←back to thread

622 points ColinWright | 1 comments | | HN request time: 0.216s | source
Show context
jabbany ◴[] No.30079546[source]
I feel like this article doesn't address the reason why people moved away from the "old" Internet --- most people only care about their _content_ not the particular technology that backs it.

In the old times, you could only host content if you were really good with the technology. The people who had websites were people who knew how to set up and run a web server, and to compose the HTML content that was being served. Similarly, communities who ran BBS forums, mailing lists, IRC, etc. all needed at least some people who could set up and manage the technology for them.

Centralized services (Web 2.0) changed that. Now all of a sudden you didn't need to run your server --- you could just type some text and click some buttons to put up a blog on a shared service (myspace? blogger? wordpress.com?). You didn't need to run your own forum or mailing list or chat rooms, you could just put your community on an existing service that provided that (Google groups, FB/Whatsapp/<insert your IM> Chat). Need a forum? You have reddit. What is Twitter if not just a collection of mailing lists that you can openly subscribe to but not post on? Centralization allowed the vast majority of people to do what they wanted to do while "outsourcing" the technology part to somebody else, because most people don't care about the technology they care about their content.

The problem with the modern decentralized web (3.0?) is that we haven't focused on solving the real problem --- making it easy to manipulate _content_. If anything, setting up a Web 3.0 presence is _harder_ than it was on a centralized service, and as long as it remains harder, the new "old" Internet will always be a niche.

replies(7): >>30080050 #>>30080328 #>>30080470 #>>30080780 #>>30083098 #>>30084128 #>>30101773 #
largely_sitting ◴[] No.30080050[source]
Its the data centers that are going to be responsible for end of web2, when they figure out they can undercut web2 and offer a continuous federated open data storage system and virtual machine. Advertisers will pay them for access directly. Google, Meta, would just be third party indexing services.
replies(1): >>30081575 #
1. lovemenot ◴[] No.30081575[source]
At best that could be only marginally profitable. Such a service would have to be a commodity in order to be interoperable. If it could have high margins the index owners such as Meta would keep it in-house. As long as they control the services users want, they control the market.

Your prediction could come to pass only in the case of the open backend being tied to a massively popular service that could compete with those giants