I thought OpenBSD was good because the OpenBSD devs seem to really know what they are doing:
libcseems great1, while
libcof choice (licensing aside)
Also, the community was pretty friendly and welcoming when I stumbled into #openbsd with my troubles, which is refreshing coming from #archlinux (terrible).
The issues I had were pretty specific, and probably could've been mitigated with a different (less modern) hardware setup.
My first hurdle was the lack of support for GPT-partitioned hard drives, making my HDD invisible to OpenBSD unless I wanted to obliterate my other operating systems. Quite a large implication of this is that OpenBSD won't work on UEFI systems, which seems to be the way things are headed. They are working on it, but it's simply not done yet; maybe it'll be done by November.
Fortunately, my BIOS supported legacy boot, so I installed OpenBSD on a USB drive instead.
But then I encountered Nvidia woes. Short answer: there is just no way to make it work right now. If you have a modern Nvidia card, you can't use the
nv 2D-accelerated driver, because Nvidia canned it a few years ago. OpenBSD has a strict no source, no inclusion for drivers rule (after all, how can you trust a binary blob in the kernel?), so the proprietary
nvidia driver doesn't work and probably won't ever be supported.
nouveau is something which would fill this hole, but it doesn't work with OpenBSD yet, and even if it did don't expect any 3D acceleration on the scale that the
nvidia driver would provide anytime soon. That leaves
vesa, which means you have a very slim chance of being able to get the native resolution on a widescreen monitor.
nouveau is clearly the way things are intending to go here, but I'm a little less hopeful that it will be integrated in the near future - the advice I got from Google and #openbsd was buy a different graphics card. Maybe in a couple of years
nouveau will be viable.
And that was about all I could take.
OpenBSD certainly has its value as a secure operating system, but there are a few things which make it suboptimal for desktop use. It really is security over all else; I once heard that generally OpenBSD takes years to implement something and then does it perfectly, which is probably why they are still using FFS which is terribly slow. Another big thing on my list was lack of support for SSD discards/TRIM, without which the lifespan and performance of an SSD will decrease due to write amplification.
I think it's fair to say that OpenBSD isn't for everyone, nor does it try to be - it just very conservatively puts security first.
Future: Bitrig looks promising.