[nSLUG] Linux Certifications

George N. White III gnwiii at gmail.com
Wed Apr 14 19:12:02 ADT 2010

On Wed, Apr 14, 2010 at 5:16 PM, Daniel Morrison <draker at gmail.com> wrote:

> On 14 April 2010 16:09, Adam Hartling <adam.hartling.ns at gmail.com> wrote:
>>>I have, for example, heard a rumour that a carpenter applied for a senior
>>>Unix sysadmin position [...] the subsequent interview lasted
>>>somewhere under 5 minutes (including introductions) before he was walked
>>>out the door.
>> That's the sort of thing I'd hope a formal licensing system would help stop.
> To what end? To avoid wasting the interviewers time, or the
> humiliation of the candidate? This is an example where I have to ask:
> what is the problem you would like to solve?
> >From my perspective, it's a free country -- anyone can apply for a
> job, and it's up to the interviewer to reject the unqualified
> candidates (or pick the most promising one and train them
> appropriately). Over-regulation results in a situation where the
> potential employees are annoyed by ever-changing certification
> processes, and the employers are annoyed by impediments to hiring the
> candidate they wish to hire, rather than the one who's good at passing
> exams.

"Can you build us a safe nuclear power plant?", or "Can you design a fail-safe
fly-by-wire engine management system?".  9 times out of of 10 the guy who
says "no problem" gets the jobs while the guy who says "That will not be
easy." is shown the door.

Exams and certification can give the bosses a defense if their widgets
goes rogue -- "our software people are all certified".

As long as bosses are rewarded in current time now without suffering
penalties when problems become apparent later, the cheapest approach
will be chosen with too little regard for the long-term consequences.
Regulation is supposed to ensure some minimal standards, but in
many industries the "regulators" are tasked with promoting the
industry and are not allowed to impede the bosses' plans for
world domination.

Engineering certifications can help engineers stand up to bosses
because many of the practices have clearly understood rules.
Attempts to design similar standards in computing have generally
failed because there is little agreement over what "best" or even
"good" practice looks like.   We do have some ISO standards that
impose some minimal practices for documentation, version tracking,
issue tracking, etc. that at least make it harder to hide problems from
the courts.

You need oversight bodies that have people who really understand
the industry and can not only point out where "regulations" are
being broken, but how to go about building a better widget that
conforms to the spirit of the regulations, which is much different
from the boosterism of many "regulatory" bodies today and also
different from the "you can't do that" attitude of many regulators,
compliance officers, etc. when confronted with anything novel.
It may be true that most novel proposals are bunk, but we need
oversight that can tell the difference between hooey and hooray.

George N. White III <aa056 at chebucto.ns.ca>
Head of St. Margarets Bay, Nova Scotia

More information about the nSLUG mailing list