About open-source software

Introduction

When does producing OSS make no sense?

If you are a commercial software company whose product is sold as is (ie. you're not making money as a hardware company selling servers, hand-helds, etc.), with no need for closed-source add-ons, and requires no after-sale service (eg. a word processor). Currently, the situation is even worse on the Linux platform since users often migrate to Linux to escape the problems linked to using closed-source, especially if it is commercial, so are very reluctant to use closed-source software, even if it's free beer. This kind of software can't be open-sources, because the company won't recoup its investment, and competitors will act as free riders, ie. ripping the result that that company without having spent any money of their own
If your source code is crappy but customers don't know how crappy: Publishing it removes all doubts and will generate bad publicity, hacking (security through obscurity can be effective if your software isn't sold to millions of users, and hence, don't attract hackers' attention), or even legal pursuits
Your code includes non-OSS code for which no OSS alternative exists. Opening the code is not possible due to copyrights and/or patents
While customers are fine with buying software (licenses, really), they are usually not so comfortable with the idea of software being freely downloadable and optional support. It'll look fishy to them (and it partly is, see the next point)
If the company makes money from support, it has no incentive to make the software easy to install and use, and well-documented...
If the whole project relies on either volunteers, or the originating company making a living just from support, the risk is high that the project sinks for lack of constant and sufficient revenue. Even with an open-source software, there is some switching cost (finding an alternative, transfering or re-enterind data, educating users with the new software, rewriting add-on's, etc.) So, unless the project is huge and/or can easily be replaced by something else, it's just as risky to use an OSS as a closed source product coming from a small team
You are a developer, and don't want to spend more of your time providing support instead of working on new features: If the money comes support, you'll likely have to sacrifice development over support just to pay the bills

When does producing OSS make sense?

You are a commercial software company, but you figure an OSS project can help you get people to use your product, and either buy commercial extensions, or are willing to pay for enhanced service (eg. Sendmail)
You sell closed-source software, but provide entry-level, free (beer or speech) tools to bar small competitors from entering your market (preemptive strike)
You are a commercial software company, and you market your product by emphasizing the benefit users get, eg. the choice of upgrading when/if they want to, and solves the issue of being held hostage to corporate greed and/or marketing teams. This could be made to work by keeping the latest version closed-source, and open-source it only when the next version comes out.

As an illustration, Microsoft customers are increasingly dropping out of the upgrade cycle because most of them get no sufficient additional benefits in new versions to offset the PITA and $ that entail upgrading, and because they see upgrades for what they really are: Just a way to keep shareholders happy since MS makes little or no profit from hardware or consultancy. Hence, users MUST keep buying new versions of the only profitable products that Microsoft churns out, namely Windows and Office, or it will go out of business at some point.

With OSS, customers aren't required to upgrade, and can be more cooperative and contributing (eg. sending bug reports or ideas for new features) since they know they aren't taken for a ride. Good marketing ploy.
Hardware companies (embedded, servers, extension cards, etc.). For those companies, software is an expense, not a revenue, and rarely a differentiating parameter. This can be called "commoditization of complements", since, to a hardware company, an OS is just a complement. Required, but just a complement to its real business which is selling iron
Non-software companies that are confronted with a de facto monopoly due to first mover advantage, proprietary standards, and vendor lock-in. In this case, the only way to see alternatives appear is to cooperate and finance OSS projects. To limit free-riding, limit availability of the latest version only to paying customers for a certain period of time, or offer priority service only to paying customers. Payment can either be direct, financial, or take the form of barter (code, bug fixing, documentation, training), or a mixture of both, so as to reward stakeholders vs. free-riders
Universities (students get to work on real-life projects, and expand their work experience and résumé, and successful OSS projects may enhance the name recognition and prestige of a university)
Professional software for which heavy post-sale service is required and budgeted, regardless of the license under which the software is produced (deployment, training, customization, maintenance - partly due to changes in legal requirements or the software/hardware used by the customer - , support, bug fixing and safe deployment, new features, etc.).

In this case, all things being equal, an OSS solution lets a company come up with a cheaper proposition since the customer saves money on the licenses, and can be a good differentiator by making the customer feel safer (avoiding vendor lock-in or migration to a new software if the editor goes out of business or decides to kill the product).

For this kind of company, OSS also has the additional benefit of increasing mind share and contributions in the form of code, bug fixing, or documentation
Non-software companies that come up with some piece of software for which they either don't see any commercial potential, not sufficient to justify launching a commercial venture to commercialize it, or to which software is just too removed from their core competency. Most software development is actually done within non-software companies, and is now easily available thanks to the zero distribution cost allowed by the Internet
Governments and the military, which for obvious reasons, would rather not depend on some commercial software provider, especially if it enjoys a de facto monopoly, and originates from a foreign country. Considering the sheer size of Windows, the fact that Microsoft grants governments (read-only) access to the source code doesn't offset the disadvantages of this dependency

Benefits of OSS to end-users

For end users, either individuals or organizations: More independence from suppliers, higher bargaining power; Only requested features are implemented (since either people will stop upgrading, or the project will fork if taken in a direction that most users don't want to go)
For developers: Enhanced productivity and more opportunities, who don't have to waste time or money on features that have already been developed; No risk of being taken in a direction they don't want to go (like users: If a feature or general direction doesn't match what most users want, it either won't happen, or a fork will occur to cater to the minority)
For both: OSS code has to be more solid and clean since it is public (ie. no way to hide hacks, which aren't good for the reputation of the project and/or developers who wrote the code), and because crappy, non documented code doesn't attract contributers. Closed-source can accept lower standards since customers have no idea how crappy the code is, and, up to a point, developers are forced to grin and bear working with bad code due to financial pressure (off-shoring, higher unemployment)

How to make a living selling open-source software...

... when either this piece of software doesn't generate any or enough consulting/customizing to put food on the table, or you simply don't want to spend your life providing paid-for support (ie. your life is developing software, not answering the phone)?

The core software is open-source, but extra features are closed-source
Free for open-source projects, but closed-source projects require a commercial license
(fill in)