He laments youtube comments and health-gadget data in silos and walled gardens, but this is entirely congruent with the original http client/server concept.
The protocols created no incentives to protect data and identities from being walled off. The original system was not "really good" at anything and arguably succeeded because it could be adapted for so many different purposes.
In contrast, email has been more successful thus far at resisting being walled off.
I would argue that financial incentives explain the Web's walling-off, and the inverse for email. There's just not that much money to be made, comparatively, from email.
But after some thought I'm coming around to your suggestion that the protocols were compatible with this outcome from the beginning. With email protocols, the messages themselves are sent from one system to another. With the web's protocols, the body of an HTTP request could be anything, or crucially it could be nothing. Walled gardens choose nothing. If email providers did the same, it wouldn't be email anymore.
> In contrast, email has been more successful thus far at resisting being walled off.
The point of e-mail was electronic mail: instantly sending text multimedia digitally. It's not necessarily been "walled off", but I think the wide spread adoption of things like Whatsapp, Facebook Messenger, Snapchat, MSN, ICQ and even SMS all happened because e-mail wasn't really convenient enough for instantly sending multimedia digitally at the time.
Now though, it would be an interesting experiment to force all chat/messaging apps to become fancy e-mail clients for e2e encrypted e-mails that they can't access.
The Web has fared better than e-mail IMO: it's far easier to find a website than it is to find an e-mail address, and people are far likely to go to something other than e-mail for the things e-mail can do.
It's worth noting that the initial proposal for WWW (https://www.w3.org/History/1989/proposal.html) was actually for a distributed/decentralized network, requiring no central authority/control:
> CERN Requirements - Non-Centralisation - Information systems start small and grow. They also start isolated and then merge. A new system must allow existing systems to be linked together without requiring any central control or coordination.
As the web grew, this obviously became less and less true. But I don't think there is anything in particular in the initial ideas for WWW that locks it into a client/server model, although that's what naturally happened.
> Non-Centralisation
I imagine this term was used because it was before everything got centralized, so there was no need to "de-"centralize yet.
> client/server model
The original design of the WorldWideWeb application was a web browser and editor, which I think implies that anyone using it could run a server as easily as browsing other people's servers.
Edit: Not totally sure, but it does seem there was an HTTP server bundled with the browser/editor.
How to make a WWW server (1992) - https://www.w3.org/History/19921103-hypertext/hypertext/WWW/...
WWW Daemon user guide - https://www.w3.org/History/19921103-hypertext/hypertext/WWW/...
Email is sort-of a walled garden: I've heard running your own email server is very difficult and many providers will consider your mail as spam.
Technically it's impossible to make a service that can't be a walled garden, specifically because the walls can be legal. Today, there are laws preventing you from sharing data you have access to (e.g. DMCA, clickwrap). Without those laws, no publicly-accessible data would be walled off, because people could just scrape and redistribute it, and distribute hacks (though without those laws, less services would exist in the first place, since they would be much harder to monetize).
> I've heard running your own email server is very difficult and many providers will consider your mail as spam.
This is just another case of monopoly abuse though. Both Google and Microsoft (the two largest email providers) make it notoriously hard to deliver regular mail to their customers. Meanwhile, you still get tons of spam that makes it through their filters so they are both blocking legit mail and allowing spam to filter through at the same time.
Yes but it is more as a result of the weakness of the protocols. It started naively as something to send message from machine to machine. There was no real concept of an identity or a user, it only has a random name (and later domain, but that part is tied to DNS and another machine abstraction). And there is the problem that sending is "free".
Which is why spam is an impossible problem to solve. The newer messaging apps are successful because they start with what really matters in the first place: who is sending the message, in other words the identity. Sending is in theory free, but in practice you pay with your data by letting the companies running the service exploit it for ads. Since they have full control of the identities of everyone using the service, a misbehaver is quickly blocked/removed depending on faults.