In defense of Ubuntu

> PS: It's Hurd, not HURD.

At least it's not Turd.  The FSF's ability to pick bad names is unmatched.  They should change
while they can, before many people are even aware of the project.

Hi Michael! =)

Part of joining Canonical was largely handing off my Hurd responsibilities, as well as a lot
of my porting ones.  My previous employer had been forgiving about me having 13 architectures
(including very noisy and power hungry Alphas and Itaniums) under my desk.  My home-office
manager, that is to say: my wife, was much less so. =)

I think that given patches, the Toronto Hurd Users Group, Talks, on-line tutorials, and such
that I did with the Hurd, it's fair to say that I worked on it even after I joined Canonical.

It's somewhat ironic - some of why I got the job at Canonical was because of my glibc
co-maintainership, which I only took on because of the Hurd. =)

Tks,
Jeff Bailey

As far as capital for Ubuntu I would imagine that given Shuttleworth's current financial
position and the apparent soft place in his heart he holds Ubuntu that it's likely that even
if Canonical ran out of money and couldn't obtain capital from the open market that
Shuttleworth himself would reinvest. I'm not speaking for him or saying it's guaranteed but
based on his public statements and the positions he takes on the company it's my belief he
would keep pumping money into the company until it does become profitable. I wouldn't worry
much about their survival. From his statements he sees the project for what it is, a long term
investment in something that is very likely to pay off substantially down the road as the
Linux desktop becomes a reality.

There is no question in my mind that Linux as a whole will take over on the desktop. Not even
MS can sustain the development effort that goes into Linux just to keep Windows on the same
level. Given how far behind Linux was even 5 years ago and how viable they are today you can
see the progression is an order of magnitude faster than commercial development.

It actually makes a lot of sense, when targetting small business as a customer, to make a
distro that employees *can* run at home, even if not in the exact same version (Professional
vs. Home, for example).  If your employees use the same OS at home as at the office, they're
more comfortable with their computers, at least in theory.  What's more, they're self-training
on the OS at home, so you need to invest less in training them in the basics on the job.

From what I read from previous interviews (not that long ago, but I don't seem to be able to
find them again) Shuttleworth has stated that Canonical was aiming for break-even in 2008.
With 130 employees at an average salary of $40k/pa (last number pulled out of my backside, but
including taxes etc. is probably conservative), they need to make upwards of $5 million
annually to cover staff costs. Including value added taxes and the other costs of running a
business, probably $7 million is nearer a base figure for turnover. 

That's nothing like Novell's or RedHat's annual turnover on "Linux", but both are growing and
RH in particular has been posting some pretty amazing figures. As far as I can tell,
Canonical's only product (outside of special deals on putting Ubuntu on OEM hardware?) is
selling support basically. A target of $7 million is a lot of support contracts...

Does that include landscape too? Can you provide a reference?

Code where mouth is and all that.

Cheers to you, Tridge.  I have one slice of pizza left here. (I really do.  It's a Super
Veggie.)  If you want, I could send it parcel post. ;-)

As with a Frank Lloyd Wright building, the support beams in our FOSS community are not always
completely obvious.
 

That's a very enlightened thing to say. Thanks for a very good comment.

It really comes down to sustainability on several different levels. 

First there's just the sustainability of the Ubuntu community itself.  If Canonical as an
organization blinked out of existence tomorrow, could the Ubuntu community pick up the pieces?
Is there enough of a non-proprietary infrastructure to keep Ubuntu going with the same
processes? Are the critical pieces of what Canonical does for the Ubuntu community replicable?
Mark Shuttleworth certainly isn't replicable..but I would hate to think that the entire Ubuntu
community hinges on him. Cults of personality are not exactly community.  The question remains
though, are the services that Canonical provides the Ubuntu community replicable if there was
ever a need for the community to take over some of those responsibilities?  That's a very
serious question that active Ubuntu community members should be thinking about.

Second, is the effort Canonical is making a sustainable in the context of the larger
ecosystem.  Are the technologies that Canonical is spending money on creating a sustainable
platform for other to do their work in a way that does not assume the continued existence of
Canonical?  Do you want your work flow to be locked in to Canonical and its business
interests?  That is a very serious question that external projects should be thinking about
when making use of Canonical's services.

Third, are the spending decisions Canonical is making as a company, leading towards a
sustainable business model for itself?  That's a very serious question that Canonical's stock
holders...oh wait...still a venture capital startup.  I guess that's a question that Mark
should be thinking about. Mark and anyone he's courting for the next venture capital
injection. 

You can make decisions that seem great in the short term, but are not necessarily sustainable.
A bumper crop of corn for 3 or 4 straight years, to take advantage of increased demand for
corn, without letting your field lie fallow, can have serious long term consequences for the
productivity of the land and the long term sustainability and profitability of your farm.  

If you are a large farm, a corporate farm, with a significant amount of land, big enough to
have your corn production count as a significant amount of national or global supply then such
short term actions would depress the price of corn for other people, making it harder for them
to turn a profit while you were flooding the market with unsustainable supplies of corn, and
affecting the sustainability of their farms as well.
Luckily, Canonical isn't that big.

-jef

In what way does providing packaging of other free software components free of charge not
constitute contribution? Packaging is surprisingly hard and huge amounts of work. Ubuntu
provides a valuable service to the community absolutely free of charge.

The number of Ubuntu users does not translate directly to commercial success, and it doesn't make much sense to measure "contributions...roughly in line with the number of their users". In fact, it's to be expected that the user community will grow much earlier and much faster than the customer base.

The fact that Canonical is compared with competitors with an order of magnitude more developers was flattering once, but when it's used as justification for this type of criticism, it's discouraging.

I'm sure that ingroup/outgroup psychology is a valid explanation of part of the phenomenon of
why some people have irrational loyalties but you're ignoring a simple, pragmatic reason why
many of the examples you mention were considered undesirable. At various stages and in various
ways they have failed to demonstrate a commitment to Free Software.

* Back in the pre-Fedora days there was a certain amount of disquiet at Red hat because they
used to distribute an extra "applications CD" which contained a Macromedia Flash player, Adobe
Acroread and some other stuff like that.  They were still aware of the issues though. This
link is from RH 8.0, but I'm nearly certain I remember this from as far back as 5.2 though:
http://www.redhat.com/advice/speaks_80mm.html

* Novell, pretty obvious and it's not ancient news

* Canonical and the Ubuntu project has been covered well here on LWN in terms of contributions
of Free Software, the closed nature of Launchpad and of course the desire of have Aunt Tilly
run non-Free software to her heart's desire

Anyway, no point in going over Linspire and Xandros, the point is that the main criticism of
all those distros or companies has been their distribution of non-Free software.

Trivializing this objection as a simple group psychology phenomenon is to ignore some very
valid objections.

Red Hat was vilified in the linux community because it took a very popular product (Red Hat
Linux) and made it proprietary. New updates to Red Hat Linux required paid subscription. They
quickly realized their folly and created Fedora to soothe angry users.

Personally I am suspicious of a distro that starts free and becomes closed. This is why I use
debian which will _always_ be free.

Fedora and Red Hat would love to ship codecs to users to allow them to play DVDs, however they
would also love to stay in business. Red Hat is currently worth suing, has money in the bank
and a sustainable income stream, and is an actual US company.

Compared to a company HQed in the Isle of Man, with no major profit or cash reserves, RH is a
quite valid target for all kinds of patent trolls and MPEG consortium to take to court if we
shipped things that we should really license.

You know, some distros do prefer quality releases to crap on exact day.  And I'd argue that
this *is* part of caring for user experience as well.

Promoting wild push (ever been at a daily gazette?) isn't gonna win much friends among
upstreams and peers, as was noted already... if you'd look at quite some Ubuntu patches or
glue code, you'd barely touch the result with a six feet pole.  But you don't care, luckily.
:)

So...

> Be happy with your distro of choice.

RH is the biggest single commercial contributor to GNU/Linux; only problem is that RH is not
mentioned in that paragraph.

OTOH Canonical open sources all the infrastructure projects of its distro just like RH; it
"just" happens that RH, as a better established company than Canonical, has better established
projects with more visibility and following.

If Mark Shuttleworth is very smart with money, or at least hires people that are smart with
money, then the profits he made selling Thawte  (something like 575 million dollars) could
support Canonical pretty much indefinitely on interest and return from investments alone. 

Think about it. If you have $500,000,000 dollars and are able to pull in a 3% return on that
money per year, then that's pretty darn close to $15,000,000 dollars a year. Unless you figure
out a way to give almost all of it a away taxes will simply take it from you. 

You could spend $40,000 a day and never go broke.

Ubuntu has the hallmarks of a vanity project. It's goal is probably to be mostly
self-supporting in the long-term. In the short term I expect their goal is to simply get 'mind
share' and market penetration. If they gain high levels of popularity then the opportunities
to make money will present themselves naturally. 

If I had a couple hundred million dollars to burn I'd do a similar thing. 

I'd would think I'd hire a team of people to simply do the 'boring' work in Linux and try to
convince them that they are going to improve humanity with their tireless efforts.
Documentation, bug fixing, software packaging, etc. Then for 'profitability' I'd farm them out
to companies that want to improve some particular piece of open source; open source software
development as a service.

Something like that. It would be one of a few different projects. 

This is one of the positive aspects of capitalism.

I'm going to guess that what Canonical is doing is similar to Google or Pets.com.  They are
building a strong brand first, making sure to stay true to the brand before getting completely
wrapped around an axle about making money right away.

My chosen examples are two extremes of this strategy at work.  With Pets.com, pretty much
everybody is now convinced (including former investors) that a brand that means "get your mail
order dog food here" doesn't have a lot of value.  However, with the right brand (e.g.
Google), it's possible to monetize a winning brand.

For Canonical, I wouldn't be entirely surprised if their model depends on Linux becoming a
mainstream desktop operating system (e.g. >10% marketshare) before they try for profitability.
If/when Linux does go mainstream, there are a lot of ways to monetize having the leading
distribution.  It's obviously a very risky strategy, but a potentially large payoff on the
other end.  It's also a risk that Shuttleworth can afford to make, because even if he's not
successful, he at least has a full-time staff working to make sure Linux works on his laptop.
;-)

> What is Canonical's business model? Long term? RedHat's agenda is fairly clear but I can't
figure out Canonical's business model at all. Perhaps I'm missing something obvious, but I
find myself pondering this every time they are brought up. 

I'm not sure it's much different than Red Hat's business model. I.e., sell support. If you
counter that Ubuntu can be used for free without support, well, there's CentOS, and I think
CentOS and Ubuntu being free are _good_ for getting paying customers, just like piracy helps
Windows - it increases market share, and once you have that market share, paying customers are
a part of it. In fact Ubuntu's model is the better one in that sense, since (just like Windows
piracy), the free users are experiencing your brand.

Aside from 'normal' paid support, there are also contracts for customized builds (in an
interview Mark or some other Canonical person said that this was more common on the desktop
actually). If you're using Ubuntu and want some modification of it, it makes sense to hire the
people who are the most expert at that, namely, the people making Ubuntu. This is also a
typical FOSS business plan.

Will this work? I don't know, but to have any chance of success Ubuntu must become a very
popular distro, and surprisingly enough it has achieved that goal. So it is at least possible,
and it'll be interesting to see what happens.

> It all boils down to... does the overall Linux community gain much 
> from the mass, somewhat transient influx of users that Ubuntu has 
> given us?  Even those that stick around... are they a plus for the
> community or a drag?  Do we really care about the numbers?  Should
> we care?  Or should we just be trying to make Linux better?

Overall, those who stick around are a plus. Even when the strength density drops, there's
still power in numbers. It really is worth the trade. 

This whole 'netbook' thing is a gainer. From my limited exposure, Linux on the SSD is fine,
but XP doesn't like SSD. That Linpus Linux on the Acer Aspire One was nice I thought. 

The Dell netbook looks interesting. Those are the ideal market for Ubuntu. As you mention,
people who move from windows to Ubuntu are going to look for familiar things. On a new form
factor, they won't be looking for the familiar as much. 

> Now having said all of that, I must clarify and say that I'm not claiming
> that all Ubuntu users are stupid or that the Ubuntu distro is for stupid
> people.

It is not.  It's marketed as "Linux by human beings, for human beings".

I do realise it's probably nothing more than a slogan, i.e. advertisement, but I can't quite
give up the fact that words have meanings even when the words themselves are turned into ads.
As such, I've always been slightly irritated by that slogan, which seems to imply that I, as a
Debian developer and user, am not a human being.  Okay, so I've got super geek powers (whoa, I
master the command-line!), but super-heroes are humans too.

People have been complaining that making Linux easier to use is bringing 
dumber newbies for at least a decade.  Ubuntu is only the latest target of 
such accusations.

As someone who has used Linux since 1995 (Slackware, Red Hat, Mandrake, 
Debian, Ubuntu) and has a tiny bit of code in there, I'm much happier 
having Ubuntu (well, Kubuntu) to run than dealing with something 
non-newbie-friendly like (for example) Gentoo or Slackware.

And as someone who ends up being asked for computer help by friends and 
family, I'd much rather help them if they're using Ubuntu than if they're 
using Windows.  Or Gentoo or Slackware.

Making Linux easier to use attracts more users.  This is a good thing, not 
a bad thing, because of network effects.  As users increase, benefits to 
those users increase exponentially.  When we have a measurably large 
market share, we have fewer problems getting support from hardware 
companies, software companies (yes including free software), media 
companies, and governments, all of which have made life difficult for 
Linux users in the past due to our small market share.  When the "dumb" 
users are using Linux, the "smart" users find it easier to do their own 
everyday tasks with it.

I'm sure there are some Windows to Ubuntu users like the people you talk about, and every day
some of them give up on Ubuntu.  However, most of the people I see on the Ubuntu forums are
making a real effort to learn Ubuntu as they encounter and fix problems.  

Some of these former Windows users then write up what they've learnt as HOWTOs on the forums -
while I'd prefer a wiki like Gentoo's (the Ubuntu one is not as up to date), the forum HOWTOs
really are an example of open source in action, and incredibly useful.  

Overall, the huge influx of Windows users to Ubuntu is a good thing - it ensures that there's
a real focus on usability.  And being able to run a few programs from Windows is very useful
for all sorts of people, including businesses migrating to Linux - if you use Virtualbox or
VMware you won't have many problems, though WINE is often harder to get working.  

Oh, you mean like the DNS flaw should have been totally obvious to anyone implementing DNS
applications?

Regarding your part 1&2 bugs, they are indeed something that not everyone (and definitely not
people coming from Windows world) would mostly notice. Anyway you have some false information
and some stuff you think that should be otherwise without any technical reason other than "it
used to be like this on SUSE/etc.". Some, though few and pretty unimportant, I agree with and
would like to get fixed at some point.

- textmode install is called "alternate" CD, and is available:
http://releases.ubuntu.com/8.04.1/
- swap partition can be modified in manual partitioning, both on the desktop cd installation
and alternate - on alternate, it's what is default on Debian
- if you cannot get IP automatically from DHCP server, please while a bug - or did you just
expect a specific DHCP client and didn't get the one you wanted automatically?
- command-not-found is annoying to me too, but very useful for anyone new to Linux/Unix
console. you can remove it if you don't like it.
- kbd is obsolete (it's not in the supported main applications, even), I think you'd want
console-setup (haven't used myself, though)
- also annoyed by framebuffer forcing, probably will be fixed anyway when kernel-based
mode-setting comes; dunno what are the details behind the problem otherwise
- who said runlevels should be like you used to have them? there is no standard, and lots of
debate of what should be where. 2 is default in Ubuntu
- xterm upstream should probably have better defaults if they are bad, or you should contact
Debian maintainer of xterm otherwise
- file a bug about less if you think it's a bug
- lack of system-wide SCIM settings is known problem I think, being addressed hopefully for
8.10
- dunno about 64-bit userland, I think there used to be somewhat of a definition problem about
what's "proper way" to have it, of course you knew best all along since it's now like that in
Ubuntu too I guess
- bridges are again probably not for everyone, but if you don't like to use terminal for
anything openSUSE has indeed a lot more of these click-click admin tools, which is great for
those who want those

Anyway, if you think the problems you encountered really make Ubuntu "absolute fail" and you'd
even recommend WinXP over Ubuntu because of these (but not WinXP over openSUSE/Fedora), _you_
are indeed better served by other distros than Ubuntu, for now. But really those problems are
very specific, I dare to say, and some are just matter of opinion.

actually, I don't think that Ubuntu does 64 bit right according to what debian says should be
done (the entire multilib mess)

as for them being behind other distros, so what. you say that other distros did it right for
_years_ before ubuntu got it right, well, those other distros also took years to get it right,
ubuntu hasn't been around that long.


I would actually be as annoyed by lesspipe being the default for less as you are by the search
for missing commands. 

overall your list strikes me more as you considering your personal prefrences (and things taht
you are used to from other distros) as being the one and only way of doing things.

get out there are use more distros, use other versions of unix, learn that they all have
differences and that there is no 'one true way' that everyone agrees is best.

quote:
Canonical has made the strategic decision to include non open source technology in the Ubuntu
distribution if such inclusion can be justified on the grounds of accelerating linux adoption.
Their aim, therefore is to get linux adopted first and fully open sourced second (Mark will
argue that the second naturally flows from the first).

there are other distros that do this and don't catch nearly the flack that Ubuntu does, so
this doesn't explain it.

as for the rest of your post that says it's due to the "publish now, don't wait to get it
upstream" approach, there may be a little of that, but I don't think it's that much. Yes,
ubuntu is worse than some of the other distros, but it's a matter of degree, not of kind.
RedHat never caught this much flack and they were _far_ worse for a while. 

The "upstream first" approach is deemed to be an advantage for the distros as it lessens their
labor of maintaining patches going forward. If this is really a win then we should see Ubuntu
falling more in line with the other distros over time.