๐Ÿ’ง Cascade
A delivery network owned by everyone

The internet's heavy lifting, handed back to the crowd.

Cascade delivers the internet's biggest content โ€” video, downloads, files โ€” from the spare bandwidth of ordinary devices, and pays each one per parcel, in real time, on Kaspa. It is a decentralized replacement for the handful of companies that own delivery today.

delivering a 500 MB video from the crowd, paying as it flows โ— live
Parcels delivered
0
Paid to the crowd
0.00000 KAS
Nodes earning
0

Each stream is a parcel โ€” one verified, paid piece โ€” flowing from a different node; each gold flash is a micro-payment settling on Kaspa. An illustration of Cascade's real behaviour, pinned by tests.

01 โ€” THE PROBLEM

Five companies deliver the internet. That's a fragility, and a chokepoint.

Almost everything you load โ€” a video, an app update, a news site โ€” doesn't come from the website you visited. It comes from a cache near you run by Cloudflare, Akamai, or Amazon. This layer, the Content Delivery Network, is most of what makes the internet feel fast. It is also owned by a handful of firms โ€” and that concentration has a cost we've all paid.

~1 in 5
of all websites sit behind a single company, Cloudflare. When it stumbles, so does a fifth of the web.
~1 hour
of a single 2021 CDN outage was enough to knock Reddit, Amazon, Spotify, the NYT and government sites offline at once.
One call
from one company's trust-and-safety desk can make a site vanish from the internet โ€” whatever you think of the site, that's a lot of power in one place.

None of this is villainy โ€” it's what happens when critical infrastructure is owned by a few. A single point of speed becomes a single point of failure, of billing power, and of control over what stays online. The delivery layer of the internet was never supposed to belong to five companies.

02 โ€” WHAT CASCADE IS

BitTorrent, grown up: the crowd delivers, and the crowd gets paid.

Cascade takes the one idea that has moved the internet's heaviest files for twenty years โ€” a Meridian, where many devices each hold pieces and you pull them from all at once โ€” and gives it the one thing it always lacked: a way to pay the people who serve. Torrents die because seeding is charity. Here, delivering a parcel earns money, so serving is a job, and the Meridian has a reason to exist. The bytes move device-to-device over the ordinary internet; only the payment touches the chain, and it's tiny.

Bytes from the crowd. Money to the crowd. No company owning the middle.

03 โ€” HOW IT WORKS

Pull every parcel from whoever's fastest. Verify it. Pay them.

  1. A file is content-addressed into parcels โ€” a manifest of hashes, one per parcel. The manifest names the file; each hash names a piece.
  2. Founts announce what they hold โ€” which parcels, so a consumer knows who to ask. No fount needs the whole file; a phone can carry a few parcels of what's popular.
  3. Parcels are pulled in parallel and checked against the manifest before anything is believed or paid. A wrong parcel hashes wrong and is refused โ€” no trust required.
  4. Each fount is paid for what it served and verified โ€” a fraction of a cent per parcel, on Kaspa. A fount that serves junk earns nothing and is routed around; the file still completes.
CAN'T BE PAID JUNK

A wrong parcel never bills

It doesn't match the hash you held before you asked, so it's refused, never paid for. The manifest is the referee โ€” no trust needed for delivery.

CAN'T BE OVERBILLED

The price is fixed in advance

A parcel costs its size from the manifest, which you had before the request. A fount can't inflate it. Stop partway and you've paid for exactly what arrived, to the byte.

04 โ€” WHAT IT REPLACES

The delivery bill, not your wallet โ€” and not the server.

Cascade replaces the CDN โ€” the part of the cloud that moves finished bytes โ€” not the server that runs an app. And it competes on the publisher's bill: today a site pays Cloudflare to deliver its video to you; with Cascade it pays the crowd's spare bandwidth instead, for less. You still watch for free โ€” the crowd is paid, not you.

Cascade delivers thisCascade can't โ€” a server must
A video, a song, a podcastYour personalized feed or inbox
A game or app download, an OS updateSearch results, a checkout, a dashboard
Images, fonts, a site's JS/CSS bundleThe dynamic HTML built per request
A big public dataset; a fully static siteAnything private or per-user

The rule is simple: Cascade carries bytes that are the same for everyone โ€” which happens to be most of the internet's actual weight (video alone is well over half of all traffic). Anything generated on the fly, just for you, stays on a server. The closest thing that exists is IPFS โ€” but with the real-time payment layer it never had.

05 โ€” THE MARKET

Publisher funds it. The crowd earns it. The viewer never sees it.

Publisher

Pays to be delivered

A site funds a delivery budget in KAS โ€” exactly as it funds a CDN account today โ€” and gets a manifest for its content.

โ€” KAS โ†’
Founts ยท the crowd

Serve parcels, earn

Spare bandwidth and disk. They cache popular parcels and serve them, earning per verified parcel. Serve junk, earn nothing.

โ†‘ parcels โ†‘
Gatherer ยท the viewer

Pulls, verifies, pays nothing

Pulls parcels, checks each against the manifest, and signs a receipt. That receipt is a fount's proof it earned โ€” junk gets none.

โœ“ receipts โœ“

The viewer is the meter and the referee; the publisher is the payer; the crowd is the earner. Cheating is caught for free, because a bad parcel never earns a receipt. It's a two-sided market where the money only ever moves for bytes that provably arrived.

06 โ€” WHY KASPA

Paying the crowd as it works needs a chain that settles in real time.

A delivery network makes thousands of tiny payments a second โ€” one per parcel, per viewer, per fount. That's impossible if settling takes minutes and costs more than the payment. It's why every decentralized-CDN attempt so far either doesn't really pay per use, or batches it and leans on a speculative token โ€” quietly putting a trusted middle back.

Real-time decentralization isn't our slogan. It's Kaspa's own doctrine.

Kaspa states its purpose as real-time decentralization: censorship-resistance and permissionless settlement in real time, not eventually โ€” on a fair-launched proof-of-work chain with no premine. It confirms in about a second at up to ten blocks a second. Cascade is that doctrine made physical: a network that pays its participants as they serve, continuously. The money keeps pace with the bytes. That's not a clever use of a fast chain โ€” it's the thing the fast chain exists to do.

07 โ€” WHY DECENTRALIZATION MATTERS

Infrastructure you own can't be taken from you.

The word "decentralization" gets worn smooth by crypto marketing, so here is the plain version. When the delivery layer of the internet is owned by a few companies, three things follow, whether those companies mean well or not: they can throttle or bill you at will, they can take you offline with a phone call, and when they break, everyone breaks at once. You are a tenant on someone else's infrastructure, and tenancy can be revoked.

Cascade changes the ownership, not just the technology. When the caches are everyone's devices, no single company holds the switch. There is no trust-and-safety desk that can un-publish a lawful site, no single outage that darkens a fifth of the web, no monopoly rent on moving bytes that cost the crowd almost nothing to move. The infrastructure belongs to the people using it โ€” which is what the internet was supposed to be before it consolidated into a few landlords.

This is what "reclaiming the internet" actually means in practice โ€” not a manifesto, a property change. The heavy content we all depend on gets carried by a commons the crowd owns and is paid to maintain, on money that no company or state issues or freezes. Sovereignty isn't a feeling here; it's whose hardware your bytes live on, and whose permission you need to be seen.

A commons the crowd owns and is paid to keep. That is the whole idea.

08 โ€” HONEST AND TRUE

What's real, what's hard, and what isn't built yet.

A claim is only worth as much as its caveats. Here they are, plainly โ€” because a project that hides its hard parts isn't one you should trust with the internet's plumbing.

Getting bytes between phones is hard
Mobile carrier networks hide devices, so two phones can't just connect. This needs relays or hole-punching, and it's the single biggest engineering risk. Early on, PCs, routers and relays carry the network; phones start as viewers and Wi-Fi cache-holders, not the backbone.
It won't beat Cloudflare on reliability
A big CDN is fast, has contracts and legal accountability; the Meridian is messier. Cascade competes on price, reach, and being unownable โ€” not on being flatly more reliable, at least at first.
Supply needs demand
Nodes won't run for pennies with nobody buying. One real publisher delivering through it matters more than a thousand idle nodes. That, not the tech, is the hard bootstrap.
Content it serves is public
The Meridian serving whatever it's handed inherits the abuse problems every CDN has. Content is addressed by hash and founts serve only catalogues they opted into โ€” but this is real work, not a solved problem.
What's actually built, today: the Meridian runs end to end and is pinned by tests โ€” a file reassembles byte-identical from several founts at once, each paid its share; a fount serving corrupted bytes is caught by the manifest, paid nothing, and routed around; stopping partway pays only for what arrived. It settles on the same Kaspa rail proven on testnet by its siblings.

What's designed but not built: the publisher-pays market above โ€” today's code is the simpler viewer-pays version. The browser loader, a real DHT to replace the central tracker, and phone-network traversal are named next steps, not hidden gaps.