text to text @ayuranslounge - Tumblr Blog

it is unfortunate that there's no reason for most people to remember high school chemistry because the best analogy I have found for "the amount of energy that it takes me to initiate a task, which can be higher than the amount of energy it takes to actually complete the task" is "activation energy" and it's not precisely perfect but

yeah. and you can even include "thing that reduces the barrier to doing the task" as a catalyst/enzyme

anyway. unfortunately this does not actually clarify anything for the average person. but #ToMe it works

cipheramnesia

Business degree holder looking at these graphs would be like "I don't understand, these can't be correct because the line goes down. Fix the numbers and make the line go up.

foxes-in-love

prokopetz

I think that realistically, the thing that kills AI data centres isn't going to be cost, regulation, or public opposition. I think it's going to be the gen-AI equivalent of the WDM fiasco.

For context, back when commercial Internet service was just taking off and the dot-com bubble was at its peak, a bunch of companies put down vast quantities of fibre-optic cables in anticipation of future demand. The vast majority of that fibre subsequently ended up laying dormant, becoming what's known in the industry as dark fibre.

The dot-com bubble bursting is often cited as the reason why, but it can't account for all of the problem, or even most of it; demand for commercial bandwidth continued to grow in spite of the bust, albeit somewhat more slowly than predicted. The real culprit was the development of a technology called wavelength-division multiplexing, or WDM.

WDM tremendously increased the amount of data that could be transmitted using existing fibre-optic cables; post-WDM, many major urban centres found that they now had over one hundred times as much fibre as they needed to accommodate present and anticipated near-future demand for Internet bandwidth. Many of the corporations that had speculatively bankrolled that fibre installation ended up selling off their infrastructure for pennies on the dollar and subsequently went bankrupt.

Basically, I suspect that's what's going to happen here. The present rush to construct more and more AI data centres is being short-sightedly driven by predictions based on the first couple of generations of general-purpose generative AI technology. Any day now there's going to be some random unpredictable breakthrough that makes gen-AI models like a hundred times more efficient, and a lot of very powerful people are going to be left holding a lot of very expensive bags.

prokopetz

The plausible endgame for gen-AI isn't "the technology stops existing". It's "the bottom drops out of the infrastructure market because the technology reaches a point where a good-enough-for-Jane-Average version of the entire model can live in your pocket".

prokopetz

(If you're in the notes going "OP isn't this just DeepSeek", you need to understand that the efficiency gains achieved by DeepSeek are basically a rounding error compared to what could plausibly be achieved once we actually get this shit figured out – and if DeepSeek caused as much anxiety in the market as it did, imagine what serious improvements in model efficiency would do!)

ayuranslounge

It already has with Google's 6x reduction in memory usage.

What's next would be context and guardrails. Specifically partition then towards specialties in order to reduce the amount of rp needed to get them to start and stay down a certain branch.

It's all vector math, so it'll likely be more vector math. Maybe similar to how you can process audio down ideal paths. The tricky part is identifying this paths in the ai blackbox of data.

hometoursandotherstuff

amila-aart

Thrid batch of Furry sketches ✨️

stas-j-fox

✨ Unicorn ✨

ionomycin

alleyway meet up

azulcrescent

me smaht

i like blender. open source software restores my faith in humanity.

cerastes

All these big companies changing their logos a dozen of times for something that looks worse every time, while the real all-time GOAT found the winning formula on the first try and then even did a victory lap with it:

bug-swarm

Nothing! Traffic cones have nothing to do with media players. However they do have to do with the students that original made VLC. The original students used to steal traffic cones when they got drunk together and had amassed a collection of them. So the traffic cone was basically their mascot

molsno

beemovieerotica

derinthescarletpescatarian

I drink two cups of coffee per day not because I have an addiction, but because my body has adapted to the presence of caffeine enough that its absence causes mild but unpleasant side effects.

liquidpaperfoundation1

I dusted the TV stand not because it was dusty but because it was covered in fine particulate matter that settled from the air.

athena-cat-wizard-official

I reblogged this post not because I liked it and wanted to share it, but because reading it elicited a pleasant sensation that I felt would be appreciated by others

happilyshanghaied

The sassy ‘hm’ by the dictionary people is what makes this post for me.

azulcrescent

[Monhun fancomic] Mon hun bows

The stealth archer archetype is infamous for popping up independently in many different games, and just generally just stays as far away as possible while plinking at their enemies. Not mon hun bows tho. They're basically melee weapons with how far up the monsters' faces they are lol. In Rise, you even counterhit with a melee when you dodge past an attack, which is pretty awesome.

#tera bows used to be the same #falloff at 8m or of range at 12m #bunch of trap and kick skills with built in movement #but people complained

muttsys-deactivated20260425

Anyone else noticing theres no websites anymore. You never go to website.com anymore its always some bullshit like "go to the smeeple app store and download the flubi app" i dont want toooo what ever happened to websites. I do not like it

ayuranslounge

Honestly I'd rather they do just make an app rather than destroy a website trying to overcomplicated things.

I would rather that but... every app is trying to be built as if it were inside its own browser. Every website is built as if it were its own program.

I'm just trying to reproduce a problem at work and my adhd is screaming at me to do anything else while i wait 10 fucking second for a page to load.

bovid-beauty

stupidest thing about computer science has gotta be whatever the fuck they do to calculate font size. why is the largest 16 pt character not remdered as 16 px tall. what the fuck magic unit did they come up with

foone

It's stupid print shit. Computer fonts are designed to work like how fonts work in, like, newspapers and magazines and shit, which makes them work stupid on screens.

Points are an 18th century(!) unit of measure, because OBVIOUSLY the number of pixels your font uses should use the same unit as machines that melted lead to shove ink on paper

brianmcnett-blog

For a brief moment in time, this actually made some sense. In typography, a "point" is 1/72nd of an inch. Why is unimportant (probably the smallest one could measure in 1517), but in 1984, the Apple Macintosh had a display in which a single pixel was one point. Measuring fonts in points was equivalent to measuring them in pixels. Meanwhile, the Apple Imagewriter printer produced a dot-pitch of 72 dpi, and again measuring fonts in points made sense. Typewriters were also measuring fonts in points, but with limited choices (10pt, or 12pt). Printer resolutions went up. Postscript, Pagemaker... Screen resolution became disconnected from traditional typography rather quickly. Within six years (about 1990 by my reckoning), one-point-per-pixel was no longer a thing anywhere meaningful (low-end Macs still did that, but it was never actually a thing on Windows, and folks doing typography seriously didn't use low-end hardware). When your printer does 600dpi, it simply doesn't matter what your screen resolution is. And... as much as I like and appreciate @foone, points aren't an "18th Century unit of measure." They're a 16th century unit of measure, being based upon the work of Milanese typographer Francesco Torniello da Novara who defined the point as the ninth part of the height of the letters or the thickness of the principal stroke (a definition we don't actually use anymore, but he was first). Something, something exhaustive Wikipedia article on the typographical point, French, Belgian, Fournier, Didot, something something Benjamin Franklin importing French printing presses... American standardizing print on the "pica" (12 points). Something something Apple Retina display, Cocoa "points" automatically scaling to nearest whole number of pixels, Android density-independent pixel, Microsoft UXP... THE REAL REASON POINTS WORK STUPID ON SCREENS!

#something something vector math hard #solid lines bad #edge smoothing expensive #fonts aren't laid out by pixels

yaldabaothadeez

(source - TateOfTot on bluesky)

kaiasky

People at work keep comparing me to various sabotouers (in a complimentary way) like. "you're a ninja, you go into these massively complicated parts of the codebase and find a fix is going to be quick and minimally invasive" or "you're like a paratrooper, we can drop you on a project with no support and trust you'll be self sufficient". today was frogmen and limpet mines and I'm convinced that somebody is playing a prank on me now. Who the fuck even thinks about frogmen

brightlotusmoon

Telekinetic with ADHD surrounded by a debris halo of assorted small objects they've levitated around their head in order to have handy, but have since stopped seeing.