Will Open Source Ever Become Mainstream?

Prabhu Ramachandran asks: "I am a graduate student at the University of California at Berkeley and as part of a course project I am trying to gather comments on the following question: Will the Open Source and Free Software communities develop software that will find widespread adoption amongst the mainstream, or is such software, by its nature, suitable only for sophisticated users? As part of my literature survey I found an academic perspective that seemed to indicate that open source projects do not reach the mainstream because the developers tend to listen only to their smartest customers. There also seems to be a lack of detailed documentation and an easy-to-use interface which normally attract the not-so-sophisticated users. I would like to hear the thoughts of Open Source developers and others on this issue. If you would like to view my references or the comments posted on a website hosted for this purpose, please visit my website." There have already been some interesting comments posted on his website. What is your take on this issue?

Comments

Re:It IS mainstream already

[RevAaron]

I would be GLAD to give several hundred dollars to any company that can make a consistent, user-friendly, non-MS OS for my x86 hardware (all of it, not just some). Is this possible? Apple - where are you?

One of the reasons Apple's OS works so well is that it's integrated with its hardware, designed to work with it not just on top of it. Apple is in the consistency biz, which is why they wouldn't be interested in selling a copy of OS X for the ugly monster that is PC hardware, unless of course it was their own x86-based design, with the advantages of current Mac hardware.

Why not take that several hundred dollars and just save it- and use it to buy a Mac when your PCs outlive their usefulness in a year or two.

Documentation

[Angst Badger]

There also seems to be a lack of detailed documentation

That's it for me in a nutshell. Forget the "mainstream" -- the lack of good documentation renders a lot of otherwise nice software useless in the IT workplace.

Look at it this way: if you pay an admin $60/hr., every hour he or she spends struggling with your fragmentary docs or (much) worse, reading the source to figure out what's going on, reduces the cost-competitiveness of your software versus a commercial product by $60. In a big project, multiply that by the multiple admins and developers who have to struggle with it, and it's not long before your free-as-in-speech software is much more expensive than the free-in-no-way commercial alternative.

Forget the broader social issues, forget the long term. Management does not think that way, and they have compelling incentives not to. And most of all, forget the dollar cost of the software. Cost of software is almost always trivial compared to the cost of the labor required to maintain it, even with expensive packages like Oracle and (may god pity you if you have to deal with it) Interwoven. The real question from a "straight" business perspective is: how long will it take us to have Package X up and running smoothly? From a business perspective, that's the whole issue.

The idea that businessmen can be persuaded on a large scale to make decisions on something other than relatively short term ROI calculations is a fantasy. That's what federal regulation is for. If you want to move product -- and that includes free software -- you must understand your customers' needs and satisfy them better than the competition. Free software, by and large, does not do this.

"Intuitive" GUIs only become a major issue when you're talking about non-technical users -- not that it wouldn't be nice for plenty of server apps. When it comes to ordinary end-users, you can probably skip the docs because they won't read them. The GUI becomes absolutely critical then. Again, stop thinking about whether the end user can figure out the interface, or whether it's documented -- ask yourself, "Is my free package as easy to use as the competition's non-free package?" If your answer is no, go fix your interface problem.

Finally -- and slightly off-topic -- the notion that point-and-drool idiot-proof interfaces will cripple a program is nonsense. Sure, some things don't translate to GUIs well, but a lot of stuff will, and you can still provide a CLI/config file interface to the advanced users.