iiinova — where your digital self keeps evolving

centralization could be addressed by letting many more institutions generate money, particularly ones without profit-seeking interests (e.g. public banks) as for inequality, many more people have written about it and I won't do justice to all that work in a tweet (or several)

000

thus MMT implicitly advocates for 2 bad things: - centralization of power - growing inequality MMT could address these issues but somehow doesn't even acknowledge them I'm not yet sure if it's intentional but it's extremely bad, thus I chose to call it "psyop-adjacent"

100

but even with the federal government, there's the problem that an increasing share of its budget is just interest payments interest is inherently a generator of inequality - those who already have money will receive more of it "radical" redistribution is then needed to offset that

100

everyone else (under the federal government) has fewer options - beg the federal government to give them money, beg someone else for a handout or a loan, or produce something that others will pay for and let them worry about getting the money

100

while I agree with the idea that austerity doesn't do anyone any good, the subject of the video (MMT) seems psyop-adjacent first of all, it only seems to apply to the organizations that can directly influence money "printing" - the federal governments

100

the situation could be fixed basically overnight if only people understood enough to create municipal (aka public/state-owned) nonprofit banks that can lend (and ideally also create) money with minimal interest rates

070

there's no money to borrow from those who need ROI but that was never sustainable it's not like there's ever an inability to print money for productive activities it's that the borrowing has been feeding the same greedy people who now collectively decided that almost nobody is worth their time

1100

well the stock market is a whole another story really the "AI" bet is all the debt, the chip purchases, construction etc. the stock market is a bunch of dumbasses thinking the line will go up and making it go up as a result, a self-reinforcing cycle

000

Arvīds Kokins @archo5.bsky.social · 03/10/2026

> Quit making everyone so anxious. they can't, they're desperate gamblers they've managed to turn a household gambling problem into a whole economy by betting the farm on people paying for the chatbot

000

wdym by "post-American" exactly? even if the federal US collapses, I'm not foreseeing any structural issues e.g. for the stock exchanges in NYC (cities tend to survive the rise and fall of regimes far better anyway) unless the US dollar collapses completely or something

000

Arvīds Kokins @archo5.bsky.social · 02/10/2026

lots of soap operas don't even mix much of anything, and people aren't even necessarily watching them attentively and yet they still seem to keep going afaict theoretically that's the easiest target for automation out of all motion picture media shouldn't they be annihilated first?

000

Arvīds Kokins @archo5.bsky.social · 02/10/2026

I hope that's not where things go lol self-centered creation trends against connective communication, and the underlying impulse arguably isn't too different from the urge to generate personalized slop the "optimization engine" still makes uncanny shit, one shouldn't take it too seriously

100

sqrt is a bit slower than floating point division, but not much www.agner.org/optimize/ins... division by 2 could also have a fast path with binary floating point because it can be done by decrementing the exponent

020

Desmos | Graphing Calculator

confirmed that the w values are also effectively the same www.desmos.com/calculator/e... after pasting the unified equation here, it became obvious trig identities aren't top of mind for me so my methods are way more crude lol either way, mystery solved

desmos.com

000

I've derived two versions of the quaternion scaling function one from slerp and one from the axis-angle functions the second one could be a bit faster if the sin/cos calls are replaced with one sincos instruction otherwise they ended up remarkably similar, yet w is very different

100

yeah, I guess there isn't one upon re-reviewing the code, I see there's 4 equivalent trig ops (1x acos, 3x sin/cos) in both code paths there are a few minor optimizations that can be done on a slerp with the first argument being the identity quaternion but that's about it

100

sorry, I just realized that I've been overlooking a really basic fact qˣ = slerp([0;0;0;w=1], q, x) so that would explain how it looks in code (already tested it too, seems to work)

000

the methods I listed are the same methods in the equation DiffAxisAngleScale is basically what the right side looks like "diff" (in my code) = a⁻¹b the angle scaling is diffᵗ mathematically there should be a simpler way to write the angle scaling though (without the angle), hence the questions 👀

100

isn't that just a different way to write exponentiation? I'm more curious about the practical aspects of it tbh e.g. quaternion components can have negative values but these operations aren't usually defined for that range (which is I assume where complex numbers and sin/cos etc. comes in)

100

wondering what qˣ looks like in code I've never seen a quaternion multiplication in scalable form I know one possible implementation (axis-angle roundtrip with scaling) but it doesn't seem to trivially relate to multiplication

100

This could legitimately end Stop Killing Games.

here's another video on the "European Commission attempting to destroy videogames" topic with even more detail: www.youtube.com/watch?v=mQRc... #gamedev #EU

youtube.com

YouTube video by Stop Killing Games

070

if there was an ownership reassertion law (e.g. you have to show up every 6 months to renew even indirect/partial ownership) then it would require lots of meetings or at least calls to make that work but that's generally the idea - making the "owners" do more work the more assets they "own"

000

open to ideas on how to succinctly legally reconcile these two principles ownership of faraway assets isn't in itself a problem (e.g. for travel) but it's being exploited ownership concentration laws would require a regulator to enumerate one's worldwide properties, which is a complex task

100

Arvīds Kokins @archo5.bsky.social · 29/09/2026

I'm increasingly of the opinion that the ability for any single person to own productive, tangible properties all over the world is one of the largest perversions of property law at the same time, I need any and all (incl. international) infrastructure to be publicly owned by the people they affect

100

"corolla" already sounds like "copilot" but for cars 👀

001

Arvīds Kokins @archo5.bsky.social · 29/09/2026

@ec.europa.eu could you stop being blatantly corrupt for just one fucking moment? cybernews.com/ai-news/ai-c... there is no "legitimate interest" for mass "training" the use cases are: - performance theft - profiling - identification - data laundering none of these are even remotely acceptable

cybernews.com

EU AI data proposal could allow training without consent

A leaked EU AI data proposal could let companies train models on Europeans’ personal data without consent. Read what privacy advocates warn

001

Reposted by Arvīds Kokins

Ed Zitron @edzitron.com · 29/09/2026

Reuters got some of the IPO docs. Anthropic is a total dog of a company. Spent $12.6bn to make $4.6bn in revenue in 2025, $7.33bn of which was compute costs. Operating loss of $8bn. Losses getting worse year over year. Amazing stuff www.reuters.com/business/fin...

Summary
Companies
• Anthropic IPO could value the company at more than $2
trillion
• Anthropic lost $42 billion in 2025
• Al lab spent $7.33 billion on compute and infrastructure last year
• Operating loss widened to $8.06 billion in 2025 from $2.98 billion in 2024
• Execs, industry analysts have expressed concerns over harms of Al tech
Sept 28 (Reuters) - Anthropic is making a massive bet that Al will transform the global economy more profoundly than industrialization, electricity and the internet, according to its IPO prospectus seen by Reuters.
But the cost to get there will be staggering. Anthropic reported a net loss of $42 billion in 2025, and plans to spend $518 billion on cloud, computing and infrastructure obligations in coming year, according to the
prospectus.

17341621400

Arvīds Kokins @archo5.bsky.social · 28/09/2026

we've never truly left the dot-com/app bubbles either there's still lots of people thinking they'll make the next big website/app one long thread of ongoing speculation

030

not quite, as neither ID verification nor regulation more broadly benefits from someone else repeatedly doing something destructive these are simply additional needs that some people start to have breaking a window wastes the time that others had previously spent on that window

000

this definition of an economy also informs one's understanding of finance the more people can express what's useful to them, the bigger is the economy finance produces concentration of wealth -> less people can ask for less things -> economy gets smaller

000

more broadly speaking, anything you can convince someone to pay you for expands the economy and anything causing the opposite shrinks it economy = how many different ways of being useful to each other we have figured out well, that, plus any extortive payments (rents)

100

one of the most recent economic boosts (strangely enough) comes from "age assurance" (identity verification) tech personally I hate all of it but the economic impact is undeniable so, contrary to what many have said over the years - regulations can expand the economy

200

in Latvia, there have been lots of conversations recently around boosting productivity as well automation is often mentioned due to the reasons above, automation is not the answer it increases relative competitiveness against other companies but also shrinks the economy

100

consequently, the idea that gen"AI" can "earn" its value? if the value is in fact negative? technically it is earning its value now by losing money the vast funding spent on the GPUs and the construction of new "data centers" is all the value gen"AI" will ever create

100

therefore the idea that gen"AI" (or anything like it) unlocks the potential for *more* revenue? it could not be more wrong - the exact opposite is the truth optimizing technologies cannot expand the economy, only completely new goals and capabilities can do that

100

at that point, relative to a larger company that hires any of these people, there's now room for competitors to reduce the price - so eventually the companies also get paid less in the end, everyone loses and the economy shrinks

100

if a person earns $ based on how much time they spend while working, reducing their working hours simply makes them earn less if a person gets paid for results and produces more, less such people get hired either way, less money changes hands for the same result

100