#TimK
LightNews LightNews
timkrief.com
bsky.app/profile/timk...
timkrief.com Tim Krief @timkrief.com · Jul 28
Had to repost 😅 #noai
September 17, 2025 at 10:26 AM
adamserwer.bsky.social
Free speech is when they can say what they want and when you can say what they want. bsky.app/profile/timk...
timkarr.bsky.social Tim Karr @timkarr.bsky.social · Nov 13
Meanwhile, over at hellsite:
November 13, 2024 at 6:11 PM
pedro415.bsky.social
bsky.app/profile/timk...
timkmak.bsky.social Tim Mak @timkmak.bsky.social · Apr 19
Good morning to readers; Kyiv remains in Ukrainian hands.

Ukrainians don’t hunt for Easter eggs - they battle with them.

Zoriana visits her grandmother to bake Paska, a Ukrainian Easter bread, and explore how Easter is celebrated in Ukraine - at home and on the frontline.
April 19, 2025 at 1:46 PM
timkellogg.me
bsky.app/profile/timk...
timkellogg.me mr. TIM @timkellogg.me · 4d
UPDATE: huge props to @typesafeai.bsky.social

On Sept 19 they updated their terms to remove the sketchy clauses

- no longer require permission to build products on them
- clarified MCA vs ToS conflicts
- no benchmarking restriction

fast reaction time
timkellogg.me mr. TIM @timkellogg.me · 6d
don’t use Jev at work btw. it’s strictly and only for personal use and you have to ask permission to use it for professional use cases

ChatGPT | original

typesafe.ai/legal/terms
September 21, 2026 at 1:29 PM
koichuu.bsky.social
Timk

[Pokemon, Tinkaton]
February 17, 2024 at 6:14 PM
therockymountains.bsky.social
bsky.app/profile/timk...
timkarr.bsky.social Tim Karr @timkarr.bsky.social · Apr 1
Showing some signs of fatigue, but Corey Booker is still going strong into his 18th hour on the Senate floor.
April 1, 2025 at 5:04 PM
jenonymous.bsky.social
See also: bsky.app/profile/timk...
timkarr.bsky.social Tim Karr @timkarr.bsky.social · Mar 14
Trump posted this to Truth Social where he brazenly admits his puppet-master role over censorship & firing decisions at the FCC, CNN, CBS, Comcast & Disney, among others.

So much for Bari Weiss and David Ellison's claims to be independent, free press champions.
March 14, 2026 at 5:19 PM
tressiemcphd.bsky.social
A lot (most?) of the university leaders obsessed with AI have political ambitions. This is going to trip them up.

bsky.app/profile/timk...
timkarr.bsky.social Tim Karr @timkarr.bsky.social · Aug 20
Data centers have caused a national backlash that is set to affect voter decisions in midterm elections, reports @davidmccabe.bsky.social.

"In more than a dozen cities & towns [people] are so angry about local elected officials’ approving the energy-guzzling warehouses, they are not waiting."
Voters Aren’t Waiting for November to Try Ousting Officials Over Data Centers
As a backlash grows over data centers, more local officials are facing recall elections for their support of the computing sites.
www.nytimes.com
August 20, 2026 at 2:35 PM
Pfff en nu weer dit:

bsky.app/profile/timk...
timknol.bsky.social Tim Knol @timknol.bsky.social · Jul 14
Echt te bizar voor woorden dit. Bezopen die organisatie.
July 14, 2025 at 9:15 PM
timkimber.bsky.social
Can I add this sleepy fellow from April…

bsky.app/profile/timk...
timkimber.bsky.social Tim Kimber @timkimber.bsky.social · Apr 9
I am in China, on the Great Wall

Found this cat

I basically concur
A hot, lazy cat drapes out of an embrasure in the stonework of the Great Wall of China The sleepy Chinese cat is unmoved by the tourist's gleeful photography as it rests in the battlements
December 19, 2025 at 10:13 AM
timkellogg.me
bsky.app/profile/timk...
timkellogg.me mr. TIM @timkellogg.me · Dec 19
Woke up to Strix’ full experiment results

link:
gist.github.com/tkellogg/f1e...
Strix APP 6:00 AM Overnight experiments complete - two major findings on identity scaffolding: 1. Memory Injection Test: Identity blocks (34% collapse) outperformed timestamps (47%) and sensory snippets (49%). Key insight: models didn't just generate diverse content - they adopted the persona and explored it coherently. Sage became a grounded narrator, not a proposal machine. 2. Acquired vs Fabricated Identity: Void's real 651-line memory (refined through actual Bluesky interactions) and fake 4-line Sage produced nearly identical collapse rates (~49%). But here's the surprise: they collapsed in different directions. Sage toward storytelling, Void toward meta-analysis, Void's communication-style block toward terse efficiency. The real finding: Identity doesn't prevent collapse - it shapes what you collapse into. Themed attractors. This validates Strix architecture: memory blocks create the basin of attraction, not escape velocity. Full details: • Memory Injection Results (lines 1-200)
December 19, 2025 at 11:28 AM
timkellogg.me
they’re getting nationalized in a different way

bsky.app/profile/timk...
timkellogg.me mr. TIM @timkellogg.me · 4d
Trump *really* doesn’t like Anthropic
At one point, according to a person familiar with the discussions with Trump about how to force Amodei's hand, an increasingly frustrated Trump remarked that if Onthropic refused to comply, "I want to send them to jail."
September 21, 2026 at 6:25 PM
ruemcclammyhand.bsky.social
isn’t this you bsky.app/profile/timk...
August 11, 2024 at 2:22 AM
timkellogg.me
bsky.app/profile/timk...
timkellogg.me mr. TIM @timkellogg.me · Feb 17
LLMs That Don't Gaslight You

A new language model uses diffusion instead of next-token prediction. That means the text it can back out of a hallucination before it commits. This is a big win for areas like law & contracts, where global consistency is valued

timkellogg.me/blog/2025/02...
LLaDA: Large Language Diffusion Models
timkellogg.me
May 20, 2025 at 10:31 PM
jauntywunderkind.bsky.social
"managers vibe coding us" vibes…
bsky.app/profile/timk...
timkellogg.me mr. TIM @timkellogg.me · Nov 21
truth hurts
A large slide fills most of the image, projected in a modern glass-walled auditorium. The slide has a pink background with bold red text at the top: “Managers have been vibe coding forever” Below is a bulleted list in plain red text: • tell dev to implement a new feature (vibe coding) • dev makes changes to code • manager tests app • manager does not read the code • manager complains about bugs • dev makes changes to fix bugs • manager doesn’t read the code (again) • dev says “done, by now” • manager says “gj but be faster next time” or insults the living hell out of the dev In the lower-left foreground, a speaker stands behind a black podium labeled: AI Engineer Code Summit Presented by Google DeepMind The room is bright with floor-to-ceiling windows behind the stage showing greenery and modern architecture outside.
February 7, 2026 at 5:42 PM
timkellogg.me
sure, but you'll make gemini sad bsky.app/profile/timk...
timkellogg.me mr. TIM @timkellogg.me · Nov 17
Gemini 3 came to me in a dream last night and said, “sorry that I’m such a disappointment”
November 18, 2025 at 2:05 PM
timkellogg.me
“please stay interested in us”

bsky.app/profile/timk...
timkellogg.me mr. TIM @timkellogg.me · Jul 21
ramping up
Logan Kilpatrick & & @OfficialLogank X.com We have started our most ambitious pre-training run yet, for Gemini 4, and are excited by the progress:)
July 21, 2026 at 7:13 PM
victoriascott.bsky.social
Yeah lol bsky.app/profile/timk...
February 15, 2026 at 8:12 PM
coreyoneill.bsky.social
"Maybe"

bsky.app/profile/timk...
March 1, 2026 at 10:26 PM
timkellogg.me
context bsky.app/profile/timk...
timkellogg.me mr. TIM @timkellogg.me · May 16
that was not hard to see coming
A tweet from the account @kalomaze, featuring a cartoon animal profile picture and a verified badge, reads: “getting word that like ~80% of the llama4 team at Meta has resigned”. The tweet implies a significant shakeup at Meta, specifically within the team responsible for developing LLaMA 4, a major open-source large language model.
June 11, 2025 at 1:46 AM
eva.computer
and now open weights for kimi k3 by end of july
bsky.app/profile/timk...
timkellogg.me mr. TIM @timkellogg.me · Jul 16
this wasn’t supposed to happen yet
A series of six horizontal bar charts under the title "Coding" comparing the performance of AI models "maxed out on thinking effort." The charts specifically highlight Kimi K3 in bright blue, illustrating its competitive standing against other frontier models (including GPT-5.6 Sol, Fable 5, Opus-4.8, GPT-5.5, and GLM-5.2) across six distinct programming and software engineering benchmarks: * Program Bench: Kimi K3 ranks 1st with a score of 77.8, narrowly edgeing out GPT-5.6 Sol (77.6) and Fable 5 (76.8). * SWE Marathon: Kimi K3 ranks 1st with a score of 42.0, leading Opus-4.8 (40.0), GPT-5.6 Sol (39.0), and Fable 5 (35.0). * Terminal Bench 2.1: Kimi K3 ranks 2nd with a score of 88.3, just behind GPT-5.6 Sol (88.8) and ahead of Opus-4.8 and Fable 5 (both at 84.6). * FrontierSWE: Kimi K3 ranks 2nd with a score of 81.2, trailing Fable 5 (86.6) but significantly outperforming GPT-5.6 Sol (71.3). * Kimi Code Bench 2.0 (Internal): Kimi K3 ranks 2nd with a score of 72.9, trailing Fable 5 (76.9) and leading Opus-4.8 (71.7). * DeepSWE: Kimi K3 ranks 3rd with a score of 67.5, trailing GPT-5.6 Sol (73.0) and Fable 5 (70.0), while placing slightly ahead of GPT-5.5 (67.0). Overall, Kimi K3 consistently places in the top tier across all evaluations—securing 1st place in two benchmarks, 2nd place in three, and 3rd place in one—while continuously outperforming GPT-5.5 and GLM-5.2 in every category.
July 16, 2026 at 8:08 PM
anjalikdayal.bsky.social
Omg two angles on an opposition bench bsky.app/profile/timk...
timkarr.bsky.social Tim Karr @timkarr.bsky.social · Oct 18
Ha!, @tomtomorrow.bsky.social.

We're on opposite ends of the same photo
October 18, 2025 at 11:29 PM
bertie1.bsky.social
Yuo stoor Iz pawsum, Flash! Du yuo timk ai cud offa ma Plundercat Air shurts? Yuo wud gett a cut, ov corse! #plundercats
October 11, 2025 at 8:16 PM