A new UI paradigm and self-sufficiency

April 9, 2022

I wonder if it’s time for a new user interface paradigm.

On one hand, we have the computer terminal, which has been around for about 50 years now. It was designed for hardware from a different era. While modern terminal emulators (I currently use foot) have things like sixel image support, colors, and features that let you click on URLs, it is still a fairly primitive design. Terminal UIs (TUIs) generally only support keyboard input, and they are pretty simple, which is good because there is a wide variety of implementations to choose from. The majority of the programs I run on my computer are in a terminal emulator.

On the other hand, we have the web browser, which supports many different communication protocols, HTML/CSS/JavaScript, and is the probably the most complicated piece of software most people use today. The same technology stack is reused to create portable client applications with frameworks like Electron. The problem with modern web browsers is that the complexity of creating one is so high, that only a handful of companies can spend the engineering resources to build (actually, maintain) one, and the barrier to entry is so high that all of the existing players have consolidated around a couple of engines. The cost of creating a new one is too high for anyone at this point.

Both of the above UI paradigms communicate sending the display data over a connection, making it simpler for the apps to run locally or remotely. In contrast, we have a bunch of GUI toolkits libraries, such as GTK, Qt, etc., which have language bindings allowing users of a specific programming language to create GUIs with them. It feels like these are waning in popularity as they are harder to learn and use. The generally suffer from being harder to make cross-platform than the alternatives. Many GUI toolkits only work well in the programming language they were designed in.

Finally, it seems like we have a fourth category comprised of new frameworks created by big companies. Unlike the UI toolkits in the paragraph above, these frameworks, such as React Native and Flutter, aren’t libraries you add to your application, instead it’s more like you’re writing an application inside their framework. For React Native you’re using JavaScript to talk over a channel and with Flutter you use Dart. While I don’t closely follow the latest UI trends, it seems like these are gaining in popularity.

Will the future of GUIs be controlled by a few big companies similar to how a few big companies control the web? Will it be possible to simplify the web browser feature set so that indy developers are able to create a compatible implementation? Will the future of mobile and desktop UIs be dominated by frameworks that only run on a single programming language? It would be nice if we could create a specification like the web that could design native UI applications. Could the UI paradigm be more like a Lisp Machine REPL? Would it be based around someone else’s choice of design aesthetic or would the user be allowed to choose the UI components?

I think the unfortunate part is that average consumer does not think about free software freedoms, or even what it might take to be self-sufficient in the software that they use. Many users would be in a bad place if their account provider (social media, email, etc.) decided to delete their account. Being self-sufficient and not worrying about that is too hard for the average computer user these days. Even if it was easier though, I’m not sure they would have the incentive.