Most ClickHouse installations run replication for availability and read scaling. If you do get corrupted data for some reason, you can read it back from another replica. That's much more efficient than trying to fsync transactions, especially on HDD. The performance penalty for fsyncs can be substantial and most users seem to be pleased with the trade-off to get more speed.

This would obviously be a poor trade-off for handling financial transactions or storing complex objects that depend on referential integrity to function correctly. But people don't use ClickHouse to solve those problems. It's mostly append-only datasets for analytic applications.