My computer science life involves dealing with mostly revisions/updates on protocols designed to address problems that have been there since a few decades ago.
Surely, there are new, albeit niche problems that must have prompted people to create protocols to solve them. I am interested in knowing these problems and how they are being solved.
Gemini is purposely designed not to be extensible, and to omit features that would allow the plagues of the 21st C, such as user tracking, fingerprinting, resource-hungry and bloated clients, excessive interactivity via JavaScript etc., to be a part of the user experience.
Despite that, it's pretty usable. Creating a Gemini site is not terribly complicated, and if you don't want to use Gemini's native markup language, you may use something like Markdown instead because many Gemini clients support that. Interactivity may be added via server-side processing, just like CGI.
One of the most technically impressive/surprising solutions in Gemini is how they leveraged TOFU [1] and TLS keypairs to provide an optional decentralised user identity protocol that effectively obsoletes the need for SSO - because the browser can automatically log you in via your TLS keypair.
[0]: https://gemini.circumlunar.space/
[1]: https://en.wikipedia.org/wiki/Trust_on_first_use