I don’t think anyone can host a relay right now aside from bluesky.
People can host their own data / Personal Data Server, which is somewhere between self-hosting a mastodon instance and creating an account on someone else’s instance. The actual equivalent would be self-hosting your masto account separately from any instance (which is just not a thing with the current state of mastodon nor activity pub).
I like the idea, but then who gets to decide who is and isn’t a credible source? Is it only intra-account verifying? Can anyone verify anyone else, or do you need to be authorized by bluesky to start verifying others?
100% agree on this recommendation. Her Young Wizards / “So you want to be a wizard?” makes almost the exact opposite choices as the Harry Potter series does (magic is a choice not inherited at birth, Wizards feel responsible towards the rest of the world instead of wanting to ignore it completely, they learn on their own with their magic tome and some informal tutors instead of going to a school).
Succinctly put, though I got some cognitive dissonance when the author wrote about bluesky being their choice of decentralized network to get involved with without even mentioning the hosting costs involved with running a bsky relay (or whichever component of the ATP network actually holds the data “firehose”).
According to this article it took a server that costs around $150/month over 4 days to spin up a working relay, most of which was spent ingurgitating half a terabyte of data (that’s what ended up on disk in any case). Far from exorbitant, yet if I want to self host for my own personal needs it’s still gobs more data and compute than any activity pub software needs.
Maybe my view of “decentralization as in democracy” is just fundamentally different from the author’s. I get the feeling that to them, as long as each friend group has 1 self-hoster in it then democracy through decentralization is preserved. This would make sense that they orient themselves towards something like bluesky and the AT protocol. Personally, I don’t think we should be satisfied with that level of decentralization/democracy - it’s a nice start, but we should strive for reaching at least 50% of people self-hosting an activity pub instance to truly achieve the type of decentralization that serves democracy. Of course, I’m not aware of any activity pub software that can be selfhosted by even 10% of the population, currently, so there’s definitely a lot of work to do before my vision is feasible.
Thank you!!!
Doo you happen to have a good, informative link? Or perhaps a company name I could look up?
This sounds incredibly cool.
Thank you both (@NinjaFox@lemmy.blahaj.zone, @ChaoticNeutralCzech@feddit.org) for taking the time to make this post not just more accessible but somewhat more bit-/link-rot-resilient by duplicating the image’s info as a text comment.
We don’t talk about it as much as authoritarian censorship, ip & copyright related takedowns, and their ilk, but image macros/memes often have regrettably small lifetimes as publicly accessible data in my experience. It might be for any number of reasons, including:
or (more probably) a combination of all three and more.
In any case as silly as image memes are, they’re also an important vector for keeping culture and communities alive (at least here on the fediverse). In 5-10 years, this transcription has a much higher chance of still hanging around in some instance’s backups than the image it is transcribing.
P.S.: sure, knowyourmeme is a thing, but they’re still only 1 website and I’m not sure if there’s not much recent fediverse stuff there yet. The mastodon page last updated in 2017 and conflates the software project with the mastodon.social instance (likely through a poor reading of it’s first source, a The Verge article that’s decent but was written in 2017).
P.P.S.: ideally, OP (@cantankerous_cashew@lemmy.world) could add this transcription directly to the post’s alt text, but I don’t know if they use a client that makes that easy for them…
It’s a bit sad, but not that surprising, that if this is true then Microsoft is clearly not tasking their most experienced engineers on the control panel (you know, that part of the OS who’s function is to allow you to tweak all the rest of the OS?).
From what I understand its origin in street racing was because japanese drivers (specifically? might have been Asian more generally) were souping up cars to look pretty but still not run great. I’m hazy on the details and my google-fu is failing me - I wish I had a more precise answer but overall I recall being bummed out at how even the origins of the term weren’t as clean as I had hoped.
Ooooh, that’s a good first test / “sanity check” !
May I ask what you are using as a summarizer? I’ve played around with locally running models from huggingface, but never did any tuning nor straight-up training “from scratch”. My (paltry) experience with the HF models is that they’re incapable of staying confined to the given context.
I’m not sure if this is how @hersh@literature.cafe is using it, but I could totally see myself using an LLM to check my own understanding like the following:
Ironically, this exercise works better if the LLM “hallucinates”; noticing a hallucination in its summary is a decent metric for my own understanding of the chapter.
My reading of the article is also that the anode is bonding with the protons (aka hydrogen nuclei) as part of the redox process to generate current.
I got really into chip defense. The concept is a really fun twist on classic enemy hp mechanisms in tower defense. However I think the game’s controls design “scales” poorly with its difficulty curve. Building a new “tower” takes too many taps for the game to continue running in the background. Same for upgrading towers. The worst is when starting a level, you kind of need to rush placing towers to not mechanically end up in a fail state.
If the game let you build and upgrade when paused (or at least queue up those actions to be executed when un-paused) I would have continued playing it.
I don’t think I was making it out to be easy at all. I expect it to be nigh-impossible. I also expect it to be worth it.
But the question was what opinion on the industry do I have that I don’t feel comfortable voicing at work, not what do I think is the most feasible way forwards.
… That’s why I ended my comment with “we should be teaching others and helping them make their own”.
We should stop making software for others.
A prerequisite for reasonable tech use is understanding the amount of energy and materials you need to “burn through” for any given piece of tech to 1) exist and 2) do its useful work. Call me naive, but I really doubt that we’d be accelerating climate change this much if every person contributing to the “X thousand hours of videos uploaded to YouTube each day” was required to write their own video hosting software first. I doubt our social networks would become so captured by propagandists of every user of one had to write their own. (Obviously as an absolute this is a bit too restrictive - it’s more the tone and direction that I’m trying to convey).
Instead, we should be teaching and helping others reach our knowledge /skill level.
Maybe the execs would stop pushing shitty UI dark patterns if they had to code the service themselves (and then use it afterwards!).
Onecan^dream…
cacher does, but cache as in “cache-toi !” (go hide!) and “je me cache” (I’m hiding) are pronounced “cash”.
Besides, “correct” pronunciation in a different language is pretty meaningless. The word may have come from French but we’re speaking English, not French.
Also, it might not be a loan word so much as a legacy-of-foreigners-taking-over word (c.f. the Normand invasion of Britain), which doesn’t tend to help the language’s users care about respecting the “original” pronunciation. I’m not certain when exactly cachet entered English.
“What was Windows even doing for us?”
Beautiful 🥲
I wonder what other applications this might have outside of machine learning. I don’t know if, for example, intensive 3d games absolutely need 16bit floats (or larger), or if it would make sense to try using this “additive implementation” for their floating point multiplicative as well. Modern desktop gaming PCs can easily slurp up to 800W.
I don’t expect ai/LLM tools to make it easier to create nice looking art, but they should make it easier for anyone to create their own placeholder graphics that at least would allow them to run the game on their own.