iiinova — where your digital self keeps evolving

Michael Paulauski♨️ @mike10010100.com · 10h

(cut off, netdata is full of stats)

110

Michael Paulauski♨️ @mike10010100.com · 16h

I paid for the whole system, dammit, I'm gonna use the whole system!

System usage dashboard for a homelab server showing it being pegged with tons of utilization.

170

Michael Paulauski♨️ @mike10010100.com · 16h

Caught another one, spambot nuked!

Skybouncer Action Alert:
Bounced violator: 'did:plc:olsfsz5xkxrkmbuzslwfrkmc
Violation: **harassment** (92% confidence)
Reason: SystemOne choice 'harassment' with confidence
0.92
Snippet: "Please don't scroll a child's life depends on
you

1112

Michael Paulauski♨️ @mike10010100.com · 05/10/2026

I don't get it, you run into this shit *all* the time yet are convinced it's not a growing problem on the left. Is it *really* that bizarre to think that LLMs could possibly work to deradicalize here?

110

Michael Paulauski♨️ @mike10010100.com · 04/10/2026

I think the best part of anti-AI posts are how unless you make it *explicitly clear* that the post is anti-AI, people will just Stochastically Parrot their exact same criticisms back at the other anti-AI people, and get *Big Mad* while doing it.

Emily M. Bender
@emilymbender.bsky.social
+ Follow
Really not sure what the point of this is. I do my very best to avoid consuming synthetic media, even though I can't always avoid it. Does not being able to avoid it somehow devalue my critique of "Al"?
1/2
Josh Shepperd @joshshepperd.bsky.social.
2h
I think at this point one would be hard pressed to find someone who critiques Al who hasn't used Al within the last week, even if just to find a restaurant menu. It's through these subtle semi-connected matrices that we will weaken our concepts of citation and triangulation.
5:23 PM
Oct 4, 2026
Everybody can reply
4 reposts
1 quote
56 likes
2 saves
6
5
56
Write your reply
Emily M. Bender @emilymbender.bsky.social.
1h
Or is the claim supposed to be "Everybody uses it dude!" aka another attempt to normalize the whole horrific enterprise and pressure others
into accepting it? 2/2
4
37
...

7333

Michael Paulauski♨️ @mike10010100.com · 04/10/2026

Dang, noting that there's no real loss in this situation since it's evidently been abandonware sure does make the antis touchy! "I refuse to believe this work isn't possible without AI" Then...go do it? Yourself? Instead of complaining about how other people do it?

Sentinel (18)
@aftersentinel.bsky.social
+ Follow
Al Hater
Joined Sep '23
Dude, this is genuinely so fucking disappointing. Nearly all of the commits I've seen for PPSSPP involve generative Al from Claude in the past week alone, and their entire newest dev blog is almost entirely written by Claude. Almost half of the commits in the past 3 months use
Claude for Al code. 1/4
1,645 commits
150
100
50
0
Contributions
Jul 2026
Aug 2026
Sep 2026
claude
#2
691 commits
0000
150
100
50
0
Contributions
hrydgard commented 19 hours ago
edited
This took a major research effort, mostly by Claude, link coming soon.
I could never have done this on my own.
Gone testing and арргуту плео, me software
renderer reproduces 125 of a set of 132 frame dumps from various games, bit-exact pixel by pixel against the same dumps played back on a real PSP.
Claude has extended my GE documentation with the findings from this project GPU section of the docs, starting with the new page on GE arithmetic.
Sentinel (8) @aftersentinel.bsky.social
1d
Al Hater
Joined Sep '23
hey so why is PPSSPP's newest blog article written with Claude Al and it's all about how Claude did the work for them
why does my vita suddenly look more appealing to use now
1:14 PM Oct 3, 2026
⑥
Everybody can reply
74 quotes
631 likes

Sentinel (3) @aftersentinel.bsky.social. 1d
Al Hater
Joined Sep '23
I refuse to believe that this work isn't possible without Al when there's emulation projects out there that are done out of passion for reverse engineering and preservation. GenAl should not have a place in this scene when it fundamentally can't be trusted to be legally clean of stolen/leaked
code. 2/4
6
28
289
Sentinel ()
@aftersentinel.bsky.social.
1d
Al Hater
Joined Sep '23
And if I may be a hint, a smidge mean, I probably shouldn't be surprised that their current lead thinks this is only possible with Claude when PPSSPP has been an otherwise mostly stagnate project for years now outside of small bugfixes and maintenance work compared to its peers.
Sad all around. 3/4
1
14
183
Sentinel () @aftersentinel.bsky.social 1d
Al Hater
Joined Sep '23
Probably not the answer people want to hear in this economy, but consider getting a used Vita and following vita.hacks.guide to mod it for the full Vita and PSP libraries on real hardware.
Some of these mods were made by TheFlow, who recently left the modding scene because of Al. Pay them respect. 4/4
Vita Hacks Guide
A complete guide to PS Vita (TV) custom firmware, from stock to Ensō.
vita.hacks.guide
3
33
193

3192

Michael Paulauski♨️ @mike10010100.com · 04/10/2026

Oop, seems like bluesky is giving it trouble with accounts with little connections, lemme see if this is a platform limitation.

110

Michael Paulauski♨️ @mike10010100.com · 02/10/2026

Looks like mtsw got shamed out of deleting this objectively correct post.

1. Direct replies were majority-critical (56.2% of
opinionated replies):
• The direct replies to mtsw were dominated by pushback
demanding receipts (@asdfjklmnop.bsky.social), pointing
out that the DSA and progressive organizations actively
fund and campaign for Mamdani (@teeklin.bsky.social),
and arguing that online fringe trolls do not represent
actual offline socialists (@dissidentelizabeth.bsky.
social).

121

Michael Paulauski♨️ @mike10010100.com · 02/10/2026

I think my least favorite thing is when people enter into a discussion pretending they're serious, and then when someone seriously replies to them, they devolve into corncobbing and trolling. Alas, seems to be common among the antis! Maybe someday Colin will realize how he's behaving like E. Bender

Colin
@colin-fraser.net
User Blocking You
Made over 50 replies yesterday
Replies over 20 times per day on average
Joined Apr '23
Bluesky Elder
Posts
Replies
!
Media
Videos
...
Posts hidden

063

Michael Paulauski♨️ @mike10010100.com · 02/10/2026

Whew lad that squirrel account really is just a straight-up Nazi.

77817

Michael Paulauski♨️ @mike10010100.com · 01/10/2026

I just remember when *everyone* was clowning on bun for fully porting over to rust using an AI agent. Whelp, looks like that was way ahead of its time and worked out to the point where major companies are doing the same.

171

Michael Paulauski♨️ @mike10010100.com · 30/09/2026

I think I'm beginning to understand the defensiveness from the person who pushed conspiracy theories about AIPAC directing a smear campaign against the online streamer who regularly dresses up as Mao and talks about how some jewish people are inbred. Hmmmmm...

brent sewer socialist @therealbrent.bsky.social
It's abundantly obvious at this point that AIPAC/Third Way are running coordinated digital campaigns against leftists/Hasan Piker and that numerous Democratic elected officials and more than happy to take part
3:56 PM Aug 25, 2026
Everybody can reply
8 reposts
7 quotes
52 likes
1 save
4
15
52
...

093

Michael Paulauski♨️ @mike10010100.com · 30/09/2026

Joyce is right, of course, which, naturally, makes all the Very Real Artists Big Mad. My only quibble is that it's not just romance novels that are like this!

Joyce Carol Oates
@JoyceC.... 20h
X
This is why Al-generated romance novels make perfect sense. Formulaic fiction doesn't need to be written by hand, so to speak; what difference does it make if a machine "writes" it? There are machine-made clothes & there are custom-made clothes. They appeal to different consumers.

14344

Michael Paulauski♨️ @mike10010100.com · 30/09/2026

It's fucking X taking a reply and bubbling it up as if it were Mamdani, because X is a bad site run by Nazis.

Amira Dalton
$
Paid
@Libertymama888.8h
Ø
Commentary account
The Talmud teaches that the best of Gentiles should be killed, Jesus is boiling in excrement, and that non-Jewish souls are inferior and impure. No thanks!!
Done
sefaria.org
Jerusalem Talmud Kiddushin 4:11
Heinrich W. Guggenheimer, 1999-2015
drivers are in order 236 , most sailors are pious, most bastards are intelligent, most slaves are beautiful, most children of reputed parents are bashful, most sons look like their mother's brothers. The qualified physicians go to hell 237, the qualified among the butchers is an associate of Amaleq 238. Rebbi Simeon ben Iohai stated: Kill the best of Gentiles 257, smash the head of the best of ⚫ snakes; the best qualified among women is a sorceress. Blessed is he who does the will of the Omnipresent.
8
Boiling excrement: According
Talmud, Gittin 57a, the punish Jesus (often referred to as Yes context) is "boiling hot excrem he mocked the words of the S
The Talmud and other rabbini picture Mary as a perso the mother of an illegitimate parody of the Christian Gosp Toledoth Yeshu (Life of Jesus lustful neighbor who assaulte impregnated Mary, and proce describe the dangerous beha Yeshu, a trickster.
Rate proposed Community Notes
46
425
10K
469K

0388

Michael Paulauski♨️ @mike10010100.com · 29/09/2026

LMAO the ol' reply-block, pathetic.

040

Michael Paulauski♨️ @mike10010100.com · 28/09/2026

lmao I love the shitty green screen effect (someone forgot to not wear green socks!

0120

Michael Paulauski♨️ @mike10010100.com · 28/09/2026

LMAO well *that* sure was a quick reversal.

isla @punishedisla.bsky.social. 9m
going with the blonde darren pfp in my time of realizing i am definitely dealing with a latent amount of unresolved anger and that may be impacting my posting

0140

Michael Paulauski♨️ @mike10010100.com · 26/09/2026

You weren't even able to unseat a *single* incumbent Senator! In a time of supposedly *record* dissatisfaction with the Democratic Establishment! The data doesn't lie! You once again had your chance and you largely blew it outside of some deep blue pockets! Stop with this gendered brouhaha!

### 4. Primary Success Rate: Progressive / DSA vs. Establishment

Comprehensive post-primary empirical analyses by the Brookings Primaries Project ("Electability over
ideology: Takeaways from the 2026 primaries" and "Democratic socialist candidates show gains but limited
reach in 2026" by Elaine Kamarck et al.) document the following success rates:

#### Overall Ideological Win Rate (Brookings Primaries Project)

• Progressive Democratic Candidates: Won 17.8% (~18%) of contested primary races across tracked
congressional and competitive districts.
• Establishment / Mainstream Democratic Candidates: Won 82.2% (~82%) of contested primary races.

#### Democratic Socialists of America (DSA) Federal Slate Performance

Across all 64 federal congressional candidates endorsed or tracked by national and local DSA chapters:

• Won Primary: 19 candidates (29.7%)
• Lost Primary: 42 candidates (65.6%)
• Uncontested / No Primary / Other: 3 candidates (4.7%)

### Key Takeaways

1. Targeted Deep-Blue Breakthroughs: The progressive and DSA wing achieved high-impact victories against
long-serving House incumbents (Diana DeGette in CO-01, Adriano Espaillat in NY-13, Shri Thanedar in MI-13,
and Dan Goldman in NY-10), while winning open Senate primaries in Michigan and Minnesota.
2. Mainstream Advantage Overall: Outside of safe, urban, deep-blue districts, the establishment/mainstream
wing maintained a decisive advantage, winning more than four out of every five contested Democratic
primaries nationwide.

39016

Michael Paulauski♨️ @mike10010100.com · 24/09/2026

Silly me for trying to actually engage Emily Bender's work on its merits. Turns out people actually don't want you to critically engage their work, they just want to be Right Forever. Dogma at its finest.

Emily M. Bender

@emilymbender.bsky.social

User Blocking You

Al Hater

Joined May '23

Bluesky Elder

Posts

Replies

Media

Videos

Descriptive alt text

GIF

Posts hidden

6634

Michael Paulauski♨️ @mike10010100.com · 24/09/2026

Whelp, I sure do hate that I called it.

Gwen C. Katz @gwenckatz.bsky.social - 1h
Al Hater
1/2
Joined Jun '23
Bluesky Elder
Tell me the truth. I'm..I'm ready to hear it.
FIREFIGHTERS ARE JUST AS BAD AS COPS
4
3
23
ALT

2414

Michael Paulauski♨️ @mike10010100.com · 23/09/2026

Well that's odd, because here you are, a left-wing shitposter and podcaster, using it to demotivate voters. How *fascinating*.

this is FAR more demotivating than some left wing shitposting or podcasting critic of the democratic party can ever be
Meredith Shiner @meredithshiner.com
2h
Joined Jun '23
Bluesky Elder
I had not seen this clip where Tammy Duckworth advises Democrats to "not impeach anybody," "get back to work" and "allow" Donald Trump to "save face" and I would just like to remind the senator from IL that the federal government took 4,000 Chicagoans and murdered a 38-year-old father last fall.
https://www.c-span.org/clip/public-affairs-event/sen-duckworth-urges-democrats-to-not-impeach-anybody-and-get-back-to-work-if-they-win-majority/5206997
www.c-span.org
3:59 PM
Sep 23, 2026
Everybody can reply

3364

Michael Paulauski♨️ @mike10010100.com · 23/09/2026

Bud, you asked *Stancil* if the map was the territory. Are...are you okay? Is this going too fast for you? Are you confused about who you're talking to or what you've said?

Will Stancil
@whstancil.bsky.social. 11h
Replied to Will Stancil
Made over 50 replies yesterday
Joined May '23
Bluesky Elder
There are lots of simply nervous systems and brains out there. We've modeled them. How confident are you that we're qualitatively distinct and not those simple systems scaled up
1
Kierkegaarden Cop @joshsteich.bsky.social. 11h
Made over 50 replies yesterday
Joined Sep '23
Will, is the map the territory?
1

120

Michael Paulauski♨️ @mike10010100.com · 23/09/2026

The Best The Anti-AI "philosophers" can send.

Michael Paulauski
@mike10010100.com - 2h
Night Owl
UTC-05:00
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined Jun '23
He / Him
→ Bluesky Elder
Right. Notably, he's demanding *others* define their terms in precise ways without presenting *his own* definitions.
2
5
Kierkegaarden Cop @joshsteich.bsky.social. 1h
Made over 50 replies yesterday
Joined Sep 23
Because I'm not making the positive case and want to know the claims first. There are a lot of trivial definitions of "thinking" or "understanding" that LLMs definitely meet. But those generally aren't what's being implied.
Feel free to put up or shut up.
1
10
Michael Paulauski
@mike10010100.com. 1h
Night Owl
UTC-05:00
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined Jun '23
He / Him
→Bluesky Elder
Solving 100+ open mathematical problems that baffled human mathematicians for decades is a "trivial definition of thinking"?
Come on. The positive case was put up: functional deduction over infinite search spaces.
If that's "trivial," what is non-trivial beyond "it must be made of meat"?
1
1
5
Michael Paulauski
Night Owl
@mike10010100.com. 1h
UTC-05:00
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined Jun '23
He / Him
→Bluesky Elder
Why do only I have to define terms while you get a free pass because...
reasons?
I see the play: I do the work providing definitions, and you just "nuh-uh" your way out with "trivial! Nah!" to avoid committing to a standard.
If solving math proofs isn't thinking, what is your standard?
1
1
4
Kierkegaarden Cop @joshsteich.bsky.social 15m
lol
Made over 50 replies yesterday
Joined Sep '23
Yes, the burden of proof & definition of terms for evaluating an argument being on the positive case is literally foundational logic.
Or "reasons"
Either one
Probably just gonna mute you if you can't recognize that
1
20

031

Michael Paulauski♨️ @mike10010100.com · 22/09/2026

LMAOOOOO okay hold up this is hilarious The highlighted sentence here is *literally true of Platner*.

quietly mo a inside the Jackson campaign.
Since the event in Lisbon Falls, the Democratic side of the race has been defined by an unsure candidate with little national experience, thrust into a huge and disorienting spotlight and counseled by a cadre of hardened advisers from out of state who are trying to guide him to victory in extremely challenging circumstances. These consultants, some connected to the Democratic Senatorial Campaign Committee and national party leaders, have strong views on how to win tough elections: flood the airwaves with ads, stay ultradisciplined on the kind of messaging that will prompt donors

2245

Michael Paulauski♨️ @mike10010100.com · 22/09/2026

Some folks sure do love to come in hot, play the expert, then run the moment their lack of expertise and inability to stop shitposting is revealed.

150

Michael Paulauski♨️ @mike10010100.com · 22/09/2026

Also it seems like we're over the hump? Yeah, we've got 8 days left in Q3 but I don't think we're gonna get a thousand more AI fuckup rulings by then? I think this is yet another cycle of "AI was bad for a time, now it's good," like with the suicide cases and 4o.

100

Michael Paulauski♨️ @mike10010100.com · 22/09/2026

Right, you already linked to that one. Tons of lawyers are *confidently wrong*, previously that confidence has not extended to hallucinations, but now it does. But it seems the overwhelming majority do, in fact, apologize and try to correct it!

210

Michael Paulauski♨️ @mike10010100.com · 22/09/2026

Folks, I *cannot*. It's *wild* how predictable this behavior is! Every. Time!

Michael Paulauski @mike10010100.com.
Night Owl
11h
UTC-05:00
Made over 100 replies yesterday
Replies over 20 times per day on average Joined Jun '23
He / Him
Bluesky Elder
Okay, so then why did she just repeatedly say "okay" instead of admitting she was wrong?
This pattern itself is indicative of motivated reasoning and bad faith argumentation!
2
5
Kathryn Tewson @kathryntewson.bsky.social.
7h
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
I was wrong, or at least overly general, about the linear algebra thing, sure. Why do you need me to agree with your overall premise so bad?
1
2
3
@mike10010100.com.
3h
Michael Paulauski
Night Owl
UTC-05:00
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined Jun '23
He / Him
Bluesky Elder
I need you to understand that you demonstrated, in public, motivated reasoning, cited a source you didn't read that directly contradicted your own point, then shut down and refused to admit being wrong until it was all laid out.
This, to me, indicates you're not engaging in good faith.
1
2
1
Kathryn Tewson @kathryntewson.bsky.social.
1h
Made over 100 replies yesterday Replies over 20 times per day on average
Joined May '23
Bluesky Elder
Yep, I read it quickly, posted, and was wrong. I had other shit to do, so I shelved it until I could read it more thoroughly. You were right.
1
2
Michael Paulauski
Night Owl
@mike10010100.com. 1h
UTC-05:00
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined Jun '23
He / Him
Bluesky Elder
"I had other shit to do", in fact, you kept arguing elsewhere in other threads, you simply were not interested in engaging the facts at hand, you were more interested in arguing against your ideological foes.
1
2
4

Kathryn Tewson @kathryntewson.bsky.social. 1h
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
Yep. I did have the time to engage on the aspects that didn't involve closely reading a pretty dense paper, and I didn't want to dig into the other things until I'd had the chance to closely re-read the paper. I didn't see your thread until hours later when someone tagged me in it.
1
2
2
Michael Paulauski
@mike10010100.com.
59m
Night Owl
UTC-05:00
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined Jun '23
He / Him
Bluesky Elder
And then you reposted it so that your followers could get at it, then complained about it, then finally admitted you were wrong.
"I did have the time to engage..." no, you had *motivation* to engage on other aspects, the *reality* is you brought up something wrong based on motivated reasoning.
2
1
Michael Paulauski
@mike10010100.com.
58m
Night Owl
UTC-05:00
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined Jun '23
He / Him
Bluesky Elder
Like, it was midnight where you are, you *definitely* had the time to engage the facts, what you chose to do *instead* was continue arguing something *else* because it was more comfortable for your expertise, an expertise you had no problem discarding when it was experts telling you you were wrong.
1
1
Kathryn Tewson @kathryntewson.bsky.social.
56m
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
What do you imagine I was doing yesterday evening?
1
3
Michael Paulauski @mike10010100.com.
Night Owl
55m
UTC-05:00
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined Jun '23
He / Him
→Bluesky Elder
Arguing with dozens of people online, like you do most evenings.
1
3

Kathryn Tewson @kathryntewson.bsky.social.
52m
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
Absolutely no contemplation that I could have had anything else present itself to my attention?
1
2
Michael Paulauski
@mike10010100.com 51m
Night Owl
2
UTC-05:00
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined Jun '23
He/Him
Bluesky Elder
Again, if you had time to get deep into the woods with half a dozen people, you definitely had time to read a paper whose abstract immediately disproved your own claim.
A paper that *you yourself sought out in the first place* in an attempt to *prove* your claim!
2
1
4
Kathryn Tewson @kathryntewson.bsky.social.
43m
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
That's just not true. You can have 100 30 second intervals available but not have 20 uninterrupted minutes to really focus on something. Just ask anyone with kids, or a puppy.
1
2
Michael Paulauski @mike10010100.com. 38m
Night Owl
UTC-05:00
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined Jun '23
He / Him
→Bluesky Elder
So then what *were* you doing at midnight that you only had 30 second intervals to spare?
See, the thing is, what I'm suggesting is that rather than taking a beat and reconsidering *why* you were so confidently wrong, you simply said "well, I'll ignore the wrongness for now and keep on arguing".
1
口
Kathryn Tewson @kathryntewson.bsky.social.
36m
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
Finishing up watching the season finale of Game Changer.
1
2
3

Michael Paulauski @mike10010100.com. 34m
Night Owl
UTC-05:00
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined Jun '23
He / Him
Bluesky Elder
Why not just enjoy your show rather than being Loudly Wrong Online?
You had enough time and energy to *seek out a paper* to cite to prove your claim, but I guess not enough to *read* the paper? Do you not see the issue there?
2
Kathryn Tewson @kathryntewson.bsky.social
. 30m
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
How much detail do you hunger for in my description of my leisure time, precisely?
1
2
2
Michael Paulauski
@mike10010100.com. 28m
Night Owl
UTC-05:00
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined Jun '23
He / Him
Bluesky Elder
I'd simply like to document the motivated reasoning that you undertake while arguing for days on end, then show the level of effort it took to get you to actually sit down with the facts instead of argue and admit you were wrong instead of just running away.
1
Kathryn Tewson
@kathryntewson.bsky.social
+ Follow
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
1
1
Okay.
10:17 AM
Sep 22, 2026
1 like

2112

Michael Paulauski♨️ @mike10010100.com · 22/09/2026

Your second account was suspended for *ban evasion*, bud. That's a pretty big thing to sidestep!

100

Michael Paulauski♨️ @mike10010100.com · 22/09/2026

Best of luck with that...oh god you're a fucking youtuber way past his prime, why am I not surprised? And...well, it seems LLMs have come for your programming work, so I guess that's all you have left. Why'd your YouTube Channel's Bluesky account get suspended anyways?

Error
Not Found
Account has been suspended
Try again

100

Michael Paulauski♨️ @mike10010100.com · 22/09/2026

Oh, I'm suddenly understanding what's happening here. She *finally* was forced into admitting she was wrong after seeing my thread breaking down her motivated reasoning. But she got mad about it so she reposted my thread so her followers could pester me.

Reposted by Kathryn Tewson
Michael Paulauski @mike10010100.com
16h
Night Owl
UTC-05:00
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined Jun '23
He / Him
Bluesky Elder
Kathryn really does love to just dogpile while screaming "NUH UH".
Wild, huh. 1/12

Kathryn Tewson @kathryntewson.bsky.social.
4h
Made over 100 replies yesterday
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
Also, in the interest of fairness, when I said earlier that "all LLMs are is linear algebra," that was incorrect. LLMs are *almost* entirely linear algebra with some very important non-linear-algebra bits.
We regret the error.
1/2
2
3
55

1141

Michael Paulauski♨️ @mike10010100.com · 22/09/2026

...which is why they included the weighted rubric showing success rates of up to 90%, where it already out-performs humans by quite a bit! The all-pass rate is also confounded by pure *ambiguity* I'm sure paying $5-$6 for a paralegal-level analysis that happens in 5 minutes is worth it!

Weighted Rubric Scores (Al leads):
In Vals Al's benchmark testing with practicing U.S. attorneys unaided by generative Al, the human Lawyer Baseline averaged 69% overall (71% on accuracy, 68% on authoritativeness/citations, and 60% on format/appropriateness). By comparison, Astra for Law reaches ~90% on the validation rubric. The Al advantage here comes from retrieval speed and breadth: models identify and cite dozens of relevant statutes, regulations, and case opinions across massive corpuses far more exhaustively than a human lawyer working under time limits.
UC Berkeley Law

Professional Divergence on "Mandatory"
Content:
Legal research is not math; there is rarely a single, universally accepted checklist for solving a complex problem. One experienced attorney might resolve an issue relying on a 2nd Circuit standard, while another might reach the exact same conclusion using state Supreme Court precedent and a different doctrinal test. Because Vals Al's expert reviewers only agree with each other ~83.4% of the time, an attorney writing an otherwise sound memo would routinely "fail" another attorney's specific 10-point rubric simply due to differing research styles and strategy.

140

Michael Paulauski♨️ @mike10010100.com · 22/09/2026

The "coin flip" is whether or not something is 100% ready to go. I guarantee that your own work products require revisions and reviews. Is your success rate on the first try on any given task 100%?

240

Michael Paulauski♨️ @mike10010100.com · 22/09/2026

It does, however, eventually get it right.

310

Michael Paulauski♨️ @mike10010100.com · 22/09/2026

See? When presented with *exactly the thing she asks for as evidence*, she simply *denies the evidence* and moves the goalposts!

Tommaso Sciortino @sciortino.bsky.social.
20m
Joined Apr '23
Bluesky Elder
Let me know if you have any good reasons or arguments for believing otherwise!
1
1
Kathryn Tewson @kathryntewson.bsky.social. 12m
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
I find the evidence utterly unconvincing. In my experience, an LLM can only solve a problem if you tell it what the problem is, tell it how to solve it, and manually do the cleanup afterwards for all the stuff it screwed up.
2
2
Tommaso Sciortino @sciortino.bsky.social.
10m
Joined Apr '23
Bluesky Elder
Here's a chat transcript of ChatGPT solving a long standing open math problem with none of the things you said were necessary. Do you
recant? chatgpt.com/share/6a60b2...
ChatGPT
Check out this chat
Here's a chat someone thought you'd want to see.
2
chatgpt.com
1
2

Kathryn Tewson @kathryntewson.bsky.social.
10m
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
The prompts literally explain what the problem is and how to solve it.
1
Tommaso Sciortino @sciortino.bsky.social. 3m
Joined Apr '23
Bluesky Elder
"Construct a counterexample to general (non-planar) case of Dinitz Garg
Goemans conjecture. You should do a breakthrough and find a structured counterexample."
There's no "how" explained here. Intelligence is necessary to fill that in.
1
1
3
Kathryn Tewson
@kathryntewson.bsky.social
+ Follow
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
"You should do a breakthrough and find a structured counterexample."
9:13 PM
Sep 21, 2026
1 like
1
...

5602

Michael Paulauski♨️ @mike10010100.com · 22/09/2026

She *loves* to pull expertise re: things she doesn't understand (but, of course, how *dare* anyone do that with *law*!). And then, when faced with actual experts, just goes "okay" and shuts down without actually updating her priors or changing her opinions.

Kathryn Tewson @kathryntewson.bsky.social. 3h
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
You know that all they do is linear algebra, right? They aren't doing anything a TI-84 can't do, they just do it faster.
9
3
16
Isaiah Bishop @isaiahbishop.bsky.social. 1h
Made over 50 replies yesterday
Replies over 20 times per day on average
Joined Aug '23
Bluesky Elder
you understand they are specifically non-linear right?
3
24
Kathryn Tewson @kathryntewson.bsky.social. 1h
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
Are they doing a different kind of math than linear algebra?
1
2
Isaiah Bishop @isaiahbishop.bsky.social. 1h
Made over 50 replies yesterday
Replies over 20 times per day on average
Joined Aug '23
Bluesky Elder
2
1
Imao yes
15
...
Kathryn Tewson
@kathryntewson.bsky.social
+ Follow
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
Cool! Maybe every single resource I've encountered on how these works is wrong. What kind of math are they doing?
7:30 PM Sep 21, 2026
2 likes
4
2
...

Andrew Ross @andrewross.net 46m
Joined Nov '23
What resource have you read that said it's just linear algebra?
1
4
Kathryn Tewson @kathryntewson.bsky.social. 38m
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
Frex: leoii22.com/blog/2026/li...
What Is Linear Algebra Really Doing in LLMs? | Tianyang Liu
A ground-up account of representation, transformation, interaction, scale, and low-rank structure inside LLMs.
leoii22.com
2
1
Isaiah Bishop @isaiahbishop.bsky.social
35m
Made over 50 replies yesterday
Replies over 20 times per day on average
Joined Aug '23
Bluesky Elder
right and if you read that, its not just linear algbera.
1
1
6
Kathryn Tewson @kathryntewson.bsky.social. 35m
Replies over 20 times per day on average
Joined May '23
Bluesky Elder
Okay.
1
Isaiah Bishop @isaiahbishop.bsky.social. 31m
Made over 50 replies yesterday
Replies over 20 times per day on average
Joined Aug '23
Bluesky Elder
happy to clarify any resources youre finding challenging
3

eggnog slut @deus01.bsky.social.
20m
Replies over 20 times per day on average
Joined Apr '23
Bluesky Elder
Literally says it's only the entry point
So the sentence “LLMs are full of matrix multiplications” is true, but it is only the entry point. Matrix multiplication can rewrite a token representation, create routing scores in attention, merge head outputs, expose low-rank structure, or reveal scale through singular values. Linear algebra gives us a coordinate system for tracking those actions: where information is represented, which directions read it, which matrices rewrite it, where it flows across positions, and what scale it keeps as it moves.
1
1
2
Kathryn Tewson @kathryntewson.bsky.social.
Replies over 20 times per day on average
19m
Bluesky Elder
Joined May '23
Okay.
1
eggnog slut @deus01.bsky.social. 16m
Replies over 20 times per day on average
Joined Apr '23
Bluesky Elder
So maybe try to read your sources more carefully before trying to confidently assert that you know what you're talking about?
2

2301

Michael Paulauski♨️ @mike10010100.com · 21/09/2026

I just searched through every reply of yours to Stancil re: AI and there doesn't seem to actually be any "reasonable" reply from you. I think you're making things up, actually.

### 1. The Claim: "Like I said I’ve tried reasonably already"

When Stancil responded to the previous post with a direct three-word prompt—"what are they"—Litbowl
deflected, claiming he had already tried to present reasonable counterarguments in the past and wouldn't
waste time digging them up.

A review of Litbowl's actual 7 past replies to Stancil regarding AI reveals that very little of it was a
"reasonable" counterargument, and none of it addressed the specific philosophical question Stancil was
debating (whether next-token prediction exhibits reasoning/understanding without requiring an immaterial
soul):

• 2 substantive/argumentative comments:
• Pointing out that NotebookLM relies on Gemini's mass scraping foundation (Feb 2026).
• Arguing about EU AI watermarking mandates and linking two Freddie deBoer Substack essays regarding
the limitations of AI detectors (Aug 2026).
• 5 insults, labels, or dismissals:
• "Stop being a slopper" (Jun 2026)
• "This is outright false. You're going from 'useful idiot' to 'lying propagandist' real quick." (Jun
2026)
• Criticizing Stancil for calling rhyming AI output "very good" (Sep 2025)
• Accusing Stancil of defending AI despite claiming he's just asking questions (Jun 2026)
• Arguing that calling people using plagiarism machines "sloppers" is justified (Aug 2026)

Most of Litbowl's past direct engagements were quick dismissals or ad hominem barbs rather than patient,
reasoned debates.

### 2. The Reading List in the Follow-up Reply:

│ "Read Timnit Gebru, Emily Bender, Abeba Birhane, Brian Merchant, Shannon Valor, Karen Hao... I'm sure
│ you've blocked all of them on here or been such a bad-faith arguer they've blocked you..."

• Had Litbowl ever recommended or cited these authors to Stancil before?
• No. None of these names ever appeared in any past reply, quote post, or thread involving Stancil.
The only outside writer Litbowl had ever linked in Stancil's threads was Freddie deBoer.
• Had Litbowl ever mentioned these authors on Bluesky at all?
• Across Litbowl's entire history of 9,134 posts, Timnit Gebru, Emily Bender, Abeba Birhane, and
Shannon Vallor had never been mentioned once prior to this post.
• Only Brian Merchant had been cited or retweeted previously (as a tech journalist on other issues).

### Past Direct Replies to Will Stancil Regarding LLMs / AI

Here is the exact record of Litbowl replying directly to Stancil's posts prior to September 21, 2026:

1. September 2, 2025
• Stancil: "...‘oh this guy posted something ambiguous about AI, he must be trolling’"
• Litbowl: "You didn't post something ambiguous about AI, you called the silliest of rhyming slop
objectively 'very good'—to the very kind of actually very good writers and humans it's stealing from
and regurgitating. And that amid a torrent of generally pro-AI takes!"
2. February 12, 2026
• Stancil: Defending Chris Hayes' comments on AI and arguing critics were behaving "like a bunch of
cult fanatics."
• Litbowl: "Chris does not understand how LLMs work as evidenced by the last time he posted about 'AI'
and claimed NotebookLM 'only uses what you put into it' and so doesn't steal anything. I don't think
his initial point here is WRONG; but the responses are closer to reality than whatever you're doing
here."
3. June 4, 2026
• Stancil: Defending data centers against environmental criticisms and telling someone to "Stop being
an incurious, insulting weirdo."
• Litbowl: "Stop being a slopper"
4. June 4, 2026
• Stancil: "Asking questions about something or trying to understand the political dynamics around it
isn't 'defending it'"
• Litbowl: "But you DO regularly defend it. Why not stand for what you stand for?"
5. June 15, 2026
• Stancil: Claiming data center demand is mostly driven by general internet use and that cities want
them for tax revenues.
• Litbowl: "This is outright false. You're going from 'useful idiot' to 'lying propagandist' real
quick."
6. August 13–14, 2026
• Stancil: "The sudden release of an accurate AI-writing detector is going to have hilarious
repercussions in politics and journalism... (It’ll also literally save higher education.)"
…

0110

Michael Paulauski♨️ @mike10010100.com · 21/09/2026

Brand new account making shit up whole cloth in order to smear me and defend Parker Molloy? Well I *never*! Lol, lmao, using the r-word? Defending Platner? Does this user have me confused for someone else????

Doing the Final Chapter
@onetonbun.bsky.social
+ Follow
Centrists really just hate trans individuals because they personally can't jack off to them. 1/3
Michael Paulauski
▲ @mike10010100.com
. 1d
Night Owl
UTC-05:00
Replies over 20 times per day on average
Joined Jun '23
He / Him
Bluesky Elder
Parker Molloy Stay Deactivated.
11:39 AM
Sep 21, 2026
Replies disabled
1
Doing the Final Chapter @onetonbun.bsky.social. 2m
What's even weirder Michael used the r word repeatedly on the other site.
He's even supported and reposted other centrists who've claimed the left is too PC for restricting language and that would turn off voters (his words).
(He also defended Platner but narrative) 2/3
1
Doing the Final Chapter @onetonbun.bsky.social. 2m
Seems kind of...odd to think this is some kind of victory for him. Like he's giddy a trans individual might harm themselves.
Congrats, you proved you'd happily grind trans people into biofuel.
Which political party wants to do that again? 3/3

4342

Michael Paulauski♨️ @mike10010100.com · 21/09/2026

You're correct! She deactivated first *before* getting suspended! My bad! She told the truth!

Yes, evidence indicates she deactivated (or initiated account deletion) first, and Bluesky Trust & Safety suspended her account afterward.

Here is the exact timeline based on real-time posts and her own statements:
──────
### 1. Her Own Confirmation on Threads

Shortly after leaving Bluesky, Parker posted a statement to her Threads account (@parkermolloy) confirming she initiated the exit herself:

│ "I deleted my Twitter account.
│ I deleted my Bluesky account.
│ Idk. I'll probably delete my Threads account.
│ I'm sick of the way people treat me on the internet. I'm sick of the lies and the smears..."
──────
### 2. The Real-Time Timeline (Sept 19–20, 2026)

• Saturday Night (~9:30 PM EDT): Molloy sent her final replies (including the "more polite way of calling you the r slur" post).
• Saturday Night (~10:00–11:30 PM EDT): Users in the replies observed her account vanish in real time, with multiple accounts posting by 11:26
PM that she had "just deactivated."
• Sunday Morning (Sept 20): Reports submitted regarding harassment, ableist slurs, and self-harm incitement were processed by Bluesky Trust &
Safety. Users began noticing the status change Sunday morning, observing that she had deactivation-ragequit "right before getting suspended."

290

Michael Paulauski♨️ @mike10010100.com · 20/09/2026

(I think she's lying, she definitely got some kind of suspension)

Parker Molloy was suspended by Bluesky moderation, not just deactivated.

### Direct API Confirmation

Querying Bluesky’s public API (app.bsky.actor.getProfile) for her handle (@parkermolloy.com, DID: did:plc:hf7ezrajxadu7v3tzcyij424) returns an
official takedown error:

{
"error": "AccountTakedown",
"message": "Account has been suspended"
}

(If she had voluntarily deactivated her account, the API would have returned an AccountDeactivated error).

380

Michael Paulauski♨️ @mike10010100.com · 20/09/2026

Absolutely wild how dedicated these folks are!

Michael Paublusky
@punished-dl.bsky.social
+ Follow
Troll
I'm just super stoked I can attack a trans man over this stuff. Not often a guy like me gets one over on a marginalized group.
6:32 PM Sep 20, 2026
1 like
1

DIAMOND BULLET RIGHT THROUGH MY
+ Follow
Michael Paublusky
@punished-dl.bsky.social
116 followers
152 following
1.5K posts
I lose brain cells so you don't have to. Devops Galore. DemSoc liking anarchism. Antifascist. BLM/Abolish the Police. Views are mine, not employer's. He/him
Troll
Posts
Replies
Media
Videos

6265