WTF Bites
-
WHO HIRED THE UNIX TWAT?!?!?
Unix doesn't use GUIDs.
...only UUIDs...
$ ls /dev/disk/by-uuid/ -l total 0 lrwxrwxrwx 1 root root 10 May 18 06:23 2cd7f16e-c9b5-4d51-a52b-a988fb31fa19 -> ../../sda1 lrwxrwxrwx 1 root root 10 May 18 06:23 5c883011-b319-4090-a150-1de1eca59c5e -> ../../dm-1 lrwxrwxrwx 1 root root 10 May 18 06:23 746b7501-2ace-4ad0-99be-e718fdc4f2a1 -> ../../dm-2 lrwxrwxrwx 1 root root 10 May 18 06:23 f7c92837-8a3d-44d9-9de2-f6db8a31dd3d -> ../../dm-3 lrwxrwxrwx 1 root root 10 May 18 06:23 f88dea22-fce1-4459-a94e-0063738b3bda -> ../../sda5 $
-
@tsaukpaetra said in WTF Bites:
Are you on the latest breakage update?
>ver Microsoft Windows [Version 10.0.17134.48]
-
@TimeBandit said in WTF Bites:
@boomzilla said in WTF Bites:
I've seen Marshmallow bags with "A fat free food!" on them.
Vegetables can't compete
Fat content:
Carrots 0.2g per 100 grams.
Broccoli 0.4g per 100 grams.One of my aunts is very health-conscious, and she wouldn't allow her kids to have high-sugar cereals, but we got her to allow them to have Marshmallow Mateys (cheap version of Lucky Charms) when we showed her the nutrition facts label in comparison to Frosted Mini Spooners (cheap version of Frosted Mini Wheats). Marshmallow Mateys has more vitamins and minerals and less sugar than Frosted Mini Spooners. My cousins were pretty happy when she relaxed that rule.
-
Someone explain to me how I'm supposed to trust an AI to do autonomous driving while at the same time Google, the purported gods of ML, use "recognize traffic signs" as
Turing testsCAPTCHAs, meaning it's a task they consider hard for computers to automate.
And apparently can't even get the expected results correct. I've tried this 5 times before starting to take screen shots, and I'm pretty damn sure at least some of them I was right, even though Google didn't accept my answer:
-
@topspin I'm never sure it if wants the sign posts selected as well. It seems to be completely random. At least identifying types of vehicles and storefronts works...
-
I'm never sure it if wants the sign posts selected as well
Since it doesn't say so, I've gone with "no". Clearly, the first one is correct either way though.
-
Someone explain to me how I'm supposed to trust an AI to do autonomous driving while at the same time Google, the purported gods of ML, use "recognize traffic signs" as
Turing testsCAPTCHAs, meaning it's a task they consider hard for computers to automate.Isn't it obvious? They're using the data from the CAPTCHAs as part of the input data to train their ML algorithms.
-
@dkf At least that was what they advertised about ReCaptcha for OCR back in the day. But they already have cars on the street and their idea of the correct answer is awfully bad. Like, if their cars were no better than that, they'd need another ten years.
-
@topspin Most of the sign spotting works fairly well, at least for critical signs like speed limits, and routing can be largely done by mapping without the need for most of the signage. Similarly for most hazards. It's temporary signage that is really tricky. Also anywhere where some jerk has put up a sign that looks vaguely official but isn't actually.
There are also some weird edge cases. One of the speed limit signs on a small residential streets near here has a dent in it that my car keeps misreading, turning a 20 mph (very slow) sign into a 120 mph (!!) sign in the eyes of the software.
-
@dkf Yeah, all of this only means that real-world scenarios are far harder to get consistently correct than idealized test cases. My first picture falls squarely (as well as circularly and triangularly) in the set of easy, idealized scenarios.
IOW, you're making my point.
-
My first picture falls squarely (as well as circularly and triangularly) in the set of easy, idealized scenarios.
It could have been a calibration image, used because they want to check that you're not giving BS answers and yet are planning to give you an image that needs more work as well. I'm not well versed in how they prepare data for these sorts of things, but they definitely need to have a mix of well labelled and poorly labelled images and each image will have to go to many different people to see if there's enough agreement, or you'd be too open to problems with asses putting in bad inputs.
-
It could have been a calibration image, used because they want to check that you're not giving BS answers and yet are planning to give you an image that needs more work as well
That used to be the case, but it no longer is. Usually, when their system isn't having a day of "I'm fucking blind", you only have to solve this puzzle once. There's no two stage design where you have to identify an image with known ground truth and one unknown image to improve their data.
They did exactly that in the ReCaptcha thing, though. You had to identify two badly garbled words, one of them was known ground truth, the other wasn't. You could game that system (for fun
and profit) by correctly guessing which is which and then answering with "$correct_first_word $expletive".
-
This error message:
I was deleting things out of the DOM at the time so I think it probably was my fault, actually.
-
Using Google's business email?
Sucks to be you.
Google is starting to look increasingly like Brazil's Central Services.
-
Using Google's business email?
Sucks to be you.
Google is starting to look increasingly like Brazil's Central Services.Wow. I was thinking of upgrading to the paid version of Google Apps; now I'm thinking fuck that, time to migrate to O365.
-
Using Google's business email?
Sucks to be you.
Google is starting to look increasingly like Brazil's Central Services."Google caught me committing fraud and then banned my account!"
-
@ben_lubar said in WTF Bites:
Using Google's business email?
Sucks to be you.
Google is starting to look increasingly like Brazil's Central Services."Google caught
meone of my employees committing fraud and then bannedmyevery single of my company's accounts!"Perspective fixed.
Plus the policy is a bit like the one for Israel's nukes: "we can't confirm the existence or nonexistence of any kind of fraud or non-fraud but rest assured there was a TOS violation that we can't tell you about, and no, you can't talk to a human about how to fix this situation"
(unless you blare your message to half the interwebs on Reddit that is)
-
@laoc this kind of story is what people write when they refuse to admit they did something wrong.
Usually when there's a Reddit post on the GW2 reddit like that, someone from ArenaNet comes by to post evidence that the person lied about what ArenaNet accused them of and explains what actually happened.
-
@ben_lubar said in WTF Bites:
@laoc this kind of story is what people write when they refuse to admit they did something wrong.
The kind of story that starts out with "I fucked up and abused a policy"?
-
@blakeyrat said in WTF Bites:
@tsaukpaetra said in WTF Bites:
Oh, sorry, did online reformatted text not appear to you?
Honestly no. Was there some?
@tsaukpaetra said in WTF Bites:
var requestIP = GetClientIpAddress(HttpContext.Request):
I don't have access to that codebase right now but the problem is my "get the client's IP helper" requires a HttpRequestBase and HttpContext.Request ain't one.
AFAIK the only way to get from the HttpContext to the HttpRequest is to walk down the chain like I did: Context.Current -> Request -> RequestContext -> HttpContext -> Request
Which is fucking ridiculous, thus the comment.
Violation of the Demeter law detected! Sound the alarm!
-
@ben_lubar said in WTF Bites:
@laoc this kind of story is what people write when they refuse to admit they did something wrong.
The kind of story that starts out with "I fucked up and abused a policy"?
The claims that the user has presented:
- They committed fraud repeatedly
- Google would not support them
Therefore, I can deduce:
- They were not on GSuite (which guarantees 24/7 support https://gsuite.google.com/support/)
- An automated system probably associated the various free GMail accounts they signed up for each user in the company with each other
- A ban evasion prevention system banned all the accounts as they appeared to all be sock puppets of the reddit OP
-
@ben_lubar said in WTF Bites:
@laoc this kind of story is what people write when they refuse to admit they did something wrong.
The kind of story that starts out with "I fucked up and abused a policy"?
The claims that the user has presented:
- They committed fraud repeatedly
- Google would not support them
Yup. Sounds like a clear admission of guilt.
Therefore, I can deduce:
- They were not on GSuite (which guarantees 24/7 support https://gsuite.google.com/support/)
Probably. He didn't say they were … although I wouldn't be surprised if the policy was "well it's a TOS violation so GTFO, no support for you"
- An automated system probably associated the various free GMail accounts they signed up for each user in the company with each other
- A ban evasion prevention system banned all the accounts as they appeared to all be sock puppets of the reddit OP
Supposedly they "signed on" to something. No idea what. So far nobody has come to refute the story though even though the GSuite guys are aware of it.
-
Using Google's business email?
Sucks to be you.
Google is starting to look increasingly like Brazil's Central Services.LOL:
A Reddit, on a phone in a bathroom stall, accidentally flushes an entire company. Hackernews does not lend any credence to this story, because Google is flawless and would never deploy an automated system that accidentally fucks over an entire organization (except for the many, many times they do precisely that). When defending the honor of Google loses its luster, Hackernews moves on to telling war stories about other shitty customer support from other software companies, presumably to make Google feel better about being a pack of incompetent stooges.
-
@laoc Wrong Terry.
-
@ben_lubar said in WTF Bites:
"Google caught
meone of my employees committing fraud and then bannedmyevery single of my company's accounts as well as personal accounts which gave a company account as their recovery address!"Further perspective fix.
-
-
-
Dafuq. Of course. Excuse me, total brainfart
Uh, this is WTDWTF. You're not supposed to admit error. DOUBLE DOWN!
-
-
-
@heterodox said in WTF Bites:
No, you're thinking of Burger King.
No, you're thinking of KFC.
The Dutch airline?
-
@heterodox said in WTF Bites:
No, you're thinking of Burger King.
No, you're thinking of KFC.
I'm thinking Arby's.
-
@heterodox said in WTF Bites:
No, you're thinking of Burger King.
No, you're thinking of KFC.
I'm thinking Arby's.
Yeah! Make a run for the border!
-
Amazon finally started delivering to Serbia, so I decided to give them a try. I ordered a fancy pair of headphones.
Gulp. 27 pounds for shipping + 40 pounds customs fee. Almost doubling the price of the item.
Well, that's just Serbian tax for you, I'm used to it. Whatever.
The thing was supposed to arrive today, but now I get a call from DHL. Apparently, there's a little bit of paperwork to fill regarding customs. Then they send me an email with like 5 different documents.
WTF is all this shit!? A TIFF image? A word and excel doc from the 90-ies? So now I have to read some legal document and digitally sign a statement and submit a copy of my ID card and proof of purchase and wha...?
I don't want to deal with this shit, I just want my freaking headphones I've paid fortune for! Is that too much to ask!?
-
@ben_lubar said in WTF Bites:
The claims that the user has presented:
They committed fraud repeatedly
Google would not support themTherefore, I can deduce:
They were not on GSuite (which guarantees 24/7 support https://gsuite.google.com/support/)
An automated system probably associated the various free GMail accounts they signed up for each user in the company with each other
A ban evasion prevention system banned all the accounts as they appeared to all be sock puppets of the reddit OPWhat a shocker. Ben believes Google is a Jesus-like holy company that can do no wrong. Who saw that coming.
-
@heterodox said in WTF Bites:
No, you're thinking of Burger King.
No, you're thinking of KFC.
The Dutch airline?
That's KLM
-
@heterodox said in WTF Bites:
No, you're thinking of Burger King.
No, you're thinking of KFC.
The Dutch airline?
That's KLM
The techno group?
-
@tsaukpaetra
Connection pooling?
-
Company founded by ex-Logitech engineers to make customized 3D headphones runs out of money and closes down. Where did their $3.2 million come from? Kickstarter, naturally.
-
@coldandtired said in WTF Bites:
@benjamin-hall Here's a bottle of gluten-free vegan water. It's got about six spoons' of sugar, though.
@coldandtired said in WTF Bites:
@benjamin-hall Here's a bottle of gluten-free vegan water. It's got about six spoons' of sugar, though.
They do sell a "zero" version, where the unhealthy amounts of unhealthy sugar have been replaced with unhealthy amounts of unhealthy aspartame, though.
-
As long as the question doesn't specify heavy concurrent traffic, I think this is safe enough.
And 640kB ought to be enough for anybody.
That's a slur, he never said that.
-
They do sell a "zero" version, where the unhealthy amounts of unhealthy sugar have been replaced with unhealthy amounts of unhealthy aspartame, though.
Twice.
-
They do sell a "zero" version, where the unhealthy amounts of unhealthy sugar have been replaced with unhealthy amounts of unhealthy aspartame, though.
Twice.
OSheetSquared!
-
@tsaukpaetra
Connection pooling?Hand-rolled, yes, but the query is ensured to either be complete or cancelled before attempting to return the connection to the pool.
At least, it's supposed to be. My fears are in threading and shared pointer ghostery that the debugger won't be able to help solve.
I'm pretty sure it may also be involved in a crash-on-exit bug that's happening with a freed statement handler...
-
@tsaukpaetra
Or could it be simple resource exhaustion? Trying to run 6 queries against 5 connections in the pool and query #6 doesn't realize it needs to take a number and sit down to enjoy the DMV's hold music?
-
@tsaukpaetra
Or could it be simple resource exhaustion? Trying to run 6 queries against 5 connections in the pool and query #6 doesn't realize it needs to take a number and sit down to enjoy the DMV's hold music?Impossibru, every query gets shoved into the queue bar none. The pool picks up queries fifo. Though thinking on it, it also tries to kill queries that are taking longer than ten seconds. That might be it, if the cancellation and GTFO code isn't working properly...
-
Someone explain to me how I'm supposed to trust an AI to do autonomous driving while at the same time Google, the purported gods of ML, use "recognize traffic signs" as
Turing testsCAPTCHAs, meaning it's a task they consider hard for computers to automate.
And apparently can't even get the expected results correct. I've tried this 5 times before starting to take screen shots, and I'm pretty damn sure at least some of them I was right, even though Google didn't accept my answer:
I could have sworn that I already replied to this. I must've been typing it up when the power went out. So now that I've found it back, here goes again.
First of all, that's "normal". Google's being a dick and it's doing that to you on purpose. You'll even notice, if you pay close attention, that the "Next" button changes to "Verify" on the last set of images.
Are you using any sort of tool or script (such as the JDownloader 2 browser extension...) that automatically clicks the checkbox for you? If so, turn that off... Google appears to detect it somehow, and gives you more obnoxious captchas. Click the checkbox manually.
-
From the updated privacy policy of a game I play:
with an average annual availability of 90% (ninety percent). This does not include periods of time during which the use of online and mobile games and other services are interrupted or affected due to urgent technical reasons or required maintenance work. [...] This also does not include periods of time in which $PUBLISHER's general servers or the servers of certain games do not allow online access due to reasons beyond $PUBLISHER's control (force majeure, third party responsibility, etc.).
If you're excluding urgent technical issues, maintenance, and external issues, what the heck is left that makes you only able to manage 90% availability?
-
@anotherusername said in WTF Bites:
You'll even notice, if you pay close attention, that the "Next" button changes to "Verify" on the last set of images.
I'll check that next time I see it.
Are you using any sort of tool or script (such as the JDownloader 2 browser extension...) that automatically clicks the checkbox for you? If so, turn that off... Google appears to detect it somehow, and gives you more obnoxious captchas. Click the checkbox manually.
No. You mean a tool to automatically solve the CAPTCHA? Or I guess you mean click the "I'm not a robot" checkbox that triggers the CAPTCHA?
Still no.I've read somewhere that they said "most of the time" it should work just by clicking the check box, without any CAPTCHA at all. Supposedly by the magic of their behavior analysis AI. That could actually detect a bot clicking that button instead of a human moving the mouse, different timings etc.
But really, I think it's just more of their stupid tracking. They know you're not a bot because they follow you around wherever you go. It doesn't work for me* because, as pretty much everywhere, I use private browsing with tracking protection on.* The "don't have to solve a CAPTCHA", I'm pretty sure their tracking me still works for them anyways.
-
as pretty much everywhere, I use private browsing with tracking protection on
I'm pretty sure they also give you easier captchas when you're logged into a Google account. So, using private browsing mode could definitely cause you to get more annoying captchas.
