If Valve discovers that you can get more performance in general out of a Linux rig, my guess is that it won't take long before the entire hardcore gaming community runs Linux. That community community is crazy about performance.
In Valve's experiments so far, Linux gets better performance (higher FPS) than Windows. We're talking 280 vs 315 FPS (or something like that), but that's still better.
"Its great that people are enthusiastic about Linux as a gaming platform but there are not many people who are interested in paying for a game and that seems to be the reality."
I might add that in all humble indie bundles linux users have paid the highest. To be fair, I don't think linux right now is a prime target for games but I do believe that it will be.
Humble Bundle typically features a set of 5-8 well known and high quality products. Windows users pay roughly $1 per product and Linux users pay $1.50 per product. Individual devs for a wildly popular bundle may see as much as $50,000 from Linux users. It is yet to be seen how many Linux users are willing to pay full price for new releases.
Linux will be prime time if Valve releases their long rumored Linux console box. Until then the year of Linux will always be "next year" just like always.
The demise of Loki was informative for a variety of reasons none of which should lead you to conclude "how few Linux users will buy full priced games". The first lesson to gather from Loki was that in order to get people to buy your games, you need a solid well-oiled distribution network. Stocking a few titles in Electronic Boutique is not a recipe for success. Steam is different as it puts Linux titles on a level playing field. I don't have to schlep to a store and hope that they will have the title I'm looking for as I can see it right there on Steam. And I don't have to worry about a store stocking just the big name stuff as Steam will stock it all and all the time.
Another issue that Loki contended with was the fact that most of their games ran at a lower framerate than the Windows counterpart. Most serious gaming fans want to use their hardware to its maximum potential especially the FPS multi-player twitch gamers. Valve has a lot more leverage in this regard than Loki ever did. We all know (because they said so) they are working with the major GPU players to get better drivers made. But also, they don't just port games, they make them. The Source engine is very popular and they have already tweaked it significantly so not only is it competitive with how it runs on Windows but it actually surpasses it in maximum frames per second.
Another thing Loki dealt with was Linux users were a bit more idealistic in the 90's. We still held out hope that AAA game titles would just "appear" on Linux. Fast forward to the present day and I doubt many actually believe this. The problem back then was that if you expected Free top tier games to just appear you had less incentive to actually buy. I think Linux users are a bit more pragmatic now and are much more willing to pay for a proprietary binary as it's pretty clear that that's the only way you're going to play. So this is a psychological barrier Valve doesn't have to overcome anymore.
Basically, I think Valve is approaching this with the same methodical precision that Tesla is approaching the electric car space. Look at the problems that exist in an objective way and devise solutions one by one until you achieve success. Valve may still falter but it won't be for ignoring any lessons Loki could have provided.
He pretty clearly was talking about their experience with QuakeLive there. Many people have said a lot of times that the Humble Bundles cannot be taken seriously if you think about Linux and its users paying for games. The bundles have a hype around them, people try to make Linux look good, etc.
I know they paid the highest, but how many of them were there? That's the key metric he's talking about. Was Linux 10% of Humble Bundle sales? 20%? What was it?
Eyeballing H.I.B.'s revenues by platform, they were more like 25%, which isn't bad at all.
Even though I prefer Windows (use & dev), I welcome cross-platform compatibility - more competition, more likely to be usable on future platforms, not locked out myself when I switch platforms, etc.
I've been wondering for a while how much of that is due to the public chart of Linux/OS X/Windows donations. Does anyone know if Linux users donate at the same amount when that information isn't made available? (ie has there been a sale where you can't see donations broken down by OS during the sale, but the figures get released later?)
The effort Valve seems to be putting into bringing the platform up to speed, including hiring a lot of Linux talent and cooperating with driver vendors, seems to imply that there is a (hidden?) agenda here that goes beyond selling a few game licenses to the set of today's Linux desktop users.
(Especially when you consider the recent gaben statements about win8 being a "catastrophe", and the windows app store stepping on steam's toes)
I think Carmack's response misses the mark. Valve ain't targeting Linux as just yet-another-platform.
Make no mistake: Steam and the App store are competitors for selling digital applications over the Internet.
It tomorrow MS and Apple "force" people to pay using their App store, they need some alternative.
MS was a great competitor destroyer. You will compete with them, for example with compilers, like Borland or Lotus or Wordperfect or Netscape and they will destroy you first going against your income generators and they will care about the product later, when there is a monopoly.
MS could destroy steam just using their OS control to make steam perform badly, they did it in the past with companies like DR-DOS, with media encoders-decoders(making them use code that they slowed down on purpose so their badly programmed software looked better) and so on.
Also, with their next "flagship" title, Dota 2, it seems like they're going the free-to-play route anyway, so people who are willing to play for the game itself is going to be irrelevant.
Valve doesn't mind investing to prime the pump. Doubly so if Microsoft looks threatening and they're working on their own hardware (which is likely linux based).
Surprising that Carmack doesn't support their long view, considering his history of technology advocacy.
One of these companies needs to put their weight into getting OEMs to pre-install Linux.
Imagine if you had to install the OS onto a Playstation or Xbox before you could play a game? No wonder nobody makes money on Linux.
Notice that nobody has a problem with Linux on their phone, because it comes pre-installed. If Valve can strong arm a few OEMs to preinstall Linux on desktops and laptops (against strong MS opposition), they stand to make a lot of money and simultaneously establish the fabled "year of the Linux desktop".
That there are very few preinstalled Linux options is the reason it hasn't taken off commercially. It's the reason Linux drivers are a huge problem. It is why "grandma" can't use Linux. And it's why John Carmack can't make money.
> One of these companies needs to put their weight into getting OEMs to pre-install Linux.
One option would be for Valve to buy an OEM. Many of them are not doing very well and things are only going to get worse with Microsoft Surface.
This would give Valve hardware talent for designing the Steam Box as well. The Steam Box could easily be just a set-top PC running Ubuntu. Game makers will no doubt see the appeal of that, and start porting games to run on the Steam Box, and therefore Linux.
> One of these companies needs to put their weight into getting OEMs to pre-install Linux.
Several OEM's do just that, but nobody is buying.
> Notice that nobody has a problem with Linux on their phone, because it comes pre-installed.
No, nobody has a problem with these because Android has a good UI on top of Linux. Desktop Linux never came close to achieving that goal, hence its repeated failure to break into the mainstream.
>Several OEM's do just that, but nobody is buying.
Who? Dell has one 11" Ubuntu laptop on their website, and you have to try really hard to find it. System 76 has no marketing budget whatsoever. Anybody else? Have you ever seen a Linux machine in stores? Ever?
And the Android UI is okay for a phone, but I greatly prefer Awesome on my laptop. Mint, in my experience, has been really great for newbies. I just don't buy the UI argument. Windows baffles all the newbies I know. The "UI" problems with Linux have to do with driver support, and lack of a few big name pieces of software, that's about it.
Well, that and way too much choice. There's no "Linux UI", unless you're talking about bash or something. There's Gnome, KDE, etc. But one or a few OEMs settling on a standard desktop environment would go a long way towards standardizing the user experience, while still allowing us nerds to geek out on different window managers.
> That there are very few preinstalled Linux options is the reason it hasn't taken off commercially.
There's a lot of truth in there, but also some complex reasons why that hasn't yet come about. A fairly large part of the problem is that Linux has attracted a lot of miserly people. They want free (as in beer) stuff, and are running on older hardware. The idea of paying for software is anathema to them because they're hoping someone will give it to them for free instead, and they're willing to hold out for that. There are others who pick Linux for technical reasons, or for free (as in speech) issues, but there are enough of the cheap bastards to skew the market, unfortunately.
If you could magically move the market so that OEMs offered Linux, and companies provided Linux versions of their software, then I think a lot of what you say would be spot on. But getting there from where things are currently, now that's tricky.
You're right that the current Linux market is quite different from the market as a whole. But a great killer app like Steam could bring a lot of "normals" over from other platforms and change that.
It's basically a chicken and egg problem that needs some established company to make a big bet on Linux to initiate change. That's why I think it could pay off big time. If Valve could draw even a tiny fraction of Windows gamers over to Linux it could kickstart a huge migration away from Microsoft over the next several years, with Valve at the center.
There is no such thing as a Linux gamer. There is no such thing as a Windows gamer. Gamers will go to where the games are. Many (most?) pc gamers build their own systems. Without free OEM Windows preinstalled, and no Office lock-in, pc gamers are a perfect beachhead for Linux. Carmack, with a limited portfolio, can't make it happen, but Valve has the library to bring gamers with them.
I still can't get past how bad package management is in linux. Yesterday I installed the latest version of Ubuntu and for example I wanted to put on MonoDevelop. Looks like only 2.8 is available for Ubuntu while 3.5 is the latest. On the MonoDevelop download page there are installers for Windows, Mac, and 4 different Linux varieties. And that was only the beginning of my problems. Yesterday was not a good day.
Your issue isn't with package management which is much, much better than OS X. Try the following three very common tasks on both platforms - list all packages installed, upgrade a package, and determine which package a file belongs to. This is trivial on both the dpkg and rpm based distributions - but I don't believe there is any general way of doing this on OS X. If the App Store takes off, that might change, but right now you are out of luck.
Your actual issue is that nobody has either (A) packaged the latest version for you or (B) placed the package in your (likely conservative) repository. You can always install a package without a repository, or, add additional repositories to your system.
I think the issue here is precisely the concept of package management, as well as the currently used implementations on Linux. Package management is curated by people not involved in the original source packages, and the maintainers of the original open source projects do not do releases into the package management repositories.
And due to the proliferation of package management systems, choosing the right packages from a project's web page is quite confusing. In addition, rpm and dpkg have absolutely horrid command line interfaces, and the documentation is about as confusing as bash's man page. These are not trivial or easy to understand tools, necessitating further layers of abstraction like yum and apt. However, even these "simple" interfaces expose a huge amount of unnecessary complexity. If you disagree, look at the number of command line options in the rpm and dpkg utilities, and the strange and unexpected way that these command line options interact. It's not a clean system at its core. Suppose I want to do something extremely simple such as list all the packages installed on my system, how many words of the man page do I have to read and re-read before I can discover this interface? These tools share none of the simplicity of original Unix utilities, and are large monolithic beasts rather than composable units.
Linux package management is truly awesome, but that doesn't mean that it doesn't have faults and that it can't be improved.
apt-* and yum are only minorly better than rpm and dpkg, and are still ridiculously over-complicated. The proliferation of extra tools to deal with package management are a testament to the number of people that are unhappy with the current options.
I guess I really don't feel you. 'yum localinstall' is absurdly simple as far as I am concerned, and the standard 'yum install' is obviously even simpler. I am not even aware of any additional arguments localinstall takes. If there are any, I have never used them.
Edit: I just checked the manpage, and it seems 'localinstall' is deprecated. Simply using 'install' now does what it used to, though 'localinstall' still works for backwards compatibility. It really could not become any simpler.
Complaining that 'rpm' is complicated is like using git without the porcelain then complaining that it is complicated.
I disagree that linux is much better than OSX in this regard. Homebrew is really amazing, and while it can't help you configure an entire system, it is pretty compelling.
It does two things very well that ubuntu's apt-get doesn't:
1) consistently has the latest versions of your software
2) handles programming languages very well (that is to say, it makes you use rvm, pip, cpan, etc, instead of packaging _any_ libraries).
Those are not ideal for everybody or every system, but it's a damn fine experience for developers' machines.
I love homebrew - but that isn't OS X. It's a third-party package manager. And is also useless for identifying files you install from the App Store. For better or worse, on Ubuntu, my packages are tracked through dpkg (and, 95% of the time, I just do a sudo apt-get install) - contrawise, on an RHEL system, we've all agreed to track our package with rpm (or sudo yum install).
No such universally agreed upon metaphor exists on OS X. They didn't provide a universal package management mechanism. In many cases, you just drag the package into your Applications directory and it's "Installed."
Right, but I think this is why it's great. Apt has a lot of disadvantages because it's the only option. By contrast, OSX had fink, and macports, and now homebrew. If Fink had been designated the winner back then, there would have been no competition, and homebrew would never have come out.
Well - I would have preferred a "good enough" package management system to the "best" system. Also - Fink, too, wasn't OS X - yet another third party system.
There is some room for improvement, in the early days, RHEL had a horrendous repository-interface system called "up2date" which was anything but. I think a lot of the uptake in debian systems, such as ubuntu, was that apt-get was so much better than up2date - which seemed to have never been updated after the initial release. Thankfully, yum came about, and now RPM based systems have basic parity with DEB system in terms of package retrieval.
The Net-Net though, is that Linux package management is fairly well defined, but hop onto any OS X system, and try and find out what package a random file belongs to, is basically hopeless.
I'm not bothered by the fact that it isn't built in. What matters is the quality - ubuntu package management can be poor in some areas, and those are areas at which homebrew excels.
Also, homebrew has a significantly easier project - OSX+homebrew is almost solely aimed at devs wanted a good experience around their dev environment. By contrast, ubuntu's packaging relies on it doing everything, making it suck for some things.
3.0.3.5 is the latest. 3.0 only came out about a month ago so it's not too surprising that it hasn't made it to the official Ubuntu repos yet. 2.8 just came out 4 months ago so it's not exactly old.
There's always a tradeoff when it comes to using your package manager vs using the native installer for an app you want.
On one hand, the native installer is almost always updated sooner with the latest features and fixes.
On the other hand, the packaged version is usually vetted against other packages in the system for compatibility and stability, not to mention having all of your updates be done from one place.
you just need to get used to it.
multiple download options are not that scary once you know what the differences are.
also, to get the latest version of a software from apt you need to find a ppa that gets update more frequently than the official ones (try https://launchpad.net/~keks9n/+archive/monodevelop-latest/ )
Valve aren't trying to make a business out of selling to Linux gamers. But Linux gamers are free, enthusiastic beta testers and bug hunters for Valve's console. I imagine most Linux using gamers already have some of Valve's games on Windows, and that Valve will throw in Linux ports as they did for the Mac. So there really cannot be many new sales there. But for beta testing, long time Linux users will do great QA, and Valve's console is their big ticket out of vendor dependence.
Kudos for this man to stand in front of an audience for 3 hours without pauses.
I'd not like being there, give me a break, put different people on stage, add more visual aids. Too tough to hear, even when I could stop the video and I'm very interested in what he says.
He is right about Linux, Linux is not there and who knows when it will. If tomorrow is, Carmack will support the platform fast. He follows the market.
The problem for Linux games by Id has been the distribution. How do I buy this Quake 4/Wars that Id has ported to linux? Certainly not from Ubuntu software center. Do they have distribution system or do I have to download some binary and house it in my /home/? I don't know and couldn't find out with 1 minute of googling.
I don't think that Valve will have this problem.
EDIT: Quake Live works really well but there is a lot of free competition with Q3 type games on linux.
valve just needs to port source games on linux.
hardcore gamers will love to get better frame rates and possibly less idiotic crashes. casters will love linux too if they can get a better encoding quality with their hardware.
once that's done part of the gaming community will follow and producers will start to take linux in more serious consideration.
Steam on Ubuntu is a great idea and has outstanding potential. Linux gaming has always been a chicken and egg problem. Steam solves the distribution problem. It also has the added benefit of Windows gamers who already paid for games, so making a Linux version makes it that much easier for them to ditch Windows.
Steam is a great step forward. Next, get World of Warcraft, Diablo 3, and Starcraft 2. Then, League of Legends, DOTA 2, etc. All of a sudden you have a legit gaming platform.
Carmack is a brilliant chap but almost everything he has said in the past several years has been of the skating to where the puck was variety. Many, many home computer gamers do two primary things on their PC: the web, and gaming. With Microsoft trying to cargo cult off of Apple, Valve has the potential to really make gaming on Linux a viable affair.
Spurred on by a down vote, let me clarify: Carmack did some big loud exercise about not developing on Android right as the Android ecosystem exploded. He rallied around Direct right as OpenGL exploded in the mobile space. He rallies against WebGL... Carmack is a developer hero of mine I have no respect for his vision positions.
Linux was written in the first place for x86. Nearly every distribution released since then, major or minor, has primarily targeted it. If a distro doesn't support x86, it's probably for a very specific reason. I don't have any numbers on this, but it's a safe bet that the number of Linux users not running on x86 is pretty close to 0% (plus-or-minus Richard Stallman). Linux runs on anything Windows does, and probably ran there first. There are probably an infinite number of mildly interesting ways one could harp on this general theme: It's not just a technicality to say that Linux already lives happily on x86.
The fact that you're familiar with the word 'distro' suggests to me this is just mild confusion. Benignly curious what your impression was. Were you thinking of Android or old Mac OS?
In Valve's experiments so far, Linux gets better performance (higher FPS) than Windows. We're talking 280 vs 315 FPS (or something like that), but that's still better.