« first day (2328 days earlier)      last day (2587 days later) » 

 
3 hours later…
6:30 AM
[ SmokeDetector | MS ] Username similar to website in answer: Where does English get the word “condom” from? by Baocaosu home on english.SE
 
6:55 AM
[ SmokeDetector | MS ] Bad keyword in body, blacklisted username, blacklisted website in body: the cost of radley watches stockists the original item by radlelgl on english.SE
 
 
1 hour later…
8:25 AM
@SmokeDetector That's probably the 10th time or what this username spams ELU
Thou shan't be uncreative.
 
 
1 hour later…
 
2 hours later…
11:31 AM
 
12:05 PM
Can anyone tell me why this answer (10K+ only) was deleted? I can't figure it out from the post history. Also why was the question closevoted?
 
@DanBron screenshot?
 
I can't see why it was deleted. community locked it, but it doesn't say who deleted it.
 
That's weird.
 
It was deleted at the same time it was locked, so I assume it was community who deleted it
 
12:24 PM
6 flags? O.O
Still waiting for a screenshot
 
How do I upload a screenshot from my phone? Do I have to go upload it to Imgur manually and copy/paste a link here?
 
@Dan you should be able to access this link. What does it say? english.meta.stackexchange.com/posts/10133/timeline
 
It was probably deleted because it was not net positive score at the time of migration.
 
Something is going weird with this post ... now the original post on ELU main which has since been migrated to meta has been bumped to the top of main front page.
 
I just reopened it.
Closing it after it was migrated bumped it back to the main site.
So I had to do stuff to it.
Which might make the whole system blow up.
 
12:35 PM
@DanBron yeah, because closed on meta
 
I see thanks @KitZ.Fox
 
No problem
 
I hate using chat from mobile! This is so painful.
 
What?
 
@M.A.R. Chat from mobile makes me hit "send" in the middle of writing a messge
 
12:39 PM
@DanBron you can disable it in the menu
The 'Full Site' option
 
But then it's too small.
 
I'm actually using the mobile app. And when I launch chat from there, it pops up an embedded safari browser overlaid over the app, which has even less functionality than if I went to chat in mobile safari directly
 
Can you navigate to it through your native safari instead?
 
Lemme try
Ok, this is slightly better. I still can't see what I'm typing because of viewport issues, but at least I'm less likely to hit send prematurely. I know TPTB have decided the mobile app isn't worth add'l investment, but it would be nice if they added a lightweight chat interface. More than 90% of my site time is via the iOS app
and I totally miss out on chat because of that
 
hmm interesting. I wonder what it uses for Android. I don't have much problem with it.
 
1:17 PM
@KitZ.Fox Electricity is going out all over Mongolia. Was that you?
@DanBron Blame goes somewhere but not necessarily to SE. I fond that the phone keyboard is easy to mash the wrong button, which might be the send button near the lower right ('n', 'm', space, and delete keys)
 
@Mitch oh damn it.
 
In language news:
> The voice recognition system in the iPhone operates entirely on the basis of acoustic physics, not at all on linguistic phonetics. ... beware of imagining that it knows when you are saying something, or even what "saying something" means. The truth is that it cannot tell the difference between a labiodental fricative and an anal fricative. Let alone distinguish between when you're saying something sensible and when you're talking out of your ass.
 
haha
 
@KitZ.Fox haha
haha jinx
 
hahaha
rolls on floor, laughing
 
1:22 PM
fart noise from under arm
"I'm sorry @KitZ.Fox I can't do that"
 
oh shit
 
It is my firm belief...
 
flies into monolith
 
No it is sort of a mild speculation...
 
@Mitch That's surely an exaggeration.
 
1:23 PM
My God, it's full of mild speculation.
2
@Mr.ShinyandNew安宇 It means it just recognizes the words you are saying not the content of what you are saying, right?
 
that there is no need for concern from black hat hackers trying to take down our civilization by bit twiddling. Or rather viruses and trojans and hackers may well be a huge concern, but before that our greatest worries are about a cascade of niggling bugs, and 'technical debt', and easily aligned corner cases that will cause the entire Jenga tower to collapse
 
Basically, if I Googled every sentence that was ever said to me?
 
@Mr.ShinyandNew安宇 mmm... from the little I know about STT it doesn't sound wrong.
 
@KitZ.Fox Well, maybe, but that's not the problem here. The problem is that it heard a sound, thought you might be speaking to it, and didn't know that the sound wasn't speech, but then couldn't recognize what you said.
I mean, I've seen the same thing happen with humans. One human farts or sneezes or coughs and another human nearby says "pardon?"
 
That's certainly the case for RNNs for sentence (wordd/syntx) construction (not sounds, just typing).
@Mr.ShinyandNew安宇 What? That's classic.
 
1:28 PM
oh that, yes.
 
@Mitch yes, it's always funny
but the point is that sometimes people hear a sound and think someone is talking to them but didn't hear what you said
 
I think more of the point of the article is that the sound analysis is not what we're used to linguistically, i.e. phonemic.
It's barely phonetic.
 
I don't think it's all that unusual for a computer to have the same problem.
I mean, we expect it to be bad at recognizing things, right? Sometimes we're speaking to it and it doesn't understand.
 
@Mr.ShinyandNew安宇 Whenever I hear the president talking all I hear are expressions of GI distress.
 
@Mitch Siri should run for president
 
1:30 PM
haha politics and poop jokes together.
John Lennon's dream.
 
Imagine all the people, pooping all at once
And echo responding "I can't find [insert least favourite song name] in your library"
 
The worst water utility disaster occurred the evening of Feb 29th, 1985 at 9:35pm EST after the series finale of the very popular MAS*H TV show, watched by more than 20 Million viewers. The supposed cause was all those millions flushing after waiting the whole show.
 
you might call it machine learning
you're not the only one
One day you'll realise it's just statistics
with a name that's a bit more fun
 
@MattE.Эллен has anybody had Alexa and Siri have a conversation?
@MattE.Эллен ML is hot, stats is for accountants
who are not hot.
more lukewarm
 
@Mitch I know there's a siri vs siri, but I don't know about any other combinations
or maybe it was the google one vs itself
@Mitch they wear jumpers to save on heating
costs
 
1:35 PM
I've seen youtube chatbots conversations. Not as funny as you'd expect. They almost make as much lack of sense as normal people.
@MattE.Эллен so responsible
They know what's going to happen.
 
"Are we all going to die?"
"Statistically? Yes."
 
Because they use predictive models (= 'actuarial tables') to determine your exact time of death.
 
Statistics show that the first immortal has probably already been born. They also show that it's probably not you
 
One of those tables says for a given population that for a given age what the expected number of years is. You know: age 70 - expected years til death 8; age 80 - 5, 85 - 3, 90 - 2.
How I read that is if I can only get one more year, it'll just keep going.
That's my plan at least.
 
at some point it'll have to start giving negative numbers, so you'll have to have lived less time than you have in order to die
 
1:43 PM
Well, if you see a negative number in your profile, I suppose that's somewhat optimistic, you made it past the expiration date.
 
so I'd be statistically immortal! the trick would be to do it every day
 
@MattE.Эллен yes. Or when you see negative, that means you're actually getting younger.
on average. it can't work out for everybody
that's math
 
until I am everybody and then ... um ...
 
@Mr.ShinyandNew安宇 yes, totally. It's just that the manner of the mistakes (or misunderstanding due to interrupting noise) will be different because of the underlying mechanism
@MattE.Эллен That took a weird turn.
 
@Mitch will it though? It seems to me that in both cases the listener is hearing a sound that is in the acoustic range of speech, emanating from where it believes a speaker to be, but then the sound cannot be understood through any means because it sounds garbled.
I guess you could argue that if Siri's mic clearly hears the fart but has the same problem, then it's not the same issue - but again, is this because Siri can't process language? or because Siri's mic isn't as good as human ears?
 
2:01 PM
@Mr.ShinyandNew安宇 It could very well be that the formant analysis of (particular) farts are indistinguishable from frcatives at given places of articulation.
 
@Mitch I think what the article is trying to say is that Siri should recognize that a fart is not words, but a fart. What I'm saying is that sometimes people hear a fart, don't recognize it as a fart or words, but think that maybe someone is trying to say something, and so they'll respond like Siri does.
 
I don't doubt that people may mistake farts as an attempt at a linguistic utterance.
But what I get from LL is that they're saying that STT isn't engineered to care ever (or to take context into account, real world or not). STT is not analysing speech according to linguistic principles (eg phonemes). It's just a function that takes input to output. And the mechanism to compute the output is not a human simulation.
 
2:24 PM
perhaps. But arguably human linguistic understanding is just a function that takes input and produces output.
 
2:41 PM
@Mr.ShinyandNew安宇 yes. I guess the import is about the implementation, that STT is not doing that function in anyway similar to humans
 
I used to pronounce 'awry' as 'AW-ree'
which, if everybody else did, would be ok.
 
maybe if you keep doing it, others will follow you
 
starts skipping to the loo
feel somewhat uncomfortable having people following there
but goes with it
@Mr.ShinyandNew安宇 for example, if, say, Siri had phonemic processing, presumably it would be better with foreign accents and if not easier to fix by making it aware of sound change rules. But instead the only real fix is to have a labeled training corpus with more variety (people with other accents).
 
3:38 PM
@Mitch Sure. But I don't think that would eliminate the problem.
 
3:50 PM
2
Q: Best word to describe "hard to overlook" or "hard to ignore"?

ivanasetiawanWhat is the best word to describe something positive that is hard to overlook or to ignore? Example: Her passion for dancing is ______ (hard to ignore)!

How is this not yet another useless fishing expedition?
 
@tchrist Are there situations like this on SO? Like people asking for suggestions about what is the best function/pattern to use?
 
4:39 PM
@Mitch I think everybody else did, at least I know I once did. But I have learned to pronounce it correctly, as I'm sure most people have.
@M.A.R. Nice.
 
5:00 PM
@M.A.R. @Mitch I misspelled relinquish for many years.
 
I mix up 'redicivist' and 'recidivist'
 
Hehe
When I type 're', 'd' is much more likely to come next rather than 'c'
 
You'd think redicivist would be the opposite of recidivist.
@Mr.ShinyandNew安宇 I often confuse them with music.
 
Today someone is busy downvoting a lot of answers on ELL
Not necessarily wrongfully so, but the standards of the two sites differ
Probably a visitor from ELU, huh
 
5:08 PM
A redicivist is someone who redites.
Or is it redicts?
 
Ridiculous.
 
Redices
 
Indict, indiction.
 
Indictive
 
Undictive.
Or is it inductive?
 
5:17 PM
Inductive already exists. Booo
 
@tchrist Maybe the question is too broad by the too many possible answers definition, but "hard to ignore" is such a common sentiment that it is reasonable to assume there might be a word for it. I am left wondering why people didn't suggest near misses like distracting, or the obvious unignorable. Most importantly the questioner included the requisite exemplary sentence which among other things, is supposed to show how needful a word is.
 
@M.A.R. Goddammit...you just made me waste a half hour trying to figure out what 'redicivist' really means.
No, I'm not telling anyone.
 
I'm telling everyone
 
@MetaEd ?? with a 'k'?
@M.A.R. I dare you.
 
HEY EVERYONE, Mitch totally wasted half an hour trying to understand what 'redicivist' means!!!
 
5:30 PM
@M.A.R. ?? You can tell that to people all you want, but what I am not telling is what 'redicivist' really means. Because it means...
 
And I linked the message above in five rooms
 
Ha... you thought you had me there for a moment.
@M.A.R. Go ahead. I just tweeted it.
instagramming... done.
snapchat is kinda a waste
Screaming into my pillow...
...
still looking for a pillow
...
Oh forget it.
just screams
Wow, that really got people's attention
 
@Tonepoet AW-ree? Yeah I figured. Only someone drunk would say it like that.
@Tonepoet Oh... haha... that's totally a missspelling.
Recidivist is someone who repeats doing criminal activity.
What your parole officer is there to prevent.
Except everyone knows, including your parole officer, that incarceration is where you learn the best ways to perform criminal activities.
 
5:44 PM
@Mitch With a "g".
@Tonepoet Here's how.
 
@MetaEd in script they're almost identical
@MetaEd Exactly. That's just taunting you to find the real definition.
The Deep Dictionary.
 
@MetaEd Google auto-corrects this though.
 
By the Illunimati
Wait... hold on...
 
@Mitch Try looking under fnord.
 
dangit... I have to call my palore officer real quick.
 
5:50 PM
@Mitch You mean your force pearl. Oif!
 
 
2 hours later…
7:31 PM
@DanBron That's inexcusable.
But dictionaries don't always give you the whole story.
Especially with close synonyms, which often give me much headache, it takes more than a definition and a bunch of examples to convey the feel of the word.
But I imagine that's not the type of question you complained about.
 

« first day (2328 days earlier)      last day (2587 days later) »