Why does a replicated dataset have different "refer" and "written" to the source?

Hello,

I am probably missing something obvious, but I’m struggling to understand something. Previously, when backing up datasets, I thought that the “refer” column was consistent between source and target. I’m a bit confused that in tests I am running within the same pool, this is not the case:

I have a simple test dataset, rpool/test/source. I replicate this using a simple syncoid command:

syncoid rpool/test/source rpool/test/target2

But then the snapshot refer is different:

$ zfs list -t snapshot -r rpool/test
NAME                                                                           USED  AVAIL  REFER  MOUNTPOINT
rpool/test/source@firsttest                                                     80K      -   228K  -
rpool/test/source@syncoid_work-XPS-9315-plucky_2025-06-15:10:15:03-GMT01:00      0B      -   228K  -
rpool/test/target2@firsttest                                                    80K      -   204K  -
rpool/test/target2@syncoid_work-XPS-9315-plucky_2025-06-15:10:15:03-GMT01:00    48K      -   204K  -

As is the “written” value:

$ zfs list -o name,written -t snapshot -r rpool/test
NAME                                                                          WRITTEN
rpool/test/source@firsttest                                                      228K
rpool/test/source@syncoid_work-XPS-9315-plucky_2025-06-15:10:15:03-GMT01:00       80K
rpool/test/target2@firsttest                                                     204K
rpool/test/target2@syncoid_work-XPS-9315-plucky_2025-06-15:10:15:03-GMT01:00      80K

The datasets/snapshots are in the same pool, so the ashift should be the same. The datasets have encryption, but this is the same on both:

$ zfs get encryption -r rpool/test
NAME                                                                          PROPERTY    VALUE        SOURCE
rpool/test                                                                    encryption  aes-256-gcm  -
rpool/test/source                                                             encryption  aes-256-gcm  -
rpool/test/source@firsttest                                                   encryption  aes-256-gcm  -
rpool/test/source@syncoid_work-XPS-9315-plucky_2025-06-15:10:15:03-GMT01:00   encryption  aes-256-gcm  -
rpool/test/target2                                                            encryption  aes-256-gcm  -
rpool/test/target2@firsttest                                                  encryption  aes-256-gcm  -
rpool/test/target2@syncoid_work-XPS-9315-plucky_2025-06-15:10:15:03-GMT01:00  encryption  aes-256-gcm  -

The guids for the datasets/snapshots match:

$ zfs list -o name,guid,refer,written,encryption -t snapshot -r rpool/test
NAME                                                                                          GUID  REFER  WRITTEN  ENCRYPTION
rpool/test/source@firsttest                                                   11449625390916305729   228K     228K  aes-256-gcm
rpool/test/source@syncoid_work-XPS-9315-plucky_2025-06-15:10:15:03-GMT01:00    3139788996491844858   228K      80K  aes-256-gcm
rpool/test/target2@firsttest                                                  11449625390916305729   204K     204K  aes-256-gcm
rpool/test/target2@syncoid_work-XPS-9315-plucky_2025-06-15:10:15:03-GMT01:00   3139788996491844858   204K      80K  aes-256-gcm

What am I misunderstanding? Thanks in advance!