Two things to fix in LO, and a benefit you forgot
First of all, it ought to support all input methods that users of that platform are familiar with.
The fact that Mac users have to forego the very easy way in which accented characters (as in "those directly accessible from a layout, for example éëêè on US/UK keyboards) can be entered on a Mac is stupid - as a matter of fact, I don't know if it's patented but if not, I'd love to see that approach in LO as what is presently there is painful, for all platforms. For Mac users, the solution tends to be installing NeoOffice instead - worth the few euros per year they ask in support IMHO - if you work in multiple languages from a US/UK layout you have earned that back in time savings in no time. I'd also love to see this approach in Linux - it's simply more practical as you can do it straight from the keyboard, with no mouse involved.
Secondly, the installer really needs work if LO is ever to make it onto end user machines worldwide. It needs to be ONE installer, and it needs to work in the language of the user from the start. To install LO, you need to install the (American) English version, then "patch" the UI into the user's language by means of a language pack, then start LO (still in American) and set it to your preferred language, then restart. Updating: ditto. This is despite the fact that someone installing a language pack is very likely to want to set it to that language, so why not ask and set it? It needs one, multilingual installer. Maybe make the installer load a language pack on startup and then run both installer and UI setup in one go, but at present it's a major barrier, also for volume deployment.
One benefit not mentioned is one of speed: LO has endeavoured to introduce spreadsheet code that supports parallel processing (i.e. using multiple threads) of cells with similar formulas. As far as I know, that support went public in the 6.2 release, and I'd welcome news from anyone who has tried it (my spreadsheets are not that complex that I'd notice a change in speed :) ).
I really *want* LO to be wider successful, but the above are serious barriers as especially the installer is the first thing a new user ever sees of a product. Asking users to change their approach to input is also not helping, but comes IMHO on a second place.