[NTLUG:Discuss] Re: Linux article -- corporate desktop adoption will slowly change the consumer

[Bryan J. Smith]

Neil Aggarwal wrote:  
> This article is completely focused on Linux as a desktop solution.  In
> that arena, I have to admit that Linux is not ready.  As much as I
> would love to run Linux on my desk, it is impossible or far too time
> consuming to get the software that I need to run my business.

People use applications, not OSes.  If your business has chosen a
Hostageware format you cannot get out of, no Freedomware solution will
_ever_ let you "get out of it."  There will always be a conversion cost
out of a "vendor lock-in solution" like common Hostageware apps.

Not my words, but Gartner's.  "Vendors will never offer a way out of
lock-in."

Note there is a _difference_ between what I call Hostageware,
Commerceware, Standardware and Freedomware.  E.g., examples of each are
in sequence are MS Office, Corel PerfectOffice, Sun StarOffice and
OpenOffice.org (OOo).

The difference between MS Office and Corel PerfectOffice that Corel has
maintained backward near-verbatim compatibility with its old document
formats, and offers full OASIS Open Office XML import/export (aka
StarOffice 6/7).  MS Office only retains 1 version back compatibility to
force upgrades (long story), and does _not_ offer any XML import/export
other than useless content-only XML (designed only for 3rd party
integration, not actual documentation of the underlying MS Office XML
implementation, which is _not_ published, so it useless for anything but
content, 0% formatting).

For those of us that started using StarOffice in the mid-90s on Windows,
we have no issue with Linux as a desktop.  It's not the OS, but the
applications.  Just ask the users of MS Office for Mac, and all the
"compatibility issues" they have.  MS Office is designed for x86, and
there are issues with it being "land locked" like there is for Americans
learning a second language.  Windows applications are "land locked."

Furthermore, Windows applications will _always_ run best on Windows. 
Attempts at emulation are poor, unless you completely emulate a hardware
virtual machine, which is almost self-defeating (we'll have to see how
far CoLinux goes, which may change this).


Some of my notes on the article ...

- Windows is _not_ ubiquitous (_never_ has been since '92!)

  "Windows. Microsoft's near-ubiquitous operating system"

Windows is _not_ ubiquitous.  Never has been, never will be.

Before XP, there was a _serious_issue_ between DOS and NT
implementations.  They were _heavily_incompatible_ before NT 5.0 (2000).
And most of the NT 4.0 "Cario" technologies never surfaced, because
Gates kept trumping the NT team, even in NT 5.1 (XP).

This is now re-hashed and continues in the Win32 v. .NET API fiasco. 
.NET is now out for the basis of NT 6.0 "Longhorn" and supposedly pushed
back until NT 6.1/7.0 "Blackcomb."  Win32 is still not being used fully
by Microsoft's own application division, which is very x86-focused and
data-alignment ignorant (not my words, but select people on the MS
Office for Mac team ;-).

Your documents today in MS Office will _continue_ to have _serious_
longevity issues in the future.  It's because of the binary format of MS
Office documents.  It is a long-published issue, documented by former MS
Office for Mac developers.  That's why I don't use MS Office -- not
because of some ethical/political delimma.  But because I _care_ about
my company's IP in the future -- I do _not_ want to see them put it in a
Hostageware format that is data-alignment ignorant like MS Office.

People say Linux is not "ubiquitous" but it is no less ubiquitious than
Microsoft.  In the UNIX world, there are typically 2-3 "standards" to
something.  Not out of "mandate," but out of competition.  Just like our
Democratic political system, there are countless ideas, but most people
just conglomerate around 2-3 "common bases" out of feasibility.  That's
what choice is all about.

Microsoft is a _single_ company, yet they have 2+ different
implementations on _everything_ with _no_ choice!  They set a direction
like .NET, and then Gates and other top decision makers don't see the
transition through.  Same deal with NT before it.  The result is a
clusterfsck of competition inside of the company, and "compatibility"
(which it is not) is chosen over "reality" (which is to just make the
move and be done with it).

VB6 is _still_ being supported by Microsoft.  Nuts!  No, it's not,
because they sold 80% of companies on it, even though it's a bastard. 
(VB.NET is much better, even Delphi developers agree if that says
something).  "Longhorn" is now a joke, just like "Cairo" was before it. 
Microsoft set out to do something, then shackled its own developers as
usual!


- Linux _is_ hardware friendly, but not "superstore hardware" friendly

  "Microsoft enjoys a considerable advantage over Linux's community of
  developers in the area of hardware compatibility and support ...
   And if your new device comes with additional drivers and software,
  they will usually be compatible with Apple computers as well."

Linux is _very_ hardware compatible.  In fact, companies typically don't
have to do much driver development for linux, they just need to release
specs.  And then they have a _perpetual_driver_ for it!

It's there lies the _root_ problem why you can't walk into a superstore
and just buy anything.

"Superstore hardware" is designed to force you to upgrade.  80% of
consumers just assume they need to upgrade their PC, OS and peripherials
at the same time, about every 2-3 years.  The superstore model is
designed around this.  It is _obvious_ why AOL and Microsoft investments
in Best Buy, Circuit City and others are made, and they are more than
happy to see it done because of the increased profit model of it all.

Peripherial hardware these days are largely software-based, with drivers
for _only_ specific Windows versions.  There are other tricks.  _No_
vendor wants to release a piece of hardware, and get the one-time
pay-off, and then turn around and release more drivers, and not see
_any_ additional revenue.  So they force you to upgrade when you change
your OS or PC.

They are in _total_cohorts_ with Microsoft, AOL and other entities on
this.  Not because they are "evil," but because of the ignorance of the
consumer.  Again, Microsoft, AOL and many others have _significant_
ownership in Best Buy, Circuit City and many others as a result.  So now
they not only control much of the PC OEM channel, but the retail channel
as well -- a _total_ lock-up of the distribution channels.

If I am a hardware OEM, why oh why would I want to support Linux?  Why?

With the Superstore model, I can make far more money.  And I get the
backing of Microsoft, AOL and others.  Even Apple doesn't make much of a
fuss, because they get closed-source drivers as well, and Microsoft
leaves scraps around for them.  Although Microsoft _did_ have all Mac
software ripped off the shelf at Best Buy, and _not_ because it "wasn't
selling" either.  But because it was a distribution control game.

Just like Linux and Dell.  Intel promised 100,000 PC orders from
semiconductor engineering firms to Dell when they offered Linux as stock
back in 1999, and Dell got them -- until engineers who tried to order
them found out that Linux was "only supported up to 128MB of RAM and the
slowest CPU Dell had."  It's a distribution control game.  And Microsoft
wins every time (even versus Dell who is funded more by Intel than
Microsoft).

Again, the hardware vendors are ready to play it.  Why?  What profit
incentive does Linux offer in return?  _None_!  So other than higher-end
hardware, where the profit margins are not razor-thin.  And those
vendors _do_ support Linux, because they have _real_ hardware in their
units.  The low-end, Superstore model is _never_ going to be "Linux
friendly."  Why?  Because Linux is not friendly to their profit model.


- NT might be better, but it was still _never_ designed for the Internet

  "Microsoft is doing a creditable job of beefing up Windows XP's
  security."

Again, people compare XP to 98, NT to DOS.  Of course NT is going to
kick DOS' ass when it comes to stability!  In DOS, you are "rebooting"
the i386 in and out of Protected386 mode thousands of times a second. 
In NT, you are _always_ in Protected386 mode, with a _separate_ kernel
and userspace memory.

Unfortuantely, the last "good" NT release was 3.50.  As of NT 3.51
"Daytona," the "Chicago invasion" (MS-DOS 7.0 aka Windows 95) occured
and NT was forever DOS' bitch.

It was bad enough that the NT 4.0 "Cairo" technologies _never_ surfaced.
The same will happen in NT 6.0 "Longhorn."

[ Kinda funny -- "Chicago" snowbirds ruining what "Daytona" was. ;-]
[ I'm in Florida BTW, NE Orlando, and 45 minutes away. ]
[ I don't do Daytona anywhere near Spring Break. ;-]

All the meanwhile, the GDI was moved into NT 4.0, _killing_ stability.
In NT 6.0 "Longhorn," the Avalon desktop will require more to be moved
into the kernel.  A lot of services were done so in NT 5.0 (2000) as
well.

With NT 5.1 (XP/2003), for the first time, the "security model" of Win32
was "cracked" for compatibility.  Plus you still must run as a "power
user" for most things, if not "administrator."  That is _unheard_of_ in
the UNIX/Linux world -- even for games!

The effects are bad enough on the desktop, but most users don't mind
rebooting.  But on the server, they are _detrimental_.  That's why
people don't see the difference of Linux on the desktop from server. 
Because a reboot is easy on the desktop, but not the server.

With Win32 continuing over .NET in NT 6.0+, Microsoft is going to slit
its own throat on the corporate desktop.  Microsoft's own security chief
even admitted that "no current Windows version was ever designed for the
Internet," yet attempts to adopt .NET, which the GNOME-Mono team _does_
think is sound, is being trumped by Gates & co for "Longhorn."


WHERE IS THIS ALL LEADING TO?

The corporate desktop will change the consumer.  That's why "consumer"
magazines just "don't get it" on Linux, and they "won't see it coming"
either.

Linux will inflitrate the consumer through corporate adoption.  As
someone who has maintained NT systems on corporate networks since the
orginal 3.1 beta (worked at the largest end-user company of the first
native NT application ;-), and Linux on corporate networks since 1995
(really '93, but "officially" '95 ;-), I call tell you with _certainty_
that Linux offers a _far_lower_ TCO than NT on the _corporate_ network.

The security issues will continue.  _Not_ because Windows is "more
popular" -- not at all.  Linux Internet servers _outnumber_ Windows
ones, yet have so fewer attacks.  That's because in Linux, you can
piecemeal stuff and turn it off -- like RPC.  In NT, if you do, it
_breaks_everything_.

Yes, "integration" is nice, but it is a security and TCO nightmare. 
UNIX/Linux systems are "piecemeal" in design.  More of a learning curve,
but far more security and lower TCO in administration.

As Linux desktop adoption, and Freedomware applications in general,
occurs in the corporate network, _consumers_ who bring work how will be
forced to use Freedomware applications.  As this occurs, they will stop
putting their documents into Hostageware apps, but cross-platform
Freedomware, Standardware and select Commerceware ones.

And that's when the "vendor lock-in" will be subsided.  That's the key.
Because people use applications, _not_ OSes.

It will take much longer for the Superstore model to be broken.  But if
we're seeing anything out of the PC OEMs right now, it doesn't matter
how much funding Microsoft does for their R&D -- they can_not_ ignore
the demand for Linux systems.  It's largely on the server right now.

But a  _lot_ of desktop PCs sell from the OEM without an OS.  And it's
not always because they have an open Windows licensing agreement either.


-- 
     Linux Enthusiasts call me anti-Linux.
   Windows Enthusisats call me anti-Microsoft.
 They both must be correct because I have over a
decade of experience with both in mission critical
environments, resulting in a bigotry dedicated to
 mitigating risk and focusing on technologies ...
           not products or vendors
--------------------------------------------------
Bryan J. Smith, E.I.            b.j.smith at ieee.org