OSDev.org

It is possible today, to build a new OS and making money ? Just like bill gate do it in the times ?

I know bill gate familly was rich, so this is a great help and in this time(like 1975-85) it was more easy to make an OS because a little OS like Dos 2 can be enough, but in 2004 if someone want to make something good his OS have to be better and great than Windows longhorn/XP. By the way, the first time, somes say bill gate have never programming his first OS, he buy it from another person for 50k$ so in this case I don't know what to think. It is possible today to start a business when we have no rich familly and no partner ?

I assume you mean personal computer operating systems? I wouldnt
say its impossible, just very difficult. Youd be competing with linux and windows, not an easy feat. Youd probebly be more successful writing
operating systems for other devices.

I think it's possible. You wouldn't necessarily complete with Windows, because those looking for something other than Windows probably don't like it. This means you don't need to write bloatware just to compete, with a whole load of useless features and APIs on top of APIs on top of really badly designed (or hacked together) APIs.

You wouldn't necessarily compete with Linux because it's well suited to technically literate people and more suited to server applications than desktop applications. It derives from the old UNIX time sharing model with multiple users, and is not a general purpose OS for PCs (although you can use it like one, most people probably do.)

There is no real general purpose OS for PCs, aside from the now defunct BeOS. A general purpose OS for PCs is probably a single user multitasking system that comes with a POSIX API layer so that programs like OpenOffice, Mozilla and others can be compiled and run on the system. You could supply a complete desktop OS out the box and the user would probably want for nothing. You can include XFree86 and Gnome or KDE for example. They can play music and watch DVDs, burn CDs and DVDs easily, etc. Such a system could complete with Windows, with a little care. It must be easy and obvious in its use.

Hardware support needs to be better than it currently is. That means a really super USB stack and PCI-X support for one. M$ and Intel are working hard to get rid of serial COM ports, parallel ports and ISA. PCI will be obsoleted soon as well in favour of PCI-X. PCs five years from now will likely be just USB and PCI-X, with serial ATA or SCSI disks and ATAPI or SCSI CD devices (probably serial ATA and ATAPI because of the cost benefits to manufacturers). USB will likely be strongly favoured except for things that are not normally plugged in and out frequently like video cards, which will use PCI-X.

In fact, you may never get to write a really great general purpose OS for IBM and compatible PCs in future if the Trusted Computing Alliance has its way and all BIOSes are TC. We will need old hardware or new PCs without TC BIOSes. Intel and AMD are both members of the TCA, and that means their disgusting Fritz chip may be integrated on all new processors in the near future. With that and the fact there are only a couple of BIOS choices for all new PCs, it may be impossible to get rid of TC on future IBMs and compatibles. Even HP and IBM are working on a TC version of Linux. Reckon IBM are spending more than a billion dollars a year on Linux and open source out of the goodness of their hearts?

Lately I've been thinking about this, and I'm ditching my goals for an OS for AMD Athlon 64 in favour of a PowerPC general purpose OS. I'm instead going to get an old iMac sometime soon and start learning PowerPC assembly language and I/O on Mac machines, all about OpenFirmware, etc.

I would strongly recommend learning about PowerPCs instead, and provide an alternative to this horrible Trusted Computing nonsense. I think there will be significant opportunities there in future on PowerPC systems IF they don't have that TC crap. People all over the world will be looking for an alternative to TC-infested PCs, and may have to look toward the Mac for that.

Hmm... I've branched off topic a little. Getting back to your original question, yes I think it is possible to make money from OS development. I think you'll make money from shrink-wrapped boxes in variety stores and supermarkets more so than selling over the internet, and there will have to be a strong brand, good marketing and clear reasons why people would want your OS instead of Bill's. I think stability, smallness, cleanliness and no TC nonsense would do it nicely, along with adding that it does everything Windows users can do (play music, watch DVDs, use file sharing apps, burn CDs and DVDs, surf the net, read email, use office apps like OpenOffice, etc.)

Excuse me, but I'd like to tell my opinion: Linux is far more suited to desktop than Windows has ever been. Why? Because with Linux you can actually get something done, as you don't need to waste all the time fighting against the operating system.

Servers you can at least drop behind firewalls, pay for someone to look after, apply patches, check that they actually work and don't crash your apps, and hit the power-button everytime something decides to stop working. Not that it wasn't easier with Linux (or *BSD, or even Solaris).

Now to answer some kernel_journeyman's points:
- most linux distributions, especially the commercial ones, include "your pretty easy-to-use desktop" out-of-box with a system installer easier than what comes with Windows.
- making a single-user POSIX system is pointless. You just make the system less secure, hence losing one of the points to get rid of Windows.
- the only hardware that really needs working on Linux is 3D acceleration of graphics cards, and the problem there is that manufacturers won't offer good drivers nor specs for writing them. All in all, I've had less trouble with drivers in Linux than in any Windows version I've tried. And I do have lots of hardware for various purposes. Especially with USB everything just works, except a few devices that don't work like the standard says.
- You can perfectly well run Linux on 64-bit Athlon. In real life!

Please, you sound like your Linux-information is second-hand knowledge from the '90s. I won't say Linux is perfect, but it a LOT better than you make it sound.

Oh, I'd like to add that I've teached people completely new to computers (maybe a bit web-surfing experience but that's it) how to use Windows, and I've teached them how to use Linux (or Unix is general) and I can tell that contrary to common belief, it takes less time to teach Linux to someone than to teach Windows. Also the resulting skillset is MUCH more complete.

mystran wrote: Excuse me, but I'd like to tell my opinion: Linux is far more suited to desktop than Windows has ever been. Why? Because with Linux you can actually get something done, as you don't need to waste all the time fighting against the operating system.

Servers you can at least drop behind firewalls, pay for someone to look after, apply patches, check that they actually work and don't crash your apps, and hit the power-button everytime something decides to stop working. Not that it wasn't easier with Linux (or *BSD, or even Solaris).

Now to answer some kernel_journeyman's points:
- most linux distributions, especially the commercial ones, include "your pretty easy-to-use desktop" out-of-box with a system installer easier than what comes with Windows.
- making a single-user POSIX system is pointless. You just make the system less secure, hence losing one of the points to get rid of Windows.
- the only hardware that really needs working on Linux is 3D acceleration of graphics cards, and the problem there is that manufacturers won't offer good drivers nor specs for writing them. All in all, I've had less trouble with drivers in Linux than in any Windows version I've tried. And I do have lots of hardware for various purposes. Especially with USB everything just works, except a few devices that don't work like the standard says.
- You can perfectly well run Linux on 64-bit Athlon. In real life!

Please, you sound like your Linux-information is second-hand knowledge from the '90s. I won't say Linux is perfect, but it a LOT better than you make it sound.

In my experience, the opporsite is true.

I've spent whole days fighting with variouse linux distros (althoug Suse 9.1 is good), or even having them crash on me (in my experience Mandrake 10 is less stable than Windows Me, but Suse 9.1 is rock solid). I gave up trying to setup a printer on slackware.

Windows IMHO is simple and to the point (once you get rid of that POS Fisher Price look).

srg

Back to topic... (*please* not another Linux-Windows flamefest!)

kernel_journeyman wrote: I think it's possible. You wouldn't necessarily complete with Windows, because those looking for something other than Windows probably don't like it.

I've heard that argument from the Amiga camp quite often, but I think it is fundamentally flawed. People usually don't like Windows (or Linux) for a reason - because it's too expensive, too unstable, too badly documented, because it doesn't work the way they expect it to.

In any case, your alternative OS would have to be better in some regard to attract users. So, yes, you do compete against Windows, Linux, MacOS, BeOS, AmigaOS, RiskOS, and yes, Clicker, Moebius, Blue Illusion, etc. etc. etc.

The market is "personal computers", and there's no way around competition. You might do things different, but you're still aiming at the same market.

The biggest problem is the chicken-and-egg of users vs. software manufacturers. Many operating systems try to solve this by being POSIX-compliant, so you could use GNU tools, X Window software etc. on them easily - making them yet another GNU tools / X Window platform, with little to distinguish them from Linux or *BSD.

The only way to get around the problem and penetrate the market as your own, recognizable brand of operating system is to invest. You will need tens of thousands of development man-hours to catch up with the big players even on the most basic level.

That can be done either by going Open Source (which basically takes away your potential revenue, so no "making money"); or by investing several million bucks up front.

Sorry, but that's market science...

how did BG made money out of an OS ? by getting to a big hardware seller and say "you know what you're missing ? an Operating System! We have one, but we won't sell it. Instead, we just ask a few bucks for every machine you'll sell equiped with the OS".

Find another big hardware seller that has the same problem and is stupid enough to agree such terms

srg wrote: In my experience, the opporsite is true.

I've spent whole days fighting with variouse linux distros (althoug Suse 9.1 is good), or even having them crash on me (in my experience Mandrake 10 is less stable than Windows Me, but Suse 9.1 is rock solid). I gave up trying to setup a printer on slackware.

Windows IMHO is simple and to the point (once you get rid of that POS Fisher Price look).

srg

Ok, this is what I can accept.
Printer setup indeed used to be quite painful, but it's getting better. I admit though, that finding decent documentation about this is a bit hard, if you're not interested in reading what gs and cups have.

Mandrake is actually legendary for it's stability. %)
One version of their kernel jammed every time any application tried to dump core, and to make it worse, ANY crash (of any program) would cause a core dump.... I wouldn't touch anything from them..

Now SUSE (or Redhat) is probably best of the "newbie friendly" distributions. Usually people switch to Debian (and nowadays Gentoo) at some point, especially if they are of the more technical sort. Both are a bit ugly to setup (gentoo doesn't even have an installer, although I hear there's a port of Redhat's installer for gentoo), but very nice once the system is running. The thing about these two is that you can install virtually anything with a single command (for debian it's "apt-get install <package>" and gentoo it's "emerge <package>", former uses binaries, latter compiles locally).

But my main point was anyway was that most of the "sort-comings of Linux" are simply different ways to look at things. I think the biggest difference is that while Windows is initially a bit easier, it gets progressively harder to use once your demands raise. With Unix you have to invest a bit more effort initially, but once you get the idea, you can usually figure out things quite easily.

I think the biggest problem (which I admit) is with Linux still is installation. The graphical installers that some distributions provide are nice, but then again the distributions with best installers are unfortunately not the distributions that are best though out. I admit I was not taking installation into account in my previous post. Still, I don't think average user installs his/her Windows either.. at least not as the first task they do with their new computer.

Futher, "installation" never ends where the installer finishes. You then need to get all the extra hardware and necessary software running. Most of the time, when you encourage people to try Linux, they grow frustrated trying to get their stuff work, before they actually have learn to be productive with the system. This is most unfortunate.

I've noticed that if I instead offer to install a system for them, get them tools they need, configure most of the common problems away.... suddenly people have a working system, which they can explore. This way one doesn't need to start tweaking the system as the first thing, but can instead learn the basics as a normal user. And people starting this way usually like the system much more.

I might repeat that most people start with a preconfigured Windows too, at least to get their first touch with it. Looking at average non-technical persons Windows installation in (say) half-a-year after it came from the store indicates that most people couldn't install Windows by that time.

But indeed, it might just be that I've used Linux so much more than Windows. Maybe it get's easier once you learn to decipher UUID's in registry without needing to search for a reference.

Finally, to get back to topic =) I might add that for any new OS to make money, one will probably need OEM pre-install deals, since big companies won't switch easily (whatever they have), most potential users are either open source supporters, people going to "just try it, so they don't need to pay", or people that couldn't care less about your profits. (yes I am generalizing). Average people most likely to be happy with your system aren't suddenly going to install operating systems in the future, and I think those are the people that would also be most likely to actually pay for the system (until you get enough marketshare to get companies switch to you =).

Now my opinion... I think it's possible...
To make a new OS, which you sell it for money, you must realize, that it should be better than another OS, since that would be a reason to switch OS's (like stability, programs, etc.)
To determine the programs problem, I'd suggest, that you create a few emulators, which can run Windows and Linux(don't have to be, you can port it your own) programs.
What also should be, is support.
Example: can they email you if there are problems with the OS itself ? Do you provide the security patches on time ? Do they get the next few releases for free, or do they have to pay an upgrade license ?
And about the "looks" of an OS, you can't ask money for an OS that only has a CLI.
Instead, use an advanced GUI..
And of course, if normal users would buy such, provide a large amount of hardware drivers.
Normal users will (most of the time) say it's crappy, it doesn't support it. And how about the internet ?
And make it attractive for people, that want to buy it....

All things that should be thought about....
HTH

Hi,

IMHO it is possible to make money with a new OS without investing thousands (or millions) of dollars - you just have to be determined, smart, skilled, extremely lucky and capable of living for extended periods of time without food

Rather than discussing if it's possible, I'd rather discuss how an OS developer can increase the chance of making some money...

First step could be to establish a market niche. If your OS is much better at a specific task than any others then you could attract a small part of the whole market to your OS. For example, if your OS is perfect for games (and comes with several extremely good games) you might get users who use your OS for games only, and other OS's for everything else.

The smart way might be to offer your "market niche" OS for free, and when it is established in that market niche (and some third party developers are writing something for it) use your position to expand. For example, the free games OS gets more features that enable it to be really good for music, video, DVD's, etc. Once you've made progress (gained more market share) in the entertainment field you expand again (3D CAD software?). You'd continually build on your market niche until you've eventually covered everything.

There's a few problems with this. You'd need to expend effort maintaining your market niche, so that people don't lose interest and go elsewhere. This in itself is possible, but the more you expand the OS the more your resources (time) is stretched, until your resources are spread too thin and you can no longer maintain interest in all areas.

Ok, that's enough rambling - let's create some lists...


WAYS TO MAKE MONEY FROM OS DEVELOPMENT

A) Give the OS away for free and ask for donations. I expect the amount of money donated wouldn't cover hardware and other expenses.

B) Take advantage of government grants. Some governements offer grants for specific purposes (with specific restrictions). In Australia there's a grant called "COMET" (COMmercialising Emerging Technologies). It's usually between $5000 and $100000 (in $AU) and is aimed at turning innovative technology into tax paying companies. You need to be willing to lose complete control of your project. A good proto-type is recommended. The project must be innovative and have commercial potential.

C) Make it shareware and/or crippleware. Don't expect large amounts of cash. Do expect difficulty getting market share.

D) Give the OS away for free, but charge for an especially good (non-essential) application for it. This approach may mean that the charging money part doesn't interfere much with the market share of the OS itself.

E) Give it away for free, and make your money as an advisor, developer, consultant, etc. You may be able to talk some companies into paying you to write support for their hardware if your OS is directly involved with the same market niche (e.g. for a games OS a manufacturer of 3D helmets might be interested). If I remember correctly this is the way Linus Torvalds is making lots of money (consultant or developer - not sure)...

F) Ask around for investors (e.g. form a public company). Investors won't invest unless they can expect some return on the investment, and it would be difficult to demonstrate any returns.

G) Use your OS and everything you've learnt to get a job in the OS development field. All your dreams and ideas will be lost, but this may be the most possible solution.


PROBLEMS WITH TRYING TO MAKE MONEY FROM OS DEVELOPMENT

A) If people have to pay for it chances are that they won't bother downloading it. You have trouble getting any market share at all.

B) While developers will donate time to an open source/GPL type of project they won't spend time to make you money. You could promise them a share of the cash (or even a flying pig).

C) Without developers donating time you will have trouble writing all applications yourself, and major hassles writing all device drivers yourself. People won't pay for (or use) an OS that doesn't have applications or device drivers, and people won't write more applications and device drivers for an OS that no-one uses (catch 22).

D) You'd need to provide some form of support. Free projects tend to end up with newsgroups, message boards and mailing lists that are maintained by volunteers.

E) The operating system must be extremely easy to install, so that people with minimal technical knowledge can install it (a lesson learned from Linux).



If anyone can think of more ways of making money, or ways around the problems.....


Cheers,

Brendan

Pype.Clicker wrote: how did BG made money out of an OS ?

Simple, your reason and he abused his monopolic position.

One of the things people seem to be forgetting is that Microsoft was already nearly billion-dollar company at the time when IBM nixed the deal with Digital Research and approached Gates for their OS. They already had a near-monopoly on certain niche fields, most notably with the BASIC interpreters and FORTRAN compilers they sold for nearly every home-computer platform of the time. Big Blue was already dealing with MS to write the ROM BASIC and BASICA interpreters, and well as for various applications such as Multiplan. As for Gates not writing it, he himself had largely stopped coding after 1979, having moved into a management role by then; he actually took a greater role in the program development later, when he and Simonyi developed the project-management techniques which were used by MS after 1983 or so.

Now, IIUC, Gates had already bought QDOS (aka SCP/DOS, aka 86-DOS, aka MS-DOS v1.0) from Seattle Computer Products, and so happened to be in a damn good position for the deal. Whether this was by chance, or by Gates guessing that Kildall wouldn't play ball with IBM (DR refused to signed the IBM NDA, wouldn't make the changes to CP/M-86 which IBM wanted them to, and asked for a very high royalty rate), or that IBM knew that MS owned an alternate OS that made it safe for them to blow off DR, I just don't know. I do know that 86-DOS was the first widely sold OS for the handful of 8086 systems already on the market when the Empire decided to take over the small-computer field.

Little did they realize that they were setting up the forces that would eventually usurp their throne...