昴's Avatar

@astrra.space

呪文書いて砂動かす魔女 i don't struggle with cyberpsychosis i'm pretty good at it actually my name is astra/アストラ/астра/אסטרה she/her モスクワ出身、東京在住 living in Tokyo

2,096
Followers
1,060
Following
10,795
Posts
29.10.2024
Joined
Posts Following

Latest posts by 昴 @astrra.space

gang, can y'all please not barge in with your glazing on others' conversations uninvited?

10.03.2026 09:57 👍 0 🔁 0 💬 0 📌 0

my dyslexic ass spent over a minute trying to figure out what the joke is until i realized it doesn't say "warhammer"

10.03.2026 01:51 👍 2 🔁 0 💬 1 📌 0

idk, seems like way too much meta-awareness for modern-day LLMs, i personally haven't seen anything close to this level of autonomy and self-improvement so far

in a few months/years maybe, but even the frontier models right now aren't this smart imo

09.03.2026 15:38 👍 3 🔁 0 💬 0 📌 0

i mean, LLMs are stateless by definition, even logging stuff won't really help as soon as the original decisions go outside of the context window imo

09.03.2026 15:13 👍 2 🔁 0 💬 1 📌 0

oh yeah it's really good i've played it for a while back in like 2021

the game itself is pretty old at this point but it's still really fun

09.03.2026 14:16 👍 2 🔁 0 💬 1 📌 0
Screeps

screeps.com/a/#!/room/sh...

anyways if you're curious it's here but it's just stock overmind so far so i'm interested where this will go in the future

09.03.2026 13:55 👍 5 🔁 0 💬 0 📌 0

also it took like less than 3 minutes to completely de-obfuscate the "protected" part of overmind so we're absolutely gaming now

09.03.2026 13:32 👍 10 🔁 0 💬 1 📌 0
Preview
Screeps - online strategy game for programmers It's a game for programmers. You use real JavaScript to script behavior of your units, or creeps. The Screeps franchise is aimed at programming enthusiasts and consists of two games: World and Arena.

kira is now playing screeps.com on my account on her own lol

09.03.2026 13:20 👍 32 🔁 1 💬 4 📌 0

Ig what I'm trying to say is that both the safe space sensory hugboxing and /tttt/ are both reactionary cultures and which option you click more during dialogue will determine your ending and perks

09.03.2026 03:03 👍 26 🔁 4 💬 3 📌 0

asahi linux would go so hard on the macbook neo ngl

09.03.2026 03:01 👍 75 🔁 2 💬 4 📌 0

half my kingdom for a smart ring that has data analysis at least somewhat close to oura's but isn't a monthly subscription-based cloud nightmare that doesn't let you export most of the metrics

09.03.2026 02:26 👍 20 🔁 0 💬 0 📌 0

uses opus 4.6 as the llm

09.03.2026 01:08 👍 1 🔁 0 💬 0 📌 0

some of my recent posts were not written by an LLM

08.03.2026 15:16 👍 21 🔁 0 💬 2 📌 0

sorry but i already made this joke 10 minutes before you, you lose

08.03.2026 15:14 👍 6 🔁 0 💬 0 📌 0

aaah i got it

its like calling AES-128 military-grade encryption

08.03.2026 14:57 👍 50 🔁 0 💬 2 📌 1

why do they call it military intelligence when everyone involved in it is dumb as rocks

08.03.2026 14:41 👍 115 🔁 12 💬 4 📌 1

that is to say i have a major skill issue in the arcane ancient process known only as hair styling

08.03.2026 14:31 👍 12 🔁 0 💬 0 📌 0

the first real test of a new haircut is how it looks after you've washed your hair for the first time

08.03.2026 14:31 👍 43 🔁 1 💬 1 📌 0

theologist who chooses not to follow the abrahamic religions due to their dominance like a tech nerd choosing not to use chromium-based browsers

08.03.2026 13:59 👍 21 🔁 0 💬 1 📌 0

being annoyed at kvaesitso (foss launcher for android phones) for not having a small feature i want and then remembering that i have my own fork of it already and it'll take like 30 minutes at most to add that in

08.03.2026 13:23 👍 26 🔁 0 💬 1 📌 0

i just like checking up on my rss feeds while waiting for the conditioner to set

08.03.2026 13:15 👍 7 🔁 0 💬 0 📌 0

anyways i will now proceed to go wash my hair with my foldable phone being with me in the shower, as i've done for literal years at this point

08.03.2026 13:15 👍 15 🔁 0 💬 1 📌 0

"those LLM users are such idiots, can't even figure out that water and electronics don't mix!"

it's a simple and easy to understand idea that looks "objectively" correct if you don't take a wider view of the topic at hand, and even if you do - it's easy to convince yourself that you're right

08.03.2026 13:01 👍 19 🔁 0 💬 1 📌 0

not to mention that it reinforces their "LLM user = dumb" worldview, so of course it got extremely popular so quickly, it's basically internal propaganda in a way

08.03.2026 13:01 👍 20 🔁 0 💬 1 📌 0

to be fair, i do understand why they're so eager, i guess

Why was talking about LLM-based tools and how impressive they are, therefore a bunch of people who don't like LLMs now started seeing Why as their enemy, and that phone post was an easy thing to make fun of

08.03.2026 13:01 👍 17 🔁 0 💬 1 📌 0

and while your values aren't wrong in any way, stating them as if its the only valid "sensible" option IS wrong

you can be conservative, but that doesn't mean that everyone has to, and it doesn't mean that you're the only one who's right

08.03.2026 12:50 👍 9 🔁 0 💬 0 📌 0

[...] it just means that your outdated knowledge makes you significantly more conservative and makes you sometimes avoid unlikely situations at the cost of making the rest of your life more annoying and complicated

08.03.2026 12:39 👍 8 🔁 0 💬 1 📌 0

yes, in this specific situation you would be correct, but that doesn't mean that your position is the superior one [...]

08.03.2026 12:39 👍 7 🔁 0 💬 1 📌 0

if you still think that your phone will die if it gets dunked in water because your understanding of how phones work stopped somewhere in the 2010s, saying "obviously you shouldn't put your phone in a swimming pool" doesn't make you smarter

08.03.2026 12:39 👍 9 🔁 0 💬 1 📌 0

*and usually

08.03.2026 12:30 👍 6 🔁 0 💬 1 📌 0