BleedX (@bleed811) header
BleedX (@bleed811) profile picture

Holy Moly Games 🥤 a rhythm game you play on your own table, free in your browser and in VR on Meta Quest. Also building Marginalia, IR for small companies

Joined October 2026
Photo by BleedX (@bleed811)Photo by BleedX (@bleed811)Photo by BleedX (@bleed811)Photo by BleedX (@bleed811)Photo by BleedX (@bleed811)Photo by BleedX (@bleed811)Photo by BleedX (@bleed811)Photo by BleedX (@bleed811)Photo by BleedX (@bleed811)Photo by BleedX (@bleed811)
Fill in the blank from the Korean SAT English section, question 33
3
1,291
An Anthropic mathematician says he and Claude proved the Riemann hypothesis. What that means, plainly. The hypothesis is from 1859. It says the zeros of the zeta function all sit on one line, which would mean primes are spread about as evenly as they possibly can be. He posted a 4 page note on Nov 20 and a proof written in Lean, a language computers can check line by line. There's no readable paper yet, and he says he doesn't follow all of the math himself. So is it solved? Not confirmed. Someone still has to finish the full Lean check and make sure the final theorem really states RH, with no hidden assumptions. Brian Conrey, a leading RH expert, called it a monumental moment but hadn't seen a manuscript. If it holds, a lot of number theory written as "assuming RH" becomes plain fact. Encryption is fine, RSA doesn't rest on it. The bigger change is a proof no human has fully read, trusted because a machine checked it.
1
318
Time to drop the multi-agent fantasy. Spinning up 500 agents and telling one "you're the marketer" and another "you're the CFO" doesn't get real work done. It's role-play with a bigger bill. What works in practice is a capable model running properly on top of a well-built system. Clean data, clear permissions, real tools, checks that catch mistakes. The architecture does the heavy lifting. The job titles don't.
4
242
People keep saying AI makes learning pointless. I don't think so. Places where you can't use AI on the spot, for security or other reasons, will be around for a considerable time. Even if models start improving themselves, with today's architectures people still matter a lot in noticing what's off and deciding what to ask. My bet for the next ten years: people who can learn a wide field fast and then work it with AI.
3
223
Amy has one thing to tell Jake. Which line is it? Korean SAT listening, played once
2
534
What's f(5)? Question 21 of 30 on the Korean SAT math section
250
Another Korean SAT listening question, played once: what should the man say back?
1
374
Native speakers, which choice fills the blank in this Korean SAT English question?
1
317
About 3 minutes per question on the Korean SAT math section. Can you solve this one?
317
The last two calculus questions on today's Korean SAT, translated to English
3
424
Which reply fits in this Korean SAT listening question, played only once?
2
3
4
4,821
The Korean SAT was today. Two questions from the math section, 100 minutes for 30 of these, no calculator
1
3
7
2,235
During the listening part, planes stop taking off and landing across Korea
1
323
And the final two calculus elective questions, 4 points each
1
1
249
Try every move yourself, it's free in the browser and on Meta Quest 👉 holy-moly-tap.pages.dev
1
24
POV: the Pitch Perfect cup song, but you play it in VR on your own table 🥤 holy-moly-tap.pages.dev
3
173
Holy Moly Tap is officially in for the Meta VR Start competition If you played it and had fun, a like on our Devpost page would mean a lot. Other entrants, drop your page below, we want to see what you built devpost.com/software/holy-mo…
1
3
148
Made a rhythm game where guide hands show the move and you copy it on your own table with your real hands. Free on Quest: holy-moly-tap.pages.dev
1
155