New IPFS bug just dropped:
https://github.com/ipfs/kubo/issues/10588

This is relevant to organizing pony images to original web paths ( >>/11314/), and I have more than one million in that 1.2 TB set. It's also relevant to arc*.sh WBM alternative, which needs commands to complete and not just hang forever in order to work. Over the past ~7 days Wayback Machine has been unable to make accessible captures. Each time it says there was a delay capturing whatever and it still doesn't show any supposed captures even days later. WBM's failures is alternatives' gains. (More on that bug: adding a 1.5 MB file = command never finished and after one or more hours used a lot of memory and oom-killed the ipfs daemon.)

Have you used WebTorrent before? I have ( >>/11310/), and it uses WebRTC to enable torrenting in web browsers. The creator of it gave a talk on it years ago - "Feross Aboukhadijeh: WebTorrent - JSConf.Asia 2014" (includes info on DHTs):
https://inv.nadeko.net/watch?v=kxHRATfvnlw
https://www.wikidata.org/wiki/Q52233158

I think PeerTube instances such as PonyTube use WebTorrent. Apparently Apple's Safari Browser still does not support WebRTC 10 years after that recorded talk. Created this torrent via Instant.io but couldn't download it in another tab (torrent payload attached):
https://instant.io/#eae9d8c4877f0595d82ccbaed460aaa67137b8a4

It's an image file which show womanized MLP characters, basically the same as this folder (instant.io tab refreshed so I lost the randomized filename in it = that torrent specifically is pretty much gone forever, which does not matter much in this case):
https://ipfs.hypha.coop/ipfs/QmdbSmeK5gzhaFsag4KzGWZpEWDWJSqWHWtoWgWyTPY7yG

Archival headache. Archive.today hits me and probably other users with many Google car training captchas. Recent update and wasn't like that before ~3 days ago. The next crappy update started ~weeks ago: further 4chan enshittification, not referring to the recent 15-minute wait time update which began at a time around the halfway post in thread #41556719  >>/11256/ (halfway post would be like post #65 out of 130 total posts in that read-only thread). 4chan's CDN site which hosts images and videos now returns HTTP 429 even with delay=2seconds + concurrency=1. This prompted efforts to change some code in scripts and reprogram GNU Wget:
https://boards.4chan.org/t/thread/1310152#p1340333 to https://boards.4chan.org/t/thread/1310152/gap-gif-archival-project-3#p1341112 so far

Reprogramming Wget (written in C) was interesting and an exercise in understanding that code. Edit to Wget which makes a certain case go from originally exit code 0 to exit code 7 - should probably be exit code 1 to be consistent and so on (but one man's bug is another man's feature):
https://github.com/ProximaNova/gnu-wget/commit/8a1a0db8b3f6b21373ba1fb9da00db7367251e01

I imagine this is a headache for 4chan archive HTTP websites (those non-WARC ones), and desuarchive.org will possibly missing many /mlp/ images. (Desuarchive also makes /g/ and /mu/ searchable, so you can find stuff like download links to a bunch of albums from The Residents.)

 >>/11326/
> stream of consciousness
Perhaps you get that notion from the level of detail in which I describe things. I sometimes get into the nitty-gritty in regards to media, technology, and data. Such as:
. weird URL to file with HTTP 200: https://ipfs.io/ipfs/🚀🪐⭐💻😅✈📣🌔😘👻☀😥🌞☹🙋🖥😪😍😥💦🙇🤢😨😈☄💞😜💔💟😊🌔😂💨😙💦🖤🤑
. latest Kubo web UI: https://gateway.ipfs.io/ipfs/bafybeibgic2ex3fvzkinhy6k6aqyv3zy2o7bkbsmrzvzka24xetv7eeadm

I know that some users aren't so interested in details. That is sometimes a bad thing or ridiculous that they are so opposed to it. Other times, I can kinda understand.