This is an important point but often it's existing docs and files that cause the most grief, with subtle issues like differences in font rendering or line spacing.

For example, I wanted to make a simple change to a Word doc in Libre Office that included a side bar/column of text in a fixed size table on multiple pages. In Word the layout looked great.

Unfortunately due to font subtleties, in Libre Office the side texts overflowed the tables and the last sentences were cropped. It took some fiddling to make things fit and look as good as the original, but in the process I had to make the font smaller which lost some clarity. I did play with line spacing but that got fiddly.

In summary, a 5 second edit of an existing document laid out 'just so' perhaps 5 yr ago ended up becoming 20 minutes of hassle.

New docs could be laid out better and differently to make future edits easy, but that ignores the large legacy of existing stuff that many will have.

The migration is an opportunity to subvert the "Office" abstraction to start with.

others have articulated it much better but I think the difference isn't in what it can do. But working with existing docs - subtle things like font differences, formatting issues, working with the rest of the world that is MS dominated, keyboard shortcuts for excel warriers. LibreOffice is very capable on it's own, as a drop-in replacement is where it doesn't work most of the time. And the difference isn't even huge, but people being people give up fast.

It’s a great idea until you realize the way people are.

For the amount of dodgy stuff thales has been caught/accused of, they don't seem great either.

Realistically a fully open ID standard for on premium and cloud isn't going to happen. You always need to get into bed with someone dodgy.