@Dzertanoj Coordinates in OSM are supposed to be WGS 84. I don’t think anyone is disputing that.

But in practice -

In Great Britain, there is no accurately aligned aerial imagery available to OSM. The only accurate source available is cadastral parcels. The parcels are derived from data produced by the national mapping agency (Ordnance Survey). So we are aligning the aerial imagery with the parcels. The reference for the parcels is ETRS89.

In GB, there are not many RTK base (correction) stations which are free to use. There are a few professional ones (https://epncb.oma.be/). One, even of these, the one in west London, has a horizontal position error of 75cm and an altitude error of 43cm (relative to ETRS89). It is somewhere near WGS 84, but it cannot be WGS 84 because the position it broadcasts has not changed by even a millimetre in five years. There are more RTK base stations free to use, on rtk2go.com but they are likely to have position errors. One had a horizontal position error of 8cm and an altitude error of 85cm. Another had a horizontal position error of 280cm, I did not try to find the altitude error.

In France, the national mapping agency (IGN) makes accurately aligned aerial imagery available to OSM. The reference for the imagery is ETRS89. There is a dense network of RTK base stations which are free to use (https://www.centipede-rtk.org/). The positions of the base stations are accurate because they have been calculated by the national mapping agency. The reference is ETRS89.

So, in practice the coordinates in OSM in both GB and France are ETRS89. Or you could regard them as WGS 84 epoch 1989.0.

We are supposed to transform the coordinates from ETRS89 to WGS 84, but which date should be used (because the result depends on the date)? I would suggest 1989-01-01. Perhaps in the future OSM will have a tag for the date on which a node’s coordinates coincided with WGS 84. Then we could tag all the nodes in GB and France with the date 1989-01-01.

I guess NAD83 is more of a problem because there isn’t a date when it coincided with WGS 84. So you do need to transform from NAD83 to WGS 84. Again, the result depends on the date. I think it would make a lot of sense to use the same fixed date throughout the conterminous United States.

You can also use JOSM to convert .ubx to .gpx. Simply change the filename extension to .nmea, open the file, and save it as .gpx. If the file contains lines which are not NMEA messages, JOSM will notify you and discard those lines. (.ubx files do usually contain non-NMEA messages.) JOSM will colour-code the trace according to the type of fix: RTK, float RTK, DGPS, etc. It preserves the colours in the converted .gpx file. As Stephane said, JOSM allows you to set offsets between GPS time and camera time which are not whole numbers of seconds. JOSM also correctly converts subsecond timestamps when converting from NMEA to gpx. Not all conversion software does this.

I managed to get the ‘attic’ (historical) data for Northwick Park Hospital at annual intervals from 2012 to 2021. There are hardly any buildings except in 2021. So any buildings that may have been there, were there for less than a year.

It wasn’t easy to get the data. I spent some time tweaking the Overpass query before I found a query that didn’t time out. It didn’t seem reasonable to try to fix the problem by extending the time-out from the default of three minutes.

I tried the method described by tyr_asd. It was easier to get the data, but I had trouble with QGIS. It just showed a blank, white map, although the objects were there. I had to quit and relaunch QGIS to get it to work properly. The use of the temporal slider was very effective. It showed there were hardly any buildings from 2012 (the start time I chose) until late 2020.

The u-blox NEO-7P does precise point positioning. I wrote about it briefly on talk-fr https://lists.openstreetmap.org/pipermail/talk-fr/2016-March/080463.html I am impressed with it, provided that you have a good view of the sky; and provided that, after startup, you wait for ten minutes for it to reach the best accuracy, before recording tracks.