Trick question. I use two distributions as my main operating system. On my desktop, I run Gentoo with Sway. On my laptop, I run Arch Linux with a heavily modified Hyprland setup. My modifications essentially remove dynamic tiling from Hyprland, as I can't stand it, and make the tiling function identically to i3/Sway. I additionally use a UI that I have written in Quickshell, shell script, and Python. For about 14 months, I used i3. I switched primarily due to the security benefits that Wayland provides over X11; I have no hard feelings towards i3, and, in fact, still feel a strong love for it. I also interact with Debian and Red Hat Linux fairly frequently, as they are used by systems that I ssh into at Brown's Sunlab and High-Performance Computing Cluster respectively.
I don't have a strict preference between Gentoo and Arch, but I do believe that for me they are the two distributions which work best. I do find that Gentoo is maintained far better than Arch is. I am active on both the mailing lists of Arch and Gentoo.
On the contrary, I do have a very strong preference for rolling release distributions (like Arch, Gentoo, et al.) and a strong disdain for frozen release distributions (like Debian, Ubuntu, et al.). There is very little to be gained in terms of "stability" from freezing software updates, yet this practice incurs outrageous security costs, not to speak for the inefficiency it creates within the development teams of those distributions. Rolling release and bleeding edge are often confused - you do not have to be on the bleeding edge to use a rolling release distribution.
One of the principal benefits of Linux and the Unix philosophy in general is customization and extensibility. When you use a desktop environment, this crucial aspect of the Unix philosophy is lost - control over your system is abstracted away to the authors of the environment, and they are most likely enthused by feature creep. On the other hand, a good window manager like Sway does nothing but handle windows, and otherwise gets out of your way. In general I am an advocate of an idea presented in the linked Wikipedia article: "Write programs that do one thing and do it well."
I have not even made my case about how much more efficient use of a system with a window manager is over a desktop environment, because the above points are enough.
In general, I prefer to compile code that I am going to run from source. I can apply the flags (much more conveniently with Gentoo and portage) that I would like, and play around with settings to see what works best for my system.
The primary reason I run Arch and not Gentoo on my laptop is just due to the constraints of the CPU. I cannot expect the laptop to compile as quickly, and the laptop is certainly not able to regulate temperature as well during compilation times. Compiling the Linux kernel on my laptop, for instance, takes 30-35 minutes.
I am an advocate of using static linking much more often than most people do. Usage of static linking is an actually beneficial way to use available RAM and system resources with results that one can profile, as opposed to doing so in the name of developer laziness (see "electron" below).
I use both systemd and glibc on my desktop and laptop, but only out of reluctant necessity.
In principal, I would much rather use something like musl for my library and OpenRC for my init system, but these would incur too much of a usability hit. Unfortunately, we have reached a point where developers operate under the assumption that all Linux systems are going to use systemd and glibc, and so many programs that I use on a day-to-day basis would not work if I say switched my library for musl.
I take other hardening measures where reasonable and possible, such as compiling my Linux kernels from source and with most modules blacklisted.
In the past I have daily-driven MacOS and Windows 8.1, Windows 10, and Windows 11. My conclusions about them are as follows:
Electron is the worst development ever to occur in the realm of software. I vehemently refuse to use any program that incorporates electron (which I say is unworthy of capitalization) wherever possible, and I urge you to join me. Not only has electron significantly contributed to pointless software bloat and remarkable inefficiency - it has also distilled laziness and terrible programming habits into the workflows of many organizations and developers. If you take pride in writing code, whether for fun or for a living, do not use electron!
Anyone who programs with electron is essentially saying "I am so lazy or incompetent as a programmer that I can't take time to read documentation about native graphics frameworks on any platform and use them, so I would rather bundle a huge codebase (Chromium) which I've probably never seen, don't know anything about the inner workings of, and don't care enough to investigate, instead of taking the time to write a program the real way. Rather than build a program which uses reasonable amounts of resources and does not constrain the system, I would prefer to use hundreds of megabytes or even gigabytes of memory, incredible amounts of CPU utilization, and massively bloated binary sizes just to save myself a little bit of effort."