Hacker News Reader

Comment by dgudkov | Hacker News Reader

The fundamental power of the internet is its interoperability. It was born out of the ability of different networks to talk to each other using common protocols.

The interoperability is what we've lost in the Web 2.0 era. Even such quintessential thing as a web API has no well defined standard or protocol, just a very vague concept of REST or RPC.

I want Web 3.0 to get interoperability back. We badly need commonly accepted standards and decentralized protocols: for web APIs, for identity management, for message queuing, for web callbacks (webhooks), for online transactions, for semantic web and ontology, etc.

Take, for instance, web APIs. It's barely usable nowadays. Imagine if you had to write a special browser every time you need to connect to a new website. But this is what already happens with APIs. Accessing programmatically any web-service requires custom coding an API. Insane!

View on HN

> The interoperability is what we've lost in the Web 2.0 era

This is a huge claim that doesn't match reality. We still use browsers, HTTP clients still respect mediatypes, and websites still return HTML. We have even more standards now than we did at the beginning of Web 2.0, with things like microdata, opengraph, etc. Even non-HTML HTTP APIs are even more standard with things like grpc, json-api.org, and graphql.

Maybe since my job is to literally do this I have a different take.

I think there is some weight to the claim.

This is especially true for public user generated content. Any public content should be accessible without limits, but that is not the case with web2(twitter, public Instagram, forums etc).

ofcourse, there are some server costs but the primary reason the walled-gardens do this is more business related than to reduce costs.

There is zero weight to the claim. Not only does it ignore an massive amount of improvements since the early 90's, it seems to completely ignore what Web 2.0 was about.

> Any public content should be accessible without limits,

This is not related to the discussion, even if I did agree with it.

> but that is not the case with web2(twitter, public Instagram, forums etc).

Twitter and "forums" are all completely accessible via a browser, which is literally what the original claim suggests we've "lost". You don't need to read twitter with a special twitter client, you read it with the any HATEOAS client.

I urge developers to actually read some history.

The accessibility of public web is what web 3.0 is about however. With semantic web and stuff.

Twitter is accessible except "this nitter instance has been rate limited", and don't forget instagram...

> The accessibility of public web is what web 3.0 is about however.

The definition of web 3.0 changes depending on who I am talking to and how deep into crypto they are.

> semantic web

That's Web 2.0, it exists just fine now and hasn't gone away like the original poster claimed.

> Twitter is accessible

Okay, glad we agree on reality.

> "this nitter instance has been rate limited"

If you think there won't be rate limiting in whatever "web 3.0" distributed service is out there, I have a bridge to sell you.

> and don't forget instagram

Multiple "web 3.0" services that exist today have their content gatekeeped to those who have accounts.

if "web3" works anything like crypto, there wouldn't be any rate limiting as long as you have control over your own node(it might be more expensive / harder to set one up though).

In the same way torrents are not rate limited.

Web 3.0 refers to the semantic web specifically, nothing to do with crypto. You are confusing it with "web3". And no, web 2.0 is not the semantic web.

"Okay, glad we agree on reality."

We don't.

> ignore an massive amount of improvements since the early 90's

That's not the point. having massive improvements since 90s does not imply that we are going to continue from the same path from here on.

> completely accessible via a browser

accessible via a browser is not the same as accessible / interoperable broadly.

> You don't need to read twitter with a special twitter client

YES, some people want to. or maybe I want to do some data analytics, maybe I want to create a localized twitter, or I want to do take public social graphs and do something with it.

web2.0 was about standardizing transport level protocols, web3 is(imo) about standardizing application level protocols using commitment guarantees.

When I use some web2 service, there is almost no commitment guarantees(you can get censored, your data might be access gated or deleted, banks can stop you from doing certain transactions).

Crypto and web3 is about fixing these issues. If Bitcoin promises to work in a certain way today, it will more likely than not work in the same way 10 years from now.

>This is a huge claim that doesn't match reality.

You have a point that the interoperability that existed pre-Web 2.0 is still here and wasn't taken away (except for things like RSS).

My point is that we didn't move significantly ahead with interoperability and stuck with centralized proprietary services and niche protocols.

>grpc, json-api.org, and graphql.

You're totally missing the point. With an email client I can connect to any IMAP service (and there are tens of thousands of them). With an SFTP client I can connect to any SFTP server (and there are thousands of them) and download files. Again, without coding, just by configuring the connection.

What public service can I connect to with a GraphQL client (do they even exist?) without coding, just by configuring the connection settings like with an email client?

    What public service can I connect to with a GraphQL client (do they even exist?) without coding

GitHub? https://docs.github.com/en/graphql

Decentralisation is not interoperability.

As far as I understand Web3 is mostly presented with descentralisation in mind.

> We badly need commonly accepted standards and decentralized protocols

In my opinion "standard" and "descentralization" does not work good to support the same argument.

Here is one among many possible ways the future might unfold: a totally descentralized internet might be completely controller by few corporations. The reason is simple in my opinion: The normal user does not care about this. The normal user cares to talk with their family and their family are all using the web3-chat from BigOrg. Because BigOrg can give it for free and it works on all devices. Now as everything is descentralized BigOrg is also pushing a new browser. But this browser is showing only content from BigOrg through BigOrg descentralized servers. They have their own Web3-Smart-Blockchain-Ready-NoNSdeletion-DNS that is used only by their own browser that displays only websites that are registered there.

Now try to switch that normal user to another provider. That will be hard thing to do.

I think users will be _more_ captive than they are now as switching between descentralized services is hard. No business yet has an incentive to make a descentralized service and in the same time use a standard.

Of course I am missing maybe some web3 tech and I am in no way an expert in what web3 is or it will be.

> I want Web 3.0 to get interoperability back. We badly need commonly accepted standards and decentralized protocols

Check out Homescreen: https://homescreen.hns.siasky.net/

It's a decentralized front-end in a decentralized cloud for Web3.0 apps.

> Take, for instance, web APIs. It's barely usable nowadays.

Pretty much every app we use, system integration, anything really that has to communicate with third-parties, is driven by web APIs. If you think that they are barely usable, then I don't know what usable means to you.

> We badly need commonly accepted standards and decentralized protocols

Maybe , but there's a bunch of protocols today that make everything work quite nicely:

* authorization/delegation between third-parties: OAuth

* identity: OpenID Connect

* data exchange format: JSON and many other media-types like XML, protobuffers, messagepack etc.

* presentation layer format: HTML5

* data schemas: JSON-Schema, GraphQL, RDF...

Integrating decentralization stacks with these technologies is totally possible, and that's what, e.g. Solid is trying to do: https://solidproject.org/developers/tools/

The desire to "start from scratch" and invent new protocols for "web3" is understandable, as the OP mentions in the blog post... but that's highly unlikely to be the best way to go about it, and inventing new technologies and throwing away all of the experience we've gained over the first few decades of Internet history is bound to fail everytime, IMHO.

> Imagine if you had to write a special browser every time you need to connect to a new website. But this is what already happens with APIs.

Well, browsers only work because they are user agents. If your API required an user agent, it would not need a special client: just assume a user-agent is on the other side and it will work exactly was with any website. However, people seem to think it's possible to automate any API like that: that's patently false. Have you tried to use REST as it's actually meant to, with hypermedia-driven discovery, zero-knowledge clients which only need to be pointed at an URL and nothing else, to implement your common backend-to-backend interactions? Do you think it's even possible to do it?

I am currently working on something on those lines, but when you think a little bit about that, you quickly realize clients need to be written specifically to know what to do with the data they are given in a pretty hard-coded way. Until we have "intelligen" clients that can think like a human , only user-agents can benefit from hypermedia-driven flows, which explains why only browsers (and a few special-purpose user applications that are analogous to web browsers, but for very specific domains) are designed on those lines to this day.

Random question, someone mentioned a similar data-pod project to SOLID that aimed to be simpler. Does anyone remember what it's called?

Reply on news.ycombinator.com