Putting a little order with everything that has been said so far, we could all deduce that more or less we all agree with these basic concepts.
Multi user, it may be useful, but not in the form that other operating systems are abusing it.
It could be useful when they are real users, and possibly, for security reasons, to allow the log-in of only one session at a time, taking into account that haiku is an os for desktop or client side use. Having said that, creating a multi-user system becomes much easier, essentially it involves creating and directing a “home” folder containing all the personal data and personal settings of the applications for each individual user, in addition a level of securityand privacy for each user should obviously be created .
For all the rest we need a modern system of protection of system files, which takes as a strong example the method of different partitions for each level, such as a system level (read only plus armored), an application level (read only but with a possibility to access it in writing more simplified), a general data management level, a personal data management level, so on … such a method remains valid, for a matter of ease of data transfer and recovery of system portions …
Personal applications that are installed in their isolated sandboxes and that can communicate with other applications through API.
Finally, the root level, this must not be considered as a user, essentially the root level must be considered as a use of the system without the armoring of the parts inaccessible to those parts of the system and armored folders, essentially with the root level it must be able to be done what Haiku is already now, where you can do everything, including self-destruct the system.