Last month at the Open Hardware Summit we announced the start of the OSHWA open source hardware certification program. Part of that program involved issuing unique IDs (UIDs) for each piece of registered hardware. Since we knew that low numbers would be hot properties, we decided to wait until the end of October to assign them. That allowed us to give every piece of hardware registered in the month of October a chance for the lowest possible number.
Today, thanks to the good people at random.org, we have assigned those UIDs to all pieces of hardware registered before the end of October. We have also started assigning numbers sequentially for all pieces of hardware registered since November started. You can see all of the registered hardware here.
There were 60 different projects registered from 9 different countries. Want to get your own UID? Register here!
When a cheerleader wears her uniform, is she expressing her identity as a cheerleader or identifying herself as a cheerleader? And could the distinction ever even matter?
This is one version of the question that the U.S. Supreme Court wrestled with today. Its answers could have a big impact on how copyright interacts with 3D printed objects.
The Cheerleader Uniform Case Goes to Court
Today were the Supreme Court oral arguments in the Star Athletica v. Varsity Brands copyright infringement case, which you may remember as the case about cheerleader uniforms we’ve been writing aboutsincelastyear. On its face, the case is about how and whether copyright should protect designs on cheerleader uniforms. We care about it here at Shapeways because it is really about how copyright interacts with objects that combine functional (traditionally not protected by copyright) and non-functional (traditionally protected by copyright) elements in one.
If you are thinking about licensing your 3D printed object or worried that someone is infringing on your work, the first thing you need to know is if the object is protected by copyright at all. The decision in this case could give us a single, clear rule to use in answering that question. Appropriately enough because this is Halloween, it will also be helpful for those of you who think about copyright in relation to costumes and cosplay.
Of course, while those larger questions linger in the backdrop of this case, the immediate question is about cheerleader uniforms. Specifically, it is about the designs on cheerleader uniforms.
As a general rule, copyright law views clothing as “functional” and therefore beyond the scope of copyright protection. In a nutshell, clothing is functional because it keeps you warm, dry, and not-naked. However, designs on clothing (think a flower print on a dress) can be protected by copyright. That is because the designs are not really functional, but rather separable aesthetic elements merely attached to the otherwise functional dress.
Why Are Stripes and Chevrons on Cheerleader Uniforms?
Much of the debate today was about the role of things like stripes and chevrons on cheerleader uniforms. Are they there for “expressive” (read: copyrightable) purposes so that the wearer can express her allegiance with cheerleaders? Or are they there for “functional” (read: not copyrightable) purposes so that the wearer can be identified as a cheerleader? And yes, this distinction also makes copyright lawyers’ heads hurt.
For a case about cheerleaders, there was a lot of discussion about camouflage in Court today. Camouflage as a design in a frame on the wall is probably protected by copyright. After all, on some level it is just an artful arrangement of colors that – at least in a frame on the wall – serves no obvious functional purpose.
However, if you print the same camouflage pattern on a shirt and pair of pants, it can suddenly have the very functional purpose of making it easier to hide in the woods. At that point the pattern would probably become functional and therefore – at least in the shirt and pants context – not eligible for copyright protection.
The stripes and chevrons on a cheerleader uniform can be thought of the same way. On a piece of paper in a frame on the wall, they are just an artistic collection of stripes and chevrons protected by copyright. On a uniform, they may serve important functional purposes: in addition to identifying the wearer as a uniform, they can make the wearer look taller or slimmer, or just reinforce various seams in the uniform itself.
Abstracted up a level, all of this discussion about putting parts of an object into functional and non-functional categories mirrors the analysis you would do to determine if a 3D printed object that is both functional and artistic would be protected by copyright.
In a previous post we flagged studiogijs’ birdsnest eggcup quattro as an easy example of this. It has functional (holding an egg) and non-functional (looks like a birds nest) properties. A rule coming out of the cheerleader uniform case could provide us with a rule to understand how – if at all – copyright might apply to the eggcup too.
What Happens Now?
It is always dangerous to make projections based on oral argument. It was pretty clear that at least some justices were wary of a decision that would pull all clothing into the world of copyright (at one point Justice Breyer worried that pushing copyright lines deep into clothing would “double the price of women’s clothes” for no real purpose). Many others struggled with how to draw the line between functional and non-functional elements of clothing.
It is no surprise that the Justices struggled drawing this line. Going into this case, there were at least ten different theoretical approaches that courts have used. Hopefully, the Supreme Court will come up with a single test that works for cheerleader uniforms, 3D printed objects, and everything in between.
We will find out once the written opinion is released during the first half of next year. Until then, you can read the transcripts or hear the audio of the arguments (don’t miss Justice Breyer – again – working to understand the relationship between clothing and the context by musing, “The clothes on a hanger do nothing, the clothes on a woman do everything.”) And let me know if you have thoughts or questions in the comments below or on twitter @MWeinberg2D.
Today the Open Source Hardware Association is excited to announce the OSHWA Open Source Hardware Certification program. The program allows hardware that complies with the community definition of open source hardware to display a certified open source hardware logo. It will also make it easier for users of open source hardware to track down documentation and more information. You can check out the program here, and keep on reading for some information about how it will work.
These companies and organizations have agreed to certify at least one piece of hardware by the end of this year.
Background and Purpose
Almost a year and a half ago, OSHWA announced its intention to explore the creation of an Open Source Hardware certification. At that point, the open source hardware community had created a community definition of open source hardware and the open source hardware open gear logo. Both of these were significant contributions to the growth in awareness around open source hardware.
By design, no one owns the term “open source hardware” or the open gear logo. This allowed both the term and the logo to be widely adopted by the community. However, it also created a challenge. In many cases, creators would label their hardware as being open source and use the open gear logo without complying with the community definition. This created confusion in the community where users were unsure what it really meant when something was labeled “open source hardware”.
The certification is designed to complement the existing open gear logo by bringing clarity to how the creator is using the term “open source hardware”. Unlike the open gear logo, the certification logo is controlled by OSHWA. In order to use the certification logo, a hardware creator must make a legally binding promise that their hardware complies with the community definition of open source hardware. That means that when users see the certificated open source hardware logo they know the hardware complies with the community definition of open source hardware.
Upon registration, you will receive the right to use the certification logo and a unique identifier for the registered version of your hardware.
That unique ID has two components. The first two characters represent a country code. Open source hardware is a worldwide community, and the country code is an opportunity to recognize and celebrate that fact. The remaining characters are numbers assigned sequentially based on when the hardware was registered.
The Importance of Your Contribution
One of the critical elements of the certification process is the idea of the creator contribution. Hardware often involves many third party components outside of the control of the designer. Therefore, it is unreasonable to require every component of a piece of hardware to be fully open before becoming certified.
The creator contribution rule is an attempt to address this issue. In order to certify a piece of hardware, the creator must fully open any elements of the hardware that is within their control. If they have the legal ability to license it, it must be licensed openly. If they created documentation, that documentation must be freely distributed.
If components were created by third parties, the creator does not have the ability to license them or distribute the documentation. In those cases, the creator merely has to disclose the parts in the bill of materials and certify that they are available without an NDA. This system allow creators to distribute legitimately open source hardware without creating a (currently) unrealistic burden to find fully open components.
OSHWA has designed this process to be as user friendly and straightforward as possible. However, it does require users of the certification to make a legally binding promise that they will only use the certification logo on hardware that complies with the community definition. We recognize that this sort of legal promise can raise concerns or cause uncertainty for some community members. Therefore, we encourage users to write in to firstname.lastname@example.org with questions.
While OSHWA will work hard to answer as many questions as possible, some questions may require legal advice. That is why we have partnered with the Cardozo Law School Tech Startup Clinic. The clinic has agreed to receive referrals from OSHWA for free and low-cost legal advice regarding the certification process. We hope to add additional clinics and resources to this program in the future. (If you are a lawyer who would like to be involved with the referral program, please contact us at email@example.com).
About Those Unique IDs
Ever since we started discussing unique IDs that were sequential, we’ve gotten variations of a single question: “how can I get a low number?” In order to give as many community members as possible the opportunity to obtain a coveted low unique ID number, we have decided not to start issuing numbers until the end of October. All hardware registered before the end of October will be combined into a single pool, from which we will issue IDs starting November 1. That means that all pieces of hardware registered before the end of the month have an equal opportunity at low registration numbers. Starting November 1, we will start issuing IDs as hardware is registered.
This is the first version of the OSHWA Open Source Hardware Certification program. It was designed with a great deal of community input and is intended to complement existing pillars of the community. We hope that you take a look, try it out, and find it useful. At the same time, we know that as the certification program grows we will find things that can be improved and applications that we failed to anticipate. That’s why we welcome your thoughts and ideas in the comments of this post, in the forums, in the firstname.lastname@example.org inbox, and online @OSHWassociation.
In order to provide that feedback, you first need to check out the certification. So stop reading this blog post, click on over to certificate.oshwa.org, and register something!
News broke today that there was a decision in the case that Defense Distributed is bringing against the State Department related to putting files for 3D printing and CNC milling guns online. While this is a decision on a preliminary question that largely punts on the super-interesting parts of this case, it still has a few noteworthy elements.
As a quick bit of background, Defense Distributed released
some files for 3D printed guns online.
The State Department sent them a letter telling Defense Distributed that
under the State Department’s interpretation of export laws known as ITAR,
Defense Distributed had to pre-clear the posting of such files with State. The theory is
that posting the files on the internet is the same as exporting them to foreign
While complying with State’s request, Defense Distributed
sued the State Department over their interpretation of ITAR, claiming that the interpretation
was unconstitutional. Before getting to
the real heart of the case, Defense Distributed asked the court to suspend the
State Department’s pre-clearance requirement until the matter was
resolved. It is that request – which took
the form of a preliminary injunction request – that resulted in the decision
this week. That means that the decision
today is not about the larger questions surrounding files as free speech, what
qualifies as “exporting” a digital thing, and when information becomes a thing.
The arguments around the preliminary injunction can be
summarized quickly. Defense Distributed
said that imposing a prior restraint on its ability to speak (in the form of
releasing files) infringed on its First Amendment rights (among other
things). The State Department said that
releasing files online harmed national security.
The majority opinion essentially took the claims of both
sides at face value. Defense Distributed
had Constitutional rights that could only be infringed upon in important
circumstances. The State Department was
showing a circumstance which, if it prevailed at trial, could very well be important enough to justify
infringing on those rights.
The majority was swayed because of what it saw as the long
term impact of those harms. If Defense
Distributed turned out to be right, the harm caused by preventing it from
posting the files online could be remedied at the end of the trial (and its
appeals….) by allowing them to post the files online (note: it is probably safe
to say that this is not a universally accepted theory of how to remedy the harms related to
However, if the State Department turned out to be right, the
harm caused by temporarily allowing Defense Distributed to post the files
online could not be reversed. This is because once a
file is online it is online forever, even if Defense Distributed took them off the
Defense Distributed site. On this narrow
point, this is a refreshingly clear-eyed understanding of how the internet
works from a court.
As a result of this balancing, the majority sided with the State Department
and declined to allow Defense Distributed to post the files pending the resolution
of the case.
The dissenting opinion took a very different view. I suspect
that, while Defense Distributed would have preferred to win this round, they
are at least happy with the fairly long and detailed dissent opinion.
Instead of taking the State Department’s claim that posting
the files online negatively impacts national security at face value, Judge
Jones’ dissenting opinion interrogates the claim and finds it sorely
lacking. It questions the
constitutionality of the ITAR framework generally, and rejects the way that the
State Department interprets it in this case specifically.
After undermining the State Department’s argument over the course of a number of pages, it is unsurprising
that the opinion sides with Defense Distributed’s constitutional claims and
argues in favor of its ability to post the files online.
This opinion may very well be appealed. If it is, the full 5th
Circuit will have to decide which analytic approach to take, and then how that
approach applies. Even when that process is over, the parties will have to go
back and actually argue the heart of the case (which, regardless of your
opinions about gun control, 3D printing, free speech, and export control law,
raises some interesting questions). In
other words, this decision is one among many before all of the questions are
That being said, the opinion is an interesting window into how courts will approach the many potentially complicated legal questions connected to this case. Since Defense Distributed has partnered with the Second Amendment Foundation to pursue this case, there is actually a reasonably good chance that that the case will last long enough to get some of them answered.
A designer releases a
new 3D printable shoulder rig for cameras under a Creative Commons Attribution
license. It quickly becomes wildly popular in the film world.
A company copies the rig but fails to give the original designer credit
as required by the Creative Commons license.
An internet freak out erupts.
company defends itself by pointing out that the Creative Commons license is not
legally binding because the rig is not protected by copyright. Upon investigation, the internet realizes
that the company is right. Legally, the
license isn’t worth the pixels used to display it on the screen.
Weeks later, another
designer releases a circuit board under an open hardware license.
Another company integrates the board into its
new product, ignoring the restrictions on the license.
The designer of the original board tweets a
complaint and a new internet freak out ensues.
Once again, the company defends itself by pointing out that hardware is
not protected by copyright, rendering the license legally irrelevant.
The internet starts calling all open licenses
The world is an incalculably richer place because of the
open source software movement and Creative Commons (collectively, for the
purposes of this post, OSS/CC). It is probably
safe to say that OSS/CC has succeeded well beyond the wildest dreams of its
early proponents, and has touched areas of society that would have been impossible to
imagine when the concepts were being developed.
Nothing that follows here diminishes that success.
However, as they become more popular, OSS/CC licenses are starting to be stretched beyond their
originally intended scope. That
stretching could ultimately lead to large tears in the licenses and the
communities that have grown up around them.
These tears appear because as the use of OSS/CC licenses
expands, they are being applied to works that technically, legally, they were never designed to be used with. In many of those cases, the licenses stop being legally enforceable at all. This lack of enforceability
may eventually reduce user confidence in OSS/CC and the vibrancy of the OSS/CC
To be fair, the problem described in this post falls squarely
into the “good problems to have” category.
After all, the tears only exist because OSS/CC has become so wildly
successful that the ethos they embody is spreading well beyond their originally
That being said, the tears are real. At some point soon the “open” community (broadly
defined) will have to come to terms with them.
Fundamentally, these tears occur when the conditional
sharing ethos of OSS/CC expands beyond the scope of what is actually protected
by copyright. Without copyright, the
conditions baked into OSS/CC become legally meaningless. The “stick” that backs them up
While I wish I had answers, this post is not intended to
resolve all of the questions raised by OSS/CC spilling beyond their original
boundaries. Instead, it is designed to help raise and frame the questions. It tries to explain what is going on, and to
outline the potential consequences of the continued informal drift in the
current direction. It also touches on
some alternative approaches, none of which are perfect. While this post does not include a tidy
resolution, hopefully it is a useful step in the collective process of working everything
Sharing “Born Closed” Works
Broadly speaking – and this entire post is going to be
“broadly speaking” – OSS/CC were created to solve a specific problem created by
copyright law. Copyright law
automatically protects everything that is categorically eligible for
protection. That means that things that
are protected by copyright – software for OSS; music, movies, photos, and
articles for CC – are “born closed.”
By default, digitally sharing born closed works created by
someone else violates copyright law.
That is because sharing works digitally involves making copies. Making copies of a work protected by
copyright (in most cases) requires permission from the original creator in order to
avoid infringing on the creator’s copyright.
Things like books and records are “born closed” and automatically protected by copyright. image: flickr user Michael Cory CC-BY 2.0.
That means that if you create software or music or movies and
want to share it with the world, you need to take affirmative steps to give
other people permission to use the work.
The easiest way to do this is with a license. “I won’t sue you for sharing this” licenses
make up the legal core of OSS/CC.
Within the world of copyrightable stuff, these limitations
are enforceable because failing to follow them voids the license. And without a license, the now-unauthorized sharing is a
violation of the creator’s copyright.
Among other things, this has allowed to OSS/CC community to impose its
ethos on people who do not care about openness.
Threat of a copyright lawsuit means people and companies who just want
to access the shared stuff have to play by the openness rules too.
The upshot of this situation is that OSS/CC trained a lot of
people to share, but to share conditionally.
This conditionality was (and is) important to creating a commons: it
gives people confidence in sharing and makes them feel like they are entering a
community with reciprocity. Conditional
sharing has created an unbelievable rich commons of code, art, and expression.
When they are used for works protected by copyright, OSS/CC licenses are more than polite requests - they are legally enforceable.
This entire system works because it just so happened that
the types of things that were relatively easy to create and share with a
computer and an internet connection – code, photographs, songs, movies, blog
posts – also happened to be the types of things that are automatically protected
by copyright. That made the entire
framework more than just a cultural norm relevant to a community of people – it
was legally enforceable.
As they built the commons, people internalized the “share
with conditions” model as normal. They
did not necessarily appreciate that the conditions were all linked to
copyright, or that copyright has limits as to what it can protect. In part, they did not realize this because
the limitations were not relevant – everything at the core of the commons was
protected by copyright so copyright’s boundary did not matter.
Shift and Expansion into “Born Open”
That is no longer the case.
A number of technical and cultural changes have begun to bring works
outside of the scope of copyright into the commons discussion. This is a testament to the success of OSS/CC
– the commons are so rich and the norms so appealing that new communities
naturally align themselves with it.
Instead of being organized around “born closed” works
protected by copyright, some of these communities are organized around “born
open” works that are categorically outside of the scope of copyright. The two highest profile (or at least the two I’m most familiar with) are the open source
hardware community and the 3D printing design community.
Neither of these communities are fully outside of the scope
of copyright. Some parts of any given
open source hardware project may be protectable by copyright, and many 3D
printed objects are protected by copyright.
However, the functional parts of an open source hardware project and more
utilitarian 3D printed objects are beyond the scope of copyright protection
(you can just accept that distinction for the purposes of this article, but
there is more information on it here
and here). That means that blanket assumptions about
copyright protection, and licenses that are built on a foundation of copyright
protection, do not apply as cleanly to born open objects (more about that here).
As a result, on some level and in a non-trivial number of
cases, using copyright-based licenses for works in these communities is
inappropriate. While they may look the
same as the licenses on software or photographs, in a significant number of
cases they will not be legally enforceable.
At the same time, they are there for both creators and users to see, and
to take into account. They also may carry cultural weight. That being said, the space between
the licenses’ visibility and enforceability creates costs (and benefits).
The Costs and
Benefits of the Current Path
To be clear, this evolution is happening right now and is actively
altering the status quo around copyright and licensing. That means that ignoring this evolution and
pretending it isn’t happening is a vote for this shift to keep happening. Doing something may be a vote for a different
kind of shift. But, as far as I can
tell, there is no vote you can cast to maintain the status quo. The only decision is between which different future you prefer to work towards.
A number of things that I would classify as negative happen
if people continue to apply copyright-based licenses to works that are not
protected by copyright:
Creators’ expectations are not met. If a creator releases a work under an OSS/CC
license, on some level they expect to be able to enforce the terms of that
license against someone who violates it.
However, if the work is not protected by copyright the terms of the
license are not enforceable. As a
result, there is likely to be a high profile instance of an OSS/CC license
“failing” to protect a work from being used in a way that violates the original
creator’s hopes. The story will appear
to have a “good” creator with a sympathetic story and a “bad” violator who acts
in a villainous, obnoxious way. However,
the good creator will be trying to enforce rights they do not have and the bad
violator will not have broken any legal rules.
Everyone will be sad that the bad violator can’t be punished and people will
start to worry about using OSS/CC licenses.
Legitimate usage is curtailed. The conditions in any OSS/CC license are
restrictions on use, and those restrictions disqualify some people from making
use of the work. If the license used is
not valid, the restrictions it attempts to impose on unwilling users are not legally
enforceable. Regardless, there will be
some number of end users who 1) want to use the work in a way that violates the
terms of the license, and 2) do not realize that the license does not legally
prevent them from using the work anyway. That means that some legally allowed uses never happen.
Copyright creeps outward. There is a deep irony in this one. Regardless of their real enforceability, if
people continue to use and see OSS/CC licenses
on works that are not eligible for copyright protection they will begin
to assume that copyright actually protects those works. They might even advocate for copyright to protect the works in order to make the licenses enforceable. The outcome of this is that OSS/CC licenses –
licenses designed to increase the pool of commonly available works – could
act to reduce the pool of commonly available works by dragging currently open works into the world of
copyright. This process could take the
form of resetting social expectations or, perhaps even more problematically from
the viewpoint of the OSS/CC community, actually be used to change the statutory
scope of copyright itself in an effort to make the licenses apply to more works.
There is also a potentially positive effect of this type of
Signals for use. The wishes of creators can be legitimate even
if they are not legally enforceable. To
that end, OSS/CC licenses can be signals to users about how creators of a work
want it to be used. There are plenty of
non-legal reasons for a user to want to respect these wishes, and using
standardized ways to communicate those wishes can make it easier to do so.
And one that could be argued either way:
May encourage more sharing. For some creators who are not naturally
inclined towards sharing, using an OSS/CC license gives them enough of an
assurance of control to convince them to make their works available to others. You can see this on how some people evolve from more-restrictive to less-restrictive licenses over time as they become more comfortable with the idea of sharing. The positive version of the use of these
licenses on non-copyrightable work is that more formerly private works will be
released into the public for others to see.
The negative version is that the creators will be mistaken with regards
to the terms those works are released under.
I withhold judgment on how those two factors balance out.
There are some ways to address these effects. None of them are perfect, or even necessarily
Expand copyright or create a new
easy-to-obtain right. The problem
described here is rooted in the fact that many works are not automatically
protected by copyright or some other easy-to-obtain right which can be used as the legal basis for a license. One way to solve it is to expand the scope of
copyright or create a new easy-to-obtain right that serves a similar purpose.
While doing so would solve the immediate issue with license enforceability, it creates many more and is a
horrible idea. Converting the universe
of “born open” works into “born closed” just to make it easier to openly
license them strikes me as counterproductive.
Also, while some of the creators of these newly born closed works might
license them openly, others would likely use the new rights to restrict what
today would be a free and open creation.
Public Education. I want to think that this alone can solve the
problem. If people really understand when to appropriately use copyright-based
licenses, this problem disappears. While
I believe strongly that public education is a key component of addressing this
problem, I’m also realistic enough to know that there will always be a massive
pool of people who have better things to fill their brains with than the
nuances of copyrightable subject matter.
Any solution has to work for those people too. That means we need more than public education.
Create New Licenses. If copyright-based licenses don’t quite cut
it, why not draft new licenses that do? Because – like creating secure
encryption that also allows the government to freely access the protected files
– doing so is a lot easier said than done.
Copyright is a great basis for a license because eligible works are
protected automatically and without cost.
People do not have to do anything to get copyright protection. All of the other obvious hooks for other
licenses – like patent and trademark – require time, money, and affirmative
decisions to obtain. Even if some people
will take the time to obtain them just to make it easier to license their work under OSS/CC, those people will never reach a critical mass or majority.
Stop Using Licenses. If current licenses do not protect all of
these new types of works, should people just stop using them? Unfortunately,
that’s not a great solution either.
While many works will not be completely protected by copyright, many
parts of many works are. The analysis of
which parts of which works are protected by copyright is somewhat work-specific
(because nothing here is easy), but it is safe to assume that many works will
be just protected enough by copyright that copyright could present a barrier to
some types of reuses. To the extent they
exist for a given work, OSS/CC licenses can clear those copyright-based barriers to sharing.
Contracts. Contracts can serve as an alternative to
licenses. While OSS/CC licenses rely on
copyright to be enforceable, contracts can just rely on the agreement of two
people. The creator offers the work to a
user on the condition that the user agrees to the terms. If the user breaks the terms she has
agreed to, she is breaking the contract with the creator. The problem with contracts is that they can
only be enforced against someone who agreed to them. That is always going to be a comparatively small universe of people. Whether they like it or not, everyone is
required by law to respect a copyright. That means
that the person who receives the work directly from the creator may be limited
by the terms of a contract, but the people who stumble upon it on a random
website would not be. This global
enforceability is what makes copyright law so attractive for licenses. Its
absence severely limits the usefulness of contracts as a substitute at scale.
The status quo is changing.
OSS/CC licenses are expanding beyond the universe of clearly
copyright-protected works that they were originally designed for. That is a testament to 30+ years of community
building in the open community. It also
creates challenges. Ideally this is the
point in the post where I would describe a clear path forward and advocate for
following it. Unfortunately, I don’t
have that path (yet). All I have is a phenomenon that
I believe is occurring and a series of ways the world is accommodating it.
Fortunately, the open world is strong because it is
collaborative. I’m already working with Creative Commons and other stakeholders to try and find some answers. Hopefully this post will pique the interest of others to join in. If it has (and let’s be honest - if you made it all the way down here it probably has) let me know.
Bonus: One thing that is connected enough to mention in a bonus but not so connected as to be worth integrating into the heart of the post is the history of the Open Game License. For all of its differences, it is another example of communities looking to licenses for openness where they probably didn’t quite fit and were not necessary. I won’t pretend to be an expert on the details of that community or the politics connected to all of it, but this article on it made a lot of sense.