Notices by Janne Moren (jannem@fosstodon.org)

@mntmn
No, not really. And that's a reason why small LLMs as language processors (not chatbots) are exciting.

@mntmn
I mean, there's been many attempts. Especially for constrained applications such as a corporate document store and things like that. As far as I know, none of those systems were ever a success.

@steve @dalias
Also, I think you want the flow to be predictable and adjustable as well. For instance, you probably want to be able to shut it off if needed, without having to prime the passive flow again.

Wonder if you could make a non-electric active system with a Stirling engine powered by the water/air temperature difference, that directly drives the pump?

@stevenaleach @anildash
I use CLIs every day and have been for 40+ years. I also teach how to use the CLI and I support beginners and advanced users alike.

And while it is powerful - and, arguably, necessary in my field - the CLI also sucks.

There's no discoverability. Users live in constant fear they will break something and never even notice.

The CLI is fine *as a complement* to a GUI. Lean into its strengths. But it was never a viable interface for computing in general.

@cstross
The movie rights may be worth more than the jewels.

@inthehands @thomasfuchs
I'm a (former) neuroscientist. A lot of our human data comes from accident victims, and I've met a few of them.

If you break your leg you might have a limp. If you break your skull you may never again go to the bathroom without help.

I always wear a bicycle helmet.

@dymaxion @suzannealdrich @cstross
I was going to say the same thing: none of this hardware is usable (never mind useful) in a homelab.

With that said, 3-5 years is, I think, not unrealistic. We have racks with 8-9 year old GPUs and they're still in constant use today.

While it might not be used for training, they could still be used for other tasks, or find second life in academic HPC centers.

With Moore's Law on its last legs, they will likely be useful for a long while.

@thomasfuchs
Did you possibly reply to the wrong post?

@brunoph @etherdiver
Many (all?) "non-creatives" are creating stuff all the time. They (we) just don't think of it as "doing art".

You're adding trills or new harmony as you sing in the shower? You figured out how to plate your pasta in that cool swirly way, with a bright red tomato in the center? You draw your boss saying "poop" during a Zoom meeting? You reorganize your Excel sheets to be more balanced and visually pleasing?

You're being creative.

@dalias
Maybe; as you say, it depends on the details. But also, it's different when someone posts a single marginal thing in an otherwise normal stream, or if their post history is full of similar stuff.

If you're not sure, report. Better too many reports than too few.

@dalias
Report them, please!

@redstarfish@freesoftwareextremist.com @jimsalter
The FSF had a choice. Stallman was already on the way out. They could have done what was good for the free software movement and properly cut ties with a divisive, increasingly problematic - and clearly no longer effective - leader.

Instead they chose to be the Stallman Appreciation Society, and make the ideas of Free Software completely irrelevant for the future.

I am convinced we need something like the FSF. I'm equally sure the FSF is no longer that thing.

@aeva
I filter out as much US news as I can. And it definitely makes my feed a lot more quiet.

Still, I prefer that to a constant flood of anxiety inducing posts I can do nothing about.

My employer #OIST ( @oistedu.bsky.social ) is opening a special grad student application round for students currently studying at US universities.

The application period begins now and is only two weeks, but if accepted you may be able to start already in September.

Please reshare to get the word out.

https://www.oist.jp/admissions/special-application-deadline-students-accepted-or-studying-united-states

#USA #University #gradschool #Japan

@me
The advice I hear is that the included Windows Defender is as good as any of the others, so no need to add something else. Especially as AV packages by their nature tend to become security issues of their own.

A random thought for no reason at all: When japanese car companies want to sell cars in the US they don't just build them there. They have design teams creating variants and whole new models to fit the market.

US cars don't sell in Japan because US carmakers aren't putting in the work. Never mind being the wrong style or type of car; never mind being too large for local roads and parking spaces. Most models on offer don't even come with right-hand steering.

#car #japan #USA

In a new poll from Asahi Shimbun about the US-Japan alliance, 68% said Japan should be independent; and 77% don't trust that US would aid Japan if needed.

Those are pretty sobering numbers from a country that has been highly supportive of this alliance for half a century.

https://digital.asahi.com/articles/DA3S16202756.html?ptoken=01JSTQAY7AJRFD0NMNY7QP6YWR

#USA #Japan

FYI, if you happen to be a scientist in a large English-speaking country with uncertain future research financing; my employer, Okinawa Institute of Science and Technology (OIST) is looking for grad students, postdocs and PIs.

It's a research institute, so no undergraduates, and the working language is English - no japanese required. And its located on a (semi) tropical island in the Pacific.

https://www.oist.jp/careers?f%5B0%5D=oist_job_type%3A819

#NSF #OIST #Japan

@tomw @loke
In normal use I would say you'd use "dag" ("jag är på resa i fem dagar" - I'm traveling for five days), and only use dygn when either there's a risk of confusion or you want to emphasize the 24-hour span for some reason.

So with "han har varit försvunnen i tre dygn" (he's been missing for three days) it emphasizes the time span and the seriousness of the situation.

@loke @cobolworx
If you see it as a precursor or alternative to excel it's really not a bad language.