* Due to their software-based nature, virtual machines are extremely flexible, and can for instance emulate features that their physical host may lack.
Relying on a virtual machine as its primarily personal environment has key advantages, such as the ability to more easily migrate to a new host computer or the ability to create multiple virtual computers out of a single physical computer. However, it also comes with several limitations in comparison to using a bare-metal operating system.
* **Suboptimal guest performance**. In most cases, running a virtual machine instead of using the physical hardware directly will come with a performance penalty. This penalty can be greatly reduced by using some techniques, such as letting the virtual machine access the underlying hardware directly, but this particular solution is by definition not scalable to multiple virtual machines.
* **Limited features set**. Some operating systems are designed to leverage hardware features that may not be accessible to an operating system installed on a virtual machine, or that would require specific developments to be taken advantage of.
* **Increased general complexity**. Instead of running just an operating system on top of some physical hardware, any Phyllome OS user would need to manage it as well as their primarily guest operating system. As a result, it might be more difficult to troubleshoot an issue, and it will add a pile of code that the user has to trust.
* **Decreased general usability**. Any physical device attached to a computer won't automatically be made to a guest virtual machine. For some users, it might be considered a hindrance. Phyllome OS relies on Linux drivers. Not all hardware fully supports Linux well, which may force users to rely on device or controllers passthrough.
Here is a table to help you pick a **desktop-oriented** operating system.
How to read this table? For instance: *If you care most about virtualization and put security first, you would be better off picking [Qubes OS](https://www.qubes-os.org/) or the upcoming [Spectrum](https://spectrum-os.org/) instead of Phyllome OS.*
| | Security | Usability |
|---|---|---|
| *Virtualization* | [Qubes OS](https://www.qubes-os.org/) or [Spectrum](https://spectrum-os.org/) | **Phyllome OS Desktop** |
| *Bare-metal* | [Sculpt](https://en.wikipedia.org/wiki/Genode#Sculpt) or [Fuchsia](https://en.wikipedia.org/wiki/Fuchsia_(operating_system)) | [Linux](https://en.wikipedia.org/wiki/List_of_Linux_distributions) or [BSD distro](https://en.wikipedia.org/wiki/List_of_BSD_operating_systems), [macOS](https://en.wikipedia.org/wiki/MacOS), [Windows](https://en.wikipedia.org/wiki/Microsoft_Windows) or [Chrome OS](https://en.wikipedia.org/wiki/Chrome_OS) |
> In general, the vast majority of users will stick to the bottom-right corner of the table, because that is the operating system that ships with their hardware.
Phyllome OS is intended to be easy-to-use, but still isn't. Out of the box, Chrome OS [^1], or even Windows [^2], might be considered more secure than most Linux desktop-oriented distributions [^3], at the price of greatly limiting user freedom and privacy, however. Due to their tight integration, some BSD distributions might be considered more secure than some Linux distributions. People might find Windows easier to use than, say, Ubuntu. Finally, just as Phyllome OS, Qubes OS is compatible with running Windows-based guest systems. In other words, using virtualization, a user might be able to access more usable operating systems, and in the case of Phyllome OS, one may even host Qubes OS inside Phyllome OS, for instance to test out Qubes OS.
Also note that macOS or Windows can also be used to host virtual machines, just as any Linux or BSD distributions.
[^1]: See for instance the paper [*Security of Google Chromebook* (PDF)](http://dhanus.mit.edu/docs/ChromeOSSecurity.pdf) by Katherine Fang, Deborah Hanus, Yuzhi Zheng.
[^2]: A common pain point for Linux security are desktop environments (DE), which have a limited user base scattered across many different DE: there is a lot of complexity due to adding desktop environments atop the Linux kernel and its associated tools. Simple bugs might still lurk in the codebase for a long time. See for instance [*Is the Linux desktop less secure than Windows 10: Or how super mario music can own your system* (PDF)](https://archive.fosdem.org/2017/schedule/event/linux_desktop_versus_windows10/attachments/slides/1730/export/events/attachments/linux_desktop_versus_windows10/slides/1730/fosdem_linux_desktop_security.pdf), by M.Hanno Böck (2017).
[^3]: Take for instance the boot process, or before an operating system effectively takes control over the hardware. Major operating systems editors that are working directly with OEM integrators have a distinct advantage over editors that aren't: these major editors have almost unlimited resources, sometimes almost perfect control over hardware, and can therefore tame the underlying hardware, effectively controlling, measuring and attesting the entire boot process. To implement a user-backed root of trust on a particular hardware platform, one would need to take several extra measures, relying on something like [Heads](https://github.com/osresearch/heads) which, among other things, involves physically flashing a more open firmware to a motherboard, a complicated process that is available for a few hardware platforms only. Fortunately, some hardware integrators like [Purism](https://puri.sm/) or [System76](https://system76.com/) are backing security measures straight into hardware platforms, while at the same time respecting user freedom.
> **Still undecided?** You can give Phyllome OS a try, as a live system booting off a USB thumb drive, without impacting the existing operating system on your machine.
