[clue-talk] Assessing technical skills?

[Nate Duehr]

rex evans wrote:

> An experienced Lead IT guy will often 
> forget how many facts he really knows
> about a system, and how this allows him
> to trouble shoot a problem. A newbie's
> head is spinning because of so many
> unknowns, he does not even know where
> to probe the system.

If proper documentation is required by management and being written, 
this shouldn't be a problem.  Rarely do organizations do any real 
serious effort to really document what their admins do, however.

And admins have a natural tendency not to want to anyway -- economics. 
If they document it, anyone could do it.  Good ones document anyway, and 
learn new skills.

> Bottom line: do some of each 
> 1)Get better trouble shooting tools 
> 2)Get better trouble shooting training

Charge the Engineering group for the development and hours spent writing 
tools to find THEIR bugs.  (Another uncommon approach, but ultimately 
follows my personal rule, if you can't hold someone MONETARILY 
responsible, things don't get that much better, long-term.  The same 
bugs come up, the same problems, and no matter what methodology you're 
using, if you bonus design engineers on releases and not on bug-free 
software, you get lots and lots of software, but also lots and lots of 
bugs.)  To put it more bluntly, there needs to be a carrot AND a 
stick... and rarely do you see a stick (bad engineer, no donut!) in 
software development shops.  Of course, it's rare to find managers who 
came up through the ranks and have any REAL idea of what it's like to do 
the job anyway, so they coddle engineers to keep them "happy".  Good 
engineers are "happiest" when their stuff hits the real-world and 
doesn't break.  But there's rarely, if ever, monetary consequences for a 
multi-year, continuous pattern of "slightly broken" code.  The world is 
convinced that bug-free code is unattainable, which is sad.

> 3)Get better applicant screening but
> do not go overboard. 

The "What kind of cheese" one someone posted seemed pretty overboard to 
me.  When is IT/Software Engineering going to grow up and act like all 
the other Engineering disciplines with real knowledge and aptitude 
requirements that can easily be tested?

You wouldn't ask you Chief Engineer you're hiring to put up a skyscraper 
this silly question.  You'd want to see his dossier of serious 
engineering achievements.  You'd want to know how he worked his way up 
from a lowly civil engineer to the Chief job he has today.  You'd want 
real facts about his abilities.  Not fluff about what type of cheese 
he'd be.

Even the lowest PE certified Engineer on the totem pole you KNOW knows 
certain State mandated knowledge.  This sort of discipline is sorely 
lacking in IT/Computer/Software "engineering".

Maybe another good example would be:  You're dying of a deadly disease 
and only three Doctors in the world are able to work on you to fix it, 
and you have to choose which one.  You going to ask what kind of cheese 
they'd like to be?

> 4)Let them know what you expect:
>   day to day work + toos improvement
>   (tools improvement increases moral)

Just normal management skills.

> 5)Do not burn them out. 

Can you be more specific?  How do you know when they're headed there?

The next time I do an online transaction with a bank, I sure hope the 
coder hired for the job was asked better questions than what kind of 
cheese they'd be!

Seriously folks -- what is UP with Computer Engineering still being so 
un-professional?  Any thoughts as to what the major brain-damage is at 
the highest levels where levels of knowledge like other Engineering 
disciplines are tested on by the State, are not expected or required? 
Can you think of any transaction that involves real money that doesn't 
touch numerous computers nowadays, but we have "Building Codes" to build 
buildings by, but only "Best Practices" to build computer systems by?

Sorry - just up on my soapbox.  Nothing personal meant by any comments 
above, especially the Cheese stuff.  It's just a wonderment to me any 
computer system works right with the lack of standards, and generally 
tested knowledge bases that all "engineers" should know.

Nate