Re: LINUX is the "one" ? .... Comments???

On Fri, Apr 04, 2003 at 01:07:39AM -0500, jt wrote:
> | Careful. ESR and RMS have fundamental differences of opinion about this
> | stuff; in particular, RMS has pretty much disowned ESR as a part of his
> | movement.
> RMS was just jealous because ESR re-packaged the formula and got widespread
> publicity by doing that.

Not even ESR believes that - and he and I have talked about it quite a bit.
Whatever else you can say about RMS, he has a very strong sense of ethics,
and very strong beliefs, and (unlike most people) tries very hard to live by
his values. Hie behavior is entirely consistent with his stated beliefs
about "free software".

> | There are more than a few advantages of the Open Source methodology.
> It is imagined that there are advantages of the "Open Source methodology"
> which aren't available to commercial developers.  Again, these are imagined.

They're not. I can show you real-world examples in the Hercules code.

> "Open" source is not EVEN a programming methodology.. it's a marketing
> gimmick.  That's my main gripe, and answers the question (I optimistically
> hope, once and for all) of why it's a problem for businesses to adopt this
> methodology.  Because THERE IS NO PROGRAMMING METHODOLOGY.

It's not a marketing gimmick, either; those companies who try to treat it as
one lose the benefits.

> Most ALL the posts I read in favor of "Open" Source STATE that it's primary
> benefit IS it's being anti-commercial.  Oh, Napsterism isn't theft...!!!
> Okay, I get it now... <sarcastic>

Nope. The primary benefit of open source is that anyone can improve the
code, be it by fixing bugs or adding features - whether or not the original
author wants the improvement. No longer must a user wait for the vendor to
fix a bug. To see the benefit of this, one must merely compare the average
lead time to fix a security flaw once reported. In Linux, that average time
is measured in small numbers of hours. In Solaris, or IRIX, or AIX, that
time is measured in weeks. In Windows, that time is measured in months, *if
they choose to fix the bug at all!* (M$ announced that they would not fix
one recent security hole in Windows NT, period, even though it's supposedly
still supported.)

This has benefits for the vendor as well as the customer. In particular, he
can get lots of help from his customers when things do break.

> Granted ESR is a marketing genius, but His theories on the economics of
> software development were not correct when He wrote them and have been shown
> to be unworkable, in actual fact.  There will be SOME companies that
> actually make some profit from Open Source.

You mean like IBM? They're embracing Linux for one reason: it's driving lots
of dollars toward their professional services arm. *BIG* money.

As I said before, not everything is appropriate for Open Source, but there's
plenty that is.

> | Unlike RMS, I do not believe it's applicable to every program
> | everywhere, as a moral stand...but there are lots of places where it
> | makes sense. (Bruce Perens once told me that he didn't think that Open
> | Source was the best model for every program, and he's one of the
> | founders of the Open Source Initiative.)
> > Saying it's not the best method for every program is not going far enough.
> There are VERY few areas that I can think of where "Open" Source is a better
> model.  For development, I'm not talking marketing here.

How about for a program that only has limited potential for commercial
success in the first place? Where do you get money for developers when you
can't get anyone to buy the product?

> You may be qualified to make the statement.  Perhaps if You can show me
> where commercial software developers that provide source code don't reap
> these same advantages...???

Show me commercial developers that provide the source code these days. Not
doing so may have been extremely controversial once upon a time (were you
around for the object code only flamewars in the mainframe world? I was, at
the tail end), but it's accepted practice now.

> | That does not mean that I believe that every progfram should be done
> | that way. There are many holes in the Open Source model. For some
> | programs, such as Apache, they're not major problems; for others, they
> | are. (Witness the less-than-great success of OpenOffice.org, for
> | example.)
> Well that's not what I've heard about Apache.  I hear they're CONSTANTLY
> looking for someone to do the dirty work of testing and documentation.

IBM's doing a lot of the testing, as are other major companies that use and
depend on it. That doesn't stop it from being the preeminent web server
platform, not only because it's freely available, but because people *can*
fix bugs and get the fixes incorporated into the codebase.

> And I can't tell You HOW many times I've heard in the press and on their
> websites, over and over again:  "Well this feature is a good one, and will
> eventually be coming, but the developers are donating their spare time,
> and.. well.. You understand..."

This has the effect of automatically prioritizing development. If a feature
is important enough to you and doesn't exist already, furthermore, you can
add it yourself! Try *THAT* with Domino.

> | > It's different in that one programming methodology DOES works and one
> | > doesn't.  I've seen the successes of OS like Apache, and Linux.. and
> | > see the drawbacks to these being OS as well.  But overall, RMS has
> | > BEEN SUCCESSFUL in His goal to destroy the software industry, albeit
> | > with the best of intentions.
> | I don't think so. It's not as healthy as it was during the dotcom
> | boom, but
> Yeah, when billions if not trillions of dollars were.. how to put it
> delicately.. wasted down the drain.

Yup. However, very little of that disappearing wealth can be laid at RMS's
doorstep. As you point out, it was mainly due to people getting tons of cash
thrown at them without such basics as product.

> | > The latter is obvious, but paying nothing for software DOES
> | > automatically make it bad.
> | Wrong. If I felt so, I wouldn't be involved with Hercules.
> I'm not familiar with the business model of Hercules.

Hercules is an Open Source emulator for IBM mainframes that runs on Linux,
Unix, and Windows. (And perhaps OS/400, under PASE, depending on just how
good AIX 5L's Linux compatibility is.) I manage the project; there's a group
of about 30 core developers (many of whom are EXTREMELY sharp) that does
most of the coding. People are added to that core group by writing and
submitting good code. As I type this, there are 3062 members of the main
mailing list for Hercules users. The home page is at
http://www.conmicro.cx/hercules .

> You get paid?  Then I'm ALL FOR IT!  Lucky You...!!  If the company develops
> software, but doesn't receive income from that software, but from secondary
> derivitive income...

I wish I got paid. So does the original author. We both have gotten jobs
because of our involvement, however. (I'm hoping to land another one, as
well.)

> Period...  Meaning it could be better...  For the reasons I stated.  Puts
> the customer back in the loop, for one thing...  No paying customers.. not a
> very efficient loop, economically speaking.

OTOH, what if there were no paying customers to begin with? Several attempts
were made to commercialize Hercules, before it went Open Source. None were
able to generate any interest at all. That's about as inefficient as you can
get.

Further, to your "put the customer back in the loop" argument: When the
customer can bypass the vendor and make his own changes to the software,
he's far more in the loop than if all he can do is submit requests for
improvements and then wait for the vendor to get sufficient round tuits to
answer the request, never mind actually implement the changes. Made a
feature request to IBM lately?

> | If you think Linux is a tinker-toy, what do you think Windows is?
> Another tinker-toy.  Why does the "Open" Source discussion ALWAYS come back
> to MicroSoft and Windows.

Because that's the terms of the current debate, like it or not. Other Unixes
are getting rapidly marginalized - you need look no farther than AIX to see
that. Other OSes are, these days, niche players. (And I say that as an
avowed MVS bigot.)

> I'm talking the economics of software innovation.  And the argument ALWAYS
> turns into, "Why are You so IN FAVOR of MICROSOFT???" when I'm not.
> "MicroSoft isn't developing GREAT software EITHER", which I already know.
> As I said previously.

There are just two business models these days: Open Source and Microsoft.

> | Linux is a far, far better OS than Windows, and I say that even though
> | neither are among my favorites.
> That could be, but I'm not even gonna try to assess that one, on the server
> OR the desktop.

I know that the folks here (if they're still reading this increasingly
off-topic discussion) will laugh, but Linux is delivering uptimes that make
Windows advocates turn away in embarrassment. It's more robust and more
scalable.

> | > Look, anyone can buy into this Open Source line if they wanna, but the
> | > consequences are still the consequences...  It's a poor method to
> | > develop and distribute reliable software and produce innovation, no
> | > matter how it's dressed up.
> | I strongly disagree. Linux is the best-known counterexample, but it's far
> | from the only one.
> Again, Linux is best-known piece of marketing hype of the past century.  It
> MAY very-well be able to survive based on the marketing hype and it's being
> a counter-balance to M$...  Innovation???  Reliability???

The biggest crime Bill Gates has committed, IMAO, is destroying people's
expectations of software reliability. The "Oh, just reboot it when it
crashes" mindset has set computing back 50 years. That said, Linux is more
than adequately reliable, in the experience of just about anyone who's used
it for production work.

> What's the ETA of these, fer as Linux goes???  I keep hearing these things
> are "just around the corner".

A system that produces average uptimes measured in years is sufficiently
reliable for the vast majority of folks. Those who need better won't adopt
Linux any more than they would adopt OS/400. (They're on Tandems or
Stratuses, for the most part; the very largest are on ACP/TPF on
mainframes.)