[lug] my latest guilty pleasure -- Linuxhaters Blog

[Nate Duehr]

On Jul 29, 2008, at 8:34 PM, Collins Richey wrote:

> On Tue, Jul 29, 2008 at 6:02 PM, Nate Duehr <nate at natetech.com> wrote:
>> Sean Reifschneider wrote:
>>
>>> It's even better than that.  Back in January I hired us an  
>>> administrative
>>> assistant.  She had only used Windows before.  I set her up with a  
>>> Linux
>>> system and walked her through what she needs to do (e-mail,  
>>> openoffice,
>>> SVN
>>> for our documentation repository even, web browsing).  She had  
>>> absolutely
>>> no problems, saying something like "This all seems like what I'm  
>>> used to".
>>
>>> She's had no problems with it.  Except that "it doesn't crash all  
>>> the time
>>> like my Windows system did at my last job".
>>
>> That part we all know is kinda a straw-man... Windows will run a  
>> long time
>> without crashing nowadays with a professional admin/support staff  
>> doing the
>> right things.  Sounds like they didn't have one at her last job.
>
> Nate, you never give up.

Nah, I'm just having a recent personal epiphany.    I've worked in  
supporting computers and telecommunication systems now most of my  
adult life.  And the bugs and problems, aren't getting any better in  
either the close OR open-source worlds.

We have had the discussion here (or perhaps over on CLUE-Talk, I don't  
remember) where a year ago I was thinking very deeply and heavily  
about the concept of software "engineering", and realizing that when  
you compare the discipline and professionalism levels of say a civil  
engineer, or an electronics or RF systems engineer, and the *typical*  
software "engineer", they're just not even close.

Civil Engineers built to real standards, or people die.  Electronics  
or RF engineers build to standards that WORK, or they can't sell their  
products.  Both also have to consider the human interface aspects of  
their work.

But VERY FEW software projects follow any serious standards.  All it  
takes is subscribing to SANS security warning lists to see that  
software engineers make the same dumb mistakes (buffer overflow) over  
and over and over again, and little is demanded of them to not do that.

This recent foray into the guy's blog really opened my eyes to what  
the "rest of the world" might view Linux as... I find it interesting,  
fun, but when I *really* think about it -- I'm not a programmer, not a  
good one anyway, and I just USE computers to get other real work  
done.  When I have a choice today (because yes, I have more dollars  
than I did when I was a young man), I buy things that work and last.

Linux seems to meet neither goal, so I'm challenging the folks I've  
come to know and trust locally to think about it.  I enjoy  
conversations with you guys, but it's unreasonable of a group to say:

  "We're fans of this thing, so go away if you're interested in asking  
us hard questions about why it still doesn't work correctly (in many  
ways) after over ten years of development.  We're the people that said  
the open-source world could "beat" the closed-source world, and you  
DARE to question that?"  It just doesn't work for me anymore.

> Of course, Linux could never be an acceptable
> alternative in your opinion.

I'm leaning that way, yes.  Tell me why it should?  I get the ROI  
thing -- I've built company's billing, ticketing/tracking, monitoring,  
and various other systems that meant either they worked or the company  
didn't make money and keep people employed with Linux.  So I'm an EASY  
to convince target.  But when it comes to the desktop, Linux just  
doesn't offer enough value yet (in my opinion) for companies to switch  
to it for internal machines.

I get it that they could lock down the machines, get things going on  
them that would be business-only, and really set up a nice environment  
for the workplace that worked well.  But I don't see it happening with  
any less people or resources in the "IT Department".  In fact, I see  
it as a shame that companies even NEED such a thing as an "IT  
Department" for desktops.  Desktops are supposed to be the "dumb- 
terminal" of the 80's.  They're supposed to run local applications  
that work, and the hard problems can be solved on a back-room machine  
that does have a staff to take care of it.  That's what I think  
anyway... it seemed to work pretty darn well in the mainframe days...  
back when I was a very young pup entering the workforce.

That business back then definitely NEEDED computers, for doing their  
core business... and they still do today, but the desktops are a deep  
pit of money spending that doesn't seem to return as much money as it  
spends.  The desktop support side of IT Departments is a huge money  
sink for most companies, isn't it?

E-mail... the "killer app" and the place most of us spent a large part  
of our day inside of has done SOME good things for companies, but it's  
created some serious social and mental problems for people too.   
Almost every "time management" guru these days talks about "getting  
control of your inbox" or "making sure you don't leave your inbox  
open" so you can get some real work done.  If that's the case, isn't e- 
mail hurting more than it's helping?  And for those of us who have to  
respond instantly to customer requests... yes, we leave our e-mail  
open 100% of the time we're at the machine and probably even use those  
pop-up things to tell us new mail arrived... what a way to work?!  The  
phone still works, and I'm getting to the point where if a customer  
calls on the phone, I'll always work that first over an e-mail, unless  
the e-mail says "system down".  The human interaction factor of  
communications is fading away.  (Says the guy posting to an e-mail  
list.  Ha.  Well I don't have all your phone numbers, but I DO have a  
conference bridge I could call you all from!  GRIN...)

Videoconferencing (yeah, I'm biased... we make stuff to do that) is  
putting the human touch back into communications, but the e-mail isn't  
slowing, the regular phone still rings, and ... well, the  
communications "overload" can be a bitch in customer support roles  
these days... is all I'll say about that.

> I would rephrase that statement. Windows
> will run longer without crashing that it did in the past with or
> without professional admins. It will not run as long as Linux without
> crashing.

Take away "crashing" and do proper security and other patches and  
compare.  How much time is my Windows box down (unusable to me) and  
how much time is my Linux box down (same metric).  If you count all  
the security patches that require reboots... kernel, etc... they  
really are "about the same" these days.  Linux developers make the  
same mistakes Windows developers do.  Apple is slow to patch, so the  
Mac might be up just a teensy-bit more than the other two when I need  
to get something done, but it's catching up rapidly to the never- 
ending stream of MISTAKES ... erhm... I mean, patches... coming out of  
ALL OS's these days.  Don't you think?

I run them all, and that's my opinion right now.  Everything needs  
MISTAKES fixed, oops.. patches... on no longer than a weekly basis,  
and often daily these days.  Isn't that somehow a sign that the whole  
OS industry has gone hideously and deeply wrong?  (Trying to decide  
still myself... but I know this... the mainframe wasn't completely  
rebooted on any kind of regular schedule, and that was many magnitudes  
of technology ago. There were a couple of hours in the middle of the  
night where batch jobs, jobs SO IMPORTANT that the company would be  
out of business if they didn't complete, were run... billing and  
such... and I'd say perhaps once every 6 months there'd be a real  
outage that affected a large user group.  The Net effect of all this  
continual security patching on desktop OS's means that at any given  
time, some quite large percentage of users is down... rebooting.   
Isn't that kinda sad in 2008?)


> We have professional admins at work, and it still crashes.

We do too, and I have had one lockup and one crash this year.  They do  
buy good well-integrated with the OS hardware (IBM -- and we all know  
somewhere in the back of our minds that IBM has a better "technical  
relationship" with MS than many PC peddlers) so that probably helps.   
But I'm running the same XP Pro image that came with the machine when  
they purchased it four years ago, with no serious problems.  I do  
spend a tiny amount of "personal time" doing manual defragmentation (I  
can't believe we still have to do that in 2008 to a filesystem,  
either... but at least it's a known quantity).

I haven't had Linux lockup or crash at all this year.  So score one  
for Linux, I guess.  But I *have* had lots of time wasted by trying to  
get it loaded.  On the flip side of that, I've loaded equal numbers of  
Linux and Windows VIRTUAL machines in various virtualization flavors  
without problem, and I've loaded Windows on home or other machines at  
least four-five times, without incident.  No problems at all.  (I  
could also count up a couple of Mac OSX loads, but they always work --  
they're not even worth mentioning.)


> In particular, the hotfixes that M$ proscribes to fight the
> continually morphing viral infections are more likely to crash the
> system than not.


Windows has virii, Linux has buffer overflows.  I've tracked the SANS  
and other security lists for quite a number of years, and I wouldn't  
say that either one is winning the "secure software" race.  Part of  
this is that I'm trying to be wholistic about it, and Linux fans want  
to only count the core OS... but Firefox bugs are the same on both  
platforms... and most of the things that can attack Linux are from the  
same vectors as Windows machines.   A Windows machine with all  
security patches loaded and a Linux machine with all security patches  
loaded are pretty much in a dead tie.  And if the user is doing all of  
the "usual" things with application software, both will have about as  
many weekly updates from their security teams.


> Based on recent experience at work, I would say that
> maybe 1 out of 100 Linux updates are flawed (RHEL), whereas Windows
> fixes get about 1 out of 10 badly wrong, and these are usually hard
> failures, not just a single broken package.

I usually don't see updates breaking things on either platform...  
usually.   Just my experience.  In fact, to be honest I've had more  
failed updates on Mac OSX than either Linux or Windows.

>> A mac can also fill that last "requirement" too...
>>
>
> Of course, it's based on BSD, an extremley reliable OS.


There's not much BSD left... most of it was ripped out and thrown  
away... especially X.  :-)  You can still do X on a Mac, but it's not  
the majority that do... or even need it.


>> Also curious:  Could she have loaded that machine herself with the  
>> distro
>> you're using and ended up with the same desktop you set her up  
>> with... with
>> a single "go do it" mouse click?
>>
>> That's the goal... if we want people to use Linux.
>>
>
> Not really. The goal is to have Linux preloaded by professionals in
> the same manner that Windows is preloaded.

Okay, I just don't think we're going to see much of that.  Maybe I'm  
wrong, but I don't see it happening at my workplace, anytime soon...


> The average Luser is no
> more capable of installing Windows than Linux. Installing Windows is
> certainly not a one-click operation, and if the installation goes
> boom, the user is less likely to find reasonably quick help (did I
> mention mostly free?)


Now are you really serious about that?  Look at the real numbers.   
Millions of people (lusers is such a derogatory term) have loaded  
Windows successfully.  And I hear them on ham radio, in conversations  
at work, and in other public places helping each other for free too.   
In fact, if Linux has a 2% desktop share (which I think is generous),  
then your chances of finding help with a botched Windows installation  
are that over 90% of computer users are using it, and 2 out of every  
100 computer users is running Linux.

I think I like the odds from a purely numbers standpoint from the  
Windows side.  I love your philanthropy and I love that our  
"community" helps everyone, but seriously -- the "community" wouldn't  
survive if it didn't.  The Windows community certainly will.   Mac?   
Maybe, maybe not... they're at about 2% also, but you can call a  
single 800 number for professional support for their product and get  
it.  So it's not fair to use that platform in the comparison.

Can we call a single 800#, plunk down some cash, and get help with ALL  
of the Linux distros?  I'm not sure... but if it's out there -- I  
haven't seen it.

Getting professional help from Microsoft is an iffy thing -- I'll give  
you that.  Home users have nothing else BUT their friends, neighbors,  
and relatives... so to say Windows doesn't have a "community" support  
feature, is just flat out wrong.  Fair?


> repairing the broken system than is the case
> with Linux.


Repairing a broken system is the same for most users on either  
platform if they can't find a guru.  Grab the installation media and  
start over.  Hope you have backups.   I see no clear winner in  
either.   True for Mac users too.


> If a user can run a windows update, he/she can certainly
> deal with one of the gui package managers under Linux. In any business
> environment, professional admins are available for Linux just as for
> windows. Besides, not one out of 50 of my users needs to deal with
> installing software; we do it.


Yep, understand.   Will the industry we've built today ever ask the  
developers to put them out of their jobs?  I'm in support... I know I  
won't... but if they could release probably even a 20% less buggy  
product -- ANY one of our products -- the company would re-task the  
support people and there just wouldn't need to be many for that  
product anymore.  I don't see support departments getting that much  
smaller in most large companies of people, though.  They grow at about  
the same rate as the number of employees.

I should share that I do have a friend who is the luckiest guy in the  
world... he's the only "techie" for a company of 500 people, all  
running Macs with a Mac server on the back-end.  He has enough time  
left over that he's also the company telecom guy, the company CIO  
handling "strategic" questions, and even helped his company (a  
marketing/printing/art type place) put together a project that wasn't  
even a normal "IT" function -- he consulted with them to build some of  
the country's first billboards with built in AM broadcast  
transmitters.  He has a LOT of free time.  I know for sure there's an  
ROI lesson to be learned in there somewhere, but I can't figure out  
why more companies don't see it.


> The three (oops four)  flaws I find with Linux that aren't likely to
> go away any time soon are:
>
> 1. The continual selfish need of developers to reinvent the wheel (aka
> API) over and over. This applies to the kernel and to most
> applications. What used to work won't work on the new release without
> a lot of tinkering. This is really the M$ approach.


This seems to me to be because the developers are less focused on the  
users and more on their tools.  New languages, new ideas on how to  
build software, but not better software.  Would you agree?  The  
unbelievable amount of raw time and energy the whole "computer"  
industry as a whole puts into creating, maintaining, and upgrading  
tools to build things, is impressive, but also economically strange.   
I think Python's cool.  I think Perl's cool.  I think Java is cool.  I  
even (in a sick and twisted way) think C is cool.  (I don't think C++  
is cool, but I had a job cleaning up someone's serious C++ mess once  
-- and they were many-year "professionals" and I was a hack, but even  
I could read their code and tell it was complete crap.  They all quit  
and got "better" developer jobs, I hung around and fixed the bugs they  
wrote.  Later a more seasoned developer took their 40,000 line CGI app  
-- yes, CGI in C++ -- and pared it down to 10,000 lines and added 14  
function calls to replace all the duplicated code.  That's about the  
time I got back out of programming for a living, when I realized if  
that level of unprofessionalism was REWARDED, I'd stay on the  
"customer service" side of things, where taking care of customers is  
the reward.  Yuck.)


> 2. The multiplicity of desktop environments and eye candy. If the
> lamebrains at GNU had put the same amount of effort into improving
> KDE/QT instead of reinventing the wheel, we would have one solid
> desktop environment by now. Yes I know that QT had a non-free origin,
> but the code was released rapidly, so that's not an acceptable excuse.


Yeah... agreed.  None of them works as well as either a Windows or a  
Mac desktop for usability and consistency for users.  I like the eye  
candy (who doesn't), but at the end of the day, if I had to walk  
people on the phone through finding something in a graphical  
environment, I'd cringe if they were to call up and say they were  
"running Linux"... I can imagine the number of graphical steps I'd  
have to remember to get them to the same place in multiple different  
desktop environments would drive me batty.  THANK GOD the Linux  
desktop hasn't taken off with any of my customers, when I really think  
about it!  It's bad enough to walk them around the Windows environment.


> 3.  The lack of support by manufacturers for peripherals under Linux.
> Not something that Linux can fix.


Not true.  Linux can get off its high horse and accept binary closed- 
source drivers from all manufacturers.  Some distros have "gone there"  
to great weeping and gnashing of teeth, but the reality is... I plug  
peripherals into any other computer and expect them to work.  I plug  
them into Linux only when I know I've set aside an hour to Google and  
screw around IN CASE they don't.  That's just not right...

> 4. As you have said. Too many applications but not enough killer
> applications in Linux land.


Ahh, but here's where I'm optimistic, ironically... it's not that  
there aren't killer apps... it's that Linux still doesn't do them  
well.  There is ZERO doubt in my mind (hang on here, someone's going  
to flame me for this one...) that Outlook was a killer app for  
business the day it was released.  The fact that they haven't changed  
anything serious about how it behaves or what it ultimately does, for  
many many years, shows that too.  There's STILL not a decent  
competitor to it that handles e-mail, calendering and all the things  
it does... in one app.  Lotus Notes is "out there" and I worked at a  
place that used it... it had the potential, but Lotus has always had a  
whole pile of problems with marketing and support... Lotus 1-2-3 and  
AmiWord (is that what it was called) blew the versions of MS Office  
available at the time of their release (Excel and really old Word) out  
of the water for usability, price, and functionality... but Lotus  
dropped the ball and couldn't bust into the business world with them.

(We all know MS has a monopoly on many things, or very close to it,  
those arguments are a bit tired... I want to look at why things aren't  
getting better, in Linux, and the Lotus thing is just an example of  
how there's still not a decent replacement for Outlook -- why I  
pointed out that their products were better in the Office app  
department, I don't know... just trying to show how Lotus can screw  
anything good up, somehow.  They always manage to cluch defeat from  
the jaws of victory.)


> Nevertheless, If I were running a small business, I would consider
> Linux or MAC for most of the functions. Who needs to deal with malware
> every day?

I would too, but many small businesses wouldn't.  Because they don't  
know about it.  Marketing problem.  But big businesses SURELY know  
about Linux, and have staff that are smart enough to do what you're  
doing with it... using it in business, and it doesn't go that way.   
I'm thinking it's because more individuals need to see Linux.  But  
how, when it often requires things like "InstallFests" to get it  
running?  No one really has time for that.  Apple and MS get this and  
work really hard on their install experience.  Linux installers still  
look like something out of 1999, almost across the board.  Ubuntu  
kinda looks nice (if you like brown), but still isn't as polished as a  
Mac or Windows installer.  Devs are out there who do care, but the  
number of devs who really know how to make a good installer are few,  
and they'd never take the time to add little "help windows" that talk  
about Linux's other features while it's installing AUTOMATICALLY,  
because they have to ask the user too many bloody questions!   We  
could talk about installer design for a while, but I don't think it's  
much of a priority for Linux users and devs, really...


> There are some exposures in Linux, but the frequency is
> infinitessamally  small for the average business that does not have
> externally exposed websites.


Two thoughts here:  All real businesses have externally exposed  
websites nowadays.  Even "Bob's One Man Plumbing" has a website.

Second thought is:  Back up there we talked a bit about this... Linux  
browsers have just as many evil bugs buried in them as MS's does  
TODAY.  Perhaps in the past, MSIE was a horrible choice for a browser,  
but there were always anti-virus, anti-malware, anti-whatever tools  
for Windows users to avail themselves of.  Not built into the OS, but  
they were there for the using and didn't require a guru to set them  
up.  Pay your $49.99 and load 'em, and most of the time, they work.   
They're set up with the same "never-ending" treadmill of Internet- 
delivered updates these days as the OS's are.  MISTAKES by  
programmers, every one of those updates... in something.

I guess I want to share that I'm NOT a negative person -- my  
"dream" (why I get up to go to work in the morning) is to help people  
use their computers first, and help them to use their computers more  
effectively second.  I chase that for a living and in my personal  
life.  But when asked by people interested, "Is Linux better?"  I ask  
them if they want the fast or the slow answer.  Those that want the  
fast answer, it's "No.  Buy a mac or Windows.  Make sure you have anti- 
virus software on it, and do all the updates it asks you to do, and  
99/100 times, you'll be just fine."  If they say they want the long  
version, we talk about the differences in Windows, Linux, and Mac and  
the plusses and minuses of each.... and then the question comes out  
that always does... "Would you use X as your ONLY computer?"  No... I  
don't have an ONLY computer.  Yeah, but if you did?  Right now... If  
I'm true to my "I hate broken stuff" motto... I buy the Mac.  If I  
have no money, I load Linux.  (And if I'm putting an application on a  
server that costs tens of thousands of dollars a minute if it's down  
for any reason, I put it on a high-end Sun box.  Seriously.  Dells are  
nice playthings, but if you want it up and running 24/7 for years...  
it goes on a Sun.   And not the little 1RU Suns, they're just PC's in  
Sun logo'd box.  A real Sun server, complete with 20 years of  
engineering in things like backplanes and hardware internal  
monitoring, and all of that.)

I'm not (by any stretch of the imagination) a Windows or Mac  
"apologist".  I use them all.  I'm just "picking" on Linux right now.   
I have nasty things I could say about Windows and Mac too, but I trust  
the PEOPLE here (and I value you all as people more than I care about  
ANY darn OS...) and your opinions, and I think it's good to "shake the  
tree" once in a while to see where everyone stands.  I don't do it to  
insult or bother anyone.  (Everyone's here on the list of their own  
free will, I hope!  And I'm talking about Linux... I'm just not  
ignoring the "bad bits" about it right now...)

Back in 1980 when I got my first computer, it didn't have a word  
processor or any storage system of any value.  It booted in an instant  
(faster than I could see until I got better at it, and decided it was  
still under 1 second and I didn't care).  They added storage (tape  
drive) and useful programs.  I had a word processor and a really bad  
printer, and it all kinda just worked.  I never once hit a bug that  
crashed the machine.  Software was simple, but I could write things,  
do math calculations, and generally do "computing" just fine.   
Advanced things required a 10 minute wait to load someone smarter than  
me's program to do it.  But those didn't crash either.

That machine got peripheral upgrades, a modem, a good printer that  
worked, and hard disk storage before it was retired.  It lasted me  
until 1990.  I was certainly using IBM PCs and other things also by  
then, but that first machine ALWAYS worked.  Slowly, but no bugs of  
any serious consequence.

Same thing with the mainframe "business" computers at my first "real"  
job.  They ran, and did what they were supposed to and had laser  
printers attached (yeah, they had a lot of money and valued a good  
looking printout... what can I say?), and I could write, and do math  
calculations, and send messages (early e-mail) and the only thing that  
broke in that environment were the dumb terminals... they'd give up  
the ghost after YEARS of operation with a puff of smoke from a flyback  
transformer, and they'd be replaced in an hour or two by some guy paid  
to keep things running.  Every once in a while a print queue would get  
hung, is about all I remember for "trouble" with that system as a  
whole.  A quick call to the mainframe underworld in Texas and stuff  
would fly out of the hideously expensive laser printers.

But then something changed... "buggy" software started to get  
apologized for, stuff didn't work properly after a while, and the  
computing world -- even the professionals, started to accept this.   
Phrases like, "Oh you tried to do something you shouldn't  
have!"  (Then why did the programmer give me the option?)  And "You  
must have done something wrong!"  (No one ever tells me this about the  
other machines in my life... unless I go ram a light pole and put  
myself at serious bodily harm with my car.)  started creeping in...  
and are still the FIRST thing most support technicians say to this day  
about anything a user has a bad experience with on a computer.

Then Linux came along, and had the promise of maybe being a way BACK  
to that 80's and early 90's computing "experience"... it even promised  
it!  (Open source means lots of eyes, and all bugs get fixed!)  Yay!   
But it hasn't worked out that way, at all...

My words are boring to read... I know.  I'm not out to shock and  
entertain like LinuxHater.  But I think we've lost something in  
computing and I'm honestly trying to go back there... overall system  
engineering discipline and enough care about the end-user experience  
that if they don't like it, they won't "buy" this thing called a  
computer.  We all know everyone's going to continue to really "buy"  
today, but Linux has this weird niche where you don't HAVE to buy...  
and people still don't use it.  That troubles me, somehow.  If it's  
FREE (as in beer) and also FREE (as in Freedom), why aren't people  
flocking to it in droves?   I mean SERIOUS numbers here... should it  
not have at least 20% desktop "share" by now?  Really?

And I think the reason why is that it just really isn't BETTER.  It  
kinda has to be to do that.  I think.

So ... yeah, I'm "hard to please" right now, but if you know me in  
person -- I'm NOT out to bother or mess with everyone when I'm asking  
these questions.  Deep down, I'm just a computer user who misses  
machines that WORK.  Would such machines put me out of a job?  Yeah  
probably.  Would I adapt and do something else, like help people with  
HARDER problems than "Help, it just doesn't work!", probably.  So I  
think there's a "better place" this whole industry could be in... if  
we could just get our act together.

But like you, I think we spend FAR too much time focusing on the wrong  
aspects of computing when we create software or integrate something  
for people.  The focus is on break/fix more-so today than ever.  We  
rate companies and free projects on how fast they PATCH, not the  
quality of their work.  We don't see computer programmers as SERVANTS  
anymore (sorry programmers, but that's the reality of engineering --  
build what other people want because you're smart enough to... ask  
anyone who's engineered a bridge or a building), we view them as  
demigods who we beg for patches and hope for "the next version" from.

It just all seems really twisted right now.  I don't think a whole lot  
of people agree with that sentiment, but they probably also have never  
had ... never had! ... a computer that just worked right every time  
they powered it on.

So if I'm too "challenging" about this, I apologize.  How can I make  
it right to everyone?  If y'all can't take it and want me to stop, I  
will... I'm just in "question the big stuff" mode right now.

For those that made it this far... want to see something ultra- 
impressive?  Here's some people who came together and made something  
ULTRA-hard to do look VERY easy.  It's inspiring on many levels... and  
wasn't done with computers...

http://www.sonnyradio.com/swanlake_0001.swf

After watching that... think... can sitting down at my computer be  
that awe-inspiring?  Can someone develop ways to use these machines  
that are so fluid, so easy-looking, that my jaw drops open using  
them?  I think they can... but I'm not a programmer and therefore have  
very little say in the current environment about how we get there.  If  
programmers want usability testing (most don't)... I can help there!

--
Nate Duehr
nate at natetech.com