Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Fine Print: Intellectual Property for Engineers...

Fine Print: Intellectual Property for Engineers (with notes)

Noah Kantrowitz

August 12, 2017
Tweet

More Decks by Noah Kantrowitz

Other Decks in Programming

Transcript

  1. I am not an attorney and this is not legal

    advice. This is for entertainment purposes only. Consult a patent attorney for specific advice. No really don't do anything in a court of law based on the advice of a guy on a stage. No joke.
  2. This is mostly US-centric. Intellectual property laws and customs vary

    around the world. Again, consult an attorney.
  3. This talk covers a lot of topics and so by

    the nature of time I can't dive super deep on any one topic. Any one of these topic could easily be a whole career. I'm happy to answer questions afterwards, but I hope this will at least give you a solid basis to learn more.
  4. Okay, disclaimers out of the way. Let's start at the

    start, what is intellectual property?
  5. First we need to define property. For our discussion here

    let's consider property to be an object over which you have some level of exclusive control. In law, this control takes the form of you having certain rights on your property, rights which are exclusive to you. If you own some land, that means you get to say who enters it or builds on it. Not all rights apply in all cases, but roughly speaking that's enough for now.
  6. So back to IP, literally speaking we have property that

    is intellectual, as opposed to physical. If you paint a painting, you probably own the physical object of the paining, you can sell the painting, rent it, destroy it, whatever. But you also own the abstract, intellectual component that is your creative work manifested as a painting. And just like with the physical object, you have some rights over the abstract part too.
  7. Before we jump in to IP as it is today,

    let's quickly talk about where it came from.
  8. For the tradition the became US IP law, it pretty

    much all started in 1624 in England. Patents existed before then, but they were one-off decrees by the king or queen to allow monopoly control over whole industries and were very unpopular. The Statue of Monopolies transferred most of the control over to the English parliament, formalizing these monopolies and making sure they stayed temporary.
  9. This was revisited again in the Copyright Act of 1710,

    more commonly known as the Statute of Anne. Whereas the Statute of Monopolies was concerned with people "inventing" new industries, this was about the rights of a creative person over their work. While frightfully vague by modern standards, it established the legal tradition of copyright that we would inherit.
  10. The US constitution addressed IP directly, saying "To promote the

    Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries". This was later refined into both law and regulations under the jurisdiction of the US Patent and Trademark Office and US Copyright Office.
  11. In some other IP traditions, like continental Europe, an authors

    rights to their creation is considered an inherent or moral right. In the US we treat all IP rights as the domain of the law, you have only the rights assigned to you by law and nothing more.
  12. Always keep in mind that this is a patchwork of

    new laws updating the old. In many situations in IP law, the true answer to a question can only be determined by a court, and will be based on not just laws and regulations, but centuries of case law and other writings. I'll try to be careful in my word choice but almost everything I say would need a dozen "except for"s if I was being 100% accurate.
  13. To say it again, if you need specific advice on

    any of these topics, talk to an attorney.
  14. Bob sells copies of that book to a bunch of

    people who want to read it.
  15. And starts selling it themselves. Obviously if we want more

    people to write books, we have to prevent this. And thus was born copyright law.
  16. There are 6 rights that are assigned exclusively to a

    copyright holder. Three are about the rights of public performance which doesn't usually apply to software so these three are the most important to us. The first is the right to make more copies of a work, the second is make new works based on this one, and the third is the right to sell or give away copies of the work to the public. If you own the copyright for a work, only you get to do these things.
  17. The biggest piece of copyright law in the US is

    the Copyright Act of 1976. It basically called a full do-over on US copyright, defining those exclusive rights we just saw, what could be copyrighted, and what exceptions would exist. But the biggest change was that copyright is now completely automatic so you no longer have to register with the copyright office. You have copyright protection from the moment your work is created.
  18. For something to qualify for copyright protection, it has to

    meet three main qualifications. First, it has to be original, meaning you created it. Second, it must be a work of authorship. There is a list of 8 categories that count as works of authorship but the two you'll generally run in to in software are literary works and graphic works. Yes, this means that as far as the law is concerned, your source code the same as a novel. Third it must be fixed in a tangible medium. This means that a dance performance itself is not covered, but a video recording or script of it is.
  19. It's sometimes easier to think about what can't be copyrighted.

    First off, facts are not subject to copyright protections. Ideas or concepts aren't allowed because they aren't in a tangible medium. Anything made by a government employee during their job is not subject to copyright protection, though it may get other protections. A work must be creative to count, the usual counter-example being a phonebook, a simple arrangement of facts is not creative enough to be covered. And finally a useful article isn't covered, like the lines on graph paper. However that last one is currently in flux due the supreme court case Star Athletica v. Varsity Brands decided earlier this year which declared some elements of a useful article could be covered if they are sufficiently creative.
  20. And as the last bit of copyright to define, how

    long the protections last. The constitution specified that protections would be "for limited Times". Glossing over some procedural details about how you used to have to manually declare and renew copyright, all works published before 1923 are no longer subject to copyright protection, anything published between 1923 and 1978 is protected for _at most_ 95 years from publication. Works published after 1978 and owned by a person get the lifetime of the author and then 70 years from their death, while works owned by a company are either 95 years from publication of 120 years from creation, whichever is shorter.
  21. The elephant in the room of copyright term is Mickey

    Mouse. Disney is certainly not the only company that lobbies for increased copyright protection, but the Copyright Term Extension Act of 1998 is, for better or worse, deeply associated with the company. As it stands, almost nothing has entered the public domain since 1923 and the impact on the arts is still being felt today. 95 years from 1923 happens to be 2018 so we may see some changes here soon, either things finally losing protection or new laws extending the term.
  22. So we know our rights, we quality for copyright, the

    next thing you'll usually see are these words: "all rights reserved". What do these mean? It's an explicit statement that you, the owner, are keeping all 6 of the exclusive rights for yourself. But remember this is also the default state, any right not shared by the owner are exclusive to the owner.
  23. If copyright is automatic, what about things I do at

    work? Those are classified as a "work for hire". In most cases the copyright in any work you are directed to produce in the course of your job automatically belong to your employer. The specifics will depend in your employment contract in most cases though, as it generally supersedes the default work for hire rules with more specific provisions.
  24. I'll cover both of these in more detail later on,

    but briefly a license is a contract between a copyright owner and some other person to allow limited use of one or more of the exclusive rights. In open-source software the other person is usually a generic "anyone". Fair use is a exception that allows the use of a copyrighted work under certain conditions, such as me using the copyrighted Mickey Mouse graphic on the previous slide as part of a comment and critique on Disney.
  25. That covers the overall concept and structure of copyright, so

    let's look at a legal case: Oracle America Inc v Google Inc.
  26. The case had 3 major claims of copyright infringement, Oracle

    claimed Google infringed on the code of the rangeCheck function, the Java API as a whole, and several other code and documentation files. It also had some claims around two patents be we haven't gotten there yet.
  27. Setting aside the hilarity of the trial, let's just look

    at how the decisions played out. The first jury found that the rangeCheck function and the overall Java API were infringing, but the documentation was not. However they were deadlocked as to if the infringement on the APIs were fair use and did not reach a verdict.
  28. The judge reversed both the documentation claim and the API

    claim. For the latter he held that APIs do not meet the requirements for copyright protection as it is a "system or method of operation".
  29. Oracle then appealed the decision, and an appellate court overturned

    the original judge, saying that APIs are a concrete statement in a tangible medium as well as being a creative work, and thus are protected.
  30. Because the first trial hadn't reached a verdict on if

    the API infringement was fair use, a second trial was needed to answer that.
  31. Aaaaaand whoops. The jury came back with a verdict that

    yes, it was fair use. As you would expect, Oracle is again appealing this decision. So as it stands in case law, APIs are probably copyrightable, but you might not be able to really enforce that copyright if copying them is considered fair use.
  32. So why bring up this case? Because it shows that

    copyright is really hard. Even though I keep saying answers can only come from a trial, this case shows that even the courts are struggling to understand copyright as applied to software and technology.
  33. I'll contradict myself a bit later on, but whereas copyright

    protects specific, tangible works, patents protect the entire concept of a given invention.
  34. But similarly to copyright, patents provide an exclusive right to

    the holder, the right to exclude. This means you can prevent anyone else from using your invention. Notably this is only a negative right for others, not a positive right for you. Even with a valid patent, it might be illegal to actually produce your invention for a variety of other reasons.
  35. For an invention to be patentable it has to meet

    three criteria. First the invention must be useful, as in actually do something. This is generally an easy bar, but rules out things like perpetual motion machines or an invention which exists but doesn't actually accomplish anything.
  36. Next it has to be novel, meaning a new invention,

    meaning it did not exist previously and is sufficiently distinct from any pre-existing inventions, called prior art in patents, that your new patent will not cover them.
  37. And it must be non-obvious, meaning that at the time

    of invention it would not have been obvious to a "person having ordinary skill in the art". This is often the most contentious as patent examiners do not know every single industry and often must rely on post-hoc establishment in the courts to know if something was obvious or not, especially in software patents.
  38. Not always, some types of inventions are deemed too categorically

    questionable for patent protection. There's a ton of subcategories on each of these, as well as totally separate things for plant patens and design patents, but most utility patents have to be one of these four things.
  39. Or to reverse it, you cannot patent a pure idea

    with no functional invention made from it, nor you can you patent universal absolutes like laws of nature or mathematical formulas. You can patent novel algorithms though, so the line there can be a bit fuzzy. And finally you can't patent any naturally occurring substance or process, even if you've put in a lot of effort to isolate it. But again, the line between a manufacturing process and isolating a natural substance can be real thin.
  40. The fundamental theory of patents boils down to enabling disclosure.

    We want to encourage inventors to make awesome new things, and so in return they get a 20 year monopoly to turn a profit on their invention, but a quid pro quo is that they must document their invention so thoroughly that after those 20 years, anyone in the same field could reconstruct it. So inventors make their buck, and society at large is improved by having more and more inventions to draw from over time.
  41. The target of enabling disclosure is this "person having ordinary

    skill in the art", which we saw before in the Non-Obvious requirement. This is called a legal fiction, a made-up person used to make certain points or assumptions in law. This also means we don't have a hard and fast rule as to what qualifies as "ordinary skill" so the courts kind of have to make it up as they go.
  42. For the most part patents last 20 years from the

    date of filing. Some extensions exist if the review process takes a long time or if it was filed before 1995, but that is usually only adding a year or two. This is a lot shorter than copyright, but companies have worked out ways to slightly tweak designs and file new patents, giving them 40 or 60 years instead.
  43. The case revolved around four patents held by Alice Corp

    covering computerized versions of payment escrow systems. Alice claimed that CLS Bank was infringing on their patents, CLS counter-claimed that the patents were invalid.
  44. Skipping over the appeals process since it isn't as interesting

    as the last case. Alice lost a bunch of times and appealed all the up to the Supreme Court.
  45. In the majority opinion, the Supreme Court held that not

    only were all four of Alice's patents invalid, that whole area of patents were invalid too. In short, taking a non-patentable abstract process or business process and putting it on a computer is insufficient to make the invention patentable.
  46. This has led to many thousands of rejections, show here

    by quarter, and hundreds of patent invalidations. That is an average of over 100 rejections a day citing Alice specifically.
  47. To call this a big deal is an understatement. The

    vast majority of software patents now have a target on their back. The world of tech patents is now firmly divided into the pre-Alice and post-Alice eras. Some have called for congress to reform section 101 to re- allow these kinds of patents, but only time will tell how things go.
  48. Roughly speaking trademarks and service marks, although I'm only going

    to say trademarks from now on because they are basically the same thing, protect brands. They identify the source of commercial goods and services so consumers can trust that a thing they purchase is what it says it is. This is a bit different from copyrights and patents as it isn't to encourage the creation of new works, but to encourage the creation of brands and thus facilitate commerce.
  49. As a concrete example, Wikipedia. The Wikipedia brand has a

    lot of value, people trust it to be a mostly correct encyclopedia and the Wikimedia foundation has put in a lot of work to develop positive feelings towards the brand.
  50. Let's say I wanted to make a competing website and

    get in on that sweet donation cash. I know I can't copy the articles directly because it is protected by copyright, but what if I just name it MyWikipedia and hope people think it's related to the real site.
  51. The Wikimedia foundation would obviously not want me using their

    brand's good will and trust, and so would serve me with a well-deserved trademark infringement lawsuit.
  52. So what can be a trademark? Just about anything, it

    turns out. Most marks are words or graphical logos, but anything with a strong brand association can probably be a trademark. This includes the shape of the classic Coca-Cola bottle and MGM lion roar.
  53. In order to register for trademark protection you need to

    either demonstrate actual use in commerce via a specimen of your product or service, or file an Intent- to-Use document stating you plan to use it in the near future, and then prove your use later on.
  54. But like with copyright, registration isn't actually required. You can

    put a ™ on anything you assert is a trademark, with no legal or official oversight, however registering your trademark with the USPTO has some advantages. First, you get to use the snazzy ® symbol, but more importantly you start off any lawsuit with a presumption of validity of your mark. With just a ™ you have to start off any infringement case by proving that your mark is valid in the first place.
  55. Unlike copyrights and patents which have a fixed term, as

    long as you are using a mark in commerce, it can last forever. For a registered trademark you do need to file renewal paperwork every 10 years to prove you are still using the mark, but a mark can last as long as the brand it protects.
  56. Trademark infringement is a complex issue with a lot of

    factors, but the 8 most common considerations are known as the Polaroid Factors after a 1961 infringement case. I won't go over them here, but roughly if a random person on the street would confuse the two brands, it is likely infringing.
  57. And one last word you may have seen in connection

    with trademarks is the concept of dilution, or the idea that a mark must be actively defended via litigation to remain valid. This is technically correct, and there are real cases where valuable trademarks have been lost due to dilution like aspirin or escalator, however it is a lot more rare than some news outlets make it sound. As in less than a hundred cases in the history of US trademark law. This is, for example, why Google continues to insist they are not a verb.
  58. Very simply, trade secret law protects things you don't want

    to disclose but are still valuable. Of course you can't make something a secret again, but you can at least sue the leaker.
  59. The canonical example is the recipe for Coca Cola. Recipes

    aren't subject to copyright, and can't be patented, so trade secret protections give the company some basis for keeping it to themselves.
  60. Why put a license on software in the first place?

    If you remember back to the copyright section, by default all rights stay with the owner of the copyright. In practical terms this means that code put somewhere public probably can't be actually used by anyone else in the same way that you can't walk in to an art gallery and start making t-shirts of their copyrighted paintings.
  61. The original, and probably still most common form of software

    licenses are commercial licenses. You give the copyright holder some money and they give you a license that allows some limited rights like copying from the install disk to your hard drive, but not the right to re-sell the software to the public or create derivative works. This is generally wrapped up in the End User License Agreement or EULA, along with a cornucopia of legally dubious contract clauses.
  62. But the place most of us care about licenses is

    in open-source software. Instead of a EULA between the copyright holder and a single purchaser, we want a license that applies to anyone.
  63. Starting from the top, the MIT license is just about

    as simple as they come. Let's walk through the various sections. We start with a declaration of who owns the copyright and what years the copyright covers. As mentioned before, copyright is automatic, we don't need to do this to be protected, but we want it to be clear who owns the copyright. Next we take some of our exclusive rights and say anyone can use them subject to some conditions. We specifically grant access to the rights of duplication, derivative works, and sale to the public, which are the important ones for software.
  64. There are three conditions you must meet to use rights

    granted by the license. First, the notice condition which requires the code always be accompanied by the original copyright statement and this license text. Then we have the yell-y part which has conditions two and three, a warranty disclaimer and limitation of liability. Together those help insulate the copyright holder from some default provisions in most state laws around products being assumed to be fit for purpose. Put together this means the user gets enough rights to use the code in exchange for some fairly minor requirements, and thus we have open-source.
  65. Usually the next question after talking about the MIT license

    is "what about BSD?". The BSD family of licenses are minimalist like MIT, but adding one or two extra conditions around publicity and advertising.
  66. Unfortunately software patents have crashed this party. Being minimalist, both

    BSD and MIT only deal with copyright licensing, not patents.
  67. The first kind of interaction with software patents that we're

    worried about is me or my company sharing some code that we claim is covered by a patent we own. This means we need to grant users of the software a license to those patents, or sharing the code doesn't accomplish much. This can be handled in the main license, or via an auxiliary patent grant like Facebook's shown above.
  68. The other case is dealing with potential future patent claims

    from contributors. The nightmare scenario is someone sending in a patch and getting it merged, and then some time later their legal department rolls up and says "well actually, that code is covered by one of our patents" and tries to sue not just you, but all your users. To the best of my knowledge this hasn't happened yet, but newer licenses try to avoid it from the start. The text on screen is part of the Apache 2.0's patent grant section.
  69. So that Apache license. It's a much more recent creation

    than than MIT or BSD licenses. It has the patent clause as well as trademark licensing, more precise definitions of the rights assigned, and has some text around contributor licensing. Like MIT and BSD, it still puts few requirements on any derivative works other than that they maintain attribution and the license text.
  70. Originally described by Richard Stallman in 1985, the idea of

    copyleft is a license which allows derivative works but requires those new works to be licensed similarly to the original. Compare this to the MIT license we just saw which grants the right to create derivative works but says nothing about what those look like. This philosophy has led to a whole family of copyleft licenses, mostly varying on what constitutes a derived work and who must be permitted access to source code.
  71. The most famous and most common copyleft license is the

    GNU General Public License, which requires all derivative works be licensed under the GPL or a compatible license and that anyone with access to the software must be provided with the source code if they request it.
  72. The conditions of the GPL inspire a lot of fear

    and confusion, so let's talk about the two main ones. The first is the viral condition shared with other copyleft licenses. This means that any derivative or combined work that is based on GPL code must be distributed under a license that places no more restrictions on the user than the GPL license itself. Unfortunately exactly what qualifies as a derived work is complex.
  73. The obvious case is taking the code of something under

    the GPL and modifying it, that is unambiguously a derived work. The tricky cases are linking. The creators of the GPL insist that any linking between software components creates a derived work, but that network or file I/O does not.
  74. There is a similar license, the LGPL, which specifically allows

    linking as creating a non-derivative work. That means any modifications to the library itself need to be GPL compatible but use of the library without modification does not.
  75. Oddly enough the most applicable case law comes from the

    decision in the case Galoob Toys v Nintendo of America where Nintendo tried to assert that the game genie created a derivative work and thus infringed on their copyright. The court held that no, it didn't. This certainly lends weight to claims that the GPL is actually much less viral than it is made out to be, if at all, as dynamic linking would not create a derived work and so would not activate the virality. As always, only a judge can say for sure.
  76. The other main bugbear is code sharing. This requirement says

    that any user of GPL code must be able to request the complete source code for software they are using. This ensures the user always has the freedom to modify any GPL'd tools they use. With the GPL itself this only applies to people with access to the actual binaries, but another license, the Affero GPL, extends it to anyone using a service over the network too.
  77. So far I've mostly been talking about licenses as applied

    to software code, but projects will hopefully include non-code portions too like documentation or images. Creative Commons offers some well-tested licenses for these non-code works that allows you to pick which kind of license you want. They also maintain the CC0 rights waiver I mentioned earlier.
  78. Here I get to put my finger on the scale

    a bit and give my personal opinion on which licenses to use for new projects. I take a hard stance that the only two options which make sense are GPLv3 and Apache- 2.0. Some people are still attracted to the simplicity of the MIT and BSD licenses, but in a world of software patents they are no longer safe for you or your users.
  79. And if I can go a bit further, I'm firmly

    in the permissive license camp. There are use cases where copyleft is important, but chances are your code isn't one of them.
  80. A CLA is a contract between a contributor and a

    project, or a company acting as caretaker of a project. This usually looks very similar to the license used for end-users, but with some additional sections where you assert that you have the legal authority to make the contributions, i.e. you own the copyright or have permission from them and are granting a patent license.
  81. In some projects, CLAs have been replaced by the Developer

    Certificate of Origin process which uses a much smaller attestation but has to be agreed to on every commit, usually by using a note in the commit message, but makes the same kind of statement that you have a legal right to give the contribution and the project can assume patent safety.
  82. To the best of my knowledge, there have not been

    any legal cases involving either CLAs or the DCO, so the best guidance I can give here is that very good lawyers have decided that one or the other of these is required to safeguard their projects, talk to your own attorney or legal department for advice on your own projects.
  83. The whole law is too complex to go over but

    there are two sections you should probably know about. First is the safe harbor provision. It exempts ISPs, caches, search engines, and websites hosting user-created content from copyright lawsuits under certain conditions. This is, for example, why YouTube can't be sued for users uploading copyrighted videos as long as they take them down when notified of infringement.
  84. In order to qualify for safe harbor protections, you must

    first make sure that you have no knowledge of copyright infringement ahead of time. This usually means making sure it is mentioned in your terms of service and that you inform your users that copyright infringement is not okay.
  85. Second, you need to register your copyright agent. This is

    done through an online portal runs by US Copyright Office. Generally your registered agent will be your company attorney, but anyone that agrees to receive your notices can be used.
  86. And finally you need a process for copyright holders to

    notify you of infringement, generally called DMCA takedown orders. You need to collect a bunch of specific info about the alleged infringement, and offer the user responsible the option to counter-claim that it is not infringing.
  87. Ensuring compliance with the DMCA safe harbor rules are pretty

    important to any modern website that allows user content. Make sure you talk to a lawyer. Really.
  88. And the second bit you've probably seen, DRM protections. It

    created criminal penalties for circumventing copyright protection systems, or helping others to do the same.
  89. There have been many attempts to use these anti- circumvention

    laws to prosecute DRM removal tools, with mixed success. The Library of Congress has made it clear some some level of personal DRM circumvention is definitely allowed as fair use
  90. The EFF and ACLU continue to challenge the DRM sections

    of the DMCA, and i hope some day soon they succeed in defeating them.
  91. I mentioned way back at the start that only the

    copyright holder of a work has the right to sell or distribute it to the public. A major exception to this is the first sale doctrine. This holds that while you continue to hold copyright over a work after selling it, you lose most control over the physical object. This means if I buy a book, I can't starting printing more copies and selling them, but I can sell that specific book.
  92. Unfortunately this right has been significantly eroded in recent years.

    The general opinion of the courts has been the most modern media sells you only a license, not an object that would be subject to first sale doctrine. A CD counts because it is a thing you own, while an MP3 file might not, depending on very specific readings of the terms and license of the sale.
  93. This came to a head in the case of Vernor

    v Autodesk, when a man tried to resell a copy of some software on eBay, and the court held that no, did not actually own the software and as such could not resell it. The same argument has been used by other cases, and has been generally upheld that digital purchases are the sale of a license, not of the work itself.
  94. Be very careful, and make sure to donate to the

    EFF so they can keep fighting for consumer rights.
  95. Fair use is a principle by which you are allowed

    to infringe on the IP of others under certain circumstances. Mostly it is used in the context of copyrights and the use of copyrighted material without the permission of the owner.
  96. The most common confusion about fair use is that all

    non-commercial use is automatically okay. While commercialization is a factor, this is not a blanket authorization to infringe on copyright. Cut it out YouTubers.
  97. The most common kind of fair use is using a

    copyrighted image or trademarked name to refer to the owner of the work or mark.
  98. For example, I can tell you I wrote this using

    Apple Keynote despite both of those being trademarked terms. Similarly, a news show can use a copyrighted logo when running a story about the company as long as it's there to inform the listener in a nominative way.
  99. As with most things in IP, as I keep saying,

    if something is fair use or not can only be established by a court, but there are four main tests used by the courts to decide.
  100. First is the purpose and character of the new work,

    usually viewed through the lens of how transformative it is. A parody song, for example, might use the melody of another copyrighted song but transforms the rest of it considerably. The more verbatim copying you do, the less likely something is fair use.
  101. Next is the nature of the original work. Copying verbatim

    from something that is entirely or mostly factual rather than creative is more likely to be fair use.
  102. Third is the amount of the copyrighted work you used.

    More copying is less likely to be fair use, but neither very little nor very much is automatically one way or the other.
  103. And last is the effect on the original work and

    its owner. Even very small, transformative copying can be not fair use if it is attempting to diminish the market for the original. Taking a parody song example again, the existence of that parody song is unlikely to reduce the market for the original and might even increase it.
  104. If you are unclear on fair use, just assume you

    probably won't qualify other than clearly nominative situations. It's safer that way. Then go ask the original author about what license their work is under and if they would consider other licenses.
  105. We talked about copyrights and how they protect new artistic

    or creative works, patents and how they encourage novel inventions, trademarks and how they facilitate commerce, trade secrets and how they let companies keep things to themselves, the public domain and the value of the commons, and how licenses allow people to use software while navigating all the protections of IP.
  106. The Cornell LII has a great web-based copy of as

    much of US law and regulation as they can manage. If you want to look up any of those USC references in this talk, go there. The book IP and Open Source by Van is a fantastic 0-60 guide on all of this with much more detail on how it pertains to software. The USPTO website has lots of educational resources as well as the TESS trademark search system. USPTO also has a patent search system but Google Patents is often easier to read and navigate. And finally the This Week In Law podcast from the TWiT network is a great way to keep up to date on new happenings in the IP world.