Nothing wrong with p2p. But often it's simply overkill. Skype started out as a p2p thing. But ultimately, video calls became a commodity where orchestrating them centrally proved to be good enough. I can make video calls with meets, facebook, whatsapp, signal, slack, etc. Not a big deal. I even still have Skype somewhere and it still works even though they ripped out the p2p bits ages ago.
I'm not sure what web3 is other than a label that people started slapping on the collective efforts of people vaguely doing anything with or depending on some kind of block chain thingy. IMHO it has very little to do with the web and little or no relevance to end users.
From a technical point of view, I don't see anything wrong with having a distributed store of ownership of stuff. There are all sorts of useful and valid uses for that. The problem is all the hyena's with dollar signs in their eyes floating around any companies active in this space. It's hard to see the forest for the trees because of this. Doing anything with a blockchain almost automatically buckets you with the fraudsters, utopians, and other people plaguing this space.
Most of the companies in this space are a combination of naive, fraudulent, or misguided. With a few notable exceptions where you might squint and see some potential of some actual economic value. But most of them are obvious duds. I actually briefly used Stellar for some actual applications. Great tech. Mostly works as advertised. Not that expensive to use. Etc.
Nothing wrong with p2p. But often it's simply overkill.
P2P may have a comeback, because home connections are becoming more symmetrical. DSL usually had relatively low uplink rates, and COMCAST cable was worse. But fiber is usually symmetrical, and 5G wireless typically has an uplink bandwidth of half the downlink rate. This makes P2P more feasible.
I mean how could it happen that such a basic utility as the router was developed with such fundamental design flaws that hosting a webserver at home is above the skills of 99% of people?
There’s a router in every home anywhere now, so this is a fact we’ll have to deal with for decades.
I’m pretty young so i was not around at the time. How did it come about?
It was NAT. We ran out of v4 addresses and self hosting / p2p became much harder. And now many iot devices depend on Nat existing as some kind of security measure.
But largely no one cares because you wouldn’t want to host any real website at home anyway. It’s extremely cheap and reliable to just rent a VPS.
So what happened with IPv6 then? Seems to me that that we solved the v4 address exhaustion problem and then just… didn’t use it? IPv6 support is mostly just a greyed out option on a buried configuration menu as far as I can tell.
I would argue many aspects of the protocol’s design made it difficult to implement, for both vendors and networks. A simple “same protocol, larger address field” and more focus on backwards compatibility probably would have fared better.
Either way it doesn’t completely solve the problem mentioned above. You need a firewall to block inbound connections to things on your LAN. So to do p2p you need a way to add rules to allow the traffic you want in. So it either remains too complex for the average person to do manually, or you run something like uPNP. But that can have its own security considerations.
v6 does remove some of the complexities for trying to run services from private address space behind NAT though. So it’s an improvement.
Reliability could be improved if all required functionality is P2P, as the program would not need to rely on a third party server. Of course, this assumes that the party/parties on the other end of the P2P connection are online, which is the downside of all P2P services.
Another underrated benefit of P2P (if implemented in good faith) is that the users will have peace of mind with regards to the ownership of their data.
> if implemented in good faith ... users will have peace of mind with regards to the ownership of their data
Famous last words! Not sure how users will establish the trustworthiness of P2P protocols (and there will be many competing versions), when now they cannot determine the trustworthiness of one service.
Messaging, chat, video calls, and gaming without the need for a centralized service. All you need is a directory so you can find the other party. Like Skype before Microsoft.
i do think this trend is happening and will unlock some use cases. Not sure about P2P though. One reason I switched to home 5G is because i can get 40 mbps up for $25, where I needed to pay $100 to comcast for similar up speed. My main use case is streaming my playstation 4 remotely so I can play on the go.
I am using tmobile's Home 5G right now. I was waiting for Verizon for the mmWave 5G because I think the latency will be lower. If Verizon comes around soon, I'll switch. I have both Verizon and Tmobile phone service so I can get the bundle discounts with home through either. I think both of them are ~$50/mo standalone or $25/mo with a 5G smartphone plan.
Skype became centralized about the time Microsoft joined the PRISM spying program. I don't think it becoming centralized had anything to do with p2p being a problem...
I think this conflates a bit p2p as a technical means and p2p as a means to organise control.
Like, there are all kinds of products which use p2p architectures internally to solve some technical problem - bit torrent is sometimes used inside updaters, elasticsearch organizes its nodes as a p2p network for greater robustness and scalability, WebRTC let's you make direct connections between browsers to save roundtrips, etc etc. But those are just technical details - the products themselves are still controlled and administered by a central entity.
In contrast, crypto, file sharing and federated systems use p2p as a technical means but they are also administered in a decentralised fashion, where no person or entity has ultimate control (or should have at least).
I think Skype was falling squarely in the first category here: They were experimenting with decentralised network protocols, but the nodes were all pretty clearly belonging to a singular and centrally controlled product. That's why p2p really just was an implementation detail that could be replaced by a different technical architecture by a simple update.
(Weird thought as an aside: I think most companies or organisations follow a sort of similar pattern: They are communist with respect to coworkers of equal status ("everyone should chip in for Carol's birthday present"), authoritarian with respect to management and owners - and libertarian/anarchist with respect to the outside world ("freedom! ...from annoying rules and regulations..."). But that doesn't mean leadership necessarily follows communist, authoritarian or anarchist principles when deciding strategy)
Skype’s quality, and subsequently popularity, dropped significantly after they removed the P2P bits.
There’s still nothing that comes close to the call quality Skype provided for long distance calls, especially when the network may not be as good on 1 end.
> Skype started out as a p2p thing. But ultimately, video calls became a commodity where orchestrating them centrally proved to be good enough.
It's my understanding that "phone" without the ability to tap is illegal in the US.
IIRC, the "must be able to let the govt listen" requirement occurred at the same time as Skpye switched models and the requirement was given as the reason for said switch.
> It's my understanding that "phone" without the ability to tap is illegal in the US.
This is not true. Else you couldn't have whatsapp or signal calls (e2e encrypted). In 2009, Skype made the argument with the ACLU that they are not a telecoms carrier, and are therefore exempt from CALEA and similar regulations. Additionally, Skype's original CSEO refused to promise that they did not support wiretaps in the original architecture. We know from Edward Snowden that "old Skype" was included in NSA data sources.
Skype changed architecture because routing calls through neighboring, always-on nodes is not practical in a mobile world where 50%+ of users pay for their data by the MB, and have spotty connections with high latency. For several years before the rearchitecture, Microsoft propped up the network by running all the supernodes itself.
Skype is not Common Carrier. It's an Internet Service which ISP's have been deadset on avoiding getting classified as covered under Common Carrier obligations because it would entail, according to them, enormous harm to the profitability of the sector due to the cost of increased regulatory compliance".
Telco's/Telephony must provide pen register/tap capability. Internet services as of yet need not. Until you become big enough for Federal LE to start harassing.
It might be some sort of opaque incentive where gov goons tell you they’ll make life difficult for you if you don’t comply, but I certainly have not ever seen this being an open requirement. Signal still exists, for example.
The article doesn't imply (let alone say) that there's anything wrong with p2p?
To the extent that it says anything positive about p2p, it's rather positive. Unless you're one of those americans for whom "communist" reflexively aliases to "bad" I guess?
>> Nothing wrong with p2p. But often it's simply overkill.
> you're one of those americans for whom "communist" ...
The GP is making some comment about the technical merits of P2P which segues naturally from the article. You can agree, you can disagree, but I really don't know what just did.
Collectivism as a means of ordering a society necessarily tramples individual liberty, which yes, generally speaking, is bad. Communism is a subset of collectivism, just one of the flavors we've seen, so it is a collection of ideas that may be judged by their outcomes as bad.
I'm not sure what web3 is other than a label that people started slapping on the collective efforts of people vaguely doing anything with or depending on some kind of block chain thingy. IMHO it has very little to do with the web and little or no relevance to end users.
From a technical point of view, I don't see anything wrong with having a distributed store of ownership of stuff. There are all sorts of useful and valid uses for that. The problem is all the hyena's with dollar signs in their eyes floating around any companies active in this space. It's hard to see the forest for the trees because of this. Doing anything with a blockchain almost automatically buckets you with the fraudsters, utopians, and other people plaguing this space.
Most of the companies in this space are a combination of naive, fraudulent, or misguided. With a few notable exceptions where you might squint and see some potential of some actual economic value. But most of them are obvious duds. I actually briefly used Stellar for some actual applications. Great tech. Mostly works as advertised. Not that expensive to use. Etc.