This may be getting too technical, but can't a block be orphaned before the next block is found? It's my understanding that when a mining node hears about two blocks that are both the same number (e.g. 182801), it will immediately reject the one that has the larger hash (lower effective difficulty). Miners accept the chain with the largest proof of work, which is the largest sum of difficulties. In the example described, the block is not orphaned until the block competing with it gets a block built on it, and then the race is immediately over. Isn't there the possibility for a race between two chains to go for more than one block (although it is highly unlikely)? For example, I remember reading about a ~50 block orphaned chain in Litecoin a few weeks ago. Bitcoin is probably past the point of having an orphaned chain longer than one, but I find the theory interesting.
Bitcoin communicates both transactions and new blocks over the p2p network by relaying - the passing of messages to all other connected Bitcoins. This is how they "stay in sync".
The first sentence could be rewritten to be correct: "a block can be orphaned when
the next block is found." If two block solutions are both published on the Bitcoin network, the solution the next miner uses when they find another block will be the one that becomes part of the block chain. The other is orphaned, that block's miner(s) gets no reward.
I was previously corrected in thinking the hash size matters (although I still haven't looked at code for myself). I think an instance of Bitcoin just uses the first block that it receives from the p2p network as the one to build on. Since relaying happens quickly on Bitcoin, the block that is orphaned was likely found after the one accepted anyway.
Orphaned chains are usually just one block. When BIP16 happened, there were often two blocks orphaned in a chain from two different versions of Bitcoin on the network playing by different rules (an old Bitcoin would build on another's bad block, and then the whole thing would get tossed by the whole network upon the next valid block).
The alternate chains have a low hashrate, so it is relatively easy for a single actor to develop a long blockchain fork by themselves, longer than the rest of the whole network, and publish it all at once.