Blockchain Sidechains

An upgrade key, in every case here — the question is only how many hands hold it and how long they have to wait. The Polygon PoS bridge, the largest escrow in this comparison, is owned by a TimelockController whose getMinDelay() returns 0; because it is the non-enumerable variant, getRoleMemberCount() reverts, so the proposer set could not be listed, and four independent public RPCs refused the archive-range eth_getLogs that would have recovered it from RoleGranted events. SKALE shows the same gap from the other direction: its bridge is described in terms of an 11-of-16 BLS threshold, but all three mainnet escrow proxies share one ProxyAdmin owned by a Safe with getThreshold() = 3 over five owners — three keys can replace the escrow logic without delay, whatever sixteen nodes signed. Chiliz publishes the vault and hides the lock: its Ethereum handler holds 1,915,281,358 CHZ and reports seven relayers, while every getter for the signing threshold reverts.

No, and Gnosis Chain is the instructive case because it is usually named as the most decentralised set in this group — by validator key count it may well be. Measured by who actually gets paid, one fee-recipient address took 328 of the 600 most recent blocks (54.7%) and five addresses took 79.8%, sampled 2026-08-07. Key counts are a registry fact; block rewards are a behavioural one, and this comparison scores the second.

Not without testing it. Chiliz Chain returns block 0 for the finalized tag while its head sits at 36,455,222, confirmed against two independent endpoints on 2026-08-07. That is worse than an outright error of the kind SKALE and Harmony return, because an exchange or bridge wiring up the standard tag receives a well-formed block that is 36.4 million blocks stale and gets no exception to catch. The two chains with the best measured finality — Cronos at zero lag and BNB Smart Chain at one second — are also two of the three with the weakest answer on bridge custody, and neither publishes an enumerable Ethereum-side signer set. Fast settlement on a chain you reached through an unspecified bridge is fast settlement of an unknown claim.

What the real security model is when it is tested. BNB Chain's own account of stopping the October 2022 Token Hub exploit is that it worked "by contacting community validators one by one" — a set small enough to phone is a set small enough to coerce. That sentence is worth more than any architecture diagram, which is why the documented failure and halt record carries the same weight here as validator concentration.