There's a large group of people in technology who don't really care how something works, as long as it works for their purposes. That's set against another group who will put up with incovenience in the name of what they think is "correctness". I genuinely don't want to suggest that either group is better than the other - but they are distinct.
Around 2010 I felt that the first group had started to dominate the decisions around the direction of the Linux userland. It was wonderful to be able to switch to BSD, which I sense is used and maintained mostly by the second group. Having an OS that is aligned with your culture is a sort of a "happiness by a thousand cuts": the defaults are already what you would have set them to, you don't have to declutter after the first install and what you do want is in the package repo, your niche use cases get attention when it comes to bug fixes because they're not considered niche in that crowd, etc. When the maintainers think the way you do it feels like they've anticipated your needs - it's presumably the same feeling that Apple-culture people feel when they use an Apple product; like it was made for them.
In a lot of ways, technology has mostly become a matrix for competing cultures. From that perspective it's not a matter of what performance edge the BSDs have over Linux, it's about whether there's a distinct enough culture to justify the costs of the duplication of effort. In my mind that distinction has only grown in the last 10 years.
There was previously a post on HN about this.
FreeBSD has ZFS in-tree as part of the base system with zero kludges (DKMS, I'm looking at you) and perfect compatibility supported directly by the core devs, as opposed to having to deal with a semi-hostile upstream (KH: "my tolerance for ZFS is pretty non-existent", Torvalds: "If somebody adds a kernel module like ZFS, they are on their own. I can't maintain it, and I can not be bound by other people's kernel changes."), and it's the default filesystem out of the box for several releases now with good tooling support around it. ZFS can be hacked into a Linux distro, but it'll always be inferior.
* https://en.wikipedia.org/wiki/List_of_products_based_on_Free...
You'll see regular "Sponsored by" lines in FreeBSD commit messages from Intel, Chelsio, Mellanox/NVidia for their products. Also:
It's interesting to see the culture clash between different licensing philosophies.
The "notoriety" portion comes from the OtherOS side of things, as the other commenter mentioned. In the PS2 era, you could order a kit from Sony containing all of the software/hardware you needed to turn your PS2 into a reasonably capable Linux machine. To make a long story short, licensing is always a concern and it blew back in Sony's face. The whole "Linux on Playstation" thing is a blemish on Sony's reputation, and it's why a lot of people call Sony the "biggest offender" license abuse.
FreeBSD was at one point (and may still be, I haven't checked in a few years) used in the bootstrap process for FB's ISP-hosted cache clusters (because potential GPL headaches IIRC).
It's also useful to test network protocol implementations across multiple platforms (e.g. Linux + FreeBSD) and can be handy if you are working with a standards body that wants to N>1 implementations of a protocol.
Linux, for all its warts, is superior to BSD in nearly every way that matters to application developers and to companies who need a server OS. The areas in which BSD is superior to Linux simply aren't important enough to justify using BSD, unless you are a corporate parasite like Sony and need an operating system with a parasitism-enabling license but don't want to pay for more than a handful of developers.
The market just doesn't seem big enough to support Linux and a handful of BSDs, and at the end of the day that's the real reason why BSD appears to be dying, because there's no reason to reach for BSD when Debian and Slackware exist.