You've come to this page because you've asked a question similar to the following:
What are licence-free softwares?
This is the Frequently Given Answer to such questions.
A licence-free software is a softwares that is copyrighted but not published under a copyright licence. As such, you are permitted to do everything that the law says does not violate the rights of the copyright holder.
What exactly this means varies from country to country, as copyright law varies from country to country. However, in very broad strokes things are much the same in most of the countries that are signatories to the Berne Copyright Convention, simply because all of those countries take a common approach to copyright.
People are generally surprised how much one can do in the absence of a copyright licence. The way that copyright licences are presented leads people to believe that they aren't allowed to do anything at all unless it is licenced. However, the law is not in fact so strict.
The copyright licences that are used by commercial software houses imply that one is not allowed to do anything at all in the absence of such a licence, and that if one does not accept the licence one cannot use the software.1 So, too, do the older "copyleft" copyright licences employed by the Free Software Foundation. (See § 4 of the GNU General Public Licence version 2, for example.2) Dan Bernstein says that one has been "bamboozled by Microsoft" if one believes this.
Whilst it is false to say that Microsoft is behind this, it is most certainly true that the people behind these copyright licences, including the drafters of the GNU GPL version 2, misrepresented the situation. The principle embodied within Hanlon's Razor says that this is best attributed to they themselves not understanding copyright law, and, knowing this, framing their copyright licences in the most rigid terms possible in order to cover themselves.3
The elimination of people owning softwares to assert copyrights over them, and thus the elimination of copyright licencing, is of course the goal of the Free Software movement. Licence-free softwares are thus the ideal. In practice, however, that ideal cannot be attained with the law as it currently stands, because although the law permits more than many people expect it to permit, it still allows copyright holders to prevent others from distributing copies of softwares.
But in the case of most licence-free softwares, the softwares are distributed directly by the copyright owners anyway. So whilst B cannot distribute A's softwares to C, C can always obtain the softwares directly from A. Furthermore, B can provide to C a set of instructions for modifying A's softwares, once C has obtained them directly from A, in the form of a patch. The patch is B's work, over which B has sole copyright.
And this is how most licence-free softwares are distributed. For example: When Dan Bernstein's qmail package was copyrighted, before it was released to the public domain, one could obtain qmail from Bernstein and a set of patches to it (a few of which can be found here) from others.
It has been argued that having to obtain the original software and a set of patches, and then apply the patches to the original, is unusual, tedious, and inconvenient. However, this argument founders on three points:
It isn't, in fact, unusual at all.
Most people obtain proprietary softwares as originals+patches. Indeed, this is the norm for softwares such as Microsoft Windows. Windows even has an entire built-in system, Windows Update, for processing the retrieval and application of hot-fixes.
In terms of determining the exact copyright status, it is a lot more convenient than the "copyleft" and BSD routes.
With licence-free softwares, the copyright statuses of the original software and each patch set is clear and unambiguous. In contrast, one of the perennial problems with "copyleft" softwares is noting which author owns the copyright on which particular part of a combined work. This is particularly problematic with multiply-licensed softwares, where it is possible, and regularly happens, that not all authors' contributions to the overall work are licensed under the desired licence. In many cases, the work has to be be picked apart line by line in order to determine who owns what; and as a consequence people just give up because tracking what changes to the software are owned by whom is too complex.
"copyleft" software, for projects that don't fail, has exactly the same original+patches system.
The Linux Kernel Mailing List, for example, sees the publication of tens or even hundreds of patches per day. Just as with licence-free software, to construct a Linux kernel one takes a base software release and picks and chooses further patches to apply as people publish them. There are people who will pick and choose patches on one's behalf, choosing what people's patches to trust and to apply, and there is a complex bureaucracy of "maintainers" and "distributors"4 that arranges to create and to supply kernels with appropriate patches, in the form of regular "releases" and "development versions"5.
But this is merely a matter of manpower, and a difference merely in degree rather than in kind. For qmail, there was exactly the same system, with people around the world publishing patches and a group of people who got together and decided what were the "best" patches, whose patches to trust, and how best to improve the software. This produced netqmail.
Indeed, several people have made the case that it is this bureaucracy, that selects and rejects patches written by the world at large to create an "official" set of patches, centralized at its apex with a "benevolent dictator", and not the copyright licencing, that is the reason that the Linux kernel and other such projects flourish. There are, after all, a lot of "free software" and "open source" projects that have failed, precisely because they lack this system of centralized patch review and selection.
The relevant part of the United States Code is 17 USC § 117. This says that:
It isn't infringing copyright to make copies or adaptations of a software if doing so is an essential part of using that software.
Since translating source code into machine code, or into an intermediate form such tokens or bytecode, is making an adaptation, and since such an adaptation is necessary in order to run the program, this means that it is not a violation of copyright to compile a program from source code, or for an interpreter to tokenize it, in order simply to run it.
It isn't infringing copyright to make archival copies of a software, as long as you destroy all archived copies when you transfer posession of the software to someone else.
Therefore you can make backups of, compile, and run all softwares, whether they have copyright licences or not.
The relevant part of United Kingdom law is The Copyrights Designs and Patents Act (as amended) § 50A, 50B, 50C, and 50D. These say that:
It isn't infringing copyright to make backup copies of a software, for purposes of its lawful use.
It isn't infringing copyright to translate (i.e. make an adaptation of) a program from a lower level language to a higher level language, as long as that translation is solely for the purpose of creating an independent program to interoperate with it, you have no other means for creating such a program, and you aren't creating a substanatially similar program in doing so.
In other words, you can "decompile" a program only as long as you are doing so to work out its API to create a different, independent, program; as long as that API isn't available to you in any other way; and as long as you don't use the information obtained by decompiling for any other purpose.
It isn't infringing copyright to make adaptations or copies of a software, if doing so is an essential part of using that software. In particular, it isn't infringing copyright to make adaptations that correct errors in a software.
So the same applies as with U.S. law. It is not a violation of copyright to compile a program from source code, or for an interpreter to tokenize it, in order simply to run it. And, over and above U.S. law, it isn't a violation of copyright to apply temporary fixes.
Therefore you can make backups of, compile, run, patch bugs in, and decompile (for specific and narrow purposes) all softwares, whether they have copyright licences or not.
1. This is why "click-through licences" are as they are. The software vendor is trying to turn the copyright licence from a licence into a contract, one that users agree to after and in addition to the contract of sale that they entered into in order to actually buy the copy of the software in the first place. The legal foundation for such one-sided contracts, where it isn't clear what consideration is actually being exchanged in one direction and where one of the parties is entirely deprived of bargaining power, is somewhat shaky.
Moreover, the legal situation where one has not agreed to a click-through contract and patched the software so that it runs anyway despite that would be rather interesting. One cannot be held to the terms of a contract that one has not agreed to, and one does not violate copyright (in the U.K. and the U.S. at least) by modifying a lawfully purchased copy of a program in order to make it actually run.
2. The revised FSF "copyleft" copyright licences attempt to make things clearer, by explicitly defining themselves to be covering just those acts that the law actually says are copyright infringements in the first place. Notice the GNU GPL version 3's definition of "propagation".
3. In its background to the creation of the GNU GPL version 3, the Free Software Foundation openly acknowledges that the GNU GPL version 2 "was a license constructed by one US layman and his lawyers".
MAINTAINERS file for Linux version 2.6.23 is 93KiB.
5. Such as Linus' kernel source code tree, the "-mm" kernel source code tree, the Debian kernel source code tree, and so forth.