Marketing ideas -- I have to strongly _disagree_ with 99% of this ...

Bryan J. Smith b.j.smith at
Sat Apr 22 06:13:02 UTC 2006

[ I know this is going to come off strong, my apologies in advance ... ]

On Fri, 2006-04-21 at 17:51 -0400, Alo Tsum wrote:
> First suggestions has to do with the software side of things. We users
> need a competent software installer which is graphical based. One that
> functions much the same way that the windows installer works.

God *NO*!  As a veteran of 13-17 years of [each] Solaris, Linux and
Windows [NT] enterprise configuration management, you've got it
_all_wrong_.  Packages and package management systems exist for a
reason.  Windows is _total_hell_ when it comes to enterprise management
compared Solaris and Linux.

BTW, there is already a GUI installer from the former Loki Entertainment
group.  It is a shell script with a binary object attached that brings
up a GTK+ GUI.  I do _not_ recommend we move to using that approach.

> This installer should track dependencies naturally

And the various GUI interfaces, including Pup, do not?
You're kinda losing me on "should track dependencies naturally."
What you want is the Windows "oh, install it anyway" attitude.
That's mutually exclusive with dependency enforcement.

Again, I must be missing your point.

> and place icons on the desktop

Who's desktop?

I think you're starting to think in Windows terms, where the Explorer
shell is infiltrated on _everyone_, most users are administrators, and
countless other "enterprise-hell."

Unless I'm really missing something here?

> or give the option to have icons for the software just installed to be
> placed on the desktop of the user as well as in the applications menu.

Ummm, are you familiar with all the capabilities of Fedora Core?
>From what you just said, it suggests you are not.

There are standards for GNOME.
I can't vouche for how Fedora Core handles KDE though.

> Also when watching the Boston Linux conference the suggestion was made
> to offer a hard disc manager much like windows offers for formating
> and receiving hard drivers etc after installation. 

They already exist.  But it's up to the vendors to bundle them.
They don't.  We can't do much about that.

I was also at Boston LinuxWorld, and a lot of the suggestions made were
clearly made from people with more background in end-user Windows than
enterprise configuration management of any OS (much less Linux).

> My second set of suggestions will deal with the over all focus and
> structure of the fedora project. I was listening to a pod cast
> interview with the head of the fedora project where in, the topic came
> up of some how generating revenue to put back into the project and
> make it more self sufficient. So I have a few suggestions which I
> think the Redhat company itself should take note of. Firstly Redhat
> while promoting Linux among enthusiast is also in the business of
> making money.

That's an oversimplification, but I won't go there.

> On that front I believe that they should not only attempt to
> evangelize Linux in the government, education, and corporate sectors;
> they should also attempt to get Linux in each and every home.

Okay, now here's where I think most Linux users are naive.

90% of home users are _not_ interested in long-term customer value.
They are interested in "new gadgets" and baited by "loss leaders" and
are more than willing to upgrade their PC, peripherals, OS and
applications every 2-3 years.  I have extensively talked about this as
the "Superstore Profit Model."  It's the primo reason Microsoft has a
stake in Best Buy, AOL in Circuit City, etc...

90% of home users put up with the fact that they buy a PC, peripherals,
OS and applications that are integrated.  And they put up with the fact
that if they upgrade one after a few years, they might as well upgrade
the other 3.  "Oh, my new printer won't work with my old system, so I'll
need a new one -- oh, that's right, I have to buy new software."  Trying
to break them of this is difficult -- when you tell them "free" they
think "free Microsoft software."  Literally.

You're not going to reach them -- and that's 90% of the home consumer
market.  It's a long, involved story, and trying to sell them on Linux
and dealing with hardware incompatibility (which will _not_ be solved
because of the "Superstore Profit Model" when it comes to drivers,
longer story), is not what they want.

I think Red Hat has the _best_ idea.

We got the techs from Day 1.  Now we're going after the Enterprises, and
that's happening.  As more and more Enterprises adopt Open Source, it
will come into the home out of necessity as people take work home.  It
won't happen overnight.  But it _is_ happening.  It's an utter waste of
their time to jump right to the home user.

I personally think an expansion of the "Fedora Ambassadors" is far more
worthwhile.  "Fedora Ambassadors" are who will reach the smaller users.
If Red Hat wanted to allocate a little budget for that, I wouldn't
argue.  But I don't think Red Hat will see much return at all from what
you suggest.  It's not worth their time, and I don't blame them.

> The reason being is if people are starting to use Linux in their homes
> and they are comfortable with it, employers will be more likely and
> willing to deploy a operating system which is different in many ways
> to windows on the interactive level.

No, you have it backwards.  Windows came into the home because of
business.  Just like DOS, 1-2-3, WordPerfect, etc... before it.

If you can show me where mass adoption of an OS in the home forced
enterprises, great!  But I don't see it.  I only see small adoption in
the consumer _until_ businesses proliferated it so all consumers

> Reason being is when people have to stop to learn new technology this
> cuts down on productivity and as a person who works for a IT
> department in a major university I can also vouch for the fact a IT
> team will not be willing to suggest an infrastructure restructuring
> when they know, supporting users on something foreign to them is going
> to increase their workload 10 fold.

IT infrastructure and personnel != average home user
Linux has the geeks, and more and more of the IT professionals.
That's already happening in the enterprises _now_.

I think you're talking about apples and oranges.

> So bottom line, more users both advanced, intermediate and beginners
> need to be converted to the Linux faithful but now the question
> becomes how?

It's _exactly_ how Red Hat is doing it.

The geeks got Linux in the corporations.
I know, I was there.
Internet services in the early '90s.
Engineering desktop and applications in the mid-to-late '90s.
Mainstream development and even desktop adoption in the 21st century.

SuSE (now Novell) and Red Hat added the traditional support behind it.
That's what enterprises want.
That's what Red Hat is giving them.

> The Fedora project is the perfect tool for this and here is how. The
> fedora team should focus SOLELY on making the operating system run as
> smoothly and as fast as possible, interacting with a HUGE number of
> hardware configurations.

Ummm, is that not the focus?
What do you think the focus is?

> Installation needs to be as smooth as silk

Installing Fedora Core on a bare hard drive is _easier_ than Windows.
Installing Fedora Core into a dual-boot is _easier_ than Windows.
Installation has _never_ been the problem.  Why?
Because 99.9% of home users get their OS _pre-installed_!

> and upgrading needs to be fail proof from version to version.

Fedora Core upgrades _easier_ than Windows too.
Heck, even Service Packs or Hotfixes to Windows.
I'm trying to find out what you are "comparing" to here?

> Previously installed drives with personal user data needs to be able
> to be retained without fail from upgrade to upgrade if the user isn't
> doing a clean install.

Okay, you've got to be _kidding_me_ here!  _All_ UNIX systems have "home
directories."  It is 100% _absolute_ and _all_ programs save config
files and data to it!

Microsoft took *2*Decades* to finally "standardize" on "Documents and
Settings" and even then, you still do _not_ have 100% (not even 90%) of
applications and other programs that handle it correctly.

I absolutely do _not_ understand this point at all.
UNIX is very standard in this, Windows is not.

Anyone who managed NT 3.1, 3.50, 3.51, 4.0 along side DOS 7.0 (95/A),
7.1 (95B/C/98/98SE) saw this clusterfsck first hand.  And it's still not
solved in NT 5.0 (2000), 5.1 (XP/2003), etc... either.  If they change
it in Vista, I'm done -- I really am.

> Now I would like to move on to "partnerships" Fedora project should
> look into making "partners" or some other creative term to define
> other Linux projects and organizations. In this partnership Fedora
> will tightly enforce standards which will ensure that any software
> created to run on fedora is following say the OIN and the GPL
> standards to the letter to ensure an user friendly and secure/stable
> operating system that runs smoothly.

Ummm, is that not the focus?
What do you think the focus is?

The GPL does _not_ guarantee anything.
In fact, open source does _not_ guarantee anything.

If anything, Red Hat's continued control over the Fedora Project and the
release of Fedora Core is more about guaranteeing proper regression and
integration testing of packages into a distribution than what individual
open source projects can do on their own.

I would _love_ for Microsoft to do the same with their own software.
God knows they rely on Altiris internally do handle their own patch
management resolutions and expertise.

> Many people in the Linux community may grumble about this suggestion

We grumble because we think you're not aware of the details you speak

> however life is about progression and when things do not change and
> evolve and progress to new levels then they are doomed to become
> extinct (think dinosaurs here) or at the very least remain niche
> applications.

I have no idea of the point you're making.

> If Fedora project implements such a model,

What "model"?  I honestly don't know what you're looking for.  You're
making assumptions that are incorrect -- about Linux, about Windows,

> they do not have to worry about making certain software for the OS
> which would take far to much time and man power to create.

Please detail this?  I'm really interested in where you are coming from.

> Prime example would be the hard disk manager

What?  I'm utterly confused.

> or even the software installer.

Again, unless you are advocating the Loki GTK+ installer (which I say
"hell no"), I'm utterly confused.

> This sort of work could be left to groups who's soul purpose is to
> make such software and by following strict guidelines they would
> become Fedora project "partners" and in turn they would be promised
> that their software will be included in the fedora core release.

That ignores regression and integration testing.
Regression and integration testing is _distribution-wide_.
Open source projects, on their own, doesn't address that.

> Also by following strict guidelines

Please define "strict guidelines."
You mention OIN and GPL.

What is the "standard" that says a distribution will ship with what
version of GCC, GLibC?  That's just 2 _core_ libraries right there that
make 75% the difference between distros and releases.

> this software could be implemented in other Linux distributions which
> are also following said guidelines.

What "strict guidelines"?

Red Hat was and still is constantly criticized for "pushing the
envelope" with GCC and GLibC versions in newer Red Hat(R) Linux and,
now, Fedora(TM) Core.  Other vendors "hold back."

I don't think you are completely aware that I'm just talking about
*2*base* and core libraries -- and that's just the start!

Even Linux Standards Base (LSB) won't touch it!
Until LSB says "this is the standard GCC-GLibC combo," there is
definitely no guarantee of distro inter-compatibility.
And that's just 2 core libraries to start!

> This would take the pressure off of Fedora and they can then focus on
> whats important which is making their OS run like silk.

What makes Fedora(TM) Core trusted, like Red Hat(R) Linux before it,
which is the foundation of Red Hat(R) Enterprise Linux, is the package
testing, regression testing and integration testing and resulting

You can't get most distros to "work together" on that, because they
handle many things differently.  Who's to say that the Debian model of
package testing, regression testing and integration testing is better or
worse than Fedoras?  Who's to say Gentoo's "ports" approach that has its
advantages and disadvantages against the formal "packages" approach of
Debian and Fedora-based distros?

> Again let me stress this approach is keeping in mind that projects
> such as Fedora and other Linux distributions desire to penetrate more
> into the home desktop market,

Which is a _pipe_dream_ because 90% of consumers are catered to by the
"Superstore Profit Model."  Consumers are not interested in hearing
excuses on why that brand-new $80 All-in-One Printer doesn't come with
Linux drivers out-of-the-box, or they have to wait 3-6 months on
drivers, when a newer, better, cheaper model will be out!

> which then also means more users will or could eventually equate to
> greater adoption of the platform in other industries as a result of
> user awareness and user comfort with the Linux platform.
> I should also mention that those software development groups that do
> not comply could be offered as Fedora extras so the community still
> has choices which is really part of the appeal of Linux.

Comply with what?

Again, to start, 50% of distro incompatibility comes down to GCC-GLibC
combination.  Red Hat pushes the envelope with Fedora Core.  It holds
back with Red Hat Enterprise Linux (based on several, previous Fedora
Core releases).

> Okay so with all that said how could this generate income? Well lets
> say Fedora project comes up with one of the first 100% standard
> enforced distributions

Again, there's that "standard enforced" phrase.
I honestly feel for you man, I really do.

> which is as user friendly or even more so than MS windows.

When 90% of American consumers can't get their brand new model, $80
on-sale All-in-One printer to work with Linux, they really aren't going
to think Linux is "easier-to-use" at all.

> Now say a "ambassador" from Fedora can start making the rounds to Dell
> and other companies and attempt to get them to start offering this
> FREE Linux distribution on some of their PC models, which would also
> allow for lower prices on the retail side for them (ie Dell, Gateway
> etc.) as the OS is FREE and that cuts down on cost which the end user
> ends up incurring.

Have you ever heard the Microsoft term, "per-model licensing"?
Please, _please_ read up on this.

> But we still have not addressed revenue for the Fedora project, and
> this can be done by following the Redhat model of offering technical
> support.

Are we back in the .COM boom or something?  ;->

First off, apparently you missed the reason why Red Hat Enterprise Linux
exists.  It's not because Red Hat was "greedy," it's because SuSE Linux
Enterprise Server (SLES) 7 _outsold_ Red Hat Linux 6.2 "E" (considred to
be, retroactively, Red Hat Enterprise Linux 1) 6-to-1!

Why?  Because most enterprises want support bundled with the product.
If this wasn't the case, Ian Murdock and Progeny would be infinitely
more successful with their service offerings than Red Hat is with RHEL.

And when it comes to home user tech support?  Forget it!  It's so
"nickle'n dime" that Microsoft itself won't touch it!  That's what it
relies on the PC OEMs for -- to handle those home user calls!

It's $135/call to call up Microsoft directly for Windows support by an
home user.  They aren't going to get any better value out of a formal
support program for Fedora.  So why would home users buy into Fedora?
Only the community can offer a better deal -- and that removes at least
90% of home users.

> Fedora project could basically offer technical support certification
> and training to Dell staff as an example so they (the PC manufacturer)
> can then take over supporting the platform for their end users, which
> also equates to revenue for these companies in the long run because
> they can offer extended tech support to end users at a premium.

Do you know what's _involved_ with building a training/certification
program?  If not, get involved with LPI!

> Fedora core could charge a VERY minimal fee for this training,

How?  I mean, building a training/certification program is
_very_costly_!  How are you going to fund it?  Oh, that's right, Red
Hat!  As if they don't already have their _own_ approach.

Again, do you understand the _reason_ why Red Hat Enterprise Linux
exists?  It's because such things cost _real_dollars_.

> so say charge enough that it would generate revenue that can then be
> pumped back into the project and at the same time would still make it
> cheaper for Dell and other companies to go with Linux on some desktop
> offerings as opposed to having a windows only offering.

The project is largely funded by Red Hat (or Red Hat pays a lot of
salaries), and that's unlikely to change anytime soon.  Just like Sun on, etc...

> The future of Linux if to be taken seriously

Seriously by whom?
The average home user?
Or the enterprise with actual money?

Learn from Microsoft.
The used each market infiltration to force product adoption.
>From DOS to MS Office, you infiltrate and proliferate.

SuSE (now Novell) and Red Hat have done it at the enterprise.
It will come, it just takes time.

> should not be relegated to just the business, government, and
> education world as far as standards, reliability, and software /
> hardware vendor support is concerned.

Again, the "Superstore Profit Model" is a _big_brick_wall_ in the face
of Linux with regards to hardware vendor support.

> The brand will grow far more rapidly if consumers are adopting the
> standard at home and at work so basically this is a bottom up
> approach.

Ain't gonna happen that way, sorry.  Enterprise adoption will push Linux
downward.  Once it reaches the point where Linux adoption is widespread
enough that it affects the "Superstore Profit Model," _then_ and
_only_then_ you'll see realistic consumer adoption.

Unfortunately, the "Superstore Profit Model" is hard to deal with.
Linux offers hardware vendors _nothing_ in increased profits.  Windows
version-specific drivers and loss-leaders offset by Microsoft dollars
does.  ;->

> This of this, more desktop users in the home also means more software
> sales for major companies because you will have more people buying
> video games

Okay, software economics 101 ...

Q:  What is the _least_profitable_ software industry?
A:  Games

Yes, you have breakaway titles.  But for the most part, it's a
cut-throat, razor-thin, often _in-the-red_ industry.

> and other such things which will also mean more companies willing to
> adopt the platform because software offering become greater. 

Top-down:  Enterprises adopt, users are stuck with
running it at home.  Evolution to do corporate collaboration, etc...

> There needs to be a consorted effort on the parts of all parties

Who are?
The projects?
The distros?

Have you read Linux Standards Base (LSB)?
Do you understand what it does _not_ address?

> involved to take Linux to that next phase of existence other wise
> Linux as a brand,

Linux is _not_ a brand, it's a technology.
Linux will _never_ be a brand.

> while it may grow some what will not see its full potential.

As long as people don't understand why Linux is not being adopted, this
will very much be the case.

> With the software being a open and free model we still have to realize
> with a flurry of hodge podge coding and no standards insight

Whoa!  Great job!  You've managed to insult the projects.

No offense, but in working on _mission_critical_ defense and financial
systems code for about 9 years of my career, I can tell you, with great
certainty, that open source is _very_robust_ and _very_standardized_.

The problem isn't coding and no standards -- that's Windows' problem!
Open source projects are _far_better_!

The problem is no standards on core libraries, release models, etc...
Things that make up a distribution as a whole.

Debian, Fedora, SuSE, etc... all differ!
Even before looking at Gentoo.
And that's _unlikely_ to change!
Because Debian, Fedora and SuSE _disagree_ on those details.

Heck, Ian Murdock and Debian-Progeny have an entirely different
philosophy than Michael Tiemann and Fedora-RHEL.  _Both_ models have
merit.  I actually believe more in Murdock's "configuration management
as a process" than Tiemann's "subscription/support as a product," but as
a professional, Red Hat has the mindshare because of what enterprises
assume from Microsoft, Sun, etc... before them.

I think you're not looking at the entire picture -- and down to those
actual, real technical and enterprise details.  And that's before we
look back at the home user and the "Superstore" issue.

> the end user ends up losing at least as far as the home front is
> concerned.


Yes, I agree the home user is bombarded by software written largely by
incompetent, oursourced or H1B Visa Indian, Irish and Israeli
programmers who have had virtually _little_ (if any) exposure to any
formal software development or engineering processes, let alone don't
care about the software they write like the passion of open source.

But for the average home user, he/she doesn't care about security.
He/she only cares why their brand new model, $80 All-in-One printer
doesn't work.  And if a new PC and software comes out, they want to run
all those latest things on it -- that old PC can be Linux, if they get

> Most people are forced to run duel boot Windows and Linux systems
> because software makers and hardware manufactures have not fully
> bought into the Linux model and we as a community only have ourselves
> to blame for that.

No, not at all.

The idea of open specifications and community-developed, perpetual
software drivers for hardware is _mutually_exclusive_ with the
"Superstore Profit Model" of releasing hardware for a limited time,
enforced by lack of drivers for the next OS version, let alone the
hardware itself is largely software-based -- licensed from a 3rd party,
so it cannot be released open source.

That's 90% of the issue with hardware right there!  At least for home
users!  The users who buy cheap hardware at the Superstore, and not
quality hardware that enterprises use for servers and workstations.

90% of users are more than willing to buy a new PC, peripherals, OS and
applications every 2-3 years.  If they replace one, and it's
incompatible with any of the other 3, they just assume they have to
upgrade all 4!  They see the pricetag and the loss-leader, and in 2-3
years, it's "not new and fun anymore" so they don't care about it not
being usable.

Welcome to the Superstore Profit Model!
Welcome to the whole reason Microsoft controls Best Buy!

> This approach will also take some evangelizing to the software makers
> of such things as yum and KDE however I believe that those who do not
> see the need and importants of doing such things will render
> themselves obsolete in the long run.

Huh?  Paying customers are supreme.  What can Linux offer hardware and
software vendors over the "Superstore Profit Model"?  _Nothing_ but

> Just look at the Unix model, and we can see what the disasters of not
> doing this can incur. 

What was the "disaster" of the UNIX model?
Licensing was about the only thing tangible.

Microsoft gained control of the PC distribution channel and went from
virtually _no_ Office marketshare to _everything_ in 2 years.

Microsoft has since gained significant marketshare in the retail
channel, and that killed Apple until Microsoft gave them a cash influx.
The main thing saving Apple on the retail shelf now is the iPod.

That's the "disaster" that keeps SuSE (now Novell) and Red Hat away.

Bryan J. Smith             Professional, technical annoyance
mailto:b.j.smith at
****** Speed doesn't kill.  Difference in speed does! ******

More information about the marketing mailing list