Block propagation, measured live Sovright relay vs native Zcash P2P, across four regions.
RELAY
ZCASH P2P
First-to-Zebra Win Rate
—best region · —
Blocks the relay delivered to Zebra before native P2P (submitblock accepted vs duplicate). Per region — seeder-adjacent regions are structurally near zero.
Network Orphan Rate
—
stale blocks that lost the propagation race — the reward faster relay saves
Mempool Backlog
—
unconfirmed bytes waiting for block space — demand that drives block size
Revenue Lost to Orphans — by Poollast 24h · an orphaned block forfeits ~1.25 ZEC of reward
Per-Pool Orphan Rate — Rolling Window
—
—
—
One claim we've proven. One we'd like your help proving.
—
Measured Measured
This part is done: our delivery cost stays nearly flat as
blocks grow (+0.110 ms/KB) while public gossip's
climbs (+1.739 ms/KB into APAC) — 16× steeper, a
306 ms head start on full blocks. In Europe, where
peering is already good, the measured gap is only
+42 ms. Wherever you mine, that's the number that
applies to you.
Modeled Modeled
This part is arithmetic, not observation: if the head start
cuts orphans as the model assumes, a 10% pool in APAC forfeits
~214 ZEC/year less; a 44% pool,
~944. From Europe, ~29
and ~130. No miner has run the experiment that
would confirm these. A pilot is that experiment.
Stale blocks that lost the propagation race — the reward faster relay saves.
—
trailing 24h
accumulating — the rate history builds over the coming days
——0%
7 days agonow
Median Propagation
—ms
relay vs P2P reach-time (not win rate) · P2P —
▬ relay vs ▬ P2P · relative reach time
7.6× faster
Coverage p75 · reach time
—ms
relay p75 first-to-last · P2P p75 —
4 regions · all reached within p75
8.1× tighter
Compact payload / forward
—KB
one compact+FEC forward · full block —
Per recipient. The relay overlays native P2P (fanout × FEC), so total relay bandwidth is higher — it trades bandwidth for latency, not bandwidth saved.
▬ one forward vs ▬ full block · relative payload
—
Lead vs Public Explorer
−2.4s
Relay observers see blocks first, before zcashexplorer
Relay holds flat and low. P2P runs high and ragged.
——0 ms
▬ relay · avg — ms▬ p2p · avg — ms
Payload per forward
One compact+FEC forward vs a full block
Per recipient. The relay overlays native P2P and fans this payload out with FEC parity, so total relay bandwidth is higher — it buys lower latency, it does not save bandwidth.
ZCASH P2P · full block1.84 MB
full serialized block + getdata round-trips
RELAY · compact + FEC14.2 KB
IDs
FEC
short IDs · prefilled coinbase · 32 parity shards
—
bytes / forward
0
round-trips to read · by design
Advantage by block fullness · last 24h
Where the relay wins: full blocks vs empty blocks
—
Regional Relay Status
4 GCP regions · Zebra mainnet attached
RegionGossip SrcsAgent
Block Arrival Feed
relay vs p2p · per block
· = empty block (coinbase only) — no transactions to reconstruct, so no relay time is measured
HeightRelayP2PΔ
Block tags by mining pool
Observed coinbase tags on the canonical chain. Tags are self-reported and do not prove which software a pool runs. Zero can inherit Zebra’s marker; only an explicit Zero tag is counted as Zero. Unknown includes absent, unrecognized, or conflicting tags.