[SPEC] Encouraging HTTP Proxies to support Gemini hosts self-blacklisting

[Bradley D. Thornton]

On 2/22/2021 6:21 PM, Jason McBrayer wrote:
> 
> Dave Cottlehuber writes:
>> 403 Forbidden is ideal for this, or one of the 50x error codes. In
>> practice most systems will retry a 50x request but not a 403.
> 
> I feel that by analogy, status code 418 might be appropriate.
> 

Oh Jason you beat me to the punch lolz.

Yes I was just this very minute playing with ASCII-art generators for a
kewl teapot response page delivered with status code 20.

I figured I might need this after checking a few  webproxies. some
respect robots.txt, some don't. I understand that some of those authors
may not view their proxy as a bot, but middleware it is. It may not have
ever occured to them to accommodate that part of the spec.

to further complicate things where my philosophy is concerned, and this
was pointed out yesterday or so with the explanation as to what exactly
a "webproxy" is, due to the necessarily vague verbiage in the spec.

What if, Someone uses something like what Bitreich has with their SSH
kiosk, to enable people to burrow Gopherholes? I've actually thought of
deploying a service like this for Gemini...

But if I authored such a thing, I myself would have to say that yes, I'm
going to have to consider my product a webproxy and not violate a
Disallow rule for webproxy user agents.

And certainly, my specific intent here is not to stymie that sort of
vehicle. My beef, is I don't want any of the unique content that I
publish in Gopher space or Gemini space leaking to HTTP space. I just
loathe the thought of that - What purpose does Gopher or Gemini have
where the client side is concerned if it's just going to be published in
HTML.

In fact, I rarely, and I mean rarely ever publish anything in HTTP space
that exists in either Gopher or Gemini space. I maintained a package
repo containing well over ten thousand slackware packages back in the
Slackware 9 and 10 days, and none of those packages were obtainable via
HTTP - only gopher.

To me (no point in debating me on this as it's simply how I feel about
it), having "Unique" content in a particular aspect of a network breeds
relevance for the usage of that space. With Gemini, I see much more
potential than simple phlogs in Gemini capsules here in the future.

But enough of why I feel that it's important enough for me to be kinda a
dik about it with respect to the way I brought it up the other day. I
try to be a lot nicer than that most of the time lolz.

Now I'm just thinking of my own properties in the following, so it may
or may not be applicable or attractive to others now or in the future -
or NOT. Either way is kewl :)

So short of /dev/null routing tables, I think a CIDR based event handler
for Vger might be worth a go, because as it has been pointed out, it's
an actual user that will receive the page for all of those URLs. Instead
of baking this into the the daemon of Vger, perhaps it would be more
elegant to forward the packets to another Gemini service with something
like go-mmproxy and simply serve a single page with that HTTP 418
graphic for any and all requests?

I'll still have to manually hunt down the offending webproxy bots that
refuse to comply with the published spec:

gemini://gemini.circumlunar.space/docs/companion/robots.gmi

But that's not so difficult, since most are eager to advertise as a
public service (and most with good intentions, to be certain).

I'm long past being angry about it. Now it just feels like a fun little
tinkering project to play around with. This problem is not unique to
Gemini space - the bat bot phenomenon has plagued the HTTP space for
decades, but as the spec points out, it's different with Gemini, because
there's no user agent to assist in the identification process.

Although I think serving a .gmi with an HTTP 418 graphic is quite
hilarious, there's an undercurrent that is sinister on another,
non-technical level.

I have wanted to believe that I can incorporate copyright law into the
things that I personally wish to share with the world by using things
like the GPL v2 (and in some instances the AGPL) or with a CC-BY-SA. But
this matter has me questioning if that's going to afford others (and
myself) the protections I choose for my works.

Do I need to do something like "Copyright 2021 all rights reserved?" or
will simply racheting things up a little bit to a CC-BY-ND legally
protect my intellectual property from being converted into HTML?

And what about the other user agents? Search engines like Gus and
Houston? Will they think that maybe they shouldn't crawl and index
servers that state that webproxies aren't welcome - I certainly don't
wish for that to happen, those are not simply valuable services to the
community, but in the coming years they're going to be vital.

And what about the archiver agetns? are they going to store their
results in space that includes HTTP servers?

It's potentially a whole can "O" worms.

My position is quite simple. I don't want ANYONE to be able to read,
surf or access content on Vger from a web browser via HTTP protocol.
This of course, excludes plugins like Geminize, because the user is
actually using native Gemini protocol to access Gemini space.

Very simple concept to me. It makes Gopher more valuable. it makes
Gemini more valuable (provided the content is worth visiting via the
native protocols).

And I really don't want people to have to adhere to a "No Derivitive"
clause in a creative commons license. I want them to be able to take my
toys and edit them and share them with others in a form that suits the
greater good (Yes, even if  that means they put that shit on a web
server - they just have to retrieve it via Gemini or Gopher client).

I really don't know the answers to all of the questions that may raise,
and maybe that's where the discussion should go, coz I don't see any
roadblocks adressing this technically, or in a neighborly fashion. It's
the people who choose an immoral approach that may beg of licensing and
copyright discussions.

But in the end... Hey man, what do I know? I'm just a teapot :)

Kindest regards,




-- 
Bradley D. Thornton
Manager Network Services
http://NorthTech.US
TEL: +1.310.421.8268