3 ms·
My experience with trying to help open-source projects with design has been quite bad. Projects that most need design help are usually not sponsored by a compan
by lekevicius 5y ago
My experience with trying to help open-source projects with design has been quite bad. Projects that most need design help are usually not sponsored by a company, a true contributor project. Design is opinionated, and for many projects it can be hard to find consensus. Much easier to find consensus around adding features.
Two very different examples: Blender and Audacity (pre-acquisition). Blender was able to pull off something spectacular with 2.8, a full rethink of UI, controls, even app structure. But Blender is basically a company, with full-time employees, it's just that their output is open-source. Audacity, as far as I know, didn't have full-time employees. And as it grew in features, it never had any significant UI cleanup. Menus just grew, icons were tacked on. It might change now that they have been acquired.
- danbolt 5y agoI sometimes wonder if UI/UX passes on open-source tools would be better with very explicit "experimental forks" that keep with upstream when they can. Over time, the successful fork starts to get more traffic from regular users and re-merges. It might save time compared to long debates without much agenda.
- indigochill 5y agoI'm fairly firmly of this opinion (though I only have a little experience working on large open-source projects). Sure, there can be productivity loss in developing in parallel vs developing a consensus, but nothing highlights the strengths and weaknesses of a design like actually implementing and using it.
- simulo 5y agoThe challenge for end-user focussed design is that it is very resource intensive to develop parallel UIs (and for many devs they are the least fun thing to implement), whereas it is not that much of a problem to add an additional parameter to a command line program.
- webmaven 5y ago> I sometimes wonder if UI/UX passes on open-source tools would be better with very explicit "experimental forks" that keep with upstream when they can. Over time, the successful fork starts to get more traffic from regular users and re-merges. Probably the best example of this approach is the Firefox (originally Phoenix, then Firebird) origin story: https://www.wired.com/2005/02/firefox/ https://www.wired.com/2005/02/firefox/
- throw_m239339 5y agoBlender also used to be closed source. Blender foundation had to crowdfund the source code to make it open. I was already using Blender in 1998 when it went freemium and it was already quite evolved, I didn't feel like the UI was bad compared to 3DS Max or Lightwave.