Comment On #NO DATA

No comment

The real WTF is that Thomas doesn't know the difference between precision and accuracy.

<quote>The real WTF is that Thomas doesn't know the difference between precision and accuracy. </quote> Exactly. Except for one minor misspelling (illconditioned), there's no WTF.

Hold the presses, terrorists stole our list data!

As for the Socratic method, knowing that you don't know why is more important than knowing why...

Thank Goodness! I was just stopping by to explain the difference between accuracy and precision, and it restores my faith in geeks everywhere that I don't need to.

As for the first one, Errorism List seems appropriate.

NThenUDie:

The real WTF is that Thomas doesn't know the difference between precision and accuracy.

That's about the first time I've seen the phrase "The real WTF" followed by a relevant failure. Seriously, I learned that as a freshman in high school.

The real WTF is that not only does thomas not know the difference between precision and accuracy, but he's the kind of person who needs to use finite element (or possibly something else using stiffness matrices). If you're an engineer/physicist/mathemetician who doesn't know that difference... you may have a problem.

due to the lack of precision in its stiffness matrix, the servers expectation has failed and the server did not know why so it printed an invalid mailing envelope which was caught by netflix staff but was still sent anyway.

Did you know that "Warning number 55302" comes with it's own tape?

http://www.drillspot.com/products/64055/Brady_55302_Warning_Tape

I know I'm illconditioned and suffer from isolated stiffness.

CAPTCHA: consequat - what happens when you try to squat down at my age?

Darren Landrum:

As for the first one, Errorism List seems appropriate.

That's right! The subways of the intertubes are being destroyed by errorists. All you programmers better stop making errors or it's off to Gitmo with you!

If there is an ill conditioned stiffness matrix, and precision is lost, the error in precision can via multiplications yield entirely inaccurate results. Anyone who has taken a course in numerical methods knows that. I'm not sure what the WTF is here.

What's the problem with the first image? The ticker SAYS it's an "Errorism list".

I think you guys are being too hard on Thomas. Although the message is indeed not as absurd as it may appear to the layman, he is still technically correct: a loss of precision does necessarily cause a loss in accuracy (but not vice versa).

For instance, let's say I have this Python code:

googol = 100.0 ** 100
x = 1000000000.0
x += googol
x -= googol
print x # prints 0.0 instead of 1000000000.0

The error is caused only by precision, but it has rendered calculations on small numbers meaningless. No matter what you're calculating, 0 and 1,000,000,000 aren't likely to be in the same ballpark... probably not even in the same sport. When the precision was thrown out the window, the accuracy went with it.

Yes, we can ... get rid of those (t)errorists, that's why the list has #NO DATA anymore. Well done, Mr. President, and much faster than I expected.

Well, if the Netflix envelope didn't arrive, then the USPS has a lot to learn from the UK Royal Mail. The Netflix one wouldn't have been a challenge at all for them.

Could not enter comment. Form does not know why

If I've learned anything from reading The Daily WTF it's that errorism is a greater threat than terrorism ever could be.

whatever:

Did you know that "Warning number 55302" comes with it's own tape?

What 55301 other warnings did they manage to avoid in order to get to that one, though?

silent d:

If I've learned anything from reading The Daily WTF it's that errorism is a greater threat than terrorism ever could be.

Precisely. I know there are those who suffer from terrorism, but speaking only for myself, Bill Gates and his followers have caused me far more harm than Osama BinLaden and his followers.

It sounds like the stiffness matrix could use some Viagra. I have approximately 3,947,234 emails that I could forward to it...

Mr G:

whatever:

Did you know that "Warning number 55302" comes with it's own tape?

What 55301 other warnings did they manage to avoid in order to get to that one, though?

Warning number 1
Warning number 2
Warning number 3
...

Do I really need to list them all?

cdosrun:

Mr G:

whatever:

Did you know that "Warning number 55302" comes with it's own tape?

What 55301 other warnings did they manage to avoid in order to get to that one, though?

Warning number 1
Warning number 2
Warning number 3
...

Do I really need to list them all?

Yes please.

Kef Schecter:

I think you guys are being too hard on Thomas. Although the message is indeed not as absurd as it may appear to the layman, he is still technically correct: a loss of precision does necessarily cause a loss in accuracy (but not vice versa).

For instance, let's say I have this Python code:

googol = 100.0 ** 100
x = 1000000000.0
x += googol
x -= googol
print x # prints 0.0 instead of 1000000000.0

The error is caused only by precision, but it has rendered calculations on small numbers meaningless. No matter what you're calculating, 0 and 1,000,000,000 aren't likely to be in the same ballpark... probably not even in the same sport. When the precision was thrown out the window, the accuracy went with it.

Hmm...maybe my 9th grade physics teacher explained precision/accuracy to me a bit wrong, or maybe I'm remembering it wrong...but, wouldn't the predictability of your scenario show the very definition of precision? I mean, that code will always produce 0.0. Always (right?). That is as precise as it gets. Accurate, not even a little. But as precise as it gets.

I think what you were demonstrating was full precision, without any accuracy...of course, accuracy is all relative anyhow. If you're margin of error is +-googol, you're not all that far off...and you'd have some modicum of accuracy, along with the best precision money can buy.

In any event, the WTF was submitted in such a manner that it assumes they are the same. Which they are not.

Oh and you absolutely can have accuracy without precision. I think my "relative" accuracy shows that well enough...

cdosrun:

Mr G:

whatever:

Did you know that "Warning number 55302" comes with it's own tape?

What 55301 other warnings did they manage to avoid in order to get to that one, though?

Warning number 1
Warning number 2
Warning number 3
...

Do I really need to list them all?

What about

Warning number 0
Warning number 'null'
Warning number is not an object
Warning number content not found

Target: 50
Precise Result set: 4, 4, 4, 4
Accurate Result set: 51, 49, 48, 52

Target: Bullseye
Precise: Always get the same outer box
Accurate: Always get near the bullseye

Target: Explanation of Precision versus Accuracy
Precise: Providing unnecessarily complex code to humans to demonstrate an obvious thing
Accurate: Providing three analogies to describe the difference for humans

Someone forgot to enter data into something? That's HILARIOUS.

Seriously, it's getting about as old as people copy-pasting from the article and changing the noun with the word "comment" and posting it so they get the first post without actually reading the article.

pscs:

Well, if the Netflix envelope didn't arrive, then the USPS has a lot to learn from the UK Royal Mail. The Netflix one wouldn't have been a challenge at all for them.

Care to summarize that? Content filtering here has given it the somewhat dubious label "web-based proxy or anonymizer".

Kef Schecter:

a loss of precision does necessarily cause a loss in accuracy (but not vice versa).

A loss of precision does not necessarily cause a loss in accuracy, but it does almost always cause a loss in accuracy. A counterexample can be seen in the following python code:

x = 10000.000000
y = round(x, 3)
print(x, y)

x = 1000.0001110
y = round(x, 3)
print(x, y)

which returns:

10000.0 10000.0
1000.000111 1000.0

we lost precision, but not accuracy in the first calculation, and lost both in the second.

(This post was written with python 3.0.1, notice the new print syntax)

pscs:

Well, if the Netflix envelope didn't arrive, then the USPS has a lot to learn from the UK Royal Mail. The Netflix one wouldn't have been a challenge at all for them.

It's amazing that they deliver anything at all. See
http://www.telegraph.co.uk/news/uknews/1436409/First-class-only-14.5m-letters-lost.html

Whenever I go to the post office to mail something, and they ask me "Is it anything urgent or valuable?", I always want to reply "If it was, I wouldn't be giving it to you"

The Expect 100 thing is not a WTF... Look at the fonts... The expected string is clearly different.

A watch that shows the time to the nearest hundredth of a second, but is five minutes slow, is precise but not accurate.
A watch that shows the time to the nearest minute, and shows the correct time, is accurate but not precise.

Someone You Know:

Care to summarize that? Content filtering here has given it the somewhat dubious label "web-based proxy or anonymizer".

Basically, someone had lost contact with a friend, so instead of putting an address on the envelope (because they didn't know it) they put the name, with a sketched map of an English county, with an arrow saying 'somewhere around here'. It took 9 days to arrive.

We regularly get post delivered to us with just our name and street name, and occasionally with just our name and town name. (Some companies' postcode databases can't handle our address, because we don't have a house number).
(This may not sound too impressive, but my surname is 'Smith', which isn't rare in the UK...)

Milligan:

It sounds like the stiffness matrix could use some Viagra. I have approximately 3,947,234 emails that I could forward to it...

That's not an approximation but a mis-estimation.

Not a grammar nazi, a terminology nazi.

CAPTCHA: Distineo. Distineo form isn't big enough for my entire comment.

The funny thing is, I bet the Netflix envelope would work. They probably move a few tens/hundreds of thousands of them through any given distro center on any given day; one that has a bad label could probably be safely routed to wherever the other mountain of mailers are headed.

pscs:

Someone You Know:

Care to summarize that? Content filtering here has given it the somewhat dubious label "web-based proxy or anonymizer".

Basically, someone had lost contact with a friend, so instead of putting an address on the envelope (because they didn't know it) they put the name, with a sketched map of an English county, with an arrow saying 'somewhere around here'. It took 9 days to arrive.

We regularly get post delivered to us with just our name and street name, and occasionally with just our name and town name. (Some companies' postcode databases can't handle our address, because we don't have a house number).
(This may not sound too impressive, but my surname is 'Smith', which isn't rare in the UK...)

No, that's actually pretty impressive.

Long-time Listener:

silent d:

If I've learned anything from reading The Daily WTF it's that errorism is a greater threat than terrorism ever could be.

Precisely. I know there are those who suffer from terrorism, but speaking only for myself, Bill Gates and his followers have caused me far more harm than Osama BinLaden and his followers.

I think you meant to say "Accurately. I know..."

Maybe the movie is so awful Netflix doesn't *want* it back...

cdosrun:

Warning number 1
Warning number 2
Warning number 3
...

Do I really need to list them all?

...
Warning number 22018
Warning number 22019
Warning number H
Warning number E
Warning number L
Warning number P
Warning number I
Warning number '
Warning number M
Warning number T
Warning number R
Warning number A
Warning number P
Warning number P
Warning number E
Warning number D
Warning number I
Warning number N
Warning number A
Warning number W
Warning number A
Warning number R
Warning number N
Warning number I
Warning number N
Warning number G
Warning number F
Warning number A
Warning number C
Warning number T
Warning number O
Warning number R
Warning number Y
Warning number 22051
Warning number 22052
Warning number 22053
...

Long-time Listener:

silent d:

If I've learned anything from reading The Daily WTF it's that errorism is a greater threat than terrorism ever could be.

Precisely. I know there are those who suffer from terrorism, but speaking only for myself, Bill Gates and his followers have caused me far more harm than Osama BinLaden and his followers.

Wow. Really? A blue screen is far more inconvenient to you than thousands of people dead? Wow. I don't want to seem like a moralist here, but how about at least thinking before typing?

JoPoser:

Milligan:

It sounds like the stiffness matrix could use some Viagra. I have approximately 3,947,234 emails that I could forward to it...

That's not an approximation but a mis-estimation.

Not a grammar nazi, a terminology nazi.

CAPTCHA: Distineo. Distineo form isn't big enough for my entire comment.

Could it be an approximation in base 3947234, converted back into decimal?

Cody:

This post was written with python 3.0.1

Wow. I didn't know you could do that!
So I got curious and after a few tries managed to write this post using .NET 3.5!

bjolling:

Cody:

This post was written with python 3.0.1

Wow. I didn't know you could do that!
So I got curious and after a few tries managed to write this post using .NET 3.5!

.. / .-- .-. --- - . / -- -.-- / -.-. --- -- -- . -. - / .. -. / -- --- .-. ... . / -.-. --- -.. . .-.-.-

Cory A:

Wow. Really? A blue screen is far more inconvenient to you than thousands of people dead?

Well, thousands of people dead doesn't inconvenience me in the least. It's a tragedy to be sure, but it doesn't inconvenience me.

The General:

Warning number '

The text of which is '; DELETE TABLE Warnings --

There is nothing wrong about the error message for ANALYSIS.DLL. That is a very technical description of a very specific problem in numerical methods for structural analysis, which has very clear consequences in the accuracy of the results of the calculation. It makes perfect sense to those who are familiar with the domain. It might look funny only if you are not familiar with the language.

And in this case, the use of "precision" and "accuracy" is... "precise and accurate".

That's what happens when you have mathematicians-structural engineers writing software for other mathematicians-structural engineers: they speak a domain-specific language common to them that outsiders don't understand.

But, wait! There really is a WTF: a component that clearly is a computational DLL, boasting its own UI.

Euro:

But, wait! There really is a WTF: a component that clearly is a computational DLL, boasting its own UI.

Or, a UI that can report exceptions thrown by a DLL! Oh wait, you were being sarcastic...

Browser (code name BRIAN): "Are you the Expect: 100-continue?"
Server: "Expect: 100-continue?! F**K OFF! We're the 100-continue Expectation! *snicker* Expect: 100-continue..."

pscs:

Well, if the Netflix envelope didn't arrive, then the USPS has a lot to learn from the UK Royal Mail. The Netflix one wouldn't have been a challenge at all for them.

Actually, the USPS is pretty good at this too. Given that many more Netflix returns will be delivered to the correct address on the same day and days before and after, it would take a special effort not to be able to deliver the given example.

monkeyPushButton:

bjolling:

Cody:

This post was written with python 3.0.1

Wow. I didn't know you could do that!
So I got curious and after a few tries managed to write this post using .NET 3.5!

.. / .-- .-. --- - . / -- -.-- / -.-. --- -- -- . -. - / .. -. / -- --- .-. ... . / -.-. --- -.. . .-.-.-

.--- .-. -.-- -.-- / --.- -... .- .-. / ..-. ...- . .-.-.- / ..- -. .. .-. / -. / .--. -... -... -..- ...- .-. .-.-.-