On Tue, Oct 07, 2008 at 03:47:32PM -0700, Bryan J. Smith wrote:
Then why are you here?
Hopefully to help make the test relevant. Possibly I am optimistic that
some day there will be a useful certification.
If you want to go that route, does public certification and
licensing mean anything? Or, I'll re-phrase, do the exams
mean anything either?
I have seen certified electricians do a shit job. I have seen
uncertified people do a lovely job.
In the state of Florida, I am a Certified Engineering Intern
(EI). That means I passed the Fundamental of Engineering
(FE) exam, formerly known as the Engineer-In-Training (EIT).
Do you think the FE exam is a good test of my education?
And what would you say if I told you I was tested on fluid
mechanics, thermodynamics and various other system interactions
and transfers? A "smartie-pants?" or "what the f-?" for an
electrical engineering graduate?
Some profesional associations have strange requirements, which sometimes
seem intent on keeping people out.
I've been lazy, so I haven't put through the paperwork to
take the Practices/Principles of Engineering, to become a
Professional Engineer (PE). Part of the reason why is that
I'm still hopeful they'll have a more Software Engineering
-centric examination. It's a long, drawn out, multi-decade
battle between the more traditional engineering disciplines
and newer fields -- not unlike how Civil Engineers prevented
Environmental Engineering from being recognized in the '70s
(but doing so immensely improved environmental actions in the
US, "behind the scenes" instead of ignorant politicians and
whistle blowers "after-the-fact").
I know quite a few people that do engineering related work but just
can't be bothered to do the paper work and exams requried to become a
profesional engineer. A few have even been profesional engineers and
dropped out and stopped paying the fees to the association because they
saw no benefit at all to being a member.
With that all said, do you think the PE exam is any different?
Having not written one, I can't be sure. Certainly passing the test is
only proof that you could pass the test, not that you are actually going
to be a competent engineer (I have met a couple with no clue of how
reality works, but who tested extremely well on engineering exams).
And that's before you consider the reality ...
Does having an electrical engineering degree help me
do my job at all?
Perhaps if your job is licensed electrician it could. If you are in IT,
then probably not.
Oh, sometimes people get all "oh, it's a computer degree, good"
because my Electrical Engineering option was "Computer." But
just because I had my 2nd semester in analysis (all that stupid,
higher math for automated, engineering applications), architecture
(you know, gate-level system design), semiconductor (you know,
intrinsic material, junctions, layout, etc...), etc..., does
that really apply to IT at all?
Because knowing nand, nor, not, or, and, 555s, flip-flops, etc is so
hugely required in IT work. :)
Again, do any exams?
I guess it depends. To some extent assignments are probably a better
indication of what you are capable of doing, but then you get into the
whole cheating potential and such, which exams have an easier time
preventing.
Does the FE/PE test the ability of certified/licensed
engineers to serve the public trust, to be held criminally
negligent for any actions that go against such? Does it
test them on how to write statues, to protect the public?
Does it teach them how to get their employers to understand
that they answer to the public more than their paycheck,
under the threat of criminal prosecution?
So, I ask again, why are you here?
Well I am not here to be a yes-man. I want to see LPIC have better more
realistic questions I suppose. I barely passed LPIC 101 the one time I
wrote it. That offended me highly.
For example there were questions about sendmail. I have used sendmail
and administrated it. What I know about sendmail as a result is that
you simply shouldn't ever use it. Use postfix or exim or something else
that works and doesn't involve the insanity of m4 to configure it and
hasn't had a long history of security disasters by design. Why would I
care how to do anything with sendmail based on that? I sure won't put
effort into remembering how to do stuff with sendmail. So there goes a
few marks. Hence I think there can be better questions. So to me a
good linux admin is one that doesn't know how to configure sendmail to
do something because they know better than to use sendmail.
You have a problem with the exams, yes. But what about the
larger aspect of what some certification programs are trying
to accomplish with exams, even if it's a very incomplete
aspect of the larger evaluation of a professional?
If I could have my way, I would rather recruiters weren't "lazy" and
didn't rely on certifications and diplomas to identify who was the
"best" candidate for a job. But unfortunately often they know little to
nothing about the task they are hiring for, and have to go by
"measureable" values that they believe they can understand. I am not
sure that problem can be solved. So perhaps making the tests better and
more fair is the only solution there is. So I try to provide my input
on that, and LPIC certainly seems the most open of the certifications
that way.
I tend to do quite well on tests that I know the material for, but I
also hate memorizing useless stuff, which some tests tend to test for.
How about computing grids at banks? Yes, Wall Street may
have tanked, but even some of my former colleagues at Lehman
Brothers have caught on at Barclay's. Why? Lehman's existing
data center of 15,000 RHEL servers was valued in the billions.
It's still quite useful for various banking.
Red Hat's MRG efforts, based on Morgan Stanley (among other)
developments, is another example. Does Red Hat's RHCDS really
prepare people for MRG deployment in huge, multi-thousand
server grid environments?
I don't know. Does it? Is it a training course or just a test?
But why use "ls -r" when "find" is better/safer for many things?
Why use "cp -dpR" when piped tar or dump is better/safer?
Because ls is simple and quick, and find's syntax is awful (very useful
tool though but it required reading the man page a few times to figure
out what it does, and even then a test run or two can be necesary for
more complex things).
And what about inode concepts? Superblock?
Ever teach these concepts, let alone test them completely?
I know what they are. I am trying to think if I would expect an admin
to need to know that or not. I suppose you could ask "What is an inode"
and let them write their own answer, but that becomes much harder to
mark than multiple choice and requires much more knowledge and objective
standards to do.
HINT: You can't, and that's 1, small aspect. ;)
You are probably right, at least not with multiple choice.
