sophie malice
@sophieactual.mitsuko.nz
1.1K followers 1.2K following 14K posts
ancient anarchist tranny bitch, or if you'd prefer slightly more polite, she/her, nd, plural, transdyke, bratty, geekette, too old, anarchic thingy, also tangata te ika a Māui or te tiriti, take your pick, and inkcap dot yadayada
Posts Media Videos Starter Packs
Reposted by sophie malice
awaterevalley.bsky.social
The planet is alternatively flooding and burning, racism and misogyny are rife and the Crown is about to steal the seabed and foreshore all over again. And the important Parliamentary discourse? Whether men should have to tie bits of cloth around their necks at work.
Reposted by sophie malice
changleen.bsky.social
#NZPol It’s fucking wild that Israel is literally bombing and shooting people in Gaza right now but our entire media is talking about the success of the ceasefire and even a good part of it is now just arguing about whether trump should get a fucking Nobel prize or not.

We have reality problems.
sophieactual.mitsuko.nz
i hope

Posie Parker was allowed in

#nzpol
Reposted by sophie malice
acgeddis.bsky.social
Ooof ... Australian Court just upheld Australian Govt decision to refuse Candice Owens Farmer (she drops the last name in her promo stuff) a visa to do a paid speaking tour. I assume that'll kill off her trip here, too (on practical grounds, even though Chris Penk gave her permission to come).
In a lengthy ruling, the high court upheld Burke’s findings. The court said:

In the absence of evidence, or agreed facts, it is not obvious that the opportunity to hear Ms Farmer speak in Australia (the so-called “lightning bolt” effect) could add anything to political communication in Australia.

Owens will have to pay the costs of the appeal.
sophieactual.mitsuko.nz
seriously worries me that ACT NAT NZF are going at breakneck speed to implement UK, and US -like health reforms in Aotearoa

i had to ask to be taken off the cervical cancer screening list, 'cos no cervix

mammograms? you have tits, of course you have them checked

#nzpol
jennralize.bsky.social
Just got a call from a lovely but very nervous Dr, who advised I wouldn't be routinely screened for breast cancer because I'm not indicated as being "biologically female" on the central NHS database and my biology wouldn't indicate I'm at risk.

My records have been altered, which is upsetting.
Reposted by sophie malice
cmdr-unifex.bsky.social
Hey, Luxon, how do you fuck up something like data.parliament.govt.nz this badly?

The site feels like a Work In Progress, and like it should be behind a BasicAuth, at minimum, at this point. Definately not allowing public signup, given that crashes in the verification links. #NZPol
Reposted by sophie malice
kiwichar.bsky.social
Living the Kiwi capitalist dream! 🤩 #nzpol #OneTermGovernment #LevelUp #Aotearoa #NewZealand #ResignNicola #Luxon #NoConfidence
sophieactual.mitsuko.nz
*happy, albeit odd 🏳️‍⚧️ girl*

also

*hugs ya*

i enjoy ur TL, i'd miss it
sophieactual.mitsuko.nz
*If it's legit then cool/s

just my personal opinion
sophieactual.mitsuko.nz
k, my apologies for the uncalled-for rant

gotta be bright-eyed and bushy-tailed in four hours, i'd better get some sleep, it's midnight now, alarm goes 04:00

laters, if u haven't blocked me
sophieactual.mitsuko.nz
get you a bunch of skeetuals from Portland, or Seattle if that's a bit much

they're all too fucking scary for me now

the contextual rules of the lexicon change almost by the day, stay offline for a week and a lotta words no longer make sense
sophieactual.mitsuko.nz
those other languages rules evolve at a different "speed", strictly it's velocity

so across language mapping needs different temporal arrays to make sense of subtleties

this, aside from rand() decisions, is why LLM's are also garbages, they're pre-trained, their rules are at a point in time
sophieactual.mitsuko.nz
that's the basis i use for languages, or strictly speaking (?) the lexicon of a language

its lexicon is static at any temporal point, but its contextual rules are in-flight beside other language's rules known by the speaker
sophieactual.mitsuko.nz
yes and yes

i, ok we, also include a not logically definable state, not amenable to "does it exist or not exist" patterns of thinking

not binary 0 or 1

it's a real mindfuck to visualise, but in topology maths it's simply a -1 domain, array, whatevs

maths goes in both directions
sophieactual.mitsuko.nz
soz

geekette in full flight

i'll shout into the void quite happily, it helps me organise my thought processes

what if someone takes umbrage about [idea]?

easy, checkpoint, restart, change the pathway from there

can't avoid being called out on socmed, esp from a skeetual
sophieactual.mitsuko.nz
the bugbear is that there can never, should make that upper case, be a rand() decision during recovery

or you're recovering to what compiler writers lovingly refer to as undefined

it'll probably work if ur not silly
sophieactual.mitsuko.nz
and we're a plurality, we happily interact irl from multiple vm's with shared memory under a hypervisor

sophiactual's one vm, i just happen to front the console

but that model, even if our hardware is capable and it probably is, doesn't support checkpoint and recovery on audit failure
sophieactual.mitsuko.nz
not gonna discount the possibility

Data audits his internal environment, there's been plot themes built around this

i've broken my brain a few times trying to implement auditable breakpoints (in assembly, gotta think in machinespace) for a constrained distributed net, not a full neural net
Reposted by sophie malice
randomancom.bsky.social
Thank you. I wish more people talked about this. As a big sci-fi fan LLMs are not in any way shape or form artificial intelligence. We are a long way off from thinking machines, if such a thing is even possible. LLMs are not it, stop calling it AI
sophieactual.mitsuko.nz
also ML within a constrained knowledge domain is incredibly useful, but very fucking hard to audit

across several domains it's just nuclear horseshit

English as with most languages though i'll give संस्कृतम् extra points, is hugely contextual, not given to lexing

people mis-contextualise all the time
sophieactual.mitsuko.nz
not you personally Luke, soz ran of chars
sophieactual.mitsuko.nz
i loathe the use of AI as a term for LLM's

intelligence is not remotely involved

an LLM is a random horseshit generator

that's all

yes i do know stuff about Machine Learning, how hard it is to do, the nigh-impossibility of auditing a neural net

LLM's are not those, ur an idiot to believe them
lukeplunkett.com
kids teacher just sent this home as part of an assignment he's doing and i want to smash every computer at the school
sophieactual.mitsuko.nz
2°C

if anyone would like to use the degree symbol and their keyboard doesn't support it directly, copy from this link

i've got my own absurdly geekettish way of dealing with not-keyboard unicode characters, ya don' wanna know the machine-specific fuckery, let's just say never in prod
Reposted by sophie malice
mehr.nz
a few of us made public comments regarding the govt's consolidation of NZ's research funding

Richard Easther @rjme.bsky.social, Lucy Stewart @lcsnz.bsky.social, Nicola Gaston, me, et al

admittedly mine is a bit less polite than the others

comments at www.sciencemediacentre.co.nz/2025/10/14/b...
Samuel Mehr, 2023 Prime Minister’s MacDiarmid Emerging Scientist; Senior Lecturer, School of Psychology, University of Auckland; and Associate Professor Adjunct, Child Study Center, Yale University, comments:
“The consolidation of New Zealand’s primary sources of science funding is a terrible idea. It will weaken our global competitiveness and accelerate our ‘brain drain’ to Australia, Canada, Europe and various other attractive destinations for our top scientific minds.

“The biggest problem in the new science funding system its disregard for the importance of basic science, which isn’t even a ‘pillar’ and barely warrants two phrases of mention in the new funding structure. This year a Nobel Prize went to three economists who demonstrated that investment in discovery science drives sustained growth.

“One wonders if anyone in government noticed. Simply put: without core investment in basic science, our economy will suffer, and New Zealand’s universities will simply be unable to recruit and retain top talent.

“The second-biggest problem is moving oversight of funding decisions to the Research Funding New Zealand Board, which will be mainly comprised of non-academics. This is a highly unusual structure relative to the world’s top funding bodies, whose funding decisions are typically made by highly trained domain specialists. “Take the US National Institutes of Health, who spends more annually on science than New Zealand spends on roads, tunnels, and bridges. There, panels of expert scientists in all areas of biomedical research confidentially review grant proposals, providing scores for proposals; then, separate panels of ‘program staff’ — most of whom have PhDs in the area they are making funding decisions about — decide which proposals should be funded.

“Moving control of science funding decisions away from the experts and toward government ministers is a recipe for under-informed funding decisions that are likely to be biased by external influence, whatever topics are currently exciting (but don’t necessarily have any long-term value), and, worst of all, the personal interests of the specific members of the board.

“Basic science in New Zealand has yielded world-changing discoveries in neonatal care, life-saving genomic advances in cancers, and the invention of R, a programming language used at every university on the planet. This is the country of Ernest Rutherford! Our government can and must do better.”