Etherscan Forked Blocks' Reorg-Depth is Incorrect?

Question

I suddenly found there is an reorg happened today whose depth is 3 on Etherscan.io here. The uncle blocks is at height 9919331 and the hash is 0x3b1161a8bdf5107f1dc2092781798d143b5ae66b6604823630b57f5d9450e1cb. However I further checked the structure of the blockchain history and found the structure is like this:

0xa0f0 => 0xc535 => 0x7ea3
        \
          0x3b11

0xa0f0,0xc535,0x7ea3 represent the blocks (first 16 bits of their hashes) in canonical chain, while 0x3b11 is the uncle block reported by Etherscan with reorg-depth 3. The hash of 0x3b11 is included as uncles in the block 0x7ea3.

I am very confused why Etherscan reports this uncle blocks as reorg-depth 3. It seem this reorg only discard one block (i.e. 0x3b11). Is this a problem in Etherscan? or are there some uncle blocks overlooked by me?

Answer 1

Let's suppose a node has this state

A -> B -> C -> D

New blocks arrive and the situation becomes:

A -> B' -> C' -> D' -> E -> F
           |
           B

(Where B is included as an uncle block in C'.)

Since 3 blocks were replaced the reorg has depth 3

The inclusion of B as an uncle in C' doesn't reflect the depth of the reorg.

For example it might have been included further down the chain in D', E or F.

To be included as uncle the block's parent has to be in the main chain. Since B is no longer part of the main chain its children C and D cannot be included as uncle blocks.