I’m not expecting to be that interested in this morning’s keynotes. The
first one is about business issues for open source in the enterprise. I’ve
heard speakers on that topic over and over and over again, and frankly, I’ve
never heard anything very deep or useful. Maybe today will be an exception,
though. The second keynote is going to be fairly interesting. I know
because I heard it a few month’s ago as a keynote at O’Reilly’s Emerging
Technologies Conference. So it’s a “rerun” for me.
By the way, I now think that the “wireless woes” I’ve reported are really
“hotel’s Internet provider woes”. It appears that the conference is
connecting to that network, and that it’s very unreliable. I’ve been using
the wired network connections in the rooms, and it comes and goes at random,
and is very slow. The network is provided by
STSN and they need to be more reliable to
be worth $10 per day. And it needs to be faster; my e-mail is downloading
about as fast as it does on a 24000b/s dial connection.
The Business and Economics of Open Source in the Enterprise by
Stormy Peters of Hewlett-Packard. The speaker manages HP’s open source
program office. This office reviews every open source product that’s used
at HP. They review 2-10 projects per week.
She starts with the basics of open source, and how software lifecycle
costs mean that purchase price is not the main cost factor. Then she
compares open source to the pharmaceutical business, an idea she got from
Martin Fink’s
The
Business and Economics of Linux and Open Source, which HP recommends
that their customers interested in Open Source read.
Why would you open source software you created? She gives a lot of
examples of good business reasons for that:
- If the open source commoditizes market you don’t control
- Would make product pervasive (e.g., Linux drivers for HP printers>
- Would promote use of proprietary software
- Lowers overall cost to company
- Promotes hardware (or other value add)
- Create custom solution for customer
- Provide services related to the product
- Allows you to exit a business (without leaving your existing customers
high and dry) - Leverage resources from others
When you might not want to open source:
- Your product controls its market
- The product should be obsoleted
- There are costs that don’t justify the benefit to you
- Could misdirect and defocus your resources
- There’s an unjustifiable intellectual property risk
- You want to compete against the open source community
- Your only motivation is that it’s cool technology
That covers releasing your software as open source. Why would use (or
not use) someone else’s open source code? You would use it when:
- you want to promote an existing standard
- it’s already an existing, pervasive technology
- it lets you refocus your resources on value add
- there’s no risk of copylefting your existing products
Conversely, don’t use open source when:
- the technology’s goals don’t match your goals
- your product’s chief architect doesn’t want to
- time to market is critical and the open source product isn’t quite ready
HP’s open source policy’s objectives include making sure that they are
honoring open source licenses, that they aren’t unintentionally exposing
their products to being “copylefted”, to have proper business controls on
decisions (those controls might currently be triggered on purchases, which
don’t happen now).
There are lots of good things businesses want from open source, and some
that aren’t there. You can freely copy the software, without tracking
licenses. You can buy support, and get manuals either freely or in
commercial books. But indemnification doesn’t exist in this world.
Well, it was a fairly interesting talk, though not very meaty. HP has a
web portal for all its
open source activities.
Open Source on the Mainstream Desktop by Mitch Kapor. This is the
person that brought us Lotus 1-2-3, the first killer application for the IBM
PC. Now he has set up the Open
Source Applications Foundation, and is creating Chandler, and open
source PIM.
But he’s not going to talk about Chandler today (so this isn’t just going
to be a rerun for me). Instead, we will hear about Linux’s Journey to
the Mainstream Desktop. It has been a long one. Interestingly, Linux
as a server is being implicitly assumed to be a success that there’s no
reason to discuss today.
Kapor was introduced to open source (not called that yet) in 1985 when he met
Richard Stallman, who was picketing his company. Lotus was suing others
over “look and feel” and Stallman was protesting that. Although Kapor
wasn’t really in favor of those lawsuits (and was exiting the company) he
though Stallman’s ideas were unrealistic and foolish. He doesn’t feel that
any more. He complained about companies whose business strategy is suing
others over alleged intellectual property violations and got a big ovation here.
One reason for Linux on the desktop is that it doesn’t make sense to
spend $200 for a PC and then $500 for basic software for it.
Internet-centrism makes things more open to alternatives to Windows.
Linux is being massively deployed on desktops now. Not only is its adoption broad,
there are some large concentrated installations, in Largo, Florida (400),
Munich, Germany (14,000); Extremadura, Spain (80,000); and Thailand (up to
one million subsidized PCs to low income people, with 160,000 already
shipped). Microsoft is responding in Thailand by selling Windows plus
Office to install on those PCs for $35!
The desktop Linux adoption cycle isn’t going to be due to a “killer app”.
Instead it will start on the periphery then get more and more mainstream.
In Phase 1 (happening now) technical users will broadly adopt it. Phase 2
will start next year and spread to significant numbers of
“transactional workers” (people who use
computers for a few very specific tasks, like call center workers) and
web-centric consumers. Finally, Phase 3 won’t start before 2007 and will
spread to some knowledge workers, small businesses, and consumers.
How does Linux score for these users? The desktop itself (KDE and GNOME)
is given a B; it’s polished and stable. But the desktop developer platform
(tools and libraries to create consistently behaving applications) just gets
a C-, so the end user doesn’t get a consistent end-user experience.
Hardware support is a B- for the computers themselves, but a D for
peripherals (he references
www.ometer.com/hardware.html
for this). Applications get a C+ (the basic apps are all there, but there
are still issues with accessing Asian web sites and some proprietary
multimedia formats). Windows connectivity gets an A.
What about the “Linux ecosystem”? Distributions and desktop platforms
each get A-. Developers get a B, but there aren’t a lot of ISVs, so that
gets a D. OEMs are weak with a C and distribution channels get D. All this
information is in a report he’s compiled. These slides and the report will
be linked to sometime later today from the
Open Source Applications
Foundation website. By the way, you get more general information on
Mitch Kapor’s web log.
He “goes out on a limb” with a few predictions. We will continue to see
Microsoft price cuts to compete with Linux. The Linux desktop will get
better and better. We will see 10% of all desktops globally running Linux
soon (the rest of the world is leading the US on this because of the
economics and governments encouraging open source adoption). Public sector
adoption will drive other adoption, as will selective enterprise adoption.
Consumer adoption will lag behind.
He closes by announcing the launch of
Opensector.org today. It’s not
from his group, but he supports the effort.
Charles Engelke's Blog 