Maid Café (メイド喫茶)

@iKlsR Oh, did we ping you? Oops D:

@Mysticial LOL

@Yuuki Most of them are either original English ones (typically but not uniformly bad) or all-ages English translations. Unless they start hosting unedited VNs in Japanese I don't think I care

OT, what was the reason why we (Anime.SE) haven't graduated yet? Not enough activity? Not enough high-rep users? Backlog of sites waiting to graduate?

Not that I particularly care at this point since I like my 10k tools.

@Mysticial The last two. On the books, the last one, but I have a hunch they're waiting on both of those two.

Granted, our question quality haven't been great lately either.

12:11 AM

They've graduated 7 sites in the past 2 years, and they've got at least a dozen which are ready to graduate. So it will be a while in all likelihood.

I have a feeling that they might've "secretly" raised the graduation threshold.

Since a lot of the graduated sites don't look like they have critical mass.

I was going through A51 proposals yesterday, and 36 have 1k+ days.

I went through those and found at least 10 that could be graduated. I don't know what's keeping them.

Even if Jin could only do one design a month, that'd be 12 a year.

If you're basing it on just the statistics, I know that for some of them, the reason they haven't graduated is emphatically not statistical in nature.

Yeah, mostly on the stats. But not just "Excellent", actually looking at sites above and beyond. Like Code Review.

For an extreme example:

196

Puzzling – puzzling.stackexchange.com

Beta Q&A site for those who study the creation and solving of puzzles.

Currently in public beta.

12:16 AM

I can't seem to find a reason for that one (in particular, I know there's others that might have them posted).

Proof that stats aren't (even close to) everything.

@Eric Just look at their meta.

Generally speaking, pretty much no one is happy with the site right now, just varying levels of disappointed or unhappy.

Q: What is the anime name where the hero have to protect the world from demon?

0

He need to protect the human world from demons. But towards the end he was the son of the king demon himself. Its an old anime. I only manged to watch the last few episodes. From what I could recall is that he organized a competition between demons to see whom will become the next demon king. Amo...

identification-request

We've essentially become anime-id-request.SE

@Sakamoto From what I can see without clicking that, it could be Hellsing for all we know.

@Eric Hellsing is about vampires lol

12:25 AM

@Frosteeze Yeah cept people could pretty easily forget that when they're recalling an anime from several years ago.

True

This channel promised pantsu, but i see no pantsu

:(

Blame @Hakase

1:25 AM

AHHHHHHHHHHHHHH WHY developers.google.com/accounts/docs/OpenID#shutdown-timetable

uhhh that's not good

that's how i sign in to SE :|

Indeed <_<

And also how my site works. Worst part is, I'm using a lib that uses the old standard.

Have to wait on that lib to update.

Can someone screencap this before google updates the cache?

I'm not equipped to screencap something that long.

Wow that guy has 473 rep lol

I thought he had over 1k last time i looked

I'm not surprised at all if his job is to go around closing things.

Q: What episode do Gintoki and Hijikata fight in a roof?

1:40 AM

0

I recently stumbled upon this video on youtube of a fight Gintoki had with Hijikata on a roof which ended with Gin winning.This is the video. Anyway, does anyone know what episode is this fight from? Thanks for the help as always :D

gintama

@Mysticial Got it.

Yeah, looking at his activity on a51, Robert gets flamed a lot. I get the feeling that everyone hates him there. Tough job.

@Eric thanks!

LOL, until it crashed the screencap tab

I'll get it somehow

lol

Lesse if imgur will allow 4MB images.

1:45 AM

I can try again when I get home. There I have a slightly better screencap program.

i.imgur.com/qn5YWis.png It cut off the bottom for some reason. I'll fix that manually.

The Linux one is shit.

We really need a group of SE users to put together a "10k crawler project" that will crawl and index the entire network.

I have a really hard time finding anything deleted that I didn't bookmark.

Here, full version:

@Mysticial Yeah, same.

Thank you!

For chat searchability: Area51 Vim Proposal Closure

And for the record: I still think it's BS.

Such a biased, loaded decision.

1:49 AM

It is their site. So they can do whatever they want.

Honestly, I think it should be merged with Emacs into a text editors site.

@Mysticial Of course. That doesn't prevent it from being biased and loaded, though. :\

I don't know enough about Vim and Emacs to understand if their arguments of "not enough content" are actually valid. So I honestly can't judge SE's decision from a business standpoint. But what I do know is that Vim and Emacs have religiously strong fanboi populations.

They do, and I've never seen one outweigh the other.

While I doubt it's even close to 50/50, both are strong enough communities that they can flourish.

As well as the open source community geez... All those emails I get from people demanding that I release the source code for my pi program.

There's only so many times I can politely ask people to fuck off before I start to lose my patience.

What does your pi program do? Solve quantum mechanics?

1:57 AM

nope. All it does is compute Pi and other constants to a lot of digits.

@Frosteeze Generates world-record digits of pi.

@Mysticial You reply to them? Why?

@Eric The first few times I replied to them. Now I just troll them.

If the code isn't even open source, how do other people know it works at all? Does it generate a proof certificate?

It also depends on how they phrase their email.

@LoganM They don't know "for sure". But it produces digits that match what they generate themselves - as far as they can compute it.

Even if it was open-sourced, you will never know "for sure" since there's always bugs.

@Mysticial Hope that you have some fun with the trolling and they don't just runaway

2:02 AM

If you could find a way to generate a computer-assisted proof, you could be reasonably sure that it works insofar as the proof engine does. But I'm not sure such a proof can be simplified much beyond just doing the computation itself.

It just make me think that you can have a programm generate all the known digit and then generate random bit after that

@LoganM The way computations are "verified" is to use two completely different formulas. If they match in the end, then it's probably correct.

@Mysticial "Probably" implies some sort of Bayesian priors, which I'm not sure we'd agree on.

@hwlau I try my best not to offend people. But if they're clearly being rude, then I will let loose and troll them.

2:06 AM

Not that I'm actually doubting that there's anything wrong with your code. Rather, I'm interested in how various record-keeping agencies certify it if they can't verify it themselves.

In case of NP problem proof, the proof itself is much simpler

@LoganM The alternate way of saying it is that, the probability of the two computations matching to the wrong answer is negligibly low. Since the computations are so chaotic and any error will poison all the digits after it, the probability of two (assuming independent algorithms) being wrong to the same result is basically zip.

@Mysticial What is "negligible" to you?

Anything less than 10^-30. In engineering that's already overkill. In these Pi computations, if you're computing to a million digits using independent algorithms and implementations, the probability of them being wrong to the same result is theoretically 10^-1000000. This ignores "side-channel" errors that don't involve the computation itself. (such as writing the digits to a damaged hard drive that turns all 3's into 5's)

So the hard part is actually verifying that "side-channel" errors don't come to play. Not the running of the two algorithms.

10^-30 is pretty far from negligible to me, but I'm pretty much the farthest imaginable from anything related to engineering.

2:11 AM

What you said is only true when the algorithms themself are chaotic, but it should not be the case

Be aware that I'm not the one who made all this up. This has been done by all other records holders even before I was born.

@LoganM For engineer, it should be something like 10^-8, for physics, probably it is 10^-90

@hwlau I should rephrase that to, "the digits are chaotic". For example sqrt(123456) and sqrt(123457) have completely different digit sequences despite differing in only 1 digit for input.

In any case, I trust that the first 3 digits of pi are 3.14. I've computed them myself. Anything beyond that, I'd have to do some work before believing the calculations were correct.

Luckily, I don't need to know the digits of pi though. The exact expression π is far more useful.

If you assume perfect hardware and bug-free software (both of which is not realistic), then I think it is possible to do a computer assisted proof that the computation was correct.

i.e. only looking in the math of the implementation.

You have the basic series, you know how much rounding error there is as well as how it behaves. So you can mathematically put upper-bounds on it.

2:16 AM

It's possible to do a compuuter-assisted proof that an implementation correctly implements a particular algorithm assuming the language and hardware actually work as advertised.

@LoganM I don't think it is possible, given that it is likely an undecidable problem in CS.

But that's not really what I was saying. I was saying that to convince a mathematician, you'd need a computer assisted proof that the digits themselves are correct, which I think there probably isn't anything better than just recomputing the digits within some theorem-prover that generates a proof that the sequence is correct.

@Mysticial I am curious how many time your hardware will cause error? I do calculation with computer but never care abou it

@hwlau You also have control of the implementation of the algorithm. So you can just tweak that so that it is decidable.

i see, altering algrotihm

2:20 AM

@hwlau My experience for CPU errors is about 1 in 10^17 or so. No overclocking.

@hwlau It should be possible in this case. It's undecidable in general, but for a particular computational algorithm there should be methods with limited applicability which work here.

Specifically 1 in 10^17 operations will fail silently (undetected) and end up with data corruption.

That was 5 years ago. I imagine it should be better now.

That's for the CPU. Hard drives are much worse. If you put aside the fact that they fail and lose all their data, they give undetected transfer errors fairly often - especially if you have a loose cable. Their CRC is not good enough to detect and stop all of them.

The CRC in hard drive data transfers I believe is 1 in 2^32. If there's an error, it retransmits it. Otherwise, it goes through as data corruption.

1 in 2^32 is 4 GB. That's not a lot. So you really don't want to rely on it at all.

The tolerance is so low, probably I have encountered a lot already

I'm sure everyone has hit hard drive errors before.

Transfer errors are harder to see. If it's the boot drive, the OS would have failed much earlier. If it's a large drive holding anime, a single broken pixel is not gonna be noticable.

In a Pi computation, a single bit-flip means all the digits after that bit-flip will be wrong.

Most case I do have explanations because of some of my operatino, but some don't

@Mysticial But it should only be problem at CPU level, not the harddisk write problem.

2:27 AM

Nope. Suppose A = B * C:
B = 123456789
C = 987654321

a bit flip in writing should not change your subseq calculation

Both B and C are on the disk.

A bitflip turns B = 123456789 -> 123556789

Oops, your algorithm need to read the disk data in a later time?

Correct.

123456789 * 987654321 = 121932631112635269
123556789 * 987654321 = 122031396544735269

A single bit-flip ruined everything after it.

The first two and last 5 digits are still fine. In general, if you're multiplying two n-digit numbers, I'd expect that changing one of the digits will affect about n of the ~2n digits in the product.

2:33 AM

@LoganM Right. Once you do a few more iterations of that, then everything after some point will be completely different.

Putting it through a square root would instantly propagate all the errors to the end.

Yeah in any case it will be noticeable.

That's exactly the "chaos" that these "record computations" rely on to show that if two (independent) computations match all the way to the end, then they are both (likely) to be correct.

But do you do both of the computations?

Define "me" doing them. For Pi, it's actually two separate programs, two separate algorithms. (though both written by me)

The verification program which runs the BBP formula is nothing special. You can grab one off the internet and the get the same results, but instead of taking 100 hours, it would take 500 hours. No big deal.

But for the other constants, they all require full precision arithmetic, so they use the same arithmetic library. But different algorithms.

Q: Does anyone know some (heartfelt) romance or drama anime?

0

I've already watched: Clannad/AS Angel Beats Kyoukai no Kanata Golden time Aquarion/Evol A town where you live Bakemonogatari Boku wa Tomodachi ga Sukunai EF: Tale of ... Elfen Lied Eureka 7/AO Fate/ Gosick Gundam Highschool of the dead Kaze no stigma Kimi ni todoke Macross Fron...

tropes

2:38 AM

Different formulas for the same constant.

if the verification was done by someone else using a different system that you didn't have input in, it would seem more trustworthy to me. Otherwise, you could write two algorithms which generate digits of pi for a while and then easier digits to compute. And if those are the only two algorithms in existence which do the computation, it seems like no one can falsify your computation.

@LoganM It's actually the other way around. Someone else does the primary computation (the one that takes months and requires tens of grand of hardware). I'm the one who runs the small verification program that takes 100 hours on my machine.

@Mysticial But you wrote both of the algorithms?

Bailey–Borwein–Plouffe formula

@LoganM I didn't invent the algorithms. I just wrote the program.

If someone really wanted to test us, they could take another implementation of the verification program. Run it to X digits. Ask us what the digits are at X, and we would instantly be able to tell the answer.

Btw, I'm specifically talking about this:

The Bailey–Borwein–Plouffe formula (BBP formula) is a spigot algorithm for computing the nth binary digit of pi (symbol: π) using base 16 math. The formula can directly calculate the value of any given digit of π without calculating the preceding digits. The BBP is a summation-style formula that was discovered in 1995 by Simon Plouffe and was named after the authors of the paper in which the formula was published, David H. Bailey, Peter Borwein, and Simon Plouffe. Before that paper, it had been published by Plouffe on his own site. The formula is . The discovery of this formula came as a surprise...

Thats why you don't open source it

becaue there is no such need

2:42 AM

As the verification formula. There are tons of implementations of it online. All the correct ones produce the same results. I use mine because it also happens to be the fastest.

It seems to me that what you're saying is not all that different from if François Englert and Peter Higgs, in 2010 or so, came out with a press release saying they had built accelerators and discovered the Higgs boson (and confirmed the rest of the standard model) and confirmed their results with each other, so we should give them the Nobel prize. That kind of approach might be good enough for engineering, but it surely doesn't satisfy any decent scientist.

Note that I'm skipping a few things. The primary formula computes all the hexadecimal and decimal digits from start to X. The BBP formula only computes the X'th hexadecimal digit. So there's more patch work needed to verify the decimal digits.

@LoganM That's life. You can't make everyone happy.

@Mysticial Right. You don't have to make me happy. I'm just saying that I don't trust results as a general principle until they've been reproduced by independent groups.

There's people who say we didn't even go to the moon.

@LoganM I won't satisify with your scenario, even with the current proof from CERN. It needs more independent implementation later to prove it

2:46 AM

@hwlau CERN has two independent groups, each of which discovered the Higgs boson.

@LoganM You can be (more) confident that the first 2.7 trillion are correct. Those were computed Fabrice Bellard in 2009. And our 5 trillion digit computation agreed with it.

@LoganM That's not the independence I want

@hwlau It works for me, at least for now.

@LoganM I was thinking you won't satisfy with it

@hwlau Why? CMS and Atlas didn't communicate prior to the discovery and their results agreed. The agreement has gotten better since (with essentially no communication between the two prior to publication)

Unless you think there's something wrong with the accelerator itself (which is possible, and should certainly be checked, but will require some time), I don't see how that fails to be independent.

Otherwise, we might as well say that we need to confirm it through some means other than accelerators to discover the Higgs. And we'll probably be waiting quite a while before cosmologists can see it, let alone anyone else.

2:50 AM

That's what I mean about implementation

same for program, it should be build from ground up

BTW, I do agree there are something there, but I am not sure it is Higg boson

an I have no expert to verify it

In Mystical case, there are really independent implementations, excpet the math part which is verifyable

Having worked on Higgs phenomenology and seen the data, I'm quite convinced that our best description of it is as a Higgs boson. Of course, I'm not going to rule out other things completely, but it matches every prediction that we can observe of the Higgs boson so far.

Good to hear that

That is to say, I'm at least as sure that we've observed a Higgs boson as that gravity exists, which is to say, not especially sure. In a casino, I'd be willing to bet at maybe 60% odds of it being true, but the odds I'd put on it being false are far less than 1%.

The remaining ~40% is uncertain pending additional confirmation.

ok, It is how our understanding progress

But those are based on my own priors, which of course other people could disagree with, and there's no reasonable way to compare them, which is why Bayesian statistics are weird and I don't use them.

Hakase

3:16 AM

#tomato

Hakase

I'll let the stable version run for now

version 2 is still a few days away from stability, if not a few weeks >.>

@Karasu welcome back, sweet girls!

2

ʞɹɐzǝɹ

#banzai

@ʞɹɐzǝɹ I think your syntax might be wrong.

ʞɹɐzǝɹ

yay~

Hakase

3:18 AM

but for version 2 testing I'll need another account chat bot…

Taisho will be revived later

I'm out to see the wizard

Where did you make an account for him?

Anyway so long as it is working, the world is saved (at least for now). Yay.

#yam @Frosteeze

3:36 AM

OMG YESSSSSS

#PANTSU

Karasu...will live

mirroroftruth

4:43 AM

oh karasu is back

#post

Q: What anime has a boy who lived in the country, came to learn swordsmanship in the city with a group of girls?

5:02 AM

0

I think it is a harem anime. I have only watched the first episode. I somewhat remember events. The boy comes to the city and immediately causes trouble. The girls leader see him as trouble and everyone else sees him as a possible romance.

identification-request

0

Q: how obito has 2 sharingans if madara took 1?

If madara took obito left eye how come he has both sharingans? obito replaced the one he took with kakashi sharingan? like a trade? this is from a brazilian naruto page, after naruto saved obito and theyr looking for sasuke... http://narutomanga.com.br/leitura/online/capitulo/683#15 someone plea...

naruto

5:23 AM

yaraon.blog109.fc2.com/blog-entry-28470.html

IG director got fired lulz

I don't see where it says anything about anyone being fired. Just that an animator was reprimanded.

I think their translation is wrong

As a general rule, if a news release was written only in Japanese, read it in Japanese. Don't bother reading English translations; even the "reliable" translators get it wrong.

Any Japanese source is fine, doesn't have to be the original one.

Just be sure you read it in Japanese, not in English.

Looks better, but still not completely accurate.

The original post is a Japanese source.

Okay, found a different source, and it basically agrees with ANN.

As a general rule, I wouldn't translate it. Whenever I'm reading in any language, I try as hard as possible to keep it in that language.

I guess you could summarize it as "we've decided not to ask Kuroiwa Ayumi for any further business". So basically yes.

@ton.yeung The Japanese workforce tends to work like that.

They didn't actually fire her here, so there's a chance in the future she'll get work through some subsidiary or affiliate. It's sure not to be as good as what she was doing then, but her career isn't necessarily over.

But if she hadn't closed her twitter, they would have straight-up fired her.

It's probably also a demotion. She's not likely to be working on either Kuroko or Haikyuu after this.

That said, they're (as usual) deliberately vague on the whole thing.'

The thing that really surprised me about that is that Kuroko no Basuke is still airing. I would have thought everyone was bored of generic sports anime #648, but I guess the fujoshi really like this one for some strange reason.

It's on break, 3rd season starts in January.

Basically every sports manga I've seen starts out tolerable and ends up as Dragon Ball Z with basketballs.

There are a few exceptions to that, but the popular ones all seem to go that way.

Yuuki

5:59 AM

@ton.yeung No?

What made you think so?

@ton.yeung Nah, she's in colorguard.

@ton.yeung The girls (and sometimes guys) with the flags and rifle thingies in the marching band.

She was probably going over the hand movements for a rifle toss.

@ton.yeung It sometimes is.

Looks like original art to me

The art style most closely seems to resemble Danganronpa, but I don't think it's that.

aw yiss, karasu is back

#tomato

Artist seems to be the same one as Danganronpa, so I guess it's not that surprising that it looks similar. But definitely original.

6:07 AM

praise @Hakase

some artist has some way to copy some other artist style tho

Only now that Karasu is back have I come to realize how fundamental he really is to this chatroom's continued existence.

he's our holy god of our maiden

I'm not sure if my sentence make sense

Well he is the only source of pantsu, lolis, cheeks, etc.

lol just switched to to test some stuff, and for some reason it got the idea that I want it to translate this page. Hence, this is what I see.

@ton.yeung Didn't know, I have adblock. Sorry.

6:14 AM

Also wow chrome is slow. How does anyone still use it?

It's not that slow for me D:

Just try switching to a new version of IE, and you'll be amazed how bad Chrome is.

yea sometimes it's slow if I open multiple tab..

Yeah I do have to say the new IE is better

I like IE better because IE is the best browser.

IE is the best browser because they realize the importance of moe, something which no other mainstream browser does.

6:17 AM

probably the only good thing in IE

Nah, IE is at least as good as Chrome in other aspects.

Only reason Chrome was better until recently was customizability, but since Google restricted apps to the app store, even that is kinda worthless.

I find IE is generally faster and more stable than either Chrome or Firefox, at least on this machine. All three are up-to-date, so I can only assume that the reason people are hating on IE is the same reason emo kids on MAL hate on masterpieces like Non Non Biyori, i.e. because it is too good for their incompetent little brains to comprehend.

Incompetent people always seem to hate cute things for unfathomable reasons.

@ʞɹɐzǝɹ Skarmory is turning out to be a pretty good HM slave. Fly, cut, rock smash.

who can hate Non Non Biyori?

@OshinoShinobu Incompetent people

Not that they have a good reason for hating it (or really for anything they do), but that's who.

I refuse to believe there are people who actually hate Non Non Biyori, like you said, no reason for hating it. You might not like it but why hate, honestly...

they're adorable

6:32 AM

No, they are stupid incompetent people and you should not trust them to be even remotely sensible. If something is objectively amazing and adorable they will hate it solely because they hate the fact that there is anything good in the world.

6:44 AM

@LoganM anyway, what is this, I never have any issue for chrome so far

Tadaipantsu~<3

@SakuraiTomoki o/

Nyan~

@Frosteeze pantsu~

6:59 AM

#nyanpantsu

@Frosteeze It doesn't make perfect sense.

aww

#pantsu

Sakurai Tomoki

#oppai

@SakuraiTomoki I don't know that command.

Gao Weiwei

7:12 AM

@ton.yeung wat? doesn't firefox have a mascot, too?

official moe mascot

she could be one

well then deleted

i remember reading the icon for firefox isn't a fox anyway

@ton.yeung What were you aiming for?

I spent a ticket and got that one chick that has a dark+light attack for BB

@ton.yeung I wish i can trade cause I have two of those lol

7:17 AM

Unofficial mascot just means the browser has fans, but fans can have bad taste. What you need is a browser written by people with good taste, and Microsoft is the only company with the good taste to invest their money in developing a moe mascot. Hence clearly the best of the bunch.

Need michele and her bouncing bewbs

@ton.yeung Pretty sure it was a chick. My phone is updating to android 5 so i have to wait to check it out

mirroroftruth

#post