Fog Creek Software
Discussion Board

what does it take for linux to beat ms?

What do you think?

Improvement in OS --> of what kind?
dev tools --> how to compete with .Net
applications --> at minimum what should be available for mass acceptance ?
pricing -->
distribution -->
etc  -->

Tuesday, August 5, 2003

Beat MS in what field?

I already prefer developing under Linux, but that's partly because the free tools I've found are better. I have seen VS.NET and seen that it has some nifty things, but that's also a $1K a year MSDN subscription ;) (not that you can't get the product w/o it, but you see my point nonetheless, I hope).

Some would argue that in network server arena linux *has* beaten MS. I wouldn't, but I wouldn't argue the reverse, either.

"Computing" is a big field. Are you talking about me writing drivers for custom hardware for extremely high-end multiprocessor machines? Or are you talking about PIM on the Ipaq? Linux can *run* on both environments, but that doesn't mean that beating MS in one beats the other 8-}

Mike Swieton
Tuesday, August 5, 2003

I don't think this question makes sense.  The press and some guys on /. have portrayed Linux vs. Microsoft as some mindless wrestling match.

What is Linux?  Are we talking about the Gnu system?  Their goal is to provide everyone with safe, free computing.  Microsoft may have helped, if you believe they drove down PC hardware prices and drove HW manufacturers to be more respectful of standards.

rohan j.s.
Wednesday, August 6, 2003

Well at least in market share -- especially in desktop OS distribution. I believe combination of those I mentioned above has to achieve certain level of competitiveness somehow with MS product to achieve greater desktop OS market share.

Wednesday, August 6, 2003

A valid reason for users to change over to it. Currently there isn't one.

Wednesday, August 6, 2003

I may be being unfair, but here's a thought:  it has to freakin' work!  Don't know if I've mentioned this before, but I bought a beater box last summer in order to install Linux and start learning the environment (after years as a VB programmer, I thought it wise to branch out).  Installed the then-latest distros from Red Hat, Debian, *and* Mandrake -- and in all cases, some critical piece of software was fubared:  in RH's case, GNOME would crash when I tried to run it, and Samba utterly refused to talk to my Win2k box on all three distros.

I'm now very happy with my iBook (on which Samba runs without incident, I might add).

My case might have been exceptional (and I haven't bothered playing with it for a year) -- but it shouldn't have happened at all.

More room for improvement:  I don't want to have to learn how to use make in order to install software, and I don't think it's too much to ask for a GUI that isn't fugly.

Now, where did I put that asbestos underwear?

Sam Livingston-Gray
Wednesday, August 6, 2003

For Linux to even compete with Microsoft on the desktop, the linux community needs to pull their collective heads out of their asses and start being salesmen instead of arrogant jackasses who shut down as soon as you mention Windows or Microsoft.

Text file configuration, access to kernel source code, twiddling with iptables and routing - when these are buried deep in "administrative tools" and merely a curiosity for 99% of the users, then Linux might have a chance.

One way to measure its progress - pick some basic tasks and search for how to do them on the web. See how hard it is to find howto's on those tasks assuming zero linux admin knowledge. When any search turns up a good howto that could take a newbie linux user through making it work (indicating that the Linux community is buying into the idea that you *don't* have to know linux inside and out to use it), then linux can START competing for public market share.


Wednesday, August 6, 2003


A study of the end-user useability of both Windows XP and Linux with KDE as a desktop environment can be found here:,10801,83708,00.html?nas=AM-83708

Decide for yourself how relevant this study is (I don't think it has a lot of value), but it is interesting to see that Windows XP with years of useability testing, consumer panels and a lot of development mony is found to be only marginally more usable than Linux/KDE.....

Wednesday, August 6, 2003

Linux will never beat MS on the desktop because Linux is crap.  The only reason Linux is "on the radar" is because of the sheer vacuum created by the lack of a competitor to Windows and the infantile hackers who fantasize about "taking down" Bill Gates.

Put it this way, if there was a real competitive third party OS, Linux would never have achieved the prominence it has. It would be a joke.

The question should be, will their ever be a REAL competitor to MS?

Wednesday, August 6, 2003

You know, when IBM originally licensed the OS, most people thought it was the hardware that made the world go round.

Billions of dollars later the market realized it was the (Microsoft) OS.

Now, with a free and stable OS out there (Linux), it might just be that instead of the OS, it's the office suite (spreadsheet, word processor, etc.) that makes the world go round; an set of programs that allow you to exchange documents between systems and still allow them to be opened and editable.

Maybe Microsoft thinks the same way. So when and if they release MS Office for Linux, it might just be the thing to kill off Windows.

My two cents....

Chi Lambda
Wednesday, August 6, 2003

so linux_sux, if you could recommend something to the Linux community to improve its market share (i.e not sux) what would you recommend?

Wednesday, August 6, 2003

I think their focus right now should be on the office desktop usage. They should provide comparable (in functionality and features) basic applications, but at cheaper cost of ownership. This will enable common business adoption, maybe starting not from US but from Europe or Asia. Once it become widespread enough for business use then hopefully it will create snowball effect to create viable reason to develop other kind of software, that in turn will enable more specific kind of business (e.g engineering company) to use it. If enough people at work use it, I believe then personal/home use will come later.

Wednesday, August 6, 2003

Re the orginal question - for Linux to be comparable to MS, for starters the Linux community needs to stop giving away everything for free. Seriously. I think this is very counterintuitive to the way the Linux and open source community works.

First, no one takes free products seriously. Not in the business sense. If something is free, it gives the impression that it is not something that will be supported in the long run. It's also not a business model that makes sense to anyone but those in the open source world. (I know, I know, open source doesn't mean free as in beer, but essentially it's worked out that way.) I'm not talking about the kernel or the few "golden apps" out there such as Apache. I'm talking about all the applications that come packaged with the typical distro.

Second, if a company is charging for something, they have revenue to pump back into product development. Perhaps Eric Sink could chime in on this one. I don't know if AbiWord was done for the benefit of mankind or if it actually makes his company money.  If so, how?

Having money would enable companies to make all the other improvements needed to make Linux compete, but that list is too long to publish here.

Wednesday, August 6, 2003

Drop the GPL cult surrounding it.
Let me explain: The GPL is compatible with any business model imaginable but one: Creating high quality, highly usable software for a wide audience.
There are those who say the GPL is anti-capitalist. No it isn't. As far as the GPL is concerned you can lock in your users to forking over heaps of cash for any instance of your software running in any way you want as long as it is done through complements: complementary hardware sales, support contract licences, training contracts, whatever. It is also not against you charging through the roof for specialised applications for small, especially single client or small closed user group niche markets, be it private or public sector. As far as the GPL is concerned you can become a billionaire running software sweatshops using these models.

The only software business model that is going to be in trouble is a pure software business model. One that does not rely on the software being a loss-leader to some complement. So if you are going to make software that is so good that it does not need much training or support, runs on non proprietary hardware, and that does something usefull for a large group of users, so that all would pay a small amount for a licence, more or less dependign on your usage, then that does not gell with the GPL.

Now I believe that the pure software business model is the best for producing the highest quality horizontal market products, since its revenue maximisation process is alligened with the benefits to its users.

The optimist in me says that to beat MS, the OSS crowd has to drop the "sofware as a complement only" business models. This will give them a focus to produce good software.
The pessimist in me says that maybe the "software as a complement only" can win ad in doing so will destroy the "pure software play" market. Then I believe we will all be very much worse off.

Just me (Sir to you)
Wednesday, August 6, 2003

Beat? Its not like either will sudddenly disappear just because the other get a certain marketshare.
What would it take for a Linux OS to get a significat market share on the desktop then... A BETTER WAY OF DEALING WITH LIBS AND DEPENDANCIES.

Thats like the only thing that truly sucks about linux. You try to install something, and discorves that it requires, so you install that an find out that it requires GCC3.3, so you upgrade and discover that it requires... you get the picture.

Now, I can manage this, but im not scared of filenames that are 54 chars long and mostly numbers, or of creative googling. But for many this is like a brick wall.

Eric Debois
Wednesday, August 6, 2003

> but it is interesting to see that Windows XP
> with years of useability testing, consumer
> panels and a lot of development mony is
> found to be only marginally more usable
> than Linux/KDE.....

MS poured a lot of money into GUI research, and really made excellent an excellent product in this area: Windows.

Now, the problem is that once the research is done, everybody can look at Windows and implement things just the same way, without doing any research themselves.

So, KDE got where it is now by imitating Windows.

Wednesday, August 6, 2003

Now *this* is a topic where I'd really like to see the JoS-vs-Slashdot death-match :)

> Now, the problem is that once the research is done, everybody can look at Windows and implement things just the same way, without doing any research themselves.

> KDE got where it is now by imitating Windows.

I guess we can all decide for ourselves how much of a "problem" this is. IMO, it's mainly a problem for Microsoft.

I know it outrages you guys, but many people will choose Linux simply because they are worried (and justifiably, IMO) about what being beholden to Microsoft will mean. If the thread about Linux in Munich is accurate, this is essentially what happened there.

There is nothing dark or mysterious about this; it's always better for a customer to have two or more suppliers for a commodity than to have only one. Thus, there will be market pressure from customers to have an identical OS with multiple suppliers, the moreso since this goal now seems achievable.

Philo, I don't think your solution ("start being salesmen") is as simple as it sounds, and it reeks of programmers-attack-other-programmers hostility. I think what's happening is that Linux is hugely successful because it has been able to marshal the resources of thousands of programmers to work essentially for free (initially, anyway). Suggesting that those same developers turn into salespeople is just not going to happen.

What usually happens in tech at this point is that a Sylvester McMonkey McBean character comes along, and figures out how to polish up the tech so that it's salable. This is in fact happening; there's a steady progression in usability, quite a bit of it coming from the distro companies. Perhaps it's just from "awful" to "not too bad", nonetheless it's quite real, and continuing. If you've installed successive versions of Red Hat or Mandrake, you know what I mean.

Finally, you all seem to be a bit horrified at the idea that customers will, or may, pick something that's "worse". This is a classic sign that customers have different priorities than you do.  I'd say this: customers can't be fooled about their interests for long. If you're right, and they're being fooled; then they'll come back; if you're wrong, well you're wrong. And if you're half-right, that is, Linux is worse at the moment, but evolves so that it's not worth switching back, then that too serves the customers.

Peter Breton
Wednesday, August 6, 2003

Better drivers!!!!!!!!!!!!!!

Linux's driver model (or complete lack thereof) really sucks.

The drivers have no real separation from the kernel.  It's almost impossible to release a Linux driver without giving out the source code.  Many hardware makers (myself included) do not want to do this.  Exposing the driver source code exposes a lot about the inner workings of the hardware.  This is not a good thing.  Why should I give my designs away to my competitors for free?

Even then, source is not enough, if it's not a GPLed driver, the kernel gets mad.  It throws a fit about how it's been "tainted".  There's even been talk of dropping support for all non-GPLed drivers.

I can't even release a binary-only driver that will work on RedHat, Debian, and Suse without a recompile!

This makes a lot of hardware vendors unwilling/reluctant to support Linux.  Most Linux drivers were cobbled together by people in the open source community.  They write them by taking guesses about how the hardware was built (a chip's data sheet is not enough).  These drivers are lower quality, and lower perfromance.  Take a look some time through the kernel source...  You keeps seeing comments like "this hardware has a bug, I can't make this work"...  Most of those "bugs" aren't really bugs, it's just that the open-source developer didn't understand, and because he didn't have access to the hardware engineer, he assumed it was a bug and went on.

Linux needs a decent driver model.  It doesn't have to be super-complicated, but it has to be standard.  It needs to support the possibility of binary-only kernel drivers.  When this happens, hardware vendors will be able to write their own drivers, which means they won't suck anymore (or suck less).

Myron A. Semack
Wednesday, August 6, 2003

I agree with Myron, but I'd like to take it a step further.  My company also makes hardware, and we've wanted to provide a driver/API under linux for some time now.  We recently started looking at it more seriously, but found that the driver model changed so frequently between kernel revisions -- without documentation of what had changed! -- that it made keeping a driver working a very time (and money) -consuming task, whether the source is supplied or not.  We've stopped deveopment on this for the time being.

This is mostly speculation, but we perceive that the problem lies in the lack of a strong management structure.  Projects need a persistent vision.  If the driver structure and related calls keep changing, that says to me that somebody can't decide what they want.  Where's the focus and leadership?

Internally we dump on Microsoft for changing all kinds of things between releases of Windows.  We can deal with it, however; releases with such changes are usually more than a year apart.  With the Linux kernel, which has a much greater release frequency, this becomes a very time-consuming issue.

Wednesday, August 6, 2003

What does it matter? Linux isn't out to beat ms. Linux is what it is, an OS. Some people will use it for ideological reasons and some won't use it for ideological reasons. For all the rest of us normal people Linux is better in some situations than ms, in others its not. I think that in the future the number of situations where Linux is a better choice will increase but ms will always be there.  I wouldn't like to replace one OS monoculture with another, and anyway, the value is, or should be, in the applications that get useful work done rather than in the OS that runs those apps.

Alex Moffat
Wednesday, August 6, 2003

I'd really like to see someone speak of the benefits of Linux without using the word "Microsoft" anywhere in their argument.

Being able to do that will win a tremendous amount of market share, IMO.

Jeff MacDonald
Wednesday, August 6, 2003

"Philo, I don't think your solution ("start being salesmen") is as simple as it sounds, and it reeks of programmers-attack-other-programmers hostility"

Maybe. It's more out of "a programmer tired of being attacked by other programmers" hostility. If I go to a Linux newsgroup, mailing list, or chatroom for help, the users there should be falling all over themselves to help me, and *recognize* that people from the Windows world are going to be frustrated.

Last time around, when I tried to get a wireless NIC running in Linux, I was asking for help in a chatroom. I made the offhand comment "heh, you know in Windows I'd be done by now"

That was the end of the help. It then became an anti-MS flamefest.

Now I've *been* in meetings with pro-MS people (including a MS consultant) and made the comment "I know how to do it in Linux" and we laughed and went back to what we were discussing.

That's the difference. That is what needs to change, especially when the linux support community is supposed to be a selling point for the OS.


Wednesday, August 6, 2003

> That was the end of the help. It then became an anti-MS flamefest.

Uh, so you insulted the tribal dieties and then you were surprised when they stuck you in the cooking pot?

And you've seen this behavior before, but you went ahead and did it anyway?

Think about it.

I don't want to defend whoever it was that you were talking to. But your anecdote doesn't seem to acknowledge that people who are being *paid to use tech*, like your MS consultant, act very differently than people who use it because they love it.

And as the programmers-attack-programmers thread that I referred to above notes, there's a lot of hostility in the tech world; it's by no means confined to the rabid Linux zealots.

Let me pay Microsoft a compliment, and note that their OSes (Win2K and XP) are much, much stabler than their predecessors. But in the old days, you could go to an NT shop and talk about how your OS never crashes, and watch everyone bristle. Or worse. More tribal dieties.

Finally, the Linux resources on the Web are mostly valuable to other geeks.  The "supposed to be a selling point" stuff is kinda silly; as a business, you can always pay for support, and it will be free from MS flamefests. As a Home user, I think the grim truth is that you are simply stuck if the easy-to-use install breaks. That's undeniably worse in Linux, but all-too-common in the Microsoft world as well.  Didn't another thread talk about the boom in "support services for non-technical users"? What OS do you think these people are using?

Peter Breton
Wednesday, August 6, 2003

There seem to be a few points that a lot of people are missing.

First of all, on the presence of quality software on Linux: there is plenty of it, and the sectors that suck are improving at a much faster rate than MS software is. In the server field, Apache, not IIS, is *the* webserver to beat. Databases on Linux have come so far that even papers like the Wall Street Journal have stories about the advantages of MySQL vs overpriced solutions like Oracle.

On the Desktop, Gnome and KDE have made huge strides in the last couple of years. Gnome in particular when tweaked by aces like the folks at Ximian looks and runs great. Imagine how far things will come in the next 5 years. Speaking of Ximian, who of you have tried Evolution? I challenge any of you to actually claim that you prefer MS Outlook to Evolution - and Evolution is getting better and better. The Gimp is a great photoshop replacement - which does 90% of the stuff people use photoshop for. Furthermore, If you really need photoshop, you can now run it very well via Wine, thanks to work done by Disney.

OpenOffice is rapidly progressing as a valid MS Office alternative: the folks at Ximian have done a bunch of work on the UI and its starting to look really good. As far as browsers go, Mozilla Firebird runs on linux - enough said. Going back to Ximian once again, the Mono project holds great promise - within a couple years all .NET software will run on Linux.

Regarding issues concerning drivers and library handling: the Linux driver model is pretty piss poor right now and needs improvement. If Linux could get something going like what is available in Darwin life would be much better - the problems with tainting etc are a pain in the ass as well. I do reject the claim however that Drivers on Linux are better than Windows drivers. On Linux we get printer drivers that don't tell us we're out of ink way before we really are. Furthermore, drivers in a lot of sectors [for example Wlan cards] are of a much higher quality than the crap a lot of companies (DLink, Linksys) have been pumping out lately.

Regarding Libraries: On linux, if you don't like dealing with bad lib versions, you can always statically link. Furthermore, the use of a good distribution / package managment system (Redhat or Ximian Redcarpet or Gentoo for instance) should prevent any problems with library versions.

So what will it take for linux to beat MS? Time. Slowly but surely, the Linux user base will grow. Unless MS really starts to innovate (when was the last time you found an MS program innovative compared to software by, say, Apple?) software on Linux will catch up and corporations and governments (MS's biggest customers) will begin to transition for economic and political reasons. After that occurs it is only a matter of time until the private market follows.

Andrew Murray
Wednesday, August 6, 2003

Peter, my point was that a huge part of the case that Linux advocates make for Linux is that there is a huge support community available online, so you don't *need* a single company behind it - that's how they say they can compete with Microsoft.

Well, if that's their pitch, then they have to put their money where their mouths are, bite back the anti-MS venom, and help clueless newbies when they come looking for help.

As I keep saying in other threads - I think that is the #1 thing holding Linux back. (there may be others, but until that changes, Linux will never be anything more than a hacker's playtoy and something governments inflict on their users)


Wednesday, August 6, 2003


While the distribution is not the most newbie-friendly, check out the forums at for a good example of the Linux Community at work. Check out the free documentation at the site. Furthermore, check out the Linux Documentation Project at .  Any newbie with sufficient interest can pick up a gentoo disk, walk through the installation and with the help of the folks at gentoo have a great system up in running and learn a whole bunch about Linux in the process. This is just one out of a bunch very helpful Linux communities out there.

The Linux community is very receptive to newbies, we want them. The Slashdot community is really a wacko subset of the Linux community as a whole.

Andrew Murray
Wednesday, August 6, 2003

The one thing Linux needs most to win on the desktop is something it will never have: consistency.

Linux users have to open an app and have it function in an expected way.  Drag and drop amongst all their applications must work identically, the big applications must follow user interface guidelines, etc.

In a community where there are two major desktop initiatives, hundreds if not thousands of window managers, and applications which use many different types of widget libraries this won't happen any time soon, if ever.  And the biggest problem is that the existing Linux users pride themselves on this 'choice'.  It is that 'choice' that will doom them on the desktop.

Mister Fancypants
Wednesday, August 6, 2003

> As I keep saying in other threads - I think that is the #1 > thing holding Linux back.

I agree with you that usability is a major problem, but I'd emphasize a few things differently:

1. It's endemic in the computer biz. The computer is one giant leaky abstraction. I know it's not fashionable to say this, but this may just be the nature of the beast. When everyone had horses, there were lots more horse trainers, grooms, and people with shovels.

Linux should be properly chastised for falling below the usability standards set by Windows and the Macintosh, but let's recognize that those standards are actually pretty low.

2.  I'm not sure usability is actually the barrier to adoption. You joke about inflicting Linux on users, but that is exactly what will happen when the economics are right. (And personally I can't see it as such a bad thing, when I remember how Windows 3.1 and DOS were inflicted on users).

3. Linux competes by harnessing the energies of lots of geeks (although now some of them are very well paid). To keep momentum, there will be more technical innovations, which means more leaky abstractions, and ultimately more things breaking.

I hope to see more it-just-works in Linux, but I'm wondering how many things are going to break on the road from here to there. In short, I think the rabid geeks will be with us for quite a while.

> It is that 'choice' that will doom them on the desktop.

Perhaps the bigger distros will mandate guidelines, and eventually just select the apps that follow them. (In fact, I think they already do this to some degree).

The word "doom" struck me as funny here, and I flashed on the bogus Malthusian prophecy that London would suffocate under piles of horse manure, given an exponential increase in the number of horses.

Why not reason the other way:

1. If there are powerful economic forces pushing Linux AND
2. The all-over-the-place GUI chaos is THE major thing holding Linux back AND
3. Random people on Weblogs are smart enough to figure this out

Then isn't someone smart enough to solve this?  *Especially* when there isn't one company sitting on the problem?


Peter Breton
Wednesday, August 6, 2003

I installed Mandrake 8.1 with much less difficulty than I had installing W2K on the same machine. And loads of apps as well, at one go. Probably a quarter of the time that it took to install W2K and Office and assorted utilities.

Stephen Jones
Thursday, August 7, 2003

*  Recent Topics

*  Fog Creek Home