Beyond HTTP

What the web host still does today

Marked packets unbundle the platform

• identity, storage, and administration become separate replaceable layers
• marked objects keep their meaning when they move
• competition appears at every layer, not only between full-stack silos

Addressing set of packets

What that buys you immediately

• copied packets keep their integrity proof
• marked packets carry their verifier claim
• history is addressable, not just remembered by a host
• caches and mirrors require no new trust
• storage and routing become infrastructure, not app identity

HAVI is an HPPR-native browser

• you still build with HTML, CSS, and JavaScript
• the browser speaks HPPR coordinates and marked packets natively
• the API changes are small: window.source, explicit named clients, <x>
• it feels like the web, but over HPPR objects

A post is written first

• the author publishes one raw post packet
• it is marked and addressable immediately
• it is not public yet

Content-Authority publication is visible page composition

• <x> is an HTML element, like <iframe>, for HPPR-native subviews
• Content-Authority publishes a post by including it in a wrapper page
• adding or removing that include changes publication state

<x src="////V.EXAMPLE~HASH.H3"></x>

<x src="/post/p-2026-03-20-voorbeeld/index.html/|/seal/V.example-userkey.H3"></x>

Comments follow the same rule

//dse.nl/forum/comment/p-2026-03-20-voorbeeld/c-0001/index.html

<x src="/comment/p-2026-03-20-voorbeeld/c-0001/index.html"></x>

• a comment is written as a raw packet first
• it becomes public only when the post page includes it, in Content-Authority-chosen order

DSE could be a repo server + Content-Authority

Raw packets plus wrapper pages.

Edit = publish a new version at the same coordinate

• same public coordinate
• new packet version
• wrappers follow the latest tip

Demo in four moves

1. write raw post
2. Content-Authority publishes the post
3. write raw comment
4. Content-Authority publishes the comment

HAVI makes this usable as a browser, not just a protocol

hppr://dse.nl/home/index.html
hppr://dse.nl/forum/index.html

• open pages by coordinate
• resolve exact packets and histories when needed
• treat local and remote repos as ordinary browser surfaces

The programming model stays small

const post = await window.source.client.get("//dse.nl/forum/post/p-2026-03-20-voorbeeld/index.html");
const page = await window.source.client.get("//dse.nl/forum/index.html");

await window.source.client.add({
  headers: [
    "Location: comment/p-2026-03-20-voorbeeld/c-0002/index.html",
    "Author: alice",
    "Published-At: " + new Date().toISOString(),
    "Content-Type: text/html; charset=utf-8"
  ],
  data: html
});

Read, list, add, watch. Small surface area. The substrate does the heavy lifting.

Committed source keeps the mental model sane

The other 50%

• repo access control lists, per key or per group of keys
• mounting an hpprd server as FUSE or NFS
• HTTP proxy in and out

That exists too. I am leaving it out today because the publishing model is the more important shape to show first.

Why this is buildable

• small, inspectable core
• ordinary client/server starting point
• distributed later without changing object identity
• no chain, no global consensus, no opaque platform backend
• HTML, CSS, and JavaScript on top of marked, addressable objects

Will this work?

• a new technology usually needs to be 10x better to win adoption

I think it's 10x better

• real communities can carry identity and history across hosts
• platform rent-seeking loses leverage
• the same substrate supports both public squares and private groups

What if the internet becomes 10x worse?

• AI floods the network with cheap content
• spam detection and source credibility degrade
• host reputation stops being enough
• trusted community becomes the only way