:** I guess what I really mean is it's not the normal case. Certainly, there are people doing WebRTC and stuff like that; even Google does parts of this stuff for Google Hangouts, but I guess I'm still waiting for the future where this is how you would build it... I don't know. That was more my sentiment t... |
**Rachel White:** The more people that are trying to do stuff with it is what it's gonna take in order to get it to something where it's more used practically. |
**Alex Sexton:** Yeah... I guess my point isn't "We should all stop working on these things, because they're pointless, they're not gonna catch on." \[laughter\] That wasn't my -- the thing that is sounded like. I'm just wondering what we can do to make it the way people think map onto these types of ideas in a way tha... |
**Rachel White:** I have a question for both of you then... Is there anything that is currently out today that uses the kind of P2P stuff that we just went over, that you're excited about and see potential with it? |
**Mikeal Rogers:** Yeah, I'm trying to look at the name of it now, but I don't recall the name of the site. |
**Alex Sexton:** ChatRelay? |
**Mikeal Rogers:** No, no... So to back up with what Alex said, there's definitely been a longer timeline with WebRTC... We've been seeing demos probably longer than NodeBots maybe, and still haven't really seen a lot of big production stuff, and it seems like kind of never-ending. I think part of that is that the way ... |
If you look at what I've had to do with Roll Call - for those of you that haven't seen it before, RollCall.audio... It's just like a little audio-only web caller. Because it's audio-only, I can take that audio that's coming out and pass it through a bunch of the other web audio APIs in order to actually do all of the s... |
Additionally, even with just audio calls - that app is never gonna work for more than nine people, because then you need to create super nodes and proxy audio data, and you don't have access to do that at the level that you need to. However, I will say that the data channel, which is actually newer than any of the audi... |
**Alex Sexton:** The failure mode there, just to be clear, is that if you have nine people you're talking to, you have to make a connection with each of them, and each person has to send every single person their audio individually. So you're uploading eight versions of your audio. Everyone is uploading eight versions ... |
**Mikeal Rogers:** Exactly. And also, everybody in the network has to be directly connected to everybody else in the network, whereas if you have, say, a data network, you could have thousands of nodes with different nodes connecting them in between, and the data flowing between them. But you don't really have access t... |
\[00:40:04.19\\\] Anyway, so WebTorrent is quite a bit further along; it's using the data channel, and there is -- I can't find the name of it, but there's a YouTube(ish) competitor that is using WebTorrent. This is actually kind of cool - if you go back to... |
**Alex Sexton:** Did you laugh at the idea of a YouTube competitor? Was that that little giggle? |
**Mikeal Rogers:** \[laughs\] Yeah, yeah. I'm somewhat skeptical of a YouTube Competitor. Popcorn Time - I don't know if you can still download it off of random, weird website or whatever, but Popcorn Time is basically like this Apple TV-like application that is just beautiful. It will go out to P2P networks, find all ... |
A lot of the people that made Popcorn Time are in hiding or arrested or something for various... He actually only worked on the lower level libraries, so that never happened and he's a free man. But he is doing amazing P2P work still. The big innovation that he figured out is that "Okay, we can actually in the BitTorre... |
Now that's been adapted to WebTorrent, and you can just use that in your browser now. With this YouTube competitor, that's how it's working. WebTorrent allows you to add a web URL basically as a sort of peer of last resort; so if you can't get any data on the P2P network, like the file's not alive, it will just fall ba... |
So you can actually have a really decent experience for watching videos on the web entirely over P2P networks. And the cool thing for the creators of this service is that their bandwidth bill is gonna be a fraction of what YouTube's bandwidth bill is, right? |
**Alex Sexton:** Yeah, but everyone else's bandwidth bill is going to be a large fraction higher than they're used to. |
**Mikeal Rogers:** No it won't, because you don't pay for bandwidth that way, as a consumer. |
**Alex Sexton:** You don't... I think a lot of people do. People definitely have caps. Most people these days - I think this is an accurate statement - most people watch videos on their phone. |
**Rachel White:** Yeah, definitely. |
**Mikeal Rogers:** Yeah, yeah... So I will also say that I don't know where this ended up and I have no kind of insider knowledge on this, but Netflix did prominently promote a position of WebTorrent expert that they wanted to hire; so Netflix was really trying to bring in somebody that knew WebTorrent, that they could... |
**Alex Sexton:** Yeah, I'd be interested -- like, "If you click that button, you get a $2 off/ month", something like that. What would the incentive be for someone to be like... This could potentially be flaky every once in a while if there's a bad Node. It's not gonna be as solid for a while, right? |
**Mikeal Rogers:** \[00:43:58.00\\\] I'm more interested in just like "How do we make it more solid?" There's a couple of techniques that you can use to make it more solid. |
**Alex Sexton:** Before that -- the point I guess I'm making is you agree to stream it to other people who you don't know, you're agreeing to be a part of that network. Like, "I get $2 for being part of the network, versus just being a consumer of the network." Something like that could be cool. Anyways, go ahead. |
**Mikeal Rogers:** I think for a lot of these though, you're only really sharing it while you're watching it; you're not still uploading in the background, or something. That's just not an experience that people are used to. |
I think there's a couple ways that you can overcome the performance issues. If you can establish the network that you're gonna get the content from before you need the content, then you're actually gonna be faster than traditional websites, because you're already gonna actually have a pretty good understanding of where... |
**Alex Sexton:** Sure, but that's the exact opposite of what you just said though, where you said you're never gonna be sharing something whenever you're not using... |
**Mikeal Rogers:** You're right, I'm just trying to iterate over them. The other is literally just to say "Okay, we want a fast start to this." Netflix already definitely knows how much of the file they need to get in order to ensure that it will play immediately and keep playing without buffering, right? So let's get ... |
**Alex Sexton:** Sure, or even like... Netflix has quite a bit of data and are very good at knowing what you're gonna watch next, probably with 70% accuracy, or something like that. |
**Mikeal Rogers:** That's really smart... Oh my god, that's brilliant. |
**Alex Sexton:** And so they could still even use the P2P to quickly download the first two minutes of the top 22 things that you're gonna probably watch next. |
**Mikeal Rogers:** That's scary. That would be really, really fast. That's a really good idea. |
**Alex Sexton:** I mean, they could already do that, for what it's worth... They don't need the P2P stuff, they could just... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.