Hi,
rdos wrote:
DavidCooper wrote:
It isn't about him being relevant. If you ask someone in the street if they've heard of Gary kildall, most of them would say "who?", but almost eveyone's heard of Bill Gates. Do you suppose that would be the case if IBM had gone with Kildall instead and if Gates didn't buy QDOS from Paterson? Maybe he'd have made some other major breakthrough, but it's more likely that he would have made insufficient impact to be widely known other than by geeks. Luck is a key component of this business, as is getting in first with something, and Gates benefited enormously from IBM's mistakes. (But while luck has a large role, there is much greater need now to provide quality if a new OS is to have any chance of making an impact, and that's why Brendan is absolutely right to throw out all the old crud and do the whole thing properly.)
I disagree to that. Luck plays a big role today too, and so does determination and positioning oneself in the right roles with the right contacts. You cannot expect to create a new, successful OS today in your garage. You need to work at the proper place, have the right contacts, and then persuade people that it is a good idea to do a new OS. You also need it to fill some function. All of this was part of Bill Gates success. Microsoft already had products (Basic), they happened to stumble on a new niche where they could introduce a new OS.
Also, a new OS today that "throws out all the old crud" (IOW, something completely new with no backwards compatibility) will never succeed in the mainstream OS area. It might succeed as a niche product, but then you need contacts and people that want to use your product in that niche.
First; think of it as attempting to maximise the chance of success (and not about trying to guarantee success). To maximise the chance of success, you must provide one or more reasons for people to use your OS instead of existing OSs, and those reason/s must be convincing enough to overcome the hassle of switching to a different OS.
For reasons:
- "more compatible" can not work (you can't have an OS that's more compatible with Windows software than Windows, or an OS that's more compatible with Linux software than Linux).
- better hardware support can not work (lack of drivers isn't something that can be solved until after the OS starts to be successful enough for a large number of people to write drivers)
- "cheaper initial price" can not work (you can't be cheaper than free)
- "cheaper running costs" can not work. In theory you might be able to reduce power consumption a tiny little bit for some computers (despite the "worse hardware support" problem), and you might be able to reduce admin/maintenance costs a little; but for people that actually care about running costs (mostly larger businesses/companies) the hassle of switching to a different OS (many computers, many users) is far greater, and "slightly cheaper running costs" is nowhere near enough.
- "better security" can not work. You can try to make it "more secure in theory" but that counts for nothing - you need to prove that it's more secure in practice, and you can't do that without a reasonably long track record of resisting attacks (which can't be done until after the OS is successful).
- "faster performance" can not work. Most existing OSs have had 20+ years of performance improvements that your OS hasn't had; and even if you do manage to improve performance it's not going to be enough to overcome the hassle of switching to a different OS.
There is only 1 reason for people to switch to a different OS that has a chance of overcoming the hassles of switching to a different OS - it does something other operating systems don't. However, "OS supports feature that other OSs don't in theory (but none of the software supports that feature)" is worthless; and if a feature is attractive (and isn't patented by you) then you can expect people to try add that feature to existing OSs (and in that case the reason for people to switch only lasts for as long as it takes other OSs to add support for the feature). Also note that doing 2 or more things that other operating systems don't is 2 or more times better.
This means that to maximise the chance of success (or to encourage eventual adoption); you want to do as many things as possible that other operating systems don't.
If you don't care about compatibility at all; it's difficult to find one thing that your OS could do (that might interest some users) that other operating systems don't do; and finding many things that your OS could do is many times harder.
Compatibility severely limits your ability to find things your OS can do that other OSs don't; and completely ruins your ability to find things your OS can do that other OSs can't support quickly. It's like attempting to win the decathlon at the Olympics while being chained to the wall in your basement, where you can't train/prepare effectively beforehand (before OS is released) and can't compete effectively after training/preparation is finished either.
The only way to maximise the chance of success is to ignore compatibility, find as many things as possible that your OS could do that other operating systems can't do now (and hopefully can't quickly support in future), and ensure that all software for your OS takes advantage of those things if/where appropriate.
After you've done this (after you know your OS isn't being limited by compatibility compromises); then you can think about compatibility - e.g. file format converters, network protocol bridges/gateways, etc. You can also spend some time thinking about whether or not these things should be "one way" or "2 way" (
should you provide tools that people can use to convert their files from your OS's file formats back into the file formats other OS's support so that users can easily switch from your OS to another OS? Surely the developers of other OSs would be responsible for their OS's compatibility and not you).
Note that succeeding as a niche product can be an extremely good start (and once your OS is established as "king of a niche" it's much easier to expand into other areas). This is probably not something you should try to avoid, it's something that you should probably try to encourage.
Cheers,
Brendan
Login
Register
FAQ
Search
