💾 Archived View for tilde.team › ~smokey › logs › 2022-10-05-sneaker.gmi captured on 2023-09-08 at 16:39:31. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2023-01-29)
-=-=-=-=-=-=-
The question at the end was a good one.
would Gopher and Gemini pages really be worth passing around a drive?
Gemini and gopher pages are largely blogs, recipes, and informative essays on passion projects/hobbies. This question is veery similar to
are most personal homepages worth sharing around?
The answer largely depends on what your personal taste in reading is, how much you value blog spam and human based content for creations sake, informational value to storage ratio which plays into the ease of archival/sharing.
Imagine if libraries and other such public informational repositiories had flash drives that could be loaned out containing things like wikipedia (like kiwix project does) or in our case, a sizable portion of the small net. Im sure the internet archive team could figure out a way to adapt their services to gopher+gemini, and archive a SUBSTANTIAL portion of both protocols with much greater ease than the web.
Lets say for simplicity sake we want to archive all of gemtext+plaintext in gemini space, how much storage would we need?
Just doing some simple math with the current gemini stats over at
276,285 total gemtext
41,201 total plain text
317,486 total text files
The average size of all gemini pages is 1,237 bytes (1.2k) or less
Average gemini page size x total text pages (gemtext + plaintext) = total memory size needed to save a large portion of geminispace
1,237 bytes x 317486 text pages = 392730182 bytes (392.7 MB)
Huh, it seems that a full text archive of gemini in its current form would be less than half a gig. Not bad, honestly. I didn't think it would be THAT low.
Of course, the above equation is a mere toy model, statistics is reality is not nearly this clean and cut, and theres huge margins of error both in the provided statistics and in our assumptions.
So lets do what real staticians do, adknowledge our assumptions are most likely bullshit and account for it by overshooting the numbers, and cross our fingers in a real mathematical way hoping its close enough.
Lets say that the real text file archive of gemini is 100 times larger than what our simplified toy model suggest. That would be about 40 gigabytes. 40 gigabytes is well enough to realistically capture most of geminispaces text based data, with well enough to save pictures to boot. Funnily enough 40GB is about what wikipedia is as a download at the moment.
40 gigabytes was an unfathomable amount of data 20 years ago, now every schmuck and their grandmother has a 250GB, if not more. Now the average server maintainer can afford many terrabytes worth of storage for a couple hundred dollars. I am willing to bet that ALL of the small net protocols data could comfortably fit within 10 terrabytes, probably a lot less.
One more thing to note. There is already somewhat of a gemini archive in the form of Delorean, the kennedy search engine's lesser known sibling which does its best to cache many capsules indexed by kennedy. I wonder how big the archive is, perhaps ill shoot acidus an email and get back to this.
Delorean: View cached capsules
I like the idea of mesh networks a lot more than sneaker nets. Recent advances in radio technology and data compression has made them a whole lot more viable as an option. I fantasize about a kind of 'mesh router' that acts both as a way to access the network, and be able to host your own 'homepage' equivalent to a capsule, website, gopherhole, ect. You would be able to acess websites that are physically closer to you and thus perhaps more locally relevant. Would help connect people together, and a mesh network means cutting out the ISP for good. So basically it would be a small, local intranet of people who pay once for the box/radio and connect with eachother on mesh protocols.
Now imagine if something like RSync was developed for mesh networks that allowed multiple users to contain up to date archival copies of local nodes, AND send those localized nodes to other nearby nodes. Were getting into cryptographic ledgering at this point, which is some advanced near technological magic type stuff thats well beyond my ability to really grasp, but I can see it. The picture is there.
Forget about sneakernets, we need to develop more sophisticated mesh networks able to run independently by the users that can cut the ISPs out for good. Even if its just good for sending text based information, small net protocols like gemini, gopher, spartan, ect, all show that there is a demand/desire for text based content among many people, and that simple technologies have the power to convey great ideas in a more natural, human way.
Archives would be VERY important in these kinds of networks, even implementing sneakernet to help move some of the big ones to a node. As each user in a node is partially responsible for storage space and data transmission rates, its important to keep things like storage space vs informational value in mind.