These past couple of days, I got caught in the "on-chain data" trap again while running new chain tasks: I clearly sent a transaction, but the block explorer wouldn't update for a long time, switching to a different RPC immediately showed success... I only then realized that node synchronization, RPC cache, and indexer database are not all synchronized; what I see as "on-chain" might be a processed version by someone else, and it can also be delayed.

After news of cross-chain bridge thefts, everyone started "waiting for confirmation," which I can understand; plus, with oracles occasionally glitching in their quotes, I increasingly feel that we shouldn't be too superstitious about a single front-end number. Honestly, I trust data more, but it has to be cross-verified from multiple sources: for the same transaction, I check several RPCs and explorers; I prefer to take my time to harvest, rather than be fooled by fake progress bars into rushing interactions. Anyway, mining is like bringing food, so I split the tasks into multiple parts to avoid crashes.