[belugacat]

A lot of open source projects suffer from “pet project” syndrome. Some maintainers have very strong opinions about how/what things should be done regardless of actual user demand or feedback, and there’s no one with the power to rebuff them because the product owners deciding what to implement and the engineers doing implementation work are one and same. Commit right makes might.

In a (healthy) company, you have PMs and executives who will tell the overly opinionated engineers to STFU and actually implement things that move the needle and solve problems users are facing.

This is also why most open source projects have terrible UI/UX and any designer who attempts to help and improve things finds themselves ignored, with no means to actually carry out any decision, and walk away soon after.

[oefrha]

In fact larger open source GUI projects also tend to suffer from the opposite: with multiple equally powerful maintainers, a good number of contributors, and no one to rally them behind a coherent design, you get a messy bag of poorly documented features spread all over the place. Design by programmer is usually bad enough; design by a loosely-connected web of programmers working almost independently is even worse. Also, a lot of the developers/users are ideologically motivated and heavily invested, when you point out how bad the software is for the average user, instead of reflecting on it they often get defensive.

[davidhyde]

> “and there’s no one with the power to rebuff them”

Except for those that discover the fork button. /s

In all seriousness I think that what you mean is that people rarely have the power to fork a project and drag all the users and development expertise over to their version. Especially if they are simply users of the system and not contributors. Users of commercial software can get companies to do stuff for them with their wallets.

[HeckFeck]

> any designer who attempts to help and improve things finds themselves ignored, with no means to actually carry out any decision, and walk away soon after.

Ah yes, remember how long it took before image thumbnails appeared in the GTK file picker? Finally in 2022, but that was after countless forks and patches submitted that implemented them.

https://jayfax.neocities.org/mediocrity/gnome-has-no-thumbna...

[prox]

It stems from a bias, developers are power/expert users and their interface behavior is likely matching, which means they are heavily reliant on keyboard shortcuts (Think Blender 1.x and Blender 2.x who leaned heavily and even exclusively on shortcuts)

This gives them the idea that their software is very powerful (and probably is) but this creates a cliff to the beginner and intermediate level users.

A lot of examples can be found in “About Face: On interaction design” by Alan Cooper.

[robertlagrant]

If maintainers are paid to produce a feature, they should produce it regardless. Otherwise they're breaking a contract.

[ShadowBanThis01]

Well, Resolve suffers from the same attitude in places. While a good and usable product, it suffers from many design gaffes that go unaddressed year after year, despite denying users functions that are expected in any similar product.

The render-job UI is pretty shambolic (as is the treacherously inaccurate timeline on the same page), for example. And BMD refuses to add simple functions like "match timeline properties to clip." Instead they have a prompt that offers to match only frame rate, in contrast to just about every other media-editing app I can think of. Fixing that should be easy; and if it isn't, the codebase must be hopelessly inept.

The integrations of Fusion and Fairlight are buggy and exhibit nonsensical and misleading UI behavior. Resolve has a node view... oh wait, now it has two node views, which are not integrated with each other or with the editor. Fixing that should be priority one, which could create a truly excellent hybrid product that some of us have asking for for years.

And now they've ported the thing to the iPad. Really? I realize that from the outside we don't know how engineering people are allocated, but nonetheless it seems absurd to spend resources on that while the desktop product suffers from serious defects that have languished for a long time.

[imtringued]

That is basically how we ended up with FreeCAD.