consistently candid, ineffective accelerationist, loss-of-function researcher, epicurean hedonist, stochocract, Pigovian, 🏆husband, anti-adultist, ex-𝕏
San Francisco, California
Joined July 2007
- Tweets27.4K
- Following2K
- Followers9.7K
- Likes39.5K
Pinned Tweet
6yo: What’s the biggest number?
Me: There are an infinite number of numbers so there’s no biggest one.
6yo: What’s the biggest number you know the name of?
Me: A googolplex.
6yo: Which is bigger, a googolplex or a googoobazillion?
Me: There’s not a number named a googoobazillion.
6yo: If there are an infinite number of numbers, there must be one named a googoobazillion.
Checkmate.
Another day, another reason to cite webkit.org/performance/
I’ll quote it below, for when I need to tap the sign in the future:
We have a zero-tolerance policy for performance regressions. If a patch lands that regresses performance according to our benchmarks, then the person responsible must either back the patch out of the tree or drop everything immediately and fix the regression.
7yo (doing his math homework): It’s asking how many boots two octopi would need. I’m pretty sure the answer is zero.
Me: …
7yo: Because they don’t have feet. And they swim in the ocean.
Me: You can also tell your teacher that “octopi” is an etymologically inaccurate plural.
6yo: What’s the biggest number?
Me: There are an infinite number of numbers so there’s no biggest one.
6yo: What’s the biggest number you know the name of?
Me: A googolplex.
6yo: Which is bigger, a googolplex or a googoobazillion?
Me: There’s not a number named a googoobazillion.
6yo: If there are an infinite number of numbers, there must be one named a googoobazillion.
Checkmate.
Branch and method coverage are coming to JRuby in version 10.1.3.0 and will be fully supported by SimpleCov!
This still takes more than 1 second. The 87.5% of you who voted for less than or equal to 18 months were all wrong.
Ruby Quiz (it’s more fun if you guess first, then check)
Reminder that in Ruby .. is the inclusive range operator.
('d'..'a').include?('d')
('d'..'a').include?('a')
(0..3).include?(1.5)
(3..0).include?(0)
0%false false true false
11%false true true false
11%false false false false
78%true true false true
9 votes • Final resultsThanks for voting! The correct answer was B, but it will be A in Ruby 4.1: github.com/ruby/ruby/pull/19…
Progress on integer multiplication has slowed from “a new kappa every few minutes.”
Amazing how there was no progress for 7 years, but now an update once an hour is considered slow.
Update: Just submitted a new conditional κ = 0.001662036304429768
Progress has slowed in the last day on integer multiplication since we're approaching a new wall.
Last weekend we got a new kappa every few minutes, but the last update was an hour ago.
A lot of research is going into testing new designs now, and anyone contribute by copying this prompt into Codex/Claude:
gist.github.com/Mister-Meese…
We need expert intuition to poke the models into out-of-distribution solutions, so if you're working in related fields(lattices, coding theory, finite geometry) you may already know the structure to break past the wall.
If you know experts like this, consider sharing the project with them.
github.com/CrocSwap/integer-…
We are now at κ > 2⁻¹⁰ in 3 days, which the plurality of you predicted, though the majority of you were wrong (33.3% under and 26.7% over): xitter.cf/0xdoug/status/21097362…
This reminds me of the early days of Bitcoin mining
You have extra Codex tokens that are about to reset in three hours. We need your help on integer multiplication. Now’s your chance to become a contributor to frontier math!
It’s as simple as opening Codex in a fresh directory, setting your model to Astra 6 Max, and copying the following prompts:
gist.github.com/Mister-Meese…
More than half of you predicted it would take more than a day. In fact, it took less than an hour: xitter.cf/sj_swapnil_jain/status…
In fact, there were five more order-of-magnitude improvements within 24 hours of me starting that poll…but they claim to have hit a wall.
Eleventh update to OpenAI problem #109 (integer multiplication): κ rises to 2⁻¹⁰·⁵⁴⁷, still past 2⁻¹¹.
Witness value: κ = 2⁻¹⁰·⁵⁴⁷ = 6.6857 × 10⁻⁴
(tightened from κ = 2⁻¹⁸²)
This κ is about 1.09 fold our previous witness value κ = 2⁻¹⁰·⁶⁶⁶ = 6.1534 × 10⁻⁴, and a 2¹⁷¹ fold improvement in κ over the original OAI result.
We seem to be approaching a wall. Gains now come in fractions of a percent. The next real jump needs a big breakthrough which we are unable to find yet.