Hey, I just finished a little hobby project to make a mini social network running on gemini. If you’re a gemini fan, it’s live here: gemini://station.martinrue.com
Hey, I just finished a little hobby project to make a mini social network running on gemini. If you’re a gemini fan, it’s live here: gemini://station.martinrue.com
Since I’m sure I’m not the only one who missed the memo on Gemini: https://en.wikipedia.org/wiki/Gemini_(protocol)
Oh neat! See also, gemini://flounder.online, though this is a little different from a first look.
Thanks!
I’m sad that people keep trying to import the worst bits of the WWW into gemini. Gemini makes it really easy to host your own server/capsule/instance, and the adopted cultural practice is to write your posts on your own site and link to others.
I hope that attempts to bring centralising web services and silos to gemini fail, to be brutally honest.
To be honest, most people don’t want to do that. There’s still costs and burdens associated with hosting anything, and (even federated) servers can become policy domains/communities of their own.
If you truly want this, use a P2P system.
The good news is that things like this usually die out pretty quickly due to lack of interest. You don’t really have to worry about it.
This is basically the problem with Gemini: it does little to nothing to solve the actual problems it sets out to solve, they just seem solved because of the conventions of the currently-small user base. If it ever actually took off it would be the same as the web very fast.
I’d add that the basic issue with Gemini is that it’s trying to address social problems with technical solutions.
Can you elaborate on what these actual problems are that you see Gemini trying to solve that are only solved because of a small user base? Looking at https://gemini.circumlunar.space/docs/faq.gmi, a lot of the design goal has to do with Privacy. Gemini isn’t going to magically get cookies just because lots of people start using it.
@gerikson I hear this all the time about addressing social problems with technical solutions being some sort of bad idea. What is the alternative? I’d argue that the only way (besides superhuman Mahatma Gandhi or MLK levels of persuasive power) to solve social problems is with technical solutions. Now, you can’t just focus on the technology by itself. That’s certainly a failure mode. But social problems often boil down to some technology permitting people to do too much (e.g. cookies). Restricting people from doing things we’ve come to realize have negative externalities, this usually requires a change in technology.
Gemini is trying to work against the current widespread and profitable harvesting of online activity to target ads.
Effectively combating this requires legislation. Individual actions will do nothing to change it.
For example, we’ve had ad-blockers for a decade or more. Has online advertising gone away? No it has not. A costly arms-race between blockers and detectors has ensued instead.
Trying to defeat tracking online by offering a solution that’s much more limited than even HTML 4-era web and expecting a significant adoption is bordering on delusional.
I don’t think they expect adoption - it feels more “pretend it isn’t real”.
The three items at the top of that page are:
Tracking: Gemini may lack some of the obvious “tracking features” like 3rd party cookies, but given the sophistication of fingerprinting solutions big tracking uses at this point, that’s at most a speedbump to them going online. Privacy of Gemini users is largely safeish due to Gemini hosts being largely “good guys” at this point.
Flashy design is bad: One can deliver any content over Gemini, and as big publishers and the users who like their flashy stuff came in, the flashy content (whether HTML-over-Gemini, SWF-over-Gemini, or some new future abomination) would follow.
Bloat is bad: this is basically the same as the last point. As people put out more and more flashy stuff, the bloat they put out would also increase.
In context, my comment was also triggered by the parent post whose point was social networks / non-self-hosting is bad, which the OP is a contradiction to in the Gemini space already.
I likely just don’t know enough about Gemini, but:
How do you do images over Gemini?! I thought there’s no way to do it.
If each page loads as a single resource, isn’t there just much less room for fingerprinting? Clients are also not required to cache anything. I have a hard time seeing how one domain can learn anything about another. The spec has one line of request going out and one line of response header. How could this possibly be fingerprinted?
There’s no contradiction here. You literally can’t insert a newline in Gemini. The bloody protocol spec limits how long posts on this social network can be. Talk of bloat sounds like a lot of pearl clutching and false equivalence to me.
I flagged @kline’s post as unkind. This is the worst part of www, really? No need for the mainstream to worry when we’re all so good at tearing each other down. I have criticisms of Gemini, but I don’t feel the need to bring them up when people are sharing projects.
On how to do images: right from the FAQ you linked you can serve any content type that has MIME type over Gemini, so images should be easy.
For ad tracking you don’t need to have the client involved, anything a server can learn about the client it can relay to another server. Advertiser just has to make doing so a requirement of sellin their ads. This assuming the most rose coloured future where client side tech somehow never gets served via another MIME type for tracking purpose.
Oh, I understand that you can serve images over Gemini. But can you embed them? I don’t believe so. Requiring images to be explicitly requested seems like a huge improvement.
You’re right that servers can just sell your information outright. Is your point that it doesn’t make sense to worry about privacy if you choose to interact with a server? That’s internally consistent but an impractically high bar, IMO.
How would you go about fingerprinting Gemini users? Gemini has no user-agents or interactivity; clients just open a TLS connection and show the resulting contents. All you really get is an IP address. What other information can be gained?
Gemtext leaves presentation in the hands of the user agent rather than the content author; authors only dictate content with per-line semantics.
What sort of bloat are you envisioning in Gemini? One of the core ideas of Gemini is that one user action should trigger no more than one request. For example, a user must click a link to load an image (inline or in another window, depending on the client).
Finally, the best part of Gemini (imo) is the fact that it isn’t mainstream, and thus won’t be plagued by corporate interests. It’s actively hostile to corporate interests like branding (emoji favicons were recently rejected) and tracking (one-way communication, one request at a time, without any identifiers beyond an IP address or optional client cert for auth).
I invite you to try out Gemini before commenting with a blanket-statement dismissal.
I agree that if you serve only Gemtext, over any protocol, bloat is unlikely.
Your “best part” is in full agreement with my post. So long as it never catches on Gemini can remain largely pure because only people who want the culture happen to use it. Don’t need a whole new protocol or format for people to choose to be good, though…
It’s like “use plain text email” - it works in their tiny bubble, fails to meet reality with i.e; your clients, non-technical friends, important orgs, etc.
I don’t think this project is trying to import anything from the Web. It’s creating it’s own thing and allowing ppl on Gemini to socialize. Or even if they are not on Gemini, to be able to participate and this IMO is the built of such projects. It can help bring more people to the protocol. IMO, for a young protocol such as Gemini, project like this can really help spread the word and I welcome them.