Hacker News new | past | comments | ask | show | jobs | submit login
Peer-to-peer serverless infrastructure, powered by blockchain technology (edge.network)
34 points by bigjump 10 days ago | hide | past | favorite | 40 comments





So this appears to be a distributed network of random people's devices, and the 'blockchain' aspect is entirely unrelated to the actually useful parts of how the tech works. It's just how people are paid for contributing their devices to the network?

Am I missing anything? Are the non-blockchain parts of this actually interesting?


You've pretty much described what a blockchain is aside from the massive crypto bubble. The tokens are fees people pay for a computation and rewarded to those devices contributing to the network to offset electrical and physical costs.

You lost me at "blockchain".

From the title alone, I thought, hey maybe they use git, which is peer-to-peer, serverless (if you want to) and powered by 'blockchain technology'...

At least, they didn't mention ML.

In late 2019 (last time I physically went to a tech conference/trade show; in this case ELCE), I actually saw a company seriously advertising "AIoT Blockchain".

I have no clue what the company really did, but I clearly remember the sales person there showing me a flow chart printed on a pillow.


> I clearly remember the sales person there showing me a flow chart printed on a pillow.

wait, what?


ML in the Blockchain!!!

"How It Works" link doesn't work

Thereby all relevant information has been conveyed.

Lol

I was curious about their claims regarding confidential computing so I checked out their Github organization (https://github.com/edge), but it seems there's very little activity there and there's not a single reference to Intel SGX (the enclave they claim to use). So I'm a bit skeptical about their claims. SGX is also not a general-purpose technology and can only run highly specialized code, to my knowledge there's currently no way to run general-purpose code that e.g. uses syscalls (there are some projects that try to fix this though).

> ...there's not a single reference to Intel SGX (the enclave they claim to use).

Unrelated, but amusingly, if they want to, they could use https://secret.network/ to host their workload in peer-to-peer serverless SGX enclaves, powered by the blockchain.


Mea culpa: Looks like secret.network is indeed a joke website... https://scrt.network/ is what I meant to share.

Most blockchain projects are just an excuse to list a related crypto. It doesn’t even appear that the underlying distributed computer network has anything to do with blockchain.

> It doesn’t even appear that the underlying distributed computer network has anything to do with blockchain.

I think the blockchain acts a transparency ledger among a distributed set of distrusting participants. Think: if PirateBay could pay you to seed magnet-links all the while knowing for sure you've not tampered with the the amount of bandwidth you've claimed to have lent, and the network then rewarding (paying) you the equivalent amount that you're owed for your troubles. "Blockchain" enables that.

The ICOs are akin to companies filing for IPOs. The ICOs don't strictly need speculation (because the utility of their network ideally would have an inherent, demonstrable value that dictates the price of the network's coin), but they do need the market to trade their coins, because that's how anyone participating in the network would be able to convert their rewards into fiat. (note: at this point, we're essentially reinventing Ehtereum, the super computer, as originally envisioned...)

You could aruge that many of these networks could avoid doing ICOs and simply use one of the existing cross-chain coins from Cosmos, Algorand, Polkadot (or even stable coins like USDC, Diem, Celo, BUSD, DAI etc), but it is the summer of crypto, and everyone's got to do an ICO.


Looking at the pricing page, this seems to be almost as expensive as AWS/GCP, while having unknown performance characteristics of random people's devices and network connections. Not sure why I would pay for this instead...

> "Looking at the pricing page, this seems to be almost as expensive as AWS/GCP, while having unknown performance characteristics of random people's devices and network connections"

What happened when utility tokens first appeared in 2017-2018 is that speculators drove the price of the tokens way above the value of the services they purported to offer, so even if they did offer a genuinely useful service (which most/all didn't) then it simply wouldn't have been cost effective to use them for their intended purpose.


Because it's blockchain and distributed!

This is like pied piper from Silicon Valley series!

The new internet.


Just not as funny.

Wait till son of anton starts replicating your data to all the devices on the world.

One thing I always look for in companies in the block chain space is the About section. I want to see real names and faces on these projects. The more information the better. Sometimes they just might have a front-man like a professor that wrote a white paper, but I want to see the people actually involved. This site has no about.

To confuse things further, there is another company called edge.app which is an app to buy and sell crypto. I don't think they're affiliated


From my rudimentary understanding (i.e., what I've just read on the homepage) I don't know how "blockchain" fits into this, but I could make a guess that api signatures are "blockchain verified"?

From that same rudimentary understanding, I would also expect performance to be comparable to Tor.


> Anyone with a laptop, mobile or any connected device can earn $EDGE tokens by providing spare computational and storage capacity to the network. $EDGE tokens can be traded for Edge services, or converted into dollars, pounds or euros.

Seems like the product is mostly an excuse to push yet another cryptocoin scheme


Ugh.

I assume it's used to pay people for offering their computational power onto the network

I have been tracking new heroku style deployments tools, CDNs and ADN.

They are part of this new trend of platform calling themselves Application Delivery Network, wherein stateless (either containerized or microvm-ed) services are deployed closed to consumer.

That spaces is getting hyper-competitive and crowded. These guys are trying to differentiate themselves by calling themselves blockchain-powered network. To be honest, you do not have to be blockchain powered, it might just work against you as you might have to cross certain thresholds to serve content reliably. I rather prefer services and content delivered by datacenters that are certified and has some security standards than random devices in the network, unless they can somehow guarantee some standards on the devices.


Some day, people will start treating their compute, everywhere as rentable commodity, and the price will go to almost zero. CPU + storage + network costs will go to zero, the "edge" will be everywhere, and organizing hosting and launching websites will never be the same. Maybe we'll even get a better federated services out of all of it (if you get people comfortable with leasing their capital, maybe they get comfortable with running services for themselves with that capital other times).

This isn't it for me (blockchain is an anchor around the neck of the idea IMO), but I look forward to the day that this idea finds it's perfect (or perfectly timed) execution.


I wouldn't expect network costs to go to zero. If computing gets distributed we'll only get more bandwidth usage - especially in P2P gossiping situations. Having a few services using that will add non-trivial load to existing infrastructure which really likes the existing CDN pop solution.

> I wouldn't expect network costs to go to zero. If computing gets distributed we'll only get more bandwidth usage - especially in P2P gossiping situations. Having a few services using that will add non-trivial load to existing infrastructure which really likes the existing CDN pop solution.

But that bandwidth usage will localize more and more and may start going over really non-traditional mediums that weren't fully integrated into the "network mesh" before. If there are 200 people in a shopping mall, with 50 of them with some LocalCDNEdge app enabled and acting as beacons -- the request may never leave the shopping mall and return data that is fresh-enough (depending on the use case of course). Yeah there's now kind of a P2P gossip storm going inside the shopping mall, but I don't think that can't be solved with some more intelligent algorithms and at the worst case, just another wifi-router-as-pop.

I mean unless I'm thinking about this wrong -- what I'm envisioning is the CDN pop solution, just with WAY more pops. Yeah you have to coordinate the pops, but delivering the same payload to two people who are right next to each other (same apartment building let's say) from the far away CDN pop is less efficient than delivering it to one of the people, and P2P for the last mile so to speak.

Also, I think I can lean on innovation emerging in the network space for how to manage this kind of thing -- "Edge" is a buzzword now but I think there's a more innovation to be had.


> If there are 200 people in a shopping mall, with 50 of them with some LocalCDNEdge app enabled and acting as beacons

Given that we still don't have phones that last more than 12h of serious usage and a big issue with xmpp was that it drained battery by not using push... Local P2P on mobile phones is really not something I expect people to use. Android went fancy with opportunistic deep sleep not that many years ago - this would undo all that effort.


Just for the record, mobile XMPP clients pretty much all support push notifications these days. But you're right, this is possible because XMPP isn't pure p2p.

I'm similarly unconvinced of the suitability of pure p2p with (at least the current generation of) mobile devices. Maybe advances in bandwidth and battery technology in coming years will make such protocols more feasible for general use.


So IPFS with incentives to run a node?

One big issue I see with IPFS is that you or someone has to pin the content or it is gone for ever. This builds centralized services that offer pinning of your content.


I think that is what Filecoin [0] addresses.

[0] https://filecoin.io/


Beat me to it.

What's cool about IPFS is that anyone can pin the content and the user doesn't have to migrate anything. This makes hosting far more liquid, which should drive down prices.

The people behind IPFS developed FileCoin[0] to provide incentives.

[0] - https://filecoin.io/



Sounds like buzzword driven development



Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: