So I would assume it has to do with the file system as well. I guess not all systems keep data written in the same way.
I don't think anyone would count file system overhead as blockchain size. And anyway filesystems wouldn't differ that much in overhead in this case.
I'm not sure about this but maybe it's related to segwit stuff?
The real blockchain includes all the segwit data. Excluding that would be very wrong, and by now segwit has existed for 10 months. I don't think these sites would be that negligent.
maybe this is the cause of the discrepancies, byte vs vbyte?
vbyte should be just the number used for fee calculations instead of the actual byte size (
weight/4).
A naive way of measuring would be to just take the size of the datadir for a bitcoind instance.
Sounds too crude to be likely?
may receive more orphan blocks than another site which means that they are storing more blocks on disk. They may be measuring this as well
That sounds possible.
So there is no site known to be fully accurate in its reporting, or at least more accurate than the rest?