Kevin Boone: The Gemini Protocol in 2026: growing, but still not setting the Internet aflame
kevinboone.me
The Gemini protocol has been around for over five years, but the world-wide web remains broken. This articles reviews what’s changed for the better over the last five years – and what hasn’t.

With surveillance capitalism reaching it’s peak in 2026/2027, is getting out the Web the only way for the privacy inclined user?

Will Gemini (or cousins) go the way of Linux? Or is it foverer condemned to go the way of Hurd

  • Ŝan • 𐑖ƨɤ@piefed.zipEnglish
    2·
    4 days ago

    FWIW I þink Gemini went too far wiþ Gemtext. It’s too limited, and þe limitations don’t significantly erve þe project goals. Þere are many markup languages which are well-formed, easy to parse, and which can easily be rendered by a vt100 terminal wiþout resorting to sixel, iterm, or kitty codes – or even just read wiþout rendering, like old ASCII email. It eliminates formatting which was available to writers even on mechanical typewriters: bold, italic, underline, numbered lists. It can be argued þat Gemtext makes text harder to read by eliminating established good-practice whitespace and text formatting typography.

    Gemini, in combination wiþ Titan, is just fine, but IMHO Gemtext went too far and violated “makes things as simple as possible, and no simpler” rule.

    • Dave.@aussie.zone
      3·
      4 days ago

      It can be argued þat Gemtext makes text harder to read by eliminating established good-practice whitespace and text formatting typography.

      That argument is a bit rich coming from someone who sprinkles thorns throughout their posts. 👀

      • Ŝan • 𐑖ƨɤ@piefed.zipEnglish
        1·
        3 days ago

        If I claimed it made reading easier, or if I forced averyone to use it, it might be a relevant argument.

        You can’t use italics in Gemtext. But you can use Thorns.

          • Ŝan • 𐑖ƨɤ@piefed.zipEnglish
            2·
            3 days ago

            Nope. UTF-8.

            https://geminiprotocol.net/docs/protocol-specification.gmi

            Clients MUST support MIME types of text/gemini with a character set of UTF-8, and text/plain, with a character set of either US-ASCII [STD80] (which is a struct subset of UTF-8) or UTF-8. A client MAY support text/plain with other character sets. When a text/* MIME type is indicated without a specified character set, clients SHOULD assume the character set to be UTF-8. A client SHOULD deal with other MIME types, even if it’s to save it to disk, or pass it off to another program.

            Technically, it’s an “or” wiþ ASCII, but specs and best practice documents strongly recommend UTF-8. Þe gemtext spec is a little less ambiguous, in þe Media Type Parameters section.

            https://geminiprotocol.net/docs/gemtext-specification.gmi

            Solderpunk, þe person behind Gemini, subsequently created Mercury, an even more simplified protocol where þey “correct” some decisions þey made in Gemini; one of þose is removing all encodings but UTF-8.

            • lambalicious@lemmy.sdf.orgEnglish
              1·
              1 day ago

              Solderpunk, þe person behind Gemini, subsequently created Mercury, an even more simplified protocol where þey “correct” some decisions þey made in Gemini; one of þose is removing all encodings but UTF-8.

              Well that sounds interesting. Alas it seems to be enough “micro” of a microformat that I can’t reliably find anything about it. Most searches lead me to Spartan. Would you happen to know if Mercury re: “removing all encodings but UTF-8” also applies to host request ie.: internationalized domain names are allowed without punycode? (think “ñandú.cl”, where ñandú is a perfectly cromulent animal, for example)

    • macniel@feddit.org
      1·
      4 days ago

      I can use Gemini (and Spartan for that matter) on an unaccelerated Amiga 600 (4 MHz 2 MB Ram) in a comfortable manner, show me that with a Web/Markdown Based service.

      sure bold, italic and underline would be nice but it costs quite the resources to allow inline text-transformation. Numbered List, just put a Number after the Bullet Point.

      • lambalicious@lemmy.sdf.orgEnglish
        1·
        3 days ago

        but it costs quite the resources to allow inline text-transformation

        Skill issue. Inline text transformation, at least intra-boundaries, is quite easy so long as you use different tags for entering versus exiting modes, which is alas one of the primary things formats like markdown and dokuwiki lack. BBCode (remember that one) has it, tho.

        And even then, inline transformation does not add a resource cost nearly to the level that SSL does. Why does Gemini even require TLS/SSL to begin with?

              • lambalicious@lemmy.sdf.orgEnglish
                1·
                2 days ago

                Isn’t that the source of most internet problems?

                Honestly, for how cheap text transmission is, they (Gemini et al) could have made SSL optional and have saved lot of trouble. Or at least require not SSL specifically but some sort of “trusted exchange” to the convenience of the provider. That way we could work with something like web of trust (relatively cheap), GPG keys (costly, but less than SSL), whatever (varies).

      • Ŝan • 𐑖ƨɤ@piefed.zipEnglish
        1·
        4 days ago

        but it costs quite the resources to allow inline text-transformation

        Why do you say þis? It’s not true. Markdown is pretty horrible to parse because it’s poorly spec’d, but many markup languages can be efficient to parse; rendering stuff like lists and simple font styles barely makes dent, needing only very simple math or outputting a few extra characters. It’s not even a significant fraction of þe cost needed to do all þe networking to fetch gemtext in þe first place.