Not many bad experiences— I looked before I leapt. It's
not a hard resource constraint in many cases— the resources just opt out of contributing. Despite the unicorn moniker, many designers know how to code! I've worked as a full-time developer for 12 years, and did comparatively simple but non-trivial systems-focused scripting in sys admin/support roles prior to that. Especially among more UX-oriented designers, most started in other careers, and many of those careers were related to software in some way. A developer I worked with until recently had a BFA in New Media from a top-tier art school. Most designers/developers I know just don't contribute design to FOSS because interfacing with FOSS projects for anything other than bug fixes and feature implementations is often a giant PITA.
Facing critique about work we aren't confident in sucks. Most people reflexively respond with defensiveness, dismissiveness, or just shut down and ignore it. Since most FOSS UIs were created by people without subject matter expertise, most who make decisions about them have the expected response. It's like giving a brand new developer a code review. Calmly explaining how to practically reason about algorithmic complexity won't stop many from insisting a thrice-embedded loop with database calls that gets called every time an event loop runs is the best possible solution— at least for a while. Sometimes you can argue them out of it. Sometimes you just have to say "trust me— it might not matter in your dev environment, but this is fundamentally broken and needs to change."
That doesn't work with FOSS projects, and even getting to that point in the design phase takes a significant amount of work!
Even at work where a project leader is assigning tasks based on quantifiable business needs, dev teams often see design as superfluous. Most FOSS projects don't have that level of oversight. Fighting glib, dismissive, patronizing arguments born from assumptions and egos of people who believe the topic is somewhere from back-burner importance to superfluous just isn't worth it. I'm certainly not going to spend a bunch of time logically analyzing user goals and figuring out the most logical paths and visual cues to help them achieve those goals when I have to fight through that to even start coding.
Sometimes a polished user flow really is superfluous— anyone who sets up procmail enough to care about its usability will soon learn their way past its inscrutable configuration. But most applications aren't like procmail, and most FOSS interfaces aren't sufficient for their purposes. It's sad to see commercial software consistently trounce functionally identical or superior FOSS because of usability. Word of mouth + free could trump damn near any marketing/advertising campaign if people actually liked using it.