5 ms·

I am seriously confused. Internet is already decentralized for the most part. I run my own web server and mail server from my home. There's my "piece of the

I am seriously confused. Internet is already decentralized for the most part. I run my own web server and mail server from my home. There's my "piece of the Internet". Sure, DNS is centralized (even though it's heavily federated), and TLS certificates are centralized (less federated). So, what we are looking to replace is simply "who is there?" and "is it the real you?" services with something less... centralized. This can live on the blockchain, sure, but blockchain is a database you sync across in its entirety, right? I am sure you can only get parts of it for some applications, but this, by definition, does not scale to the size of the Internet. (Hum, this is probably a decent dapp proposal: domain name system with certs, but it'd still grow very large if it was to cover entire Internet) I don't get the connection with Google/Amazon/Facebook either: they are humongous services operating on the decentralized Internet today. Yes, people flock to them, but that's exactly the Internet you can avoid (a recent discussion on search.marginalia.nu gives a pointer as to how).

necovek 5y ago
> domain name system And it's already there it seems: https://ens.domains/ https://ens.domains/ I wonder how does one integrate ENS into an existing DNS-based system — is there a DNS-compatible resolver that will look up names on ENS (and an IPv4/IPv6 proxy perhaps)? It's also a missed opportunity to provide hierarchical certificates at the same time, signed by the registry, though in a sense, a smart contract is exactly that, but I am more interested in simple name-resolution and TLS communication for traditional web applications like HTTP and mail (name-to-IP).
throwing_away 5y ago
We've done lots of federations over the years, and the internet was absolutely built with some degree of decentralization/federation in mind. The radical change here is that now anyone with some pocket change in crypto can deploy a decentralized application. We don't know how that's going to be used yet, it's still so early, but it's certainly exciting. Web3 is a terrible name (as everyone else has already pointed out), but the magic that shouldn't get lost here is that a new decentralized internet is being born.
- rapnie 5y ago
  I find that most mentions of Web3 and DWeb relate to something with blockchain in it somewhere, which I am not interested in atm. So to me encountering these terms signifies "Probably not worth checking out". But YMMV and blockchain fans should probably be triggered by them.
- giantrobot 5y ago
  > The radical change here is that now anyone with some pocket change in crypto can deploy a decentralized application. Oh man it'll be awesome. Decentralized applications running on someone else's hardware with no SLA or even a meaningful relationship between the application author and person(a) executing it. Terabytes of data will just sort of float on the blockchain...sorry, immutable data store. I'm sure it'll be super fast when uploading from people's residential Internet connections. So exciting!
  - meheleventyone 5y ago
    And lets not forget the high cost of executing these apps not least to the environment.
- Zababa 5y ago
  > The radical change here is that now anyone with some pocket change in crypto can deploy a decentralized application. Genuine question, how is that different from renting a VPS or buying a Raspberry Pi and running your website on it?
- thesuperbigfrog 5y ago
  >> The radical change here is that now anyone with some pocket change in crypto can deploy a decentralized application. We don't know how that's going to be used yet, it's still so early, but it's certainly exciting. What about decentralized malware? Web3 seems like a perfect opportunity for malware applications and hosting illicit content. Have some CSAM you want to host? Put it on Web3! Want to dox someone or leak sensitive info captured in a hack? Put it on Web3! Edit: This also begs the question: what legal liability might one face for hosting illegal content on Web3?
frabcus 5y ago
It's only decentralised at the web layer - we basically stopped inventing shared protocols after that point in the 90s. Email interoperates, but messaging doesn't. There's no adopted standard for identity (except centralised Google/Apple), or for payment (except centralised PayPal/Apple, and the bad UX of entering credit cards). Turning these (and more) into protocols would allow more innovation at a layer above that. This innovation just can't happen right now, as FAANG have to approve of everything you do that works with systems the populuation already use. web3 doesn't seem good enough yet to help, but it is at least working on it. And it could potentially solve the lack of identity and payment standards.
- HatchedLake721 5y ago
  Standard for messaging - XMPP Standard for identity - not sure what you exactly mean by this, but there’s WebAuthn / OpenID Connect / SAML Standard for payments - W3C web payments
- Zababa 5y ago
  I'm not sure how having a standard for identity would help decentralization. Each website having their own identity managment allows you to have different identities on each website.
- acdha 5y ago
  > we basically stopped inventing shared protocols after that point in the 90s. This is incorrect but educationally so: there are many new distributed protocols (RSS/Atom, XMPP, various things around blog interactions like FOAF and ping backs, GraphQL, OpenID, etc.) but the ones which successfully saw adoption were adopted by a few large companies who saw advantages from doing so. Being open or distributed not only wasn’t enough to be competitive with the centralized platforms, it often prevented that since you needed to coordinate improvements across a range of competitors while Google or Facebook could just ship an update. Consumers didn’t value client decentralization, especially when that meant incompatibility, lower performance, and fewer people they know using a service. Putting things into a blockchain not only doesn’t help with any of the real problems but makes all of them worse: performance and reliability are completely out of your control, everything costs money (good luck getting users when the competition not only has better features, perform, and social graph but also doesn’t charge them to sign up or send messages!), and since you centralized onto a network run by finance companies the only way to get traction on any of that is to have enough money to be taken seriously.
pornel 5y ago
Blockchain-based identity revolves around a user having a private key, which doesn't even need the blockchain to work. We've had TLS client certs (with awful UX, but blockchain doesn't magically fix UX either). We've decentralised transfers with BitTorrent pretty well, and it's actually reasonably fast and cheap to operate. Opera used to have it built-in, and there are some WebRTC versions. The web has a decentralised database of domain ownership - Certificate Transparency. The same approach could have been done for NFTs if they were really about owning digital assets, and not gambling on a pump and dump scheme. Basically there are lots of energy efficient and lower-latency ways to decentralise the web if we wanted to. But the web tends to clump around large services mostly due to economies of scale and user convenience (average user can't be told they've lost their private key and they're totally screwed).
tapoxi 5y ago
It’s not trying to solve an actual user problem, it’s trying to solve a blockchain investor problem because right now their investments are speculative without any real-world use.
- thesuperbigfrog 5y ago
  Exactly. Making money on blockchain requires retrofitting the web to use blockchain to do what the web can mostly do already. In "Blockchains Are a Bad Idea", James Mickens gives several arguments against blockchain-based systems like Bitcoin: See his presentation here: https://youtu.be/15RTC22Z2xI https://youtu.be/15RTC22Z2xI 1. People have out-of-band trust relationships in real life which reduce the likelihood of malice. Bitcoin-style anonymous identities undermine trust relationships and are not needed for legitimate (not illegal) transactions. 2. Real life legal systems encourage good behaviors. If you have a dispute with someone, you can sue them. Bitcoin and related systems lack these protections. 3. Existing tools such as public-key cryptography and digital signatures can provide most of the functionality that applications need without the problems that blockchain-based systems have.
  - rfd4sgmk8u 5y ago
    I think you are missing the forest for the trees. 1. Zero trust is highly optimal for business. When it is systemically impossible to cheat your vendor/customer, nobody even tries. Happy path only. 2. Good luck suing someone in a foreign jurisdiction with incompatible legal systems. Maybe that sex toy would be embarrassing on ones credit card and tagged to ones identity. Cryptocurrency (especially private cryptocurrencies like Zcash and Monero) offers an alternative to doxing yourself. 3. You miss the great innovation of SN: Preventing Double Spends. You cannot enforce double spend without a source of truth -- which was traditionally central server/banks. It was the Proof of Work that made the ledger distributed and censorship resistant, and it is the ledger that prevents double spends. And with Bitcoin, you have a limit of 21 million coins, which addresses SN's concern that central bankers ALWAYS inflate currency. So, no, existing tools did not suffice. That is why we have Bitcoin today.
    - 5y ago
dmak 5y ago
If you run your own web server then that’s still centralized. If you run your own web server that runs code which abides to the protocol rules like miner software then your server is contributing to a decentralized network.
grognak420 5y ago
You should take a look at handshake. A lot of GitHub users can claim using their ssh or PGP keys. https://handshake.org/ https://handshake.org/

necovek 5y ago

> domain name system And it's already there it seems: https://ens.domains/ https://ens.domains/ I wonder how does one integrate ENS into an existing DNS-based system — is there a DNS-compatible resolver that will look up names on ENS (and an IPv4/IPv6 proxy perhaps)? It's also a missed opportunity to provide hierarchical certificates at the same time, signed by the registry, though in a sense, a smart contract is exactly that, but I am more interested in simple name-resolution and TLS communication for traditional web applications like HTTP and mail (name-to-IP).

throwing_away 5y ago

We've done lots of federations over the years, and the internet was absolutely built with some degree of decentralization/federation in mind. The radical change here is that now anyone with some pocket change in crypto can deploy a decentralized application. We don't know how that's going to be used yet, it's still so early, but it's certainly exciting. Web3 is a terrible name (as everyone else has already pointed out), but the magic that shouldn't get lost here is that a new decentralized internet is being born.

rapnie 5y ago
I find that most mentions of Web3 and DWeb relate to something with blockchain in it somewhere, which I am not interested in atm. So to me encountering these terms signifies "Probably not worth checking out". But YMMV and blockchain fans should probably be triggered by them.
giantrobot 5y ago
> The radical change here is that now anyone with some pocket change in crypto can deploy a decentralized application. Oh man it'll be awesome. Decentralized applications running on someone else's hardware with no SLA or even a meaningful relationship between the application author and person(a) executing it. Terabytes of data will just sort of float on the blockchain...sorry, immutable data store. I'm sure it'll be super fast when uploading from people's residential Internet connections. So exciting!
- meheleventyone 5y ago
  And lets not forget the high cost of executing these apps not least to the environment.
Zababa 5y ago
> The radical change here is that now anyone with some pocket change in crypto can deploy a decentralized application. Genuine question, how is that different from renting a VPS or buying a Raspberry Pi and running your website on it?
thesuperbigfrog 5y ago
>> The radical change here is that now anyone with some pocket change in crypto can deploy a decentralized application. We don't know how that's going to be used yet, it's still so early, but it's certainly exciting. What about decentralized malware? Web3 seems like a perfect opportunity for malware applications and hosting illicit content. Have some CSAM you want to host? Put it on Web3! Want to dox someone or leak sensitive info captured in a hack? Put it on Web3! Edit: This also begs the question: what legal liability might one face for hosting illegal content on Web3?

frabcus 5y ago

It's only decentralised at the web layer - we basically stopped inventing shared protocols after that point in the 90s. Email interoperates, but messaging doesn't. There's no adopted standard for identity (except centralised Google/Apple), or for payment (except centralised PayPal/Apple, and the bad UX of entering credit cards). Turning these (and more) into protocols would allow more innovation at a layer above that. This innovation just can't happen right now, as FAANG have to approve of everything you do that works with systems the populuation already use. web3 doesn't seem good enough yet to help, but it is at least working on it. And it could potentially solve the lack of identity and payment standards.

HatchedLake721 5y ago
Standard for messaging - XMPP Standard for identity - not sure what you exactly mean by this, but there’s WebAuthn / OpenID Connect / SAML Standard for payments - W3C web payments
Zababa 5y ago
I'm not sure how having a standard for identity would help decentralization. Each website having their own identity managment allows you to have different identities on each website.
acdha 5y ago
> we basically stopped inventing shared protocols after that point in the 90s. This is incorrect but educationally so: there are many new distributed protocols (RSS/Atom, XMPP, various things around blog interactions like FOAF and ping backs, GraphQL, OpenID, etc.) but the ones which successfully saw adoption were adopted by a few large companies who saw advantages from doing so. Being open or distributed not only wasn’t enough to be competitive with the centralized platforms, it often prevented that since you needed to coordinate improvements across a range of competitors while Google or Facebook could just ship an update. Consumers didn’t value client decentralization, especially when that meant incompatibility, lower performance, and fewer people they know using a service. Putting things into a blockchain not only doesn’t help with any of the real problems but makes all of them worse: performance and reliability are completely out of your control, everything costs money (good luck getting users when the competition not only has better features, perform, and social graph but also doesn’t charge them to sign up or send messages!), and since you centralized onto a network run by finance companies the only way to get traction on any of that is to have enough money to be taken seriously.

pornel 5y ago

Blockchain-based identity revolves around a user having a private key, which doesn't even need the blockchain to work. We've had TLS client certs (with awful UX, but blockchain doesn't magically fix UX either). We've decentralised transfers with BitTorrent pretty well, and it's actually reasonably fast and cheap to operate. Opera used to have it built-in, and there are some WebRTC versions. The web has a decentralised database of domain ownership - Certificate Transparency. The same approach could have been done for NFTs if they were really about owning digital assets, and not gambling on a pump and dump scheme. Basically there are lots of energy efficient and lower-latency ways to decentralise the web if we wanted to. But the web tends to clump around large services mostly due to economies of scale and user convenience (average user can't be told they've lost their private key and they're totally screwed).

tapoxi 5y ago

It’s not trying to solve an actual user problem, it’s trying to solve a blockchain investor problem because right now their investments are speculative without any real-world use.

thesuperbigfrog 5y ago
Exactly. Making money on blockchain requires retrofitting the web to use blockchain to do what the web can mostly do already. In "Blockchains Are a Bad Idea", James Mickens gives several arguments against blockchain-based systems like Bitcoin: See his presentation here: https://youtu.be/15RTC22Z2xI https://youtu.be/15RTC22Z2xI 1. People have out-of-band trust relationships in real life which reduce the likelihood of malice. Bitcoin-style anonymous identities undermine trust relationships and are not needed for legitimate (not illegal) transactions. 2. Real life legal systems encourage good behaviors. If you have a dispute with someone, you can sue them. Bitcoin and related systems lack these protections. 3. Existing tools such as public-key cryptography and digital signatures can provide most of the functionality that applications need without the problems that blockchain-based systems have.
- rfd4sgmk8u 5y ago
  I think you are missing the forest for the trees. 1. Zero trust is highly optimal for business. When it is systemically impossible to cheat your vendor/customer, nobody even tries. Happy path only. 2. Good luck suing someone in a foreign jurisdiction with incompatible legal systems. Maybe that sex toy would be embarrassing on ones credit card and tagged to ones identity. Cryptocurrency (especially private cryptocurrencies like Zcash and Monero) offers an alternative to doxing yourself. 3. You miss the great innovation of SN: Preventing Double Spends. You cannot enforce double spend without a source of truth -- which was traditionally central server/banks. It was the Proof of Work that made the ledger distributed and censorship resistant, and it is the ledger that prevents double spends. And with Bitcoin, you have a limit of 21 million coins, which addresses SN's concern that central bankers ALWAYS inflate currency. So, no, existing tools did not suffice. That is why we have Bitcoin today.
  - 5y ago

dmak 5y ago

If you run your own web server then that’s still centralized. If you run your own web server that runs code which abides to the protocol rules like miner software then your server is contributing to a decentralized network.

grognak420 5y ago

You should take a look at handshake. A lot of GitHub users can claim using their ssh or PGP keys. https://handshake.org/ https://handshake.org/