It's just a bunch of script kiddies trying to make a quick buck. Ignore them.

It's just a bunch of script kiddies trying to make a quick buck. Ignore them.

Really? I didn't understand the video at all because I'm not into networking and web infrastructure and all that stuff, but I can see that many people (who actually understood what the video said) have said that this isn't well planned and that it's just a bunch of random stuff throw together. Also, looking at how much money was donated, it seems they kept like $18,000 from the $50,000 donation and the rest went to bounties. So I think you're right.

I understand where you’re coming from regarding the use of HTTPS and Layer 1 filtering.

This spec indeed uses HTTPS (TLS 1.3 over TCP/QUIC port 443) as a covert transport layer specifically because it’s hard for censors to block without causing major collateral damage to the wider Internet. The design aims to mimic genuine TLS fingerprints and HTTP/2/3 behavior very closely to blend in.

Regarding Layer 1 filtering — yes, any network operator can trivially drop or throttle all traffic on a given IP or physical link. However, Betanet’s approach relies on SCION routing (L1) and multi-path selection, enabling path diversity across multiple independent ISDs and ASes. This means that even if one path or link is blocked, traffic can be rerouted through others, reducing the efficacy of simple Layer 1 filtering.

Also, gateways performing HTX-tunnelled transitions encapsulate traffic covertly, avoiding legacy transition headers that might reveal metadata on the wire.

It’s fair to say this is a complex system balancing stealth, resilience, and practicality. The spec is designed with implementability in mind, detailing concrete packet formats, handshake flows, replay protections, and calibration steps for origin fingerprint mimicry.

I think the claim that it’s “written by ChatGPT and non-sensical” might be an understandable reaction to its density and ambition, but the design leverages multiple established research and protocols — SCION routing, Noise handshakes, mixnets, federated payments, and more — integrated thoughtfully.

Happy to clarify any particular parts or discuss implementation challenges if you’re interested.

Right. So it can be filtered ? or it cant?
I think @nyirec is spot on. I would be extremely wary of this project.

This spec indeed uses HTTPS (TLS 1.3 over TCP/QUIC port 443) as a covert transport layer specifically because it’s hard for censors to block without causing major collateral damage to the wider Internet. The design aims to mimic genuine TLS fingerprints and HTTP/2/3 behavior very closely to blend in.

Anything that just mimicks TLS but doesn't serve valid certificates & pages is already blocked in countries with heavy internet censorhip (Russia, China, Iran, etc.) and it doesn't really cause any collateral damage.

To me, this looks like a spec written by chatgpt and its very non-sensical, esp for implementation (which it claims it is).

I agree, the spec looks like LLM word salad. Every normal spec starts with the glossary, describing what the spec is about, what entities are involved, often includes message sequence charts. Not with endianness or choice of crypto hash or other miniscule details.

Why is SCION layer 1, while whatever HTX is, is layer 2? Does HTX run over SCION? If HTX runs over TCP/QUIC, shouldn't layers be IP/TCP/QUIC/HTX? If SCION isn't encapsulated in HTX, does this mean ISP can just block SCION control plane traffic? How are clients bootstrapped in the first place?