Fog Creek Software
Discussion Board




Technology Review on Linux

MIT's Technology Review is one of my favorite magazines, they're always talking about super-freaking cool technological and scientific stuff that is expected to emerge in the marketplace in the next 5 to 10 years or so.

The latest issue has some articles on Linux, one of them on its prospects in the desktop market.  Currently single-digit percentage market share, but gaining, apparently.

Since I'm ignorant on Linux, my question for y'all is:  What are the factors affecting how quickly Linux gains in the desktop market?  I'm suspecting the article missed some big points.  Such as:  peripheral devices always come with Windows drivers, but often not with a Linux equivalent ... isn't that a big issue?  And many ISV's don't target Linux ... such as many of the latest games.  And isn't a lot of stuff really hard to configure in Linux?  I've been developing Windows software for 11 years, but I'm having a hell of a time configuring my first Debian server at home.

Doesn't the market need to get to a tipping point where *many* vendors (software and hardware) target Linux with really user-friendly stuff, before it becomes a really compelling option?  If so, will we get to that tipping point, and when?  Or do I have it all wrong, and there is no such tipping point, or have we already passed it, or ...?

profound insights galore
Tuesday, August 24, 2004

I find that the biggest barrier to the mainstream growth of Linux is the dichotomy found in the members of the community.  It sometimes exists in the same individuals.

It is simply that they want to gain in the desktop market (kill Bill, so to speak), but they don't want to let Joe Knownothing into their fraternity of cryptic keystrokes and complex UI.

That's not to say everyone has the frat opinion, but some have a real sense of 'holier-than-thou' and if they made the system to appeal to the masses, they'd lose that.

Conspiracy Anti-Theorist
Tuesday, August 24, 2004

"Since I'm ignorant on Linux, my question for y'all is:  What are the factors affecting how quickly Linux gains in the desktop market?  I'm suspecting the article missed some big points.  Such as:  peripheral devices always come with Windows drivers, but often not with a Linux equivalent ... isn't that a big issue?"

Fuck yes it is. Device support is growing by leaps and bounds in Linux but I think that until vendors implement Linux drivers on their own it will never be able to be really competitive with Windows in this arena. Waiting for an OSS developer to decide to implement a driver for Device X is always going to cause a delay.

"And many ISV's don't target Linux ... such as many of the latest games."

True. I don't know that this is as much of an issue -- but then, I'm not much of a gamer. Certainly average desktop applications are not quite there yet with their Windows equivalents, and specialized applications (I ended up having to switch back from a pure-Linux system to a dual-boot system to do desktop publishing) are often not available at all.

"And isn't a lot of stuff really hard to configure in Linux?  I've been developing Windows software for 11 years, but I'm having a hell of a time configuring my first Debian server at home."

Yes and no. Theoretically, anyway, the things that the average desktop user might want to configure ought to be fairly easy; I don't think it's a huge problem that setting up a server is painful and difficult; something like that is expected to have a high learning curve and once you get over that learning curve I don't think a Linux server is any more painful to work with than a Windows server.

E. Naeher
Tuesday, August 24, 2004

"It is simply that they want to gain in the desktop market (kill Bill, so to speak), but they don't want to let Joe Knownothing into their fraternity of cryptic keystrokes and complex UI."

The only problem with this argument is that most development is being done by COMPANIES not individual hackers working on their own.  Perhaps the individual hackers don't want Joe Knownothing in their fraternity but business out to make money do.

Almost Anonymous
Tuesday, August 24, 2004

I think a better question is will the desktop/laptop computer be the average joe computing device of the future, so far it looks safe, but cell phones, pda's are gaining, as are in dash service for cars (who needs mapquest if you have GPS). Maybe there will be enough reason in the future to rent/subscribe to centrally administered computers (terminals ...), this is especially true if MSFT doen't stop with the embarrasing security flaws.

So, to get back to the original question if the desktop computer goes away, maybe there will be room for more OS's for every-day use, and afcourse if different vendors start using different OS's then interpreted/byte code style language will see more yes too. In this case linux market share will soar.

the artist formerly known as prince
Tuesday, August 24, 2004

We go through trends of breaking thing down and consolidating. 5 years ago I saw this prediction too... That everything would be a seperate machine dedicated to one purpose, but now we're seeing that people want consolidated things.

Why carry around an mp3 player, cell phone, pda, and video game system when you can get them all in one?

I don't think the computer will go away, but it will be continually redefined.

Anyone remember when you played video games on your computer and listend to music on your stereo? Now it seems to be the other way around, you play video games on a console and listen to music through your computer.

I saw a computer about a year or two ago that was supposed to act as a central organizing computer for the home. It would keep track of chores and shopping lists and things like this. I can't remember what it was called but it had a woman's name. This was definately not ready for prime time. Maybe when we have RFID in our homes there will be some use for this - it'll print out a shopping list for us without us even having to know we're running out of something.

Anyway, back to Linux. I don't know much about Linux, but here are my impressions as a consumer:

There are too many flavors for me to know which to get. (Red hat, Mandrake, etc.)

It will be impossible for me to configure it properly. I won't even know whether I have root or not, and if I have root, I may pose a security risk to myself.

Is installing programs difficult?

Are there enough good programs for it? I mean ones that aren't just for geeks.

Will I be able to do everything I want? Will it work with all those Windows Media Player / DivX videos so when I come across a video I can watch it?

Is the browser good enough that if I go to my bank's website it will render properly?

Will I ever need to go to the command line? Because I know very little about Linux commands.

Will my devices work? my iPod/mp3 player/printer/scanner/sound card/usb mouse/etc.

Will it take me longer to do what I want to do?

If I want to do my finances, does a Quicken/MS Money/Turbotax like program exist for it?

Why can't I just get a Mac and get the benefits of Linux with a wider install base, better support, and better programs?

Are the programs hard to use - i.e. assume you're a Linux Geek already?

etc.

www.MarkTAW.com
Tuesday, August 24, 2004

No. Er, I mean yes. Well, what I really mean is maybe...

anon-y-mous cow-ard
Tuesday, August 24, 2004

What we're seeing is that companies hoping to run businesses around Linux simply have to start doing the things that traditional software houses do. That is, make money.

As that happens, the little purist darlings get quite a shock. There are already quite a few cases of this. Nothing changes really.

.
Tuesday, August 24, 2004

I run several Linux desktops. The big headaches are:

1) Hardware, especially on laptops or for weird peripherals. If you're encountering hardware problems, you may want to try running a Knoppix CD on your system. If Knoppix boots, and your distro doesn't, your distro is junk (Debian suffered from this until fairly recently). If Knoppix doesn't boot, don't bother.

2) Application availability. For the major desktop apps, you may want to try a copy of CrossOver from CodeWeavers; it runs office tolerably well, and CodeWeavers has excellent technical support.

3) Immaturity. The improvement in the Linux desktop over the past five years is *tremendous*. Usability has improved by orders of magnitude; graphical admin tools have started to work; and the entire developer culture has improved. Still, a bunch of Unix geeks will need a while to grok UI, write lots of code, and polish the system. MacOS X has really impressed a lot of key developers, and--quite frankly--embarassed the living daylights out of a lot of people in the open source world. But improvement takes years.

Coporate desktops are a highly segmented market, and most Linux desktop companies are targetting a few key niches: developer workstations, move-editing systems, locked-down task stations, and so on. The desktop market is an extremely challenging market, and the Linux community will need to start in easier niches.

If you're a Windows developer, you should definitely look into Mono (or wxWidgets/Qt if you need cross-platform C++). This will cut your learning curve.

And let me repeat: If your distro doesn't work with your hardware, get a better distro or different hardware. Don't make the mistake of building a production server from Tier N semi-supported hardware and installing an older Debian on it; you'll hate yourself. Use Teir 1 and 2 hardware, and a distro with a decent hardware autoprober.

J. Random Hacker
Tuesday, August 24, 2004

Nobody serious in the Linux community minds making money, as such. Linus Torvalds seems pretty fond of it. Miguel de Icaza appears to be doing moderately well--he went from a university employee to a startup CTO to a VP of Novell. Even Richard Stallman is rumored to have usurious consulting rates. Every major open source project seems to have a couple of specialist consultants who integrate it--granted, not all these folks make money, but they certainly *want* to.

The only major disagreements I've seen among serious developers is whether it's better to make money through services or proprietary add-ons. Some Slashdot posters may whine, but who are they that I care?

J. Random Hacker
Tuesday, August 24, 2004

"MacOS X has really impressed a lot of key developers, and--quite frankly--embarassed the living daylights out of a lot of people in the open source world"

I think its safe to say Apple has been embarassing both the open source and Microsoft worlds for quite a while. But J. Random's point is right on the money. People have assumed forever that Unix is difficult to install and use. OSX proves that it can be easy, friendly, and beautiful. If you've seen a Mac recently, you've seen the future of Unix on the desktop.

Tom H
Tuesday, August 24, 2004

"People have assumed forever that Unix is difficult to install and use. OSX proves that it can be easy, friendly, and beautiful."

... When a major, stable, and visionary company is behind it.

www.MarkTAW.com
Wednesday, August 25, 2004

".. When a major, stable, and visionary company is behind it."

I couldn't agree more.  That is what a lot of the OSS community is missing.  OSS tends to lean to design by committee.  Like that old joke "a camel is a horse designed by a committee."  Some projects like Linux itself or Python for that manner have a leader that ultimately says what goes and what stays.

Part of the problem is that OSS is about stroking your ego.  So what if that one developer in your project insists the menu be layed out in a way that you know 99% of sane people would reject?  What if that person wrote some really good code elsewhere in the project that you need him to maintain? 

Having a leader with full control take charge of a lot of these projects would do them worlds of good.  This is one area where proprietary software tends to be much better off.  "No, marketing says even though your proprosed name with a clever recursive acronym pleases you, it baffles our intended customer base.  So the name is X instead."

Mike
Wednesday, August 25, 2004

+1 for  J. Random Hacker point number three.

OffMyMeds
Wednesday, August 25, 2004

"Even Richard Stallman is rumored to have usurious consulting rates."

RMS is engaging in Capitalism?!?!

*Oh the scandal*

Disillusioned Lemming
Wednesday, August 25, 2004

on the marketing brocure "...and for a few dollars more, Richard will show up at your company in the wizard outfit."

Mike
Wednesday, August 25, 2004

For me, once Linux becomes mainstream it will be ruined.  The one thing I don't like about Windows is how it tries to all things to all people.  I personally use Slackware and don't want a product built for the masses but the computing elite!

Bill Rushmore
Wednesday, August 25, 2004

> What if that person wrote some really good code elsewhere in the project that you need him to maintain?

If it is really good code then anyone can maintain it.


Wednesday, August 25, 2004

"OSS tends to lean to design by committee... This is one area where proprietary software tends to be much better off."

My recent experiences make me think this is less and less true. Although primarily a Windows user/developer, I've installed various linux distributions out of curiosity over the last few years. This weekend I put Fedora Core 2 on a spare machine. It's default desktop is Gnome 2.6, and I must say this is leaps and bounds over what Linux users have had to suffer in the past. If you stick to good gnome apps, it's  better designed and *consistent* than any windows desktop. The Gnome designers, precisely because they don't have to pander to a conventional market, can insist that everything is done right, even when this confounds some current gnome users' expectations (check out the fight over the 'spatial nautilus' change in 2.6).

It's still a bit difficult for the average user because of the applicaiton installation issue. But once there's a good GUI for yum (the main package manager for fedora) included, it'll be nearly there. I've actually started using it for my daily stuff at home. Am having a whale of a time with apache/python/webware, and with Mono.

Noisyjazzman
Wednesday, September 01, 2004

*  Recent Topics

*  Fog Creek Home