The usability of open source operating systems

I often see newcomers to 9front asking about something along the lines of "Can I use this as a daily driver?", usually the implication being that if they had nothing but a 9front and a internet connection could they do all of what they tend to do on their day-to-day computing on some other OS. To which the reply is usually "well..." with the context that no we don't have a browser, no you can't play minecraft, and we only really have C (yes we have other languages here and there but C is the lingua franca).

People often add that you can indeed use more than one OS at a time, and indeed a large majority of 9front developers do as such. However I think it's important to point that at least for a decent chunk of people the notion of "is this a viable daily driver" is deeply tied to the grace of companies of making their (often for-profit) software running on your machine. Linus once mentioned that Chrome was a boon for desktop linux support because to a web site a Windows chrome instance and a Linux chrome instance are (or really should be) indistinguishable. But never forget this was only possible because Google thought it was worth their time to make Chrome work on Linux. This is not a new problem the whole "I can't run linux because I need my Adobe" was all over the place 15 years ago.

Non profit does get a bit better, there is of course a maintenance cost for supporting a different platform, and the more that system differs from the norm the more of a burden it becomes. The work for making web browsers continue to work on systems like the BSDs are largely held up by users and developers by that system. Things are better in that someone with the dedication and time can do the work for enabling their system and post the patch for inclusion, but it is still largely by the grace of the upstream project to accept it. You can of course maintain a fork but that only grows the maintenance.

At worst you have no avenue for making applications that users want work on your system, at best you can maybe roll up your sleeves and do what is needed, but the more you diverge the harder that work becomes and the more eyebrows you raise when posting the upstream patches. The BSD's likely only have support for various "mainstream" software because they're similar enough to Linux that things don't bit rot in to oblivion, projects like Haiku can get a lot of software running because they use gcc.

I'm sure for many a viewer I'm stating the obvious, none of this is exactly hidden. However as linux gets more "mainstream", I think its important to point out that the core problem here has not been fixed. In fact things have gotten worse I would argue. Without a solution, software has continued to grow more complex, more things to port, more functionality to emulate or implement. At this point writing a new system entirely from scratch and working up to supporting chrome is patently hilarious. Not to mention that the fact that ubiquity of chrome has lead to even more conceptually simple programs use it (or electron to be more precise) as their shortcut to cross platform.

How do we solve this? How do we get people to not make another god forsaken messaging app implemented in electron? I'd argue we need to stop allowing programmers to stand on the shoulders of giants, if you can't wrap your head around the black-box tech you use, you are not going to understand which parts of the interface are wrought with complexity under the hood. Complex pieces of code should naturally-select themselves out of existence the 50th time an engineer has a near panic attack working on trying to fix a bug in it. But we've gotten to clever, we've gotten really good at balancing the house of cards to keep things breaking just enough to not piss someone off that bad. The path to hell is paved with good intentions after all.

For operating systems, we've addressed the symptom not the cause. Linux got big enough for companies to start caring about targeting it, but the reliance on companies and allowing them to define what computing is did not go away. Just being "open source" is not enough, we need to support systems that can be implemented by one nerd in a weekend. If not, usability is going to continue to be defined by features outside of a developers control.