OSDev.org

The big part of OS adoption that most people miss isn't the feature set or the design being unencumbered with legacy cruft, or the intelligence of the programmer(s). It's the developer's soft skills. No one will adopt an operating system from a developer with no concept of how to provide users/customers with a pleasant personal experience. Would you want to use a product from someone who calls you an idiot if you have a question about a specific function of it? Would you want to do work with that product, or create products that work with it?

Think of the legions of support people that a company like Microsoft has, and that each person has been trained to do what they can to solve your problem in a manner that is both effective and satisfies you. Customer sat is huge to them. So is developer outreach -- read Raymond Chen's stories about working with customer liasons to solve strange programming issues and helping customers find effective and proper solutions to their development problems.

Now imagine what would happen if the last 15 years of their support system was a handful of people on a mailing list grumbling at newcomers and the project manager calling users and contributors "f**king idiots".

It's not feasible to do "my way or the highway". You're going to certainly find that everyone's going to choose the highway.

Isn't it more like building your idea in software in seek of hardware to make it shine? Why not also make the hardware?

Imho it helps to think about a problem from the perspective of what you have, rather then what you need.

So this is how Cassandra felt.

Well, before I leave this thread, then, let me fling the last bolt in my quiver; if this one doesn't scare you, nothing will:

If you continue down the path you are going, you will be like me fifteen years from now, a bitter, cynical, burnt-out shell who is afraid to take any chances at all, not because you are wrong, but because you can't see the possibility that you might not be right.

What, you didn't realize by now that I was just like you twenty years ago, and made exactly the same mistakes you are making now?

Anyway, I concede the field to you, Brendan. Enjoy the ashes of your victory while they last.

Interesting, experience: Can you enlighten on what you were trying to do?

Ali

As a customer, I'm only interested in what a software does and that it works, and I'm not interested in if it is "pleasant" at all. I'm not interested in if it contains the latest "fads" and frameworks that will take me ages to learn. I see that as a problem, not a feature. I also want to be able customize things in an easy way, and not with a bloated GUI that lacks details I want to change. Naturally, if I find a software function I like, I want to be able to integrate it into something I make myself in an easy way.



I can imagine that a company like Microsoft, which makes awfully bad software, would need a huge support division. Having a huge support division is a sign that you are making bad software, not that you are "customer oriented". With a properly designed OS, public code and fewer bugs, they would only need a small support staff.



I prefer if they are honest, competent and direct over being creepy people pleasers that don't know anything.

Hi,


The truth is somewhere in the middle between quality and marketing. A quality OS without marketing will not get known to many people, but it will be installed by the handful of people that know about it. A non-quality OS with marketing will get known to many people, but it won't be installed because it lacks quality.

Solving user problems is a third, but very important aspect. Future users looking for your OS will probably look at how effectively are the issues of current users resolved. If they are satisfied, they will maybe install the OS. Current users will probably ask for solutions to their issues. If they are not satisfied, they will most likely uninstall the OS.

--------

The vast majority of users can't read the source code. Even if the OS was properly designed and there were fewer bugs, there would still be users that can't use the OS.

Insulting users that can't read the source code is a perfect first step to have your OS uninstalled from 90% of computers (and I'm being very optimistic).


Regards,
glauxosdever

It worked for Bill Gates. He managed to sell a totally horrible OS (MS-DOS), that even based on the standards of the time, was hopelessly retarded. For instance, CP/M was a lot better, and UNIX had been available for quite some time. He then was able to let customers "move" to Windows, which in the beginning was even worse than MS-DOS, constantly crashing. Microsoft eventually got out something decently stable with Windows XP.

So, I'd say that the approach that appears to have worked in the past is to launch some crappy OS in a niche-market, and if you are lucky and the market grows, you can eventually deploy something that works decently. Actually, this was the way Linux got moving too.



That's all wrong. Users should not have problems with an OS. Users should have problems with APPLICATIONS, which eventually might be due to OS issues, and which the application developer will need to address (possibly by consulting the OS developer). Actually, an OS should not be bundled with a huge amount of applications. Users want to chose those themselves, and not be forced to use the OS companies applications.



They won't since the end-users are customers that use a public service. The supplier of the device might look for another supplier if the system doesn't work properly, but they would not be uninstalling anything. Actually, users don't even install the OS. It's pre-installed.

Just going to leave a probably unpopular opinion here: if you want to create a successful OS today (where "successful" means "reasonably widely adopted"), you're going to need to couple it with your own hardware platform. You're not going to have much luck trying to take over a platform that's already hogged by one of the huge operating systems. This is probably why Linux distros keep going nowhere (at least outside servers), they insist on piggybacking on a hardware platform hogged by Windows. Even OSX, which runs on what's basically PC hardware, relies on it being sold with Apple's own provided hardware (even if it's just a fancy brand).

The flipside is that at least you won't have to worry that much about drivers when trying to take off since you'd only need to worry over whatever specific hardware you provide and maybe common peripherals that can be attached to it.

Of course there's way more to this, I could probably write a full essay over this topic, this is just the basic gist.

It strikes me that there is a fundamental problem that the meaning of the word has dramatically changed over time..
Most machines on the market today come with a pre-installed 'execution environment' which consists of operating system/hardware services/compulsory software services/standard applications/almost compulsory update services (sometimes absolutely compulsory)/compulsory advertising of new products and features/patronising messages and a very high garden fence.

That is not what I mean by an operating system!

There are also two fundamental types of user:

(a) the user that wants to get on with the standard jobs of life and therefore wants a machine that handles these tasks well: sending receiving and playing emails, photographs, music and videos, looking up stuff on the internet and writing documents and pamphlets. They need an executing environment.

(b) the user who wants a computer to harness for specialised tasks . These are the minority. They require a machine with a reliable and fast operating system that is not bloated with daily change of pretty picture and ages indexing every occurence of the word 'banana' on their hard disks.

This thread seems to be discussing the question as if there is only one class of user and as if an executing environment is the same as an operating system.

And I'm sure others would probably be able to subdivide my two categories above even further and possibly add some totally different ones! Net effect - everyone that has contributed so far is right in regard to a tiny bit of the computing world and everybody is wrong when they try to maintain their view applies to the whole computing world.

Maybe if your OS is targeted towards a more technical audience.
For most users, choice is a bad thing and the success of the iPhone/iOS shows this.

Hi,


While it worked for Bill Gates, it didn't work for anyone else except a select few, and neither of us is Bill Gates or one of these select few. Furthermore, Bill Gates managed to persuade hardware manufacturers to pre-install Windows on most computers, and these select few supplied their own hardware.

But some of the design issues are harder to fix later and eventually the OS is replaced by some other OS that doesn't have these design issues. (Note: This may not be the case with Windows since hardware manufacturers have been persuaded to pre-install it on most computers.)

I didn't mean the OS would have problems. I meant the users would have problems to do something on the OS. So support is still needed, even in the form of good documentation or built-in tutorials.

That happens only in the case of Windows and these select few. Do you produce your own hardware? Are you able to persuade hardware manufacturers to pre-install your OS?


Regards,
glauxosdever

No, my company sells a complete solution (payment terminal / controller). We have somewhat of our own hardware, but the computer is a standard x86-based SBC (actually, we support several models, but we don't support installing it on any computer).

You have to remember how much luck is involved in it. If Gary Kildall hadn't put the phone down on IBM and virtually told them to **** off because he was busy, CP/M could have been the big player all the way through and few of us would ever have heard of Microsoft. Bill Gates was then contacted by IBM, and although he didn't have an operating system to offer them, he had an idea about where he could buy one. He acquired QDOS (which copied all the functionality of CP/M) for $50,000 from Tim Paterson who didn't realise what it was really worth, and then he played IBM for fools with one of the best business deals of all time (for him, not them). There are quite a few Tim Patersons here, but they've missed the boat. Things have moved on so far now that it almost always takes a big company with huge resources to build an OS with enough features to boost it into orbit, although there may still be room for a superior OS to take off if someone like Brendan can provide it. No lesser OS would have a chance, unless someone in a big company that wants its own OS likes the look of it and decides to buy it and develop it further, with or without the original designer's continuing involvement, but it's more likely that they'd just copy your ideas and hire people to write their own OS from scratch to cut you out of the loop entirely. The only other chance you have is if you're able to provide some new functionality that everyone wants and no other OS provides, but what is there left that hasn't already been done? (Don't answer that though - if you have an answer, keep it to yourself and get on with building it.)

Microsoft was a very large player in selling programming languages in the late 70s through 80s. Perhaps you've heard of a computer called the Apple II.

We would definitely know of Microsoft if CP/M-86 had been what IBM bundled with their system. I suspect the Windows environment would still have been created, and Dave Cutler's former VMS team would likely have still gone to Microsoft to design NT OS/2, spawning what now exists.

I shouldn't be surprised about how much ignorance about and hate for Microsoft is on this forum, yet I continue to be.

They did? All the programming languages ever manufactured by Microsoft that I know about were pure crap that wasn't cross-platform. For instance, I had to use TASM and later WASM, in order to compile my OS, as Microsofts compilers bundled a lot of crap into the executables that made them impossible to use for anything else than what they were designed for, whether it was for MS-DOS or Windows. As you probably know too, it wasn't Microsoft that provided compliers for DOS-extenders, it was other companies. So, Microsoft actually has a very long history of strongly connecting their compilers to their crappy OSes, and they did this by using undocumented features.

Such a strategy would have been unsuccessful in the absence of an OS that almost everybody used.

This was continued with their own commercial Windows applications that used undocumented Windows "features" so nobody should be able to run them on Windows emulators.



I see no reason whatsoever why that would have happened. Especially not for the OS/2 case. VMS was a really good OS, much superior to anything Microsoft ever have done. What they needed was the desktop part, as DEC had neglected that users would want to run a lot more on their workstations than they had anticipated.