On @boerst’s Stratum Work , the last jobs with an F2Pool coinbase tag were broadcast at
- F2Pool:
1782981814 - ntminerPool:
1782981828 - DogPool:
1782981828
Both ntminerPool and DogPool have a slightly higher coinbase output value of 3.14003902 BTC as the winning and loosing blocks (3.13976509 BTC), so my guess is that they were indeed generated later.
The page also shows:
Which indicates that big parts of the network were mining on the loosing block. Some pools switched to the winning (but 9s later header-timestamp) block:
- nthminerpool and DogPool after 800ms for very short time period.
- BitFuFu after 3.7s
- Poolin, Ultimus, Binance, CloverPool after about 7s
F2Pool themself never switched.
So the timestamp of the winning block might indeed have been rolled to 1782981823, while the loosing block didn’t roll. The loosing block was probably found about 800ms before the winning block. No idea why Antpool’s friends decided to switch, but Antpool not.