 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
OK, so help me out here...
I just took the LPIC-101 exam. For reasons unknown, the test is scored
from 200 to 800. My score was a piffling 710.
If my arithmetic is right, (710-200)/(800-200) = 510/600 = 85%.
The test consists of 60 questions. 1/60 = 1.666% So that means that I
must have got between 9 and 10 questions completely wrong. (?)
And to think the guys in the office were taking bets on me getting a
100% score... Not even close!
(I'd love to know which questions I got wrong...)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Am 12.03.2013 19:30, schrieb Orchid Win7 v1:
> OK, so help me out here...
>
> I just took the LPIC-101 exam. For reasons unknown, the test is scored
> from 200 to 800. My score was a piffling 710.
>
> If my arithmetic is right, (710-200)/(800-200) = 510/600 = 85%.
>
> The test consists of 60 questions. 1/60 = 1.666% So that means that I
> must have got between 9 and 10 questions completely wrong. (?)
... or 60 of them slightly wrong. Or a mix of both.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Tue, 12 Mar 2013 18:30:31 +0000, Orchid Win7 v1 wrote:
> OK, so help me out here...
>
> I just took the LPIC-101 exam. For reasons unknown, the test is scored
> from 200 to 800. My score was a piffling 710.
That's actually a fairly standard scoring range for IT certification
exams. I had the formula explained to me once, and it's not as simple as
it seems (I wish I still had the e-mail that explained it). The scoring
methodology also takes into consideration partial credit for multiple-
select multiple choice questions (ie, "pick two of the following five
items that correctly describe X").
This doesn't have the formula, but it does explain the methodology pretty
well:
http://certification.acsm.org/faq17-exam-scoring
> If my arithmetic is right, (710-200)/(800-200) = 510/600 = 85%.
>
> The test consists of 60 questions. 1/60 = 1.666% So that means that I
> must have got between 9 and 10 questions completely wrong. (?)
>
> And to think the guys in the office were taking bets on me getting a
> 100% score... Not even close!
Questions are sometimes weighted as well, so it's not a straight "x"
points for "y" questions.
> (I'd love to know which questions I got wrong...)
You know, of course, that there's no way they'll tell you that. Bad
enough that testing centres in the far east are complicit with screen
scraping the exams and producing "braindumps" that give you the answers -
and result in "paper certifications". No test sponsor is going to
release a list of what you got wrong because that just helps the braindump
companies (and dealing with them is like playing whack-a-mole already).
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> I just took the LPIC-101 exam. For reasons unknown, the test is scored
>> from 200 to 800. My score was a piffling 710.
>
> That's actually a fairly standard scoring range for IT certification
> exams. I had the formula explained to me once, and it's not as simple as
> it seems (I wish I still had the e-mail that explained it). The scoring
> methodology also takes into consideration partial credit for multiple-
> select multiple choice questions (ie, "pick two of the following five
> items that correctly describe X").
Yeah, I'm curious to know what happens for the ones where you have to
enter text. How do they measure "how close" to being right you are?
>> The test consists of 60 questions. 1/60 = 1.666% So that means that I
>> must have got between 9 and 10 questions completely wrong. (?)
>
> Questions are sometimes weighted as well, so it's not a straight "x"
> points for "y" questions.
The LPI website seems to indicate that /subjects/ are weighted, with
more important subjects getting more questions. This seems to imply that
each individual question is worth the same amount.
>> (I'd love to know which questions I got wrong...)
>
> You know, of course, that there's no way they'll tell you that.
Yeah. Heaven forbid that I might actually *learn* from my mistakes...
As an aside, I had to remove my watch before I was allowed to take the
test. Because, you know, I might have programmed the answers into the
cogs and gears somehow...
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> The test consists of 60 questions. 1/60 = 1.666% So that means that I
>> must have got between 9 and 10 questions completely wrong. (?)
>
> ... or 60 of them slightly wrong. Or a mix of both.
The questions range from laughably easy to painfully obscure. Although I
have to admit, if you do a Google search for "LPIC example questions",
most of what comes up is *drastically* harder than the actual exam.
I'm pretty certain I got several questions right. But unfortunately
there were quite a few questions where I have absolutely no clue what
the answer might be. (Which distributions use Yum? I've never even heard
of it!)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Tue, 12 Mar 2013 19:23:07 +0000, Orchid Win7 v1 wrote:
> Yeah, I'm curious to know what happens for the ones where you have to
> enter text. How do they measure "how close" to being right you are?
Well, those tend not to be essay questions, but rather "what command does
'x'" - and that's fairly straightforward unless you overthink it.
Typically, an exam like this is looking for the most commonly known way
to do 'x', not the most obscure way you can think of, or the "well, in
98% of cases, I can use this, but in that last 2%, it doesn't work".
That's one of the reasons that before I worked in training and exam
development, I never too certification exams. I know I tend to overthink
things. :)
>>> The test consists of 60 questions. 1/60 = 1.666% So that means that I
>>> must have got between 9 and 10 questions completely wrong. (?)
>>
>> Questions are sometimes weighted as well, so it's not a straight "x"
>> points for "y" questions.
>
> The LPI website seems to indicate that /subjects/ are weighted, with
> more important subjects getting more questions. This seems to imply that
> each individual question is worth the same amount.
Yeah, it's also common to weight by "knowledge area". :)
>>> (I'd love to know which questions I got wrong...)
>>
>> You know, of course, that there's no way they'll tell you that.
>
> Yeah. Heaven forbid that I might actually *learn* from my mistakes...
That's not the purpose of an exam, though. An exam's purpose is to
measure what you know, not to help you assess what you don't know.
That's what an "assessment" or "technical skills assessment" (or
"technical knowledge assessment") is for. :) An exam is a measurement
tool, nothing more.
> As an aside, I had to remove my watch before I was allowed to take the
> test. Because, you know, I might have programmed the answers into the
> cogs and gears somehow...
Actually, I'm not surprised by that. Not because they might think you're
using the exam to answer questions, but rather that you might use it to
*record* your exam session. There are *many* ways that people try to
cheat on exams like this, not the least of which is to try to copy the
questions and then use that as a study guide.
I had a situation where a candidate was alleged to have taken a USB flash
drive in to get screengrabs of the exam in question (software like VUE's
test driver doesn't allow that, but for this particular exam, we weren't
using that system and the workation wasn't locked down as tightly as a
result). The drive was confiscated and examined, and it allegedly had
screengrabs from not just the exam the candidate was sitting, but other
exams as well.
These days, you can buy watches/watch bands that have USB storage in
them, or miniature cameras. It's easier to require the candidate just
not take it into the testing room rather than to examine each candidate's
personal items prior to taking the exam.
Having proctored certification exams and had candidates do things like
appearing to feign not understanding a testing objective (scenario-based
exam in that case) in order to try to get me to give him the answer, not
much surprises me these days when it comes to requirements and
restrictions.
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Orchid Win7 v1 <voi### [at] dev null> wrote:
> >> The test consists of 60 questions. 1/60 = 1.666% So that means that I
> >> must have got between 9 and 10 questions completely wrong. (?)
> >
> > ... or 60 of them slightly wrong. Or a mix of both.
>
> The questions range from laughably easy to painfully obscure. Although I
> have to admit, if you do a Google search for "LPIC example questions",
> most of what comes up is *drastically* harder than the actual exam.
>
> I'm pretty certain I got several questions right. But unfortunately
> there were quite a few questions where I have absolutely no clue what
> the answer might be. (Which distributions use Yum? I've never even heard
> of it!)
yum is the command line package manager used by Fedora and Red Hat. Debian,
Ubuntu and Mint use apt-get. I think Gentoo uses something strange, but I'd
have to look it up to be sure.
Regards,
A.D.B.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Tue, 12 Mar 2013 18:38:43 -0400, Anthony D. Baye wrote:
> yum is the command line package manager used by Fedora and Red Hat.
> Debian,
> Ubuntu and Mint use apt-get. I think Gentoo uses something strange, but
> I'd have to look it up to be sure.
emerge.
And openSUSE uses apper and/or zypper and/or YaST.
Which can use Yum, raw RPM directories, and a number of other sources.
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> I'm pretty certain I got several questions right. But unfortunately
>> there were quite a few questions where I have absolutely no clue what
>> the answer might be. (Which distributions use Yum? I've never even heard
>> of it!)
>
> yum is the command line package manager used by Fedora and Red Hat. Debian,
> Ubuntu and Mint use apt-get. I think Gentoo uses something strange, but I'd
> have to look it up to be sure.
I've never used Fedora or Red Hat. (No, wait, that's a lie. I have a
disk somewhere for Red Hat 2.1, circa 1996 or so. I don't think Yum
existed yet.)
Debian uses .deb packages, which can be installed using dpkg, dselect,
apt-get or aptitude. The exam requires you to know ALL FOUR. (!) Even
though in reality you'll only use one of these tools. (Usually aptitude,
because dselect is utterly incomprehensible!)
Most distros derived from Debian use the same packaging system. Most
other distros use RPM. But Gentoo uses the Emerge system, because it
builds absolutely everything from source. It's a great way to expend
compute cycles and disk space...
I never use RPM itself. I always use Zypper, but that's specific to
OpenSUSE and isn't on the exam. Whereas Zypper can uninstall a package
using the "remove" command, the exam requires you to know that this is
the "-e" switch to RPM. Not u for uninstall or r for remove, but e - for
no defined reason whatsoever.
But hey, that's Unix all over. Don't expect there to be any reason or
rhyme to it, just memorise this crap.
PS. It especially amuses me how many commands let you specify the number
of bytes of operate on with a -c switch. Because, as you know, 1
character = 1 byte. Oh, wait...
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> I just took the LPIC-101 exam. For reasons unknown, the test is scored
> from 200 to 800. My score was a piffling 710.
>
> If my arithmetic is right, (710-200)/(800-200) = 510/600 = 85%.
>
> The test consists of 60 questions. 1/60 = 1.666% So that means that I
> must have got between 9 and 10 questions completely wrong. (?)
Were they all proper questions with only one correct answer, or did it
include stupid wordy answers that you can only get all the marks if you
write exactly all the keywords and phrases they have on the answer sheet
(of course you get no marks for valid answers not on their sheet)?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Orchid Win7 v1 wrote:
> But hey, that's Unix all over. Don't expect there to be any reason or
> rhyme to it, just memorise this crap.
No need to memorise such things. There is always the manual. There are
more important things to learn about UNIX than that. And if an exam
testing your UNIX knowledge in general _requires_ you to know -e without
giving you access to the manpage of RPM, then this is a clear indication
of the quality of the exam itself.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Le 2013-03-12 15:23, Orchid Win7 v1 a écrit :
>
> The LPI website seems to indicate that /subjects/ are weighted, with
> more important subjects getting more questions. This seems to imply that
> each individual question is worth the same amount.
>
Back in the days where I was Cisco certified, not only were the sections
weighted, but the exam questions focused on the stuff you didn't know.
2-3 questions per main area... then delve into the ones you failed to
see if you were just unlucky with the first questions, or if you really
don't have a clue.
This is why you hear:
"it's all about subnetting"
"I had 20 questions on MPLS switching, you?"
"Me? every second question was about IPv6"
etc...
--
/*Francois Labreque*/#local a=x+y;#local b=x+a;#local c=a+b;#macro P(F//
/* flabreque */L)polygon{5,F,F+z,L+z,L,F pigment{rgb 9}}#end union
/* @ */{P(0,a)P(a,b)P(b,c)P(2*a,2*b)P(2*b,b+c)P(b+c,<2,3>)
/* gmail.com */}camera{orthographic location<6,1.25,-6>look_at a }
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On 13/03/2013 12:53 PM, Urs Holzer wrote:
> Orchid Win7 v1 wrote:
>> But hey, that's Unix all over. Don't expect there to be any reason or
>> rhyme to it, just memorise this crap.
>
> No need to memorise such things. There is always the manual. There are
> more important things to learn about UNIX than that. And if an exam
> testing your UNIX knowledge in general _requires_ you to know -e without
> giving you access to the manpage of RPM, then this is a clear indication
> of the quality of the exam itself.
Well, that's the thing. In the Real World, it's trivial to look up what
some specific obscure option does. It's arguably important to realise
that a particular command offers a feature, but remembering the actual
switch to turn it on is something you can easily look up if and when you
ever actually need it.
On the other hand, what else can they put in the exam? They have to ask
you questions about *something*...
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On 13/03/2013 10:37 AM, scott wrote:
> Were they all proper questions with only one correct answer, or did it
> include stupid wordy answers that you can only get all the marks if you
> write exactly all the keywords and phrases they have on the answer sheet
> (of course you get no marks for valid answers not on their sheet)?
The majority are multiple-choice, with 5 options. (And if one of the
options is "none of the above", this is almost ALWAYS the wrong answer!)
There were about 6 questions which require you to type something - the
name of a command or the absolute path to a file, or perhaps one command
switch. Stuff which is extremely easy for a machine to verify.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Wed, 13 Mar 2013 20:28:07 +0000, Orchid Win7 v1 wrote:
> Well, that's the thing. In the Real World, it's trivial to look up what
> some specific obscure option does.
Removing a package is, arguably, not an obscure option one would use
occasionally.
I had a similar debate with a physics prof once in college - the course
was for engineering students, but I was a CS student, so I'd have
reference materials available to ensure that I was coding the forumulas
properly in my simulation - and if I didn't remember the exact formula
for calculating lift based on a particular airfoil shape, I wouldn't
guess, I'd look it up.
"Not good enough".
And of course, I lost the debate, because it was the professor's class
and he got to decide what was important and what wasn't.
Years later, it occurred to me that he did have a point (though I still
maintain he was applying it incorrectly by assuming all students were
engineering students and grading everyone as such, but that's a separate
issue). There are some things that, while you *can* look them up, if
you're competent, you should never *need* to look up.
Remember that a certification exam is a measure of a minimally qualified
candidate to do a particular job or task. A minimally qualified
candidate on Linux *should* be able to install/remove packages without
having to look the command up every time they do it.
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Wed, 13 Mar 2013 15:43:50 -0400, Francois Labreque wrote:
> Back in the days where I was Cisco certified, not only were the sections
> weighted, but the exam questions focused on the stuff you didn't know.
Some companies still use such adaptive testing, but it requires a very
comprehensive beta on the exam.
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> Well, that's the thing. In the Real World, it's trivial to look up what
>> some specific obscure option does.
>
> Removing a package is, arguably, not an obscure option one would use
> occasionally.
Sure. But usually you wouldn't use rpm -e, you would use one of several
far more user-friendly tools such as Yum or Zypper or whatever.
> There are some things that, while you *can* look them up, if
> you're competent, you should never *need* to look up.
>
> Remember that a certification exam is a measure of a minimally qualified
> candidate to do a particular job or task. A minimally qualified
> candidate on Linux *should* be able to install/remove packages without
> having to look the command up every time they do it.
The problem being that the exam requires you to know how to use every
possible package manager of every possible distro, not just the one you
actually use every single day.
If you ask me how to use Zypper, I can answer that. I use it all the
time. But if you ask me how to work, oh, ANY OTHER SYSTEM, I'm probably
not going to know the answer. But if some day I need to use one, I can
easily look this stuff up, and if I end up using it a lot I'll quickly
stop needing to look this stuff up.
I understand that it's important to know how to use different package
managers. I just disagree it's necessary to know every single one of
them to the level of detail where you've memorised every single option
and flag available. You're only going to do that for the one particular
system you use regularly.
Then again, I guess I should just be thankful there weren't more
questions about Vi. Do you know, the other day I had 3 LPIC-1 certified
staff crowding around my monitor, and between the four of us we couldn't
work out how to copy and paste a chunk of text using Vi...
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Wed, 13 Mar 2013 22:19:25 +0000, Orchid Win7 v1 wrote:
>>> Well, that's the thing. In the Real World, it's trivial to look up
>>> what some specific obscure option does.
>>
>> Removing a package is, arguably, not an obscure option one would use
>> occasionally.
>
> Sure. But usually you wouldn't use rpm -e, you would use one of several
> far more user-friendly tools such as Yum or Zypper or whatever.
Except those tools aren't common across multiple distributions. 'rpm' is.
And if you're trying to manage it in a distro-agnostic way, then you'd
use the more common tool.
>> Remember that a certification exam is a measure of a minimally
>> qualified candidate to do a particular job or task. A minimally
>> qualified candidate on Linux *should* be able to install/remove
>> packages without having to look the command up every time they do it.
>
> The problem being that the exam requires you to know how to use every
> possible package manager of every possible distro, not just the one you
> actually use every single day.
It's a generic Linux exam - so yes, it's going to ask about things
relating to different distros.
If you want a certification that's for a specific distro, you get that
from the maker of the distro (a CLP from SUSE, a RHCE from RedHat, etc.)
> If you ask me how to use Zypper, I can answer that. I use it all the
> time. But if you ask me how to work, oh, ANY OTHER SYSTEM, I'm probably
> not going to know the answer. But if some day I need to use one, I can
> easily look this stuff up, and if I end up using it a lot I'll quickly
> stop needing to look this stuff up.
The point of the exam is to measure a minimum set of necessary skills,
though, and for LPI, that's a cross-distro certification, so it's - by
its very nature - going to look for things that are broader than "on
openSUSE, what command would you use to remove a package". That kind of
question doesn't meet the goal of being about general Linux knowledge.
"What general command do you use to remove an RPM?" does meet that
objective.
It's important to learn according to the learning objectives for an exam
when preparing to take the exam. There is, in fact, an entire ISO
standard devoted to creating learning objectives so an exam can be
created independently from the content - both derive from the same
learning objectives. (ISO/IEC 17024 in case anyone's interested.)
> I understand that it's important to know how to use different package
> managers. I just disagree it's necessary to know every single one of
> them to the level of detail where you've memorised every single option
> and flag available. You're only going to do that for the one particular
> system you use regularly.
It's not "every single option", though, it's one of the more common
options one would use.
> Then again, I guess I should just be thankful there weren't more
> questions about Vi. Do you know, the other day I had 3 LPIC-1 certified
> staff crowding around my monitor, and between the four of us we couldn't
> work out how to copy and paste a chunk of text using Vi...
Or indeed on how to exit it when you've finished. <scnr> ;)
BTW, well done on the exam score. :)
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On 13/03/2013 10:19 PM, Orchid Win7 v1 wrote:
> Then again, I guess I should just be thankful there weren't more
> questions about Vi. Do you know, the other day I had 3 LPIC-1 certified
> staff crowding around my monitor, and between the four of us we couldn't
> work out how to copy and paste a chunk of text using Vi...
It sounds like you are enjoying your job.
And doesn't everyone outside of academia know that the exam system
sucks? Now, the education system "teaches" to pass exams not to educate.
A vicious circle that dilutes any benefit that being able to read
without moving your lips, gives.
--
Regards
Stephen
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Wed, 13 Mar 2013 22:43:45 +0000, Stephen wrote:
> It sounds like you are enjoying your job.
That's certainly true.
> And doesn't everyone outside of academia know that the exam system
> sucks? Now, the education system "teaches" to pass exams not to educate.
Not always, but usually in academia it's been my experience that that's
true. In uni, most exams are not vetted as well as a certification exam,
and the professors/teachers aren't trained in psychometric analysis, item
analysis, or how even really how to write good questions that measure
learning effectively.
> A vicious circle that dilutes any benefit that being able to read
> without moving your lips, gives.
LOL
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On 13/03/2013 10:51 PM, Jim Henderson wrote:
> On Wed, 13 Mar 2013 22:43:45 +0000, Stephen wrote:
>> And doesn't everyone outside of academia know that the exam system
>> sucks? Now, the education system "teaches" to pass exams not to educate.
>
> Not always, but usually in academia it's been my experience that that's
> true. In uni, most exams are not vetted as well as a certification exam,
> and the professors/teachers aren't trained in psychometric analysis, item
> analysis, or how even really how to write good questions that measure
> learning effectively.
>
Arrrg!<Runs away wrapping his arms about his head.> :-P
>> A vicious circle that dilutes any benefit that being able to read
>> without moving your lips, gives.
>
> LOL
>
> Jim
>
??? I am being serious. Well half serious, anyway. ;-)
--
Regards
Stephen
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> I had a similar debate with a physics prof once in college - the course
> was for engineering students, but I was a CS student, so I'd have
> reference materials available to ensure that I was coding the forumulas
> properly in my simulation - and if I didn't remember the exact formula
> for calculating lift based on a particular airfoil shape, I wouldn't
> guess, I'd look it up.
Guess it varies from place to place, in my engineering exams we had
official data books that contained pretty much all the standard
formulas. The exams were more about how to solve particular problems,
which usually involved using several equations (and often a good bit of
calculus) rather than seeing who had spent the most time memorising the
formula.
But of course the shorter ones you used most often you remembered
anyway, but the point was you didn't have to waste hours trying to
remember them exactly.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On 14/03/2013 08:29 AM, scott wrote:
> But of course the shorter ones you used most often you remembered
> anyway, but the point was you didn't have to waste hours trying to
> remember them exactly.
Formulas can sometimes be derived from first principles if necessary.
(Provided you remember enough background around them.) Unix commands and
their switches follow no reason or rhyme. Much like the spelling of the
English language, the only solution is to memorise large chunks of data
and be able to recall it later. It's not so easy to do that for parts of
the syllabus that you don't actually need to use regularly...
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Jim Henderson wrote:
> On Wed, 13 Mar 2013 20:28:07 +0000, Orchid Win7 v1 wrote:
>
>> Well, that's the thing. In the Real World, it's trivial to look up
>> what some specific obscure option does.
>
> Removing a package is, arguably, not an obscure option one would use
> occasionally.
>
> I had a similar debate with a physics prof once in college - the
> course was for engineering students, but I was a CS student, so I'd
> have reference materials available to ensure that I was coding the
> forumulas properly in my simulation - and if I didn't remember the
> exact formula for calculating lift based on a particular airfoil
> shape, I wouldn't guess, I'd look it up.
>
> [...]
You make an unfair comparison. He wanted you to know and have understood
enough about physics to reconstruct the formula on the spot. But I bet
it wouldn't have been a requirement to remember the name of every
variable as it was used in the lecture. In fact, bound variables in a
formula can be named anyway you want. For example, integrating f(s) over
s in [0,1] or f(t) over t in [0,1] is the same.
My comparison:
- Knowing a formula is like knowing how to connect dmesg and grep with a
pipe in order to find out why the graphics card's module didn't load.
- Knowing the right (by convention) name of a variable is like knowing
that -e is for removing a package.
The first thing is essential to learn, while the second you learn
automatically over time.
> Remember that a certification exam is a measure of a minimally
> qualified
> candidate to do a particular job or task. A minimally qualified
> candidate on Linux *should* be able to install/remove packages without
> having to look the command up every time they do it.
Note that the certification exam doesn't require you to be able to
answer every question. So it only checks whether you know the names of a
lot of options and commands while still allowing you to fail at some of
them. This arguably tests how much experience you have with Linux,
because you memorise these names automatically if you use common Linux
commands often.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> Formulas can sometimes be derived from first principles if necessary.
> (Provided you remember enough background around them.) Unix commands and
> their switches follow no reason or rhyme. Much like the spelling of the
> English language, the only solution is to memorise large chunks of data
> and be able to recall it later. It's not so easy to do that for parts of
> the syllabus that you don't actually need to use regularly...
The exam would be a more realistic and useful test then if it simply
gave you a real linux prompt and timed how long it took you to get the
correct answers to do what it wants. The pass rate would then be on how
long it took you to get all the answers. Knowing where and how to find
stuff efficiently is far more important than actually knowing it
(because you could never hope to know even a tiny fraction of all the
information you have access to). As stated already the stuff you use
frequently you will remember anyway.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Thu, 14 Mar 2013 13:29:17 +0100, Urs Holzer wrote:
> Jim Henderson wrote:
>
>> On Wed, 13 Mar 2013 20:28:07 +0000, Orchid Win7 v1 wrote:
>>
>>> Well, that's the thing. In the Real World, it's trivial to look up
>>> what some specific obscure option does.
>>
>> Removing a package is, arguably, not an obscure option one would use
>> occasionally.
>>
>> I had a similar debate with a physics prof once in college - the course
>> was for engineering students, but I was a CS student, so I'd have
>> reference materials available to ensure that I was coding the forumulas
>> properly in my simulation - and if I didn't remember the exact formula
>> for calculating lift based on a particular airfoil shape, I wouldn't
>> guess, I'd look it up.
>>
>> [...]
>
> You make an unfair comparison. He wanted you to know and have understood
> enough about physics to reconstruct the formula on the spot. But I bet
> it wouldn't have been a requirement to remember the name of every
> variable as it was used in the lecture. In fact, bound variables in a
> formula can be named anyway you want. For example, integrating f(s) over
> s in [0,1] or f(t) over t in [0,1] is the same.
I don't think it's an unfair comparison. It's about understanding part
of a body of knowledge. In the physics case, it's about how to perform a
calculation (one that, due to the nature of a CS degree with that
specialization, would be used once per program at most, because it'd be
coded and forgotten - and a more appropriate test item for a CS student
would be to code a function that calculates the value rather than recall
of the formula).
In the RPM case, it's about knowing how to use the rpm command to remove
an installed package.
> My comparison:
> - Knowing a formula is like knowing how to connect dmesg and grep with a
> pipe in order to find out why the graphics card's module didn't load.
> - Knowing the right (by convention) name of a variable is like knowing
> that -e is for removing a package.
>
> The first thing is essential to learn, while the second you learn
> automatically over time.
I disagree with the second point. You can calculate a formula using
different variables. You cannot remove a package with rpm using any
option other than -e.
>> Remember that a certification exam is a measure of a minimally
>> qualified candidate to do a particular job or task. A minimally
>> qualified candidate on Linux *should* be able to install/remove
>> packages without having to look the command up every time they do it.
>
> Note that the certification exam doesn't require you to be able to
> answer every question. So it only checks whether you know the names of a
> lot of options and commands while still allowing you to fail at some of
> them. This arguably tests how much experience you have with Linux,
> because you memorise these names automatically if you use common Linux
> commands often.
Of course it doesn't require a perfect score - a properly designed exam
doesn't, because it's a tool to measure knowledge, skills, or
understanding.
I spent about 6-7 years working with certification exam creation and
design, and have some background in testing theory (gained through
experience, not explicit schooling). I understand the basics of
psychometric analysis (but if you start asking me about lambda values,
you'll get a blank look from me).
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Thu, 14 Mar 2013 08:29:48 +0000, scott wrote:
>> I had a similar debate with a physics prof once in college - the course
>> was for engineering students, but I was a CS student, so I'd have
>> reference materials available to ensure that I was coding the forumulas
>> properly in my simulation - and if I didn't remember the exact formula
>> for calculating lift based on a particular airfoil shape, I wouldn't
>> guess, I'd look it up.
>
> Guess it varies from place to place, in my engineering exams we had
> official data books that contained pretty much all the standard
> formulas. The exams were more about how to solve particular problems,
> which usually involved using several equations (and often a good bit of
> calculus) rather than seeing who had spent the most time memorising the
> formula.
>
> But of course the shorter ones you used most often you remembered
> anyway, but the point was you didn't have to waste hours trying to
> remember them exactly.
What you describe is exactly how I envisioned working with it as I was
developing simulations. But the class was 90% engineering students and <
10% CS students (and a few others).
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Thu, 14 Mar 2013 08:59:14 +0000, Orchid Win7 v1 wrote:
> Unix commands and their switches follow no reason or rhyme
Sometimes it certainly appears that way. In this instance, I think it's
a case of "-r" and "-R" were already used as a secondary option for the
query parameter, and 'e' is the next letter in "remove". -U was already
used for "upgrade".
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Thu, 14 Mar 2013 13:54:11 +0000, scott wrote:
> The exam would be a more realistic and useful test then if it simply
> gave you a real linux prompt and timed how long it took you to get the
> correct answers to do what it wants.
Performance-based testing is a better way to test on this sort of thing,
but it requires a lot more infrastructure to deliver exams like that.
I spent time managing a program that was specifically about this kind of
practical testing. It is perceived (rightly so) as being a better
testing methodology, but none of the testing providers currently have the
infrastructure to support it. Forms-based exams are far easier to
administer and deploy.
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Wed, 13 Mar 2013 23:07:43 +0000, Stephen wrote:
> On 13/03/2013 10:51 PM, Jim Henderson wrote:
>> On Wed, 13 Mar 2013 22:43:45 +0000, Stephen wrote:
>
>>> And doesn't everyone outside of academia know that the exam system
>>> sucks? Now, the education system "teaches" to pass exams not to
>>> educate.
>>
>> Not always, but usually in academia it's been my experience that that's
>> true. In uni, most exams are not vetted as well as a certification
>> exam,
>> and the professors/teachers aren't trained in psychometric analysis,
>> item analysis, or how even really how to write good questions that
>> measure learning effectively.
>>
>>
> Arrrg!<Runs away wrapping his arms about his head.> :-P
I sometimes surprise myself with stuff like this. ;)
>>> A vicious circle that dilutes any benefit that being able to read
>>> without moving your lips, gives.
>>
>> LOL
>>
>> Jim
>>
>>
> ??? I am being serious. Well half serious, anyway. ;-)
Well, then maybe that should be "LC" ;)
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On 14/03/2013 01:54 PM, scott wrote:
> The exam would be a more realistic and useful test then if it simply
> gave you a real linux prompt and timed how long it took you to get the
> correct answers to do what it wants. The pass rate would then be on how
> long it took you to get all the answers.
Don't forget that it's not just about being able to /write/ commands to
do stuff; you probably need to be able to /read/ this stuff as well. ;-)
> Knowing where and how to find
> stuff efficiently is far more important than actually knowing it
> (because you could never hope to know even a tiny fraction of all the
> information you have access to). As stated already the stuff you use
> frequently you will remember anyway.
Indeed.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On 14/03/2013 04:07 PM, Jim Henderson wrote:
> On Thu, 14 Mar 2013 08:59:14 +0000, Orchid Win7 v1 wrote:
>
>> Unix commands and their switches follow no reason or rhyme
>
> Sometimes it certainly appears that way. In this instance, I think it's
> a case of "-r" and "-R" were already used as a secondary option for the
> query parameter, and 'e' is the next letter in "remove". -U was already
> used for "upgrade".
Like I said, half the text-processing commands let you specify how many
bytes to process using -b, and half call it -c. No reason, they just do.
But then again, we're not talking about a coherently designed system,
we're talking about a random collection of tools and utilities
independently developed by unrelated individuals over the course of 70+
years. Add in a few layers of backwards compatibility and YUCK! >_<
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Thu, 14 Mar 2013 19:59:05 +0000, Orchid Win7 v1 wrote:
> On 14/03/2013 04:07 PM, Jim Henderson wrote:
>> On Thu, 14 Mar 2013 08:59:14 +0000, Orchid Win7 v1 wrote:
>>
>>> Unix commands and their switches follow no reason or rhyme
>>
>> Sometimes it certainly appears that way. In this instance, I think
>> it's a case of "-r" and "-R" were already used as a secondary option
>> for the query parameter, and 'e' is the next letter in "remove". -U
>> was already used for "upgrade".
>
> Like I said, half the text-processing commands let you specify how many
> bytes to process using -b, and half call it -c. No reason, they just do.
"b" = Bytes
"c" - Characters
Developer's choice/prerogative. :)
> But then again, we're not talking about a coherently designed system,
> we're talking about a random collection of tools and utilities
> independently developed by unrelated individuals over the course of 70+
> years. Add in a few layers of backwards compatibility and YUCK! >_<
Try about 50 years. UNIX came about in 1969. Some of the development
work obviously would've been before it was released.
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> Like I said, half the text-processing commands let you specify how many
>> bytes to process using -b, and half call it -c. No reason, they just do.
>
> "b" = Bytes
> "c" - Characters
>
> Developer's choice/prerogative. :)
Nice choice of "-" and "=" there. :-P
But yes, that's why the names. Because, obviously 1 character = 1 byte.
Oh, wait...
And hey, even if that were true, it's *still* an inconsistent naming of
command switches. Some commands expect -c and some expect -b, and you
just have to memorise which is which. (A tiny few commands accept BOTH,
and actually do the correct thing with multi-byte characters... But not
very many!)
>> But then again, we're not talking about a coherently designed system,
>> we're talking about a random collection of tools and utilities
>> independently developed by unrelated individuals over the course of 70+
>> years. Add in a few layers of backwards compatibility and YUCK!>_<
>
> Try about 50 years. UNIX came about in 1969. Some of the development
> work obviously would've been before it was released.
That's a bit before my time. But the main point still stands - it's a
collection of tools independently developed by mostly unrelated
individuals over a period of several decades. There never was any
overarching design plan.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Thu, 14 Mar 2013 22:42:39 +0000, Orchid Win7 v1 wrote:
> Nice choice of "-" and "=" there. :-P
Typo. ;)
> But yes, that's why the names. Because, obviously 1 character = 1 byte.
> Oh, wait...
When you're dealing with ASCII (and EBCDIC, IIRC), that is most certainly
true.
> And hey, even if that were true, it's *still* an inconsistent naming of
> command switches. Some commands expect -c and some expect -b, and you
> just have to memorise which is which. (A tiny few commands accept BOTH,
> and actually do the correct thing with multi-byte characters... But not
> very many!)
Like I said, programmer's prerogative.
>>> But then again, we're not talking about a coherently designed system,
>>> we're talking about a random collection of tools and utilities
>>> independently developed by unrelated individuals over the course of
>>> 70+ years. Add in a few layers of backwards compatibility and YUCK!>_<
>>
>> Try about 50 years. UNIX came about in 1969. Some of the development
>> work obviously would've been before it was released.
>
> That's a bit before my time. But the main point still stands - it's a
> collection of tools independently developed by mostly unrelated
> individuals over a period of several decades. There never was any
> overarching design plan.
There was for the underlying architecture, but yes, developers who wrote
different tools didn't coordinate. You sound surprised by that for some
reason.
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Le 2013-03-14 09:54, scott a écrit :
>> Formulas can sometimes be derived from first principles if necessary.
>> (Provided you remember enough background around them.) Unix commands and
>> their switches follow no reason or rhyme. Much like the spelling of the
>> English language, the only solution is to memorise large chunks of data
>> and be able to recall it later. It's not so easy to do that for parts of
>> the syllabus that you don't actually need to use regularly...
>
> The exam would be a more realistic and useful test then if it simply
> gave you a real linux prompt and timed how long it took you to get the
> correct answers to do what it wants. The pass rate would then be on how
> long it took you to get all the answers. Knowing where and how to find
> stuff efficiently is far more important than actually knowing it
> (because you could never hope to know even a tiny fraction of all the
> information you have access to). As stated already the stuff you use
> frequently you will remember anyway.
>
Just like the CCIE lab test.
Day 1: Here's a pile of network gear*. Here's a Visio diagram of a
corporate network (with missing info, of course). All the cables and
all the manuals are over there on the shelf. You have 8 hours.
Day 2: Guys, there, was a major incident on your network last night, the
CIO has been screaming at your boss for the last half hour. Fix it.
You have 8 hours.
*Including some that's DOA, or whose firmware has been erased.
--
/*Francois Labreque*/#local a=x+y;#local b=x+a;#local c=a+b;#macro P(F//
/* flabreque */L)polygon{5,F,F+z,L+z,L,F pigment{rgb 9}}#end union
/* @ */{P(0,a)P(a,b)P(b,c)P(2*a,2*b)P(2*b,b+c)P(b+c,<2,3>)
/* gmail.com */}camera{orthographic location<6,1.25,-6>look_at a }
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Sat, 16 Mar 2013 16:01:28 -0400, Francois Labreque wrote:
> Just like the CCIE lab test.
> Day 1: Here's a pile of network gear*. Here's a Visio diagram of a
> corporate network (with missing info, of course). All the cables and
> all the manuals are over there on the shelf. You have 8 hours.
>
> Day 2: Guys, there, was a major incident on your network last night, the
> CIO has been screaming at your boss for the last half hour. Fix it. You
> have 8 hours.
>
> *Including some that's DOA, or whose firmware has been erased.
Which is a very good exam - one of the best in the IT industry, if not
the best.
But it's also not something you'd do on an entry-level exam. It's a
fairly expensive exam to take, to administer, and requires a fair amount
of setup to pull off. If you fail, Cisco will charge you to review the
results (you get the review fee back if the scoring was incorrect,
however).
IT certification exams fall into two categories, generally: Introductory
and Advanced.
Introductory exams are almost /never/ practical exams (RedHat is the only
exception to this that I'm aware of - and the guy who's in charge of
their exam and certification programs has no interest in doing anything
other than performance-based testing. I've met him and he and I talked
at length about it). The LPIC-1 is an introductory exam. The purpose of
an exam at this level is to define a minimally-qualified entry level
candidate - a first level admin, or similar. Those certifications tend
to be marketing fodder for the exam sponsors (Cisco, SUSE, RedHat, etc).
They have to measure a valid candidate in order to have some validity,
but the purpose is to build a large following of candidates who have
familiarity with the product(s) the exams are about.
The benefit shifts when you get to more advanced exams. Those exams are
targeted at a smaller audience - and if (IMHO) the sponsor is doing their
breakdown properly, the advanced exams are /not/ revenue-generating. At
best, they should break even. (If you're making an "advanced" mass-
market certification exam, you don't understand the economics of
certification programs, again IMHO). The purpose of an advanced
certification program is to have a small number of people (relatively
speaking) who have advanced skills and knowledge - the "creme de la
creme" of the mass-market certification audience are the ones who will
take it, and they should be difficult to pass (like the CCIE).
The benefit to the candidate of a properly defined advanced certification
is significant - demonstrably higher pay, higher regard/recognition, and
higher demand for their skills.
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Le 2013-03-17 00:03, Jim Henderson a écrit :
> Introductory exams are almost /never/ practical exams (RedHat is the only
> exception to this that I'm aware of - and the guy who's in charge of
> their exam and certification programs has no interest in doing anything
> other than performance-based testing. I've met him and he and I talked
> at length about it). The LPIC-1 is an introductory exam. The purpose of
> an exam at this level is to define a minimally-qualified entry level
> candidate - a first level admin, or similar. Those certifications tend
> to be marketing fodder for the exam sponsors (Cisco, SUSE, RedHat, etc).
>
Unless I'm mistaken, the CCNA (Introductory Cisco Cert) has some amount
of simulated hands-on, now. When I took it, it was just multiple choice
answers, inlcuding sections where you had a list of 100 commands, and
you needed to pick the one that applied to each of the 20 questions in
that section, but that was before the days of Flash and Java, so
simulators were not that common.
> They have to measure a valid candidate in order to have some validity,
> but the purpose is to build a large following of candidates who have
> familiarity with the product(s) the exams are about.
Agreed. their as much a sales pitch as a technical certification. A kid
just out of school who got her MiCkey mouSE certification will probably
not recommend Solaris as her new employer's server platform.
>
> The benefit shifts when you get to more advanced exams. Those exams are
> targeted at a smaller audience - and if (IMHO) the sponsor is doing their
> breakdown properly, the advanced exams are /not/ revenue-generating.
They may not be revenue-generating, but they are also expense-cutting.
For example, if a company has a certain number of CCIEs on staff (I
think it's 4), they get a huge deal on maintenance contracts, because
Cisco assumes they won't be calling TAC twice a week for silly stuff.
--
/*Francois Labreque*/#local a=x+y;#local b=x+a;#local c=a+b;#macro P(F//
/* flabreque */L)polygon{5,F,F+z,L+z,L,F pigment{rgb 9}}#end union
/* @ */{P(0,a)P(a,b)P(b,c)P(2*a,2*b)P(2*b,b+c)P(b+c,<2,3>)
/* gmail.com */}camera{orthographic location<6,1.25,-6>look_at a }
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Mon, 18 Mar 2013 09:12:44 -0400, Francois Labreque wrote:
> Le 2013-03-17 00:03, Jim Henderson a écrit :
>
>> Introductory exams are almost /never/ practical exams (RedHat is the
>> only exception to this that I'm aware of - and the guy who's in charge
>> of their exam and certification programs has no interest in doing
>> anything other than performance-based testing. I've met him and he and
>> I talked at length about it). The LPIC-1 is an introductory exam. The
>> purpose of an exam at this level is to define a minimally-qualified
>> entry level candidate - a first level admin, or similar. Those
>> certifications tend to be marketing fodder for the exam sponsors
>> (Cisco, SUSE, RedHat, etc).
>>
>>
> Unless I'm mistaken, the CCNA (Introductory Cisco Cert) has some amount
> of simulated hands-on, now. When I took it, it was just multiple choice
> answers, inlcuding sections where you had a list of 100 commands, and
> you needed to pick the one that applied to each of the 20 questions in
> that section, but that was before the days of Flash and Java, so
> simulators were not that common.
Simulation was up-and-coming, I know there's a company in China that does
a lot of the simulation construction for Citrix, and the test drivers
used in testing centers are better able to handle that type of hands-on
because there are limited possibilities.
The way that's actually done is interesting, because the test developers
actively seek out unqualified candidates during development to determine
possible answers that someone who doesn't know the material might try.
>> They have to measure a valid candidate in order to have some validity,
>> but the purpose is to build a large following of candidates who have
>> familiarity with the product(s) the exams are about.
>
> Agreed. their as much a sales pitch as a technical certification. A kid
> just out of school who got her MiCkey mouSE certification will probably
> not recommend Solaris as her new employer's server platform.
I was thinking along the lines of being able to say "we've got x
certified individuals", but yes, it is something used to build customer
loyalty as well, for customers who invest in the courses and exams -
there's a training investment, so the customer wants to get a positive
return out of that investment. The interesting thing is that that's
something that can actually be measured (in broad terms) by the exam
sponsor, and some sponsors do make claims like "you'll recoup the
training expense in 3 months" with a fair degree of certainty.
It requires a fair amount of course eval data to be able to do so - and
not just an eval at the end of the course, but a follow-up eval 6 months
after the course.
>> The benefit shifts when you get to more advanced exams. Those exams
>> are targeted at a smaller audience - and if (IMHO) the sponsor is doing
>> their breakdown properly, the advanced exams are /not/
>> revenue-generating.
>
> They may not be revenue-generating, but they are also expense-cutting.
> For example, if a company has a certain number of CCIEs on staff (I
> think it's 4), they get a huge deal on maintenance contracts, because
> Cisco assumes they won't be calling TAC twice a week for silly stuff.
Oh, yes, absolutely. The first IT certification exam - the Certified
NetWare Engineer - was originally only going to be granted to a fixed
number of candidates who worked for partners - the purpose was to allow
the partners to be an extension of the product support organization, and
too many certified candidates would have diluted that pool to the point
that nobody made any money (actually, there's a similar problem in IT
training centers today - many organizations have too many training
partners to meet the demand, and the end result is that classes get
canceled not because there isn't demand, but because there isn't demand
to fill the course - say - the 6 times a week it's being held at
different training centers in an area. Well, that and the face of
training has changed to the point that most people won't sit a 5-day
class any more).
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
