← Back to context

Comment by userbinator

13 hours ago

Maybe this is just me showing my age, but I don't understand why reinvent everything when you could just go back to something like HTML 2.0 or even 3.2 with some minor changes. I probably hate what happened with the "modern web" as much as the Gemini developers, but going full NIH is unlikely to be a good solution when there's an existing "unmodern web" to develop for, and as a bonus, can be experienced even with a modern browser.

Never underestimate interoperability.

Gemini is a reaction to all the negative changes that happened to the web. Their premise is that the design of the web was fundamentally flawed because it was extensible. You see this in both the protocol and the community where the dominant idea is basically, make it super hard to ever extend or change Gemini. HTML is clearly a failure from this perspective, because HTML changed.

Personally I don't have much use for this attitude, the main problem I have with the web is when faced with the choice of empowering the publisher vs empowering the user, we kept on choosing to empower the publisher. The standards and browsers were owned by the web publishers, no one represented the user, and now instead of having a "user agent" installed on your machine to browse the web, you have a piece of spyware, better referred to as "Google's agent."

I don't really need to trade Google's Agent for Drew DeVault's Agent, give me software that does whatever I want it to do, fuck the publishers. But what do I matter, I'm not building any of this stuff.

  • Gemini is technically extensible. The only reason servers and clients don't add "unofficial" features which eventually become official, like HTML did, is because of the community.

    But a community could form around early HTML, make clients and servers that only support early HTML, and vow to never support later HTML features. Such a community wouldn't be much different from Gemini's. Psychologically, they would have more difficulty rejecting new features (that have already been implemented), be less exciting initially (since they have less novelty), and have more trouble distancing themselves with mainstream HTML. But psychologically, Gemini is apparently reinventing the wheel without any advantages over HTML...and that disadvantage, at least to me, feels worse than the aforementioned advantages.

    EDIT: I actually think gatekeeping a community with a different protocol may be a good idea. But I haven't heard about any technical advantages of Gemini (e.g. a protocol design that would be especially hard to extend, like a bloated spaghetti protocol on purpose) and I think that's a wasted opportunity. Nor have I heard about anything particularly interesting in the Gemini community, which makes me think the psychological benefit of a separate protocol isn't enough for an effective community, Gemini's community would need some other advantage, then perhaps a separate protocol wouldn't be necessary.

The advantage of a separate protocol is that you know that every available site will be limited. If you use a search engine, it will give you a result your limited browser can read. If you use an old or limited web browser (Dillo, e.g.), then you will still have the problem of discovery of Smolweb content.

HTML 2.0 and NOSCRIPT are very hard to enforce both server and client side.

  • Are they? On client side, technically it is easier not to execute scripts then to do so (despite the UX of https://addons.mozilla.org/en-US/firefox/addon/noscript/ might let you think the opposite). Technically, with DOM inspection one can also easily filter out elements you don't like, both on client and server side. It is literally one XQuery/XPath away.

    The problem is that most modern "apps" stop working once you prevent them from exchanging data with third parties or using nowadays-standard APIs such as XHR or Websockets. This is why a radical cut was chosen by Gemini.

Interesting outro. Interoperability is presumably one very big reason for this protocol.

As for why, all I can say is, download Lagrange, go to gemini://bleyble.com/cgi-bin/random, and see for yourself. It's one thing hearing about it and a completely different experience browsing the geminispace.

  • The different experience is largely thanks to different content, not different protocol. The protocol just serves a gatekeeping role to keep the community small enough.

  • That was a terrible experience. For a start that site has an expired certificate, as do many of the pages it suggested, and of the pages that worked it was mostly people that dipped a toe in a few years ago and never came back or other broken function.

If you reinvent something then people get to be a part of it. They get to help with implementation and maybe design. People like being a part of stuff.