The "Proprietary" Software Licensing Category
Q: Doesn't the term proprietary merely mean "characterised by being subject to ownership claims"?
A: Not in the context of software, no. If it did, then pretty much 100% of usable software (other than the tiny minority that's genuinely public domain would be proprietary, rendering the term meaningless.
If the notion of word meanings being context-dependent seems strange, consider the word "base" and its differing meanings in numerous fields of life, including baseball and the military.
Q: So, what does the word mean in the context of software?
A: It means "issued under licence terms not qualifying as open source AKA free software". The latter is a concept that, speaking in general terms, amounts to any conditions guarantee and perpetuate any person's right to fork the software's source code — the right to independently update it and redistribute new versions, conveying those same rights to further recipients integrally to the new versions. If those rights aren't provided with a software instance, then it's proprietary by definition.
Please note that that is the default condition of software issued to people, by automatic operation of copyright law, absent action to the contrary: In the USA and similar jurisdictions, the copyright statutes reserve to the work's copyright owner the rights to create derivative (variant) works and to redistribute, among other things. So, it's not possible to issue an instances of a copyright-covered work (such as a piece of software) without the copyright owner specifically granting those rights. Technically, that rights grant needn't be in writing, let alone be embodied in a licence statement accompanying the software: It could even be an oral grant, or conceivably through conduct. However, for practical reasons, it's traditional (verging on pragmatic necessity) to include the grant in writing with the software.
Thus, software merely found in public with source code cannot be assumed "open source" merely because the source code escaped into public. It may be (and probably is) code for which the owner never made an identifiable licence grant, rendering people's redistribution and further development technical violation of copyright law.
Q: By "proprietary", don't you just mean commercial software?
A: No. Absolutely not. "Commercial" means "characterised by being the object of commerce". Commercial vs. free-of-charge refers to pricing; proprietary vs. open source refers to maintainability. Those are different concept categories entirely.
Many people resist these notions because they come from a world-view where the issue of software's maintainability never arises — or appears not to: You buy or download software; when it suddenly disappears or the project dies, you simply lose, and nobody thinks of fixing that situation because nobody sees it as a problem.
If a software package is available under open-source terms, then that is the case regardless of whether you or anyone else paid money for it (thus, regardless of whether it is "commercial" or not). Conversely, if it's not available under such terms, then it's proprietary even if always downloadable for free (and thus not "commercial").
Q: But don't non-commercial and open-source equate the the same thing, since you can always get a copy of any open-source package free of charge or token amounts covering little more than the cost of duplication?
A: First of all, the premise is wrong: The freedoms in question involve people who already happen to have copies being allowed to sell you a copy at any price they wish, including zero, but they don't have to. In general, you lack any inherent right to receive any copy in the first place, let alone one for little or nothing.
Second, the question assumes that the only significant economic aspect of software is original acquisition; that nothing after acquisition qualifies as commerce. As any business that develops and uses -- or just uses -- software knows (and remember that the majority of software used in business is developed and maintained in-house), that's flat-out false. Most of the cost of software is maintenance, debugging, support, upgrading, training, and customisation. Those cost money for both proprietary and open-source software.
Q: Hey, wasn't this FAQ supposed to be about proprietary software?
A: Right, it is. We'll get back to that, then.
Q: Are you saying that the term "proprietary" refers to closed-source software, then?
A: No. Software with source code provided is often furnished under proprietary terms (making it what we term "source-available" or "viewable source" proprietary software). This matter is a frequent cause of confusion, occasioned by people thinking the opposite of open source must be "closed source", and that the key issue is purely and simply access to source code.
The issue has never been access to source code. The key distinction, rather, is this: Does the public have the ability and legal right to keep the software alive by developing new versions, that are lawful to use for any purpose, or not? If so, it's open source. If not, it's proprietary.
Because it serves only to confuse the issue, and doesn't map to any truly useful concept, the term "closed source" is deprecated.
Q: So, are you saying that all licensing schemes are alike (in being lumped together as "proprietary") if they don't meet the open-source test? Are you calling them bad?
A: Not at all. Proprietary is a spectrum, just as "monochrome" is a spectrum all the way from blinding white to pitch black, united only by not having colours. In the same sense, proprietary licences run from incredibly generous and permissive to extremely restrictive.
On the liberal end, one has licences like that of the much-admired "xv" graphics utility. xv may be used for any non-business purpose at no cost, source code is freely available, and the program may be modified in any way desired that leaves the permission notices intact. The author's pristine code may be freely redistributed (but not any modified versions) in any non-commercial fashion. Any commercial use requires a fee.
University of Washington's proprietary Pine mail program and Pico editor have a similar licence.
Also at the liberal end are most packages by Prof. Daniel J. Bernstein, such as SMTP mail daemon Qmail and DNS nameservice daemon djbdns. These may be used for any purpose whatsoever without fee, and may be redistributed in source or binary form if unmodified. (There are some minor variations and addenda to these permissions.)
All of those permissions grants are generous and commendable. Their owners feel they have excellent reasons for issuing those rights and none further. In calling them proprietary, nobody aims to equate them to (e.g.) hyper-restrictive Microsoft software that you pay for (in essence, rent) by the year, that you're not allowed to criticise in print or benchmark without permission, that you're not allowed to reverse-engineer, that you must authorise the copyright holder to remotely disable for ill-defined reasons, and that's available in binary form only. There is a vast gulf between the two extremes.
And yet, we call both ends proprietary, not to suggest they're the same in users' experience of them, but rather in lacking open source software's defining characteristic of third-party developers' right to fork.
Q: What about software like Tripwire, that seems to be sometimes under restrictive licences, and also is spoken of as open source? Is it open source or proprietary?
A: Yes. That is, it's both. To explain, licensing is the terms a copyright owner specifies for an instance of his creative work. Nothing stops the owner from having a proprietary package plus an open-source one, and that's what Tripwire, Inc. does. It uses the GPLed version to improve brand recognition and
RM: Stuff to cover: Proprietary licensing piece needs to detail: Redistributable vs. non-redistributable. Code with no licence is proprietary by default. Permission attaches to instance; ergo instances can have different licensing. External factors can make something proprietary (e.g., patents). Also. What was the IIRC "no bullshit" or "no nonsense" license that came out in the early 1990s. CRC's favorite house. Borland. Allowed for install on both work and personal systems. Hard to find a representative copy on the Net. Have tried. Also: user vs. install licensing, and concurrency vs. instance licensing. MSFT CALs for example. But when I think of proprietary licenses, one of the points I hammer home is that you aren't getting a standard, or even a common-sense bundle of rights. You're specifically getting just those rights the licensor grants you. ...subject to change. And not necessarially making any degree of sense. There are restrictions to what you can do with the software. Making disparaging comments about the vendor is an example. Or what you can say about it. Much enterprise stuff proscribes performance comparisons or data. There's the use on multiple systems issue. More recently imposed by copy controls. Viz: TurboTax and/or Quickbooks. Both of whom took immense goodwill hits for totally fucking up users. To the point of modifying the MBR to audit installations. Windows typically limits use of product, not to someone with access to a system (which would seem sensible)... ...but to someone accessing the system from a Microsoft OS. ...eg: using VNC or the remote access thing I keep forgetting the name of, from a Linux box, may technically void your EULA. ...at least per MSFT's terms. There's the whole Windows Refund bit. There's installs vs. concurrencies vs. licensed users. G: WTS (windows terminal services) on Win2K server, which wasn't a concurrency or accessed based thing, but a specific remote host. You had IIRC ten default CALs with the server product. After which you had to buy more. And the CALs were considered consumed when a particular remote user ID connected. Initially, never to expire. All of the many aforementioned terms subject to change upon not-entirely-voluntary upgrades. After much user rebellion, there is now a timeout on the CAL expiry. On a random schedule. Not to be be less than n months or more than m, for some values n and m (3 and 10 come to mind, but check me). ...wrt updates. Which was another point. Generally: there *is* a common thread: The licensing terms are on a take-it-or-leave-it basis, selected by the vendor, and subject to change without notice at any time. Other than that, terms vary. Widely. Within a given vendor. Within a given product line. For a given product. For a given sale. Over time. You can quote me if you'd like ;- ...and without notice. There's also a long history of vendors not upholding their end of an agreement. There's also a highly uniform disclaimer of warranty and/or liability or indemnity, which is highly worth mentioning. And there's ye olde spectre of UCITA / 2B which are currently mostly dead, but worth mentioning. Ed Foster from InfoWorld, Cem Kraner from badsoftware, and... I think that's most of it off the top. gripeline.com, IIRC, is Ed's new blog. http://weblog.infoworld.com/foster/2003/09/01.html "closed source"