causality is superior to this, blockchains themselves use cryptography to discover causality of ordering between blocks, and in the tx payload, the chain of UTXOs being spent into new UTXOs

the property of plain time sequence in distributed systems cannot be strongly synchronous without low latency of message delivery, even 100ms is pretty much getting too slow, and global internet signals of maximum average distance are 400ms.

this is why there is a pretty wide tolerance of subjective time between nodes on the bitcoin blockchain network, why blocks can be up to something like an hour ahead or behind between one node and another, and why bitcoin's consensus allows out-of-order timestamps between blocks.

this property is called "weak synchrony" in distributed systems theory, and in general rules the operation of all distributed systems in much bigger a geographical distribution than the size of a large city.

the reason why it's used is because causality is a lot more computationally expensive, the timestamps are a proxy for total ordering, but can't be trusted. this is also why when people have made bitcoin forks that have "time consensus" that the most famous case of this going wrong was the Verge blockchain which had a successful timewarp attack, that allowed a single peer to take control of the entire chain history and prevent other nodes from making blocks, while the other nodes accepted these fake blocks.

check out Lamport Clocks and Interval Tree Clocks for an example of a hypothetical protocol for pure causality.

oh yeah, there is another reason why bitcoin needs timestamps, because of the issuance schedule, and the difficulty adjustment regime. this is part of the parameters of bitcoin that are based on Austrian Economic Theory, specifically Mises assertion that a theoretical finite supply of currency only needs precision increases (subdividing units) to serve a market, and all new issuance of units has a proportional drag effect on the economy pricing structure.