Satoshi first introduced checkpoints to prevent large reorgs, thereby limiting the amount of damage that miners can do. Heuristics show us that accidental reorgs can happen up to 2 or 3 blocks. But a reorg much larger than that would almost certainly be an intentional attack. Checkpoints limit the effectiveness of that attack, and so feeds into the game theory to make it less worthwhile. Checkpoints however introduce another weakness in that a node that has left the network and rejoins needs a source of checkpoints, and if there are two sources the node wont know what to do. Perhaps nostr could fix this as people are publishing blocks already.