Nabokov's Lolita really shouldn't be pigeonholed as merely that, but I guess the movies are another story.
Architeuthis
Dolores in Lolita was like twelve though, at least in the book.
edit: also I don't think Yud recommending The Softcore Adventures Of A Six-year-old In A Thirteen-year-old's Body as a Very Normal Book to his considerable audience fits this particular discourse.
Got stuck forever on 2-2 because of an edge case that only showed up in 7/1000 reports, ended up just brute forcing it, just ran the fitness function after removing one element at a time sequentially.
Then solved 3.x in like minutes because I could be worse at regex, posting code mostly because no-one else posted F# yet.
edited to fix spoiler header formatting
3-2 in F#
"./input.actual"
|> System.IO.File.ReadAllText
|> fun source ->
System.Text.RegularExpressions.Regex.Matches(source, @"don't\(\)|do\(\)|mul\((\d+),(\d+)\)")
|> Seq.fold
(fun (acc, enabled) m ->
match m.Value with
| "don't()" -> acc, false
| "do()" -> acc, true
| mul when enabled && mul.StartsWith("mul") ->
let (x, y) = int m.Groups.[1].Value, int m.Groups.[2].Value
acc + x * y, enabled
| _ -> acc, enabled )
(0, true)
|> fst
|> printfn "The sum of all valid multiplications with respect to do() and don't() is %A"
comments
Not much to say, the regex grabs all relevant strings and the folding function propagates a flag that flips according to do/don't and an accumulator that is increased when a mul() is encountered and parsed.
In case anybody skips the article, it's a six year old cybernetically force grown to the body of a horny 13 to 14 year old.
The rare sentence that makes me want to take a shower for having written it.
...with a huge chip on his shoulder about how the system caters primarily to normies instead of specifically to him, thinks he has fat-no-matter-what genes and is really into rape play.
The old place on reddit has a tweet up by aella where she goes on a small evo-psych tirade about how since there's been an enormous amount of raid related kidnapping and rape in prehistory it stands to reason that women who enjoyed that sort of thing had an evolutionary advantage and so that's why most women today... eugh.
I wonder where the superforecasters stand on aella being outed as a ghislain maxwell type fixer for the tescreal high priesthood.
EA started as an offshoot of LessWrong, and LW-style rationalism is still the main gateway into EA as it's pushed relentlessly in those circles, and EA contributes vast amounts of money back into LW goals. Air strikes against datacenters guy is basically bankrolled by Effective Altruism and is also the reason EA considers magic AIs (so called Artificial Super Intelligences) by far the most important risk to humanity's existence; they consider climate change mostly survivable and thus of far less importance, for instance.
Needless to say, LLM peddlers loved that (when they aren't already LW/EAs or adjacent themselves, like the previous OpenAI administrative board before Altman and Microsoft took over). edit: also the founders of Anthropic.
Basically you can't discuss one without referencing the other.
It's complicated.
It's basically a forum created to venerate the works and ideas of that guy who in the first wave of LLM hype had an editorial published in TIME where he called for a worldwide moratorium on AI research and GPU sales to be enforced with unilateral airstrikes, and whose core audience got there by being groomed by one the most obnoxious Harry Potter fanfictions ever written, by said guy.
Their function these days tends to be to provide an ideological backbone of bad scifi justifications to deregulation and the billionaire takeover of the state, which among other things has made them hugely influential in the AI space.
They are also communicating vessels with Effective Altruism.
If this piques your interest check the links on the sidecard.
NASB does anybody else think the sudden influx of articles (from kurzgesagt to recent wapo) pushing the idea that you can't lose weight by exercise have anything to do with Ozempic being aggressively marketed at the same time?
It's also the place where you go to to download models to use by yourself instead of sending all your data to the most unscrupulous people possible, so at least they've got that going for them.
discussion
Same, except in 4-1 I used a recursive function to traverse each direction according to the offset decided by the selected direction (like SW is row++,col--) , due to functional programming induced brain damage.Would have been pretty useful too if 4-2 turned out to be about finding longer patterns, instead of smaller and in half the directions.
4-1
Inlined some stuff to fit everything in one function:Then you run this for every appropriately pruned 'X' times every direction and count the trues.