Sovereignty — Proof, Not Promise

Claim: packets from you reach this server through consumer ISP → backbone → Verizon FiOS (AS701). No AWS (AS16509), Azure (AS8075), GCP (AS15169), or Cloudflare (AS13335) on the path.

Run it:

traceroute direct.cochranblock.org
# then for each hop IP:
whois -h whois.cymru.com " -v $IP"

Pass if: the last few hops resolve to AS701 (Verizon) and your ISP's AS. Zero hyperscaler ASNs in the trace.

Why this proves it: traceroute runs on your machine, not ours. We cannot fake hops. ASN ownership is public RIPE/ARIN data, independent of anything the server says.

Claim: the cert terminating TLS is from ISRG Root X1 (Let's Encrypt), issued via ACME DNS-01 to a Rust binary that speaks the protocol itself. No Cloudflare Origin CA, no AWS ACM, no Azure Key Vault.

Run it:

openssl s_client -connect direct.cochranblock.org:443 \
  -servername direct.cochranblock.org < /dev/null 2>/dev/null \
  | openssl x509 -noout -issuer -subject -dates

Pass if: issuer= contains O = Let's Encrypt. Anything else — ACM, Origin CA, DigiCert, etc — fails the proof.

Why this proves it: the cert bytes come from the TLS handshake your client already does. We cannot rewrite what your client sees. The issuer is signed by a root you already trust, not by us.

Claim: the A record for direct.cochranblock.org resolves to an IP allocated to Verizon Communications, not to a cloud provider's prefix.

Run it:

dig +short direct.cochranblock.org
whois $(dig +short direct.cochranblock.org) | grep -iE "orgname|netname|country"

Pass if: OrgName is Verizon / MCI / Cellco, NetName contains a FiOS or residential tag. Fail if it's Amazon, Google, Microsoft, Cloudflare, DigitalOcean, Linode, Hetzner, OVH, or any hosting provider.

Why this proves it: WHOIS records are registry-maintained at ARIN/RIPE and not under our control. The IP→Org mapping is public record.

Claim: the three Rust binaries on the path (approuter-acme → approuter → cochranblock) are all published and their SHA256 matches what you get when you cargo build --profile=diamond from the same commit.

Run it:

git clone https://github.com/cochranblock/cochranblock
cd cochranblock
git checkout 8f7446e781
cargo build --profile=diamond
sha256sum target/diamond/cochranblock
# Compare with the SHA256 emitted above: 33d9c55fed2b0be0...

Pass if: your locally-built SHA256 matches the one this page reports. (Perfect reproducibility requires matching toolchain + target triple — see P27 Diamond for the canonical profile.)

Why this proves it: no supply-chain middleman can inject code that doesn't show up in your local build. If the hashes match, you've verified there is no build-time vendor on the path either.

Claim: the site has no runtime dependency on any hyperscaler. Drop their entire IP space at the firewall on the origin box and the site keeps serving.

Run it (as origin operator — destructive, do in a maintenance window):

# On the gd origin node:
# Blackhole AWS, Azure, GCP, Cloudflare CIDRs
for cidr in $(curl -s https://ip-ranges.amazonaws.com/ip-ranges.json | jq -r '.prefixes[].ip_prefix'); do
  sudo iptables -I OUTPUT -d $cidr -j DROP
done
# ... repeat for Azure, GCP, Cloudflare IP lists ...

# Then hit the site from an external probe:
curl -v https://direct.cochranblock.org/health

Pass if: 200 OK, response body identical. The site keeps serving because nothing on the request path needs to reach those CIDRs.

Why this proves it: ablation is stronger than inspection. It's one thing to trace and see no vendors. It's another to actively sever the connection to every vendor and watch the site stay up.

Claim: no vendor has a power switch on this site. The only people who can take it offline are the owner and Verizon (by cutting residential internet).

Run it (as origin operator):

1. Unplug ethernet from the gd node. Site goes down. Note timestamp.
2. Wait 30 seconds.
3. Plug ethernet back in. Site comes back up. Note timestamp.
4. Record the total downtime. No cloud console was opened. No vendor was contacted. No account was recovered.

Pass if: recovery loop involves exactly one action (plug the cable back in) and zero vendor interaction.

Why this proves it: recovery scope equals sovereignty scope. If nobody else is needed to bring the site back, nobody else can take it down.