Thesis, Automattic, and WordPress

Chris Pearson and Matt Mullenweg have hardly communicated with one another in the last five years, but they are ideological enemies. They have very strong personalities and unshakable beliefs on business and software. This is a story of their dispute, their idealism, and the implications it will have on the WordPress project.

Chris Pearson and Matt Mullenweg have hardly communicated with one another in the last five years, but they are ideological enemies. They are both wealthy individuals (though of different magnitudes) thanks to their online endeavors, with very strong personalities and unshakable beliefs on business and software. This is a story of their dispute, their idealism, and the implications it will have on the WordPress project.

Matt Mullenweg co-founded WordPress, founded Automattic, and is one of the most successful entrepreneurs of his generation. He runs a billion dollar “unicorn” startup centered on a culture of embracing open source technology and has achieved incredible success embracing principles counterintuitive to either Silicon Valley or big corporate culture. He’s paving a new path for how to create a valuable software company while religiously defending and promoting open source software.

Chris Pearson founded DIYthemes and helped pioneer the early WordPress commercial theme industry. He has run his business successfully for over seven years, despite unique hurdles that result from a very public dispute with Mullenweg in 2010. He vehemently defends his work as his own non-derivative achievement and rejects the religiosity and cult mentality that he believes exists in the WordPress ecosystem. He views WordPress as a huge chunk of the web, available to be monetized — which he has done so to the tune of millions of dollars — but he does not believe he must adopt Matt Mullenweg’s principles in order to meet customer demand, run his own business, and protect his own inventions.

By all normal definitions, Mullenweg and Pearson have done incredibly well for themselves. However, from a pure size perspective and principles aside, Mullenweg is the big nation army and Pearson is the small revolutionary militia. Mullenweg views Pearson as a threat to everything he stands for and has worked to accomplish, and Pearson views Mullenweg as an overbearing figure with no true authority over his decisions.

Mullenweg has the motivation, resources, and ability to squash Pearson — and indeed most thought he’d done so already. While he has far fewer resources, Pearson has some tools available to protect his business or to potentially even disrupt the entire WordPress ecosystem as we know it today.

During their first conflict in 2010, and in the resurgent one going on now, Mullenweg and Pearson have both at times made mistakes, acted childishly, or been in the wrong. Both also have merit in various aspects of their positions. Neither conflict, so publicly debated, reflects well on the WordPress ecosystem as a whole — even though I believe it is right that each conflict is best observed under a public eye, as the results can affect so many other businesses and potentially even WordPress itself.

With this post, I aim to outline the entire conflict; to describe the implications past, present and future; to highlight non-WordPress comparisons for precedent and potential implications; and to share my own thoughts on who is in the right and who is in the wrong, as viewed for the good of the global WordPress community.

A history of conflict

The commercial theme movement started in 2007 and took off in 2008. Thesis was one of the pioneers of commercial WordPress themes. The theme industry was young and evolving rapidly, and many sellers hardly considered or understood licensing issues at all.

Many of the sources for this period are from Siobhan McKeown’s excellent account in the book, Milestones: The Story of WordPress (which I’ll refer to as Milestones).

Themes as derivative works of WordPress

WordPress is licensed by the GNU General Public License (GPL), version 2 or later. The GPL ensures certain freedoms that protect both WordPress and those that utilize it. The “four freedoms” that are the heart of the GPL are as follows:

  • The freedom to run the program as you wish, for any purpose (freedom 0).
  • The freedom to study how the program works, and change it so it does your computing as you wish (freedom 1). Access to the source code is a precondition for this.
  • The freedom to redistribute copies so you can help your neighbor (freedom 2).
  • The freedom to distribute copies of your modified versions to others (freedom 3). By doing this you can give the whole community a chance to benefit from your changes. Access to the source code is a precondition for this.

As WordPress co-founder Mike Little phrased it in the Post Status Slack, “The GPL is meant to be restrictive for developers and permissive for users.” The GNU philosophy page and subsequent articles are a good resource for understanding the nature of the license.

The GPL is a Copyleft license, which creates the “stipulation that the same rights be preserved in derivative works down the line.” In an immature theme market, licensing was given relatively little notice, and many theme authors provided their themes with no license or proprietary licenses.

Matt Mullenweg, to many, would be considered BDFL, or Benevolent Dictator for Life, of WordPress. It’s a common term for folks that lead open source projects and have final say on project decisions. In his role as WordPress BDFL, he now has a reputation — at least within certain circles of folks that pay close attention — for making large, impactful decisions with little description of why he has done so.

In late 2008, more than 200 free WordPress themes were removed from the WordPress.org theme repository. While many of the themes were removed due to spammy links, some were pulled due to GPL violations within the themes or within the theme upsells that were linked from the theme listings.

The move, which was made without announcement, shocked many theme providers that felt they were unfairly included in the group of removed themes. The situation created a spark and initiated a serious debate about theme licensing.

Authors were concerned that GPL licensed themes would mean that their themes would be bought and freely distributed, removing their ability to make money from their works. A few, such as Brian Gardner with his Revolution theme, changed their licensing as a result of conversations with Mullenweg and Toni Schneider, Automattic’s CEO at the time. In Brian’s case, he made his theme free and offered paid support services.

Eventually though, most authors “selling” themes started actually selling support, access to download, and updates for their themes. This model was both GPL compatible, as well as workable for authors to get paid.

In mid-2009, Matt Mullenweg also posted on the official WordPress blog that he was introducing a new commercial theme listing page on WordPress.org, and he shared an opinion he requested from the Software Freedom Law Center (SFLC), where they determined that the two themes packaged with WordPress were derivative works.

The SFLC opinion did leave room for a “split license” where the WordPress and PHP code must inherit the GPL, and the CSS, Javascript, and images could be under a proprietary license:

In conclusion, the WordPress themes supplied contain elements that are derivative of WordPress’s copyrighted code. These themes, being collections of distinct works (images, CSS files, PHP files), need not be GPL-licensed as a whole. Rather, the PHP files are subject to the requirements of the GPL while the images and CSS are not. Third-party developers of such themes may apply restrictive copyrights to these elements if they wish.

Finally, we note that it might be possible to design a valid WordPress theme that avoids the factors that subject it to WordPress’s copyright, but such a theme would have to forgo almost all the WordPress functionality that makes the software useful.

“Split license” is the colloquial term the community has assigned to this statement, but in fact the actual splitting of which parts are GPL and which parts are not matters, so it may not do the reality of the situation justice. Perhaps it should be termed “PHPGPL” or “Non-Assets GPL”.

A number of prominent theme sellers were unhappy with Mullenweg’s insistence that their themes maintain a 100% GPL license, but they were not willing to shake the boat over it. At this point, themes were becoming big businesses and making new millionaires (or close to it) of some of these shop owners. This settled the issue for nearly all theme sellers, and most moved to either 100% GPL or a PHPGPL license, and the doomsday scenarios never came; the theme industry continues to thrive.

Thesis holds out

But not everyone agreed to go either 100% GPL or PHPGPL license. Chris Pearson kept his Thesis theme under a proprietary license.

Thesis was one of the most popular and flexible themes in the world, and Pearson boasted on Andrew Warner’s Mixergy podcast of revenues of $1.2 million+ over the 12-18 month period prior to the interview. Mullenweg and Pearson criticized one another publicly, and Warner invited them both to Mixergy where they debated the merits of GPL licensing.

By most accounts, Mullenweg had the better argument on the Mixergy episode, and also came off as a calmer and more collected personality — in contrast to Pearson’s often heated, and sometimes very strange, statements.

The debate continued between Mullenweg, Pearson, and a variety of WordPress community members and their blogs. Mullenweg was extremely aggressive, to the extent that he offered to buy alternative commercial themes for users of Thesis that agreed to switch. Mullenweg tells me that many took him up on his offer, but it was, “less than a thousand.” In my opinion, this was a step too far by Mullenweg, though for him the issue was already personal.

Pearson held his ground over the following days until an admission by one of his own team members of wholesale copying of code in Thesis from WordPress code, which violates the WordPress copyright.

At this point, Pearson finally capitulated and announced that Thesis would be a split license GPL compatible theme, and the debate died down. Pearson put his head down and started working on Thesis 2.

He released Thesis 2 in late 2012, and by this time the debate was cool — the community had moved on to other drama (yes, even more GPL drama) — and the release of Thesis 2 was largely ignored outside of DIYthemes’ audience, which was quite large but also largely isolated from the WordPress “community” that cares about stuff like licensing.

Therefore, not many people paid attention to the new Thesis or the licensing it contained. Mullenweg, however, was still paying attention.

The battle over thesis.com

If you consider the word thesis, what do you think of?

If you are a regular person walking down the street, you probably think of the general concept for stating a theory, or perhaps you think of the long papers that university students write as part of their programs.

If you are in the WordPress world, you may also consider the Thesis WordPress theme by DIYthemes.

Good, single word domain names are hard to come by. Thesis.com, if you visit it now, redirects to the ThemeShaper blog. ThemeShaper is owned and operated by Automattic, and frequently posts articles about WordPress themes.

ThemeShaper is not a dedicated commercial property, but it does link to Automattic’s primary website, and Automattic does make and sell WordPress themes.

Automattic buys thesis.com

Automattic hasn’t always owned thesis.com. Matt Mullenweg met a third party owner of the domain at a conference — a guy named Larry — and inquired about the domain by email in January of 2014. Chris Pearson had already attempted to purchase the domain from Larry, and did not agree to pay the $150,000 that Larry requested.

According to Pearson’s accounts, he and Larry had a few exchanges that stalled with Pearson unwilling to bid beyond $37,500 for the domain, and Larry sticking to $150,000. With a $100,000 offer on the table from Mullenweg, Larry gave Pearson an opportunity to buy the domain for $115,000, which he didn’t do — in part because he thought it too expensive, but also because he suspected Larry didn’t really have the offer from Mullenweg.

As we now know, Larry did have the offer and Automattic became the owner of the thesis.com domain name.

Pearson didn’t know that Mullenweg actually bought the domain until November of 2014, when he was notified by a friend that Mullenweg’s WordCamp San Francisco State of the Word Q&A session included a section where Mullenweg noted that he owned the domain (more on that later).

Pearson attempts to force domain transfer

On April 9th, 2015, Pearson and his lawyers filed a UDRP (Uniform Domain-Name Dispute Resolution Policy) complaint, which is a formal method for resolving domain disputes, recognized by ICANN. UDRP isn’t a formal government court, but serves as arbitration for domains due to the recognition by ICANN.

There are many, many examples of UDRP complaints in regard to trademark infringement. There are a number of criteria that come into play that guide the UDRP panel’s decision making process.

The three basic tenets that must be met are as follows:

(i) your domain name is identical or confusingly similar to a trademark or service mark in which the complainant has rights; and

(ii) you have no rights or legitimate interests in respect of the domain name; and

(iii) your domain name has been registered and is being used in bad faith.

The panel reviews the initial complaint (in this case, by Pearson) and gives the respondent (Automattic) an opportunity to respond. All correspondence is in writing and not in person. The panel has two weeks after everything has been submitted to reach a decision.

In this case, which is available publicly, the panel denied Pearson’s complaint.

Pearson’s complaint cited that he fulfilled each of the three criteria:

  • By noting his trademark of the word “thesis”.
  • By noting that Automattic was using the domain with a commercial interest (by redirecting it to ThemeShaper).
  • By noting the bad faith clause by citing that Automattic, “purchased the disputed domain name to confuse and redirect customers and potential customers to Respondent’s competing webpage.”

In the response, Automattic did not contest Pearson’s trademark on the word thesis. However, Automattic also noted that the word is very generic, and also that ThemeShaper was not a commercial part of Automattic, but a “blogging site.”

For the bad faith argument, Automattic claimed that the redirect to ThemeShaper furthers their, “purpose in providing a blogging site,” and highlights that the intention for the domain is not as a commercial entity or one to be confused with Pearson’s trademark.

Automattic wins dispute

Automattic won the dispute against Pearson. As noted, the panel had two weeks to deliver the decision, and Automattic proposed a settlement with Pearson before the decision was handed down.

Pearson was considering the settlement when the decision came a day before the two week deadline, which is apparently not a common occurrence. Had the decision not come early, Pearson may have saved himself some trouble, especially in regard to eliminating the trademark cancellation requests by Automattic on thesis and related terms.

In Pearson’s blog post, The Truth about Thesis.com, he notes the general terms of the proposed settlement:

Automattic’s attorneys drafted the original settlement, which included the following terms:

  • Automattic would keep thesis.com
  • Automattic would withdraw the federal trademark cancellation request
  • I would withdraw the UDRP
  • Both parties would mutually release one another (agree not to sue over this issue in the future)

Nothing in the original settlement addressed the trademark infringement, and since this was the reason I took action in the first place, I added a requirement that Automattic no longer infringe upon my mark (which would mean they stop forwarding the domain).

At this point in the proceedings, I agreed to the settlement.

However, since the decision came early, the settlement was never binding. It’s also worth noting that Mullenweg commented to me that Pearson’s stated terms are actually not the terms of the settlement:

In the settlement Automattic offered Chris we agreed not to infringe his trademarks (which is the law, regardless of what the settlement said). He never asked us to change the redirect of thesis.com in the settlement, and if he asked after, we would have said no. There were no restrictions on thesis.com in the settlement.

He also said, “I wish he had reached out before litigating,” and noted that it would have “definitely” changed the outcome of the entire situation. Whether it truly would have is neither here nor there.

Possibility for appeal

The UDRP doesn’t have an official appeals process. Instead, they are willing to not make the changes that a ruling states, if indeed a domain transfer or other action is ordered, if the affected party files a suit in court within ten days of the ruling.

In this case, Pearson did not file in a court within the given timeline, and since the ruling did not require a change in domain ownership, there is no further recourse with the UDRP. However, there is no time limit if Pearson wishes to file in court — but that is the only path remaining if he truly wants to go after the domain.

Legitimacy of the trademark dispute

Trademark law has a long history. Trademarks follow a categorical system, meaning words and terms must be trademarked within a particular category to be applied to.

Pearson has three trademarks for the Thesis and DIYthemes brands, all under international trademark class 42, for computer, scientific, and legal purposes:

  • THESIS
  • THESIS THEME
  • DIYTHEMES

The applications were filed in 2010 and registered in 2011. On June 16th, 2015, Automattic filed cancellation requests for all three trademarks, which were instituted June 25th. Pearson has 40 days from the date the application was instituted (which would be August 4th) to file a response to the cancellation requests. His answers, “must contain admissions or denials of the allegations in the petition for cancellation, and may include available defenses and counterclaims.” It is his burden as the defendant to establish his defense, and, “Failure to file a timely answer may result in entry of default judgment and the cancellation of the registration.”

Basically, Automattic is holding his feet to the fire to defend the trademarks, which their counsel feels should not apply for two primary reasons, as listed in their formal filing.

  • The trademarks were registered in Pearson’s own name, but are used by DIYthemes, and Automattic claims that, “The Pearson Applications were improperly filed in the name of an individual, who did not have the requisite intent-to-use the marks as of the filing date, and the underlying applications are void ab initio.” Their claim notes that US Code section 1051(b) offers this justification.
  • Furthermore, section 1052(e)(1) requires that a trademark not be “merely descriptive,” as Automattic claims his trademarks are.

If upheld, the trademarks will be deregistered by the US Patent Office, further limiting Pearson’s options to defend his claims to the thesis.com domain name in a formal court suit.

I don’t know how good of a case Automattic has, but purely on the surface it looks pretty good. I spent time reviewing the application and the US Code and the arguments appear fairly sound — especially the argument that Pearson applied for the trademarks as an individual and utilizes them as DIYthemes, despite DIYthemes already having been registered as an LLC.

Automattic’s justification for the domain

During the WordCamp San Francisco Q&A, Mullenweg noted the existence of the redirect of the domain name with a sense of pride, and a bit of a side-eyed smirk. In response to a question about relationships with commercial theme sellers and marketplaces, he states:

With the premium theme and plugin folks? … We have had some ups and downs, particularly with marketplaces that didn’t follow the GPL, for example, or violated WordPress’s license — themes that violated WordPress’s license. Um, you can go to thesis.com to learn more about that. Type it in, seriously.

I was at this Q&A in person, and don’t remember him saying this, as it was so buried in a much larger conversation, and I was simultaneously writing a wrap-up post about the State of the Word. However, once the UDRP ruling surfaced publicly, a number of WordPress community members recalled Matt’s statement and it brought a new dimension to the ruling and Mullenweg’s motivation for the domain.

While Automattic bought the domain, Matt Mullenweg was clearly the driving force behind the decision. When WP Tavern prompted Automattic for a comment on their motivation for purchasing the domain, they responded with the following:

We’re happy the panel ruled in our favor. We think Thesis.com is a cool, generic .com that could be used for a variety of things. Just because you have a small WordPress theme doesn’t mean you have a right to seize generic English word .com domains.

We can accept Automattic’s case that they had a general interest in a generic .com domain, but in reality we know better. Mullenweg was clearly presented with an opportunity by this Larry character that checked all of the right boxes for him.

He could get a domain he obviously knew Pearson would want, and deny him.

It has a side benefit that it is a high quality generic domain that will likely maintain or increase its value. And he probably thought it was funny.

I doubt Mullenweg even knew what kind of trademarks Pearson held, but despite Pearson holding the trademarks, it seems Automattic’s attorneys now have the upper hand, and it is highly unlikely Pearson will ever own the domain now.

Mullenweg commented on his refusal to give up the domain to Pearson on WP Tavern:

I’m not going to give a domain worth several hundred thousand dollars to the worst actor in the entire WP ecosystem, someone who keeps repeatedly violating the GPL and now has gone beyond that into patents. Why reward that? I wouldn’t sell it if he offered a million dollars.

There are so many people doing amazing things in the WP community, and 100% GPL! I can and have supported them almost every opportunity I can, and one of the things I’m most proud of in the world is how many fantastic open source businesses have been built on top of WordPress.

And it’s just the beginning — if you remembered in 2010 Chris said that going GPL would destroy businesses and sticking to the principles of our license would destroy investment in WordPress — we all know how that’s worked out since then.

Such a statement, combined with the WCSF video, highlights that the issue is about far more than the domain and its investment potential — a 10x return in less than a year would make for an excellent investment, in the near impossible situation Pearson would offer that.

No, the move was quite clearly a personal one — if also convenient — for Mullenweg, and that’s why terms like “bully”, “petty”, and “spiteful” have been used by many WordPress community members surprised by his actions. They expected more. They expected better, even when directed toward someone as controversial as Chris Pearson.

Pearson’s patent

While a tantalizing story, the battle over thesis.com is not the story here. It has simply been the spark to reignite old disputes with new fervor with potentially much bigger consequences than the 2010 affair ever reached.

One of two additional large components of this story is an active patent application by Pearson that was submitted in 2012 and published in 2014. Keep in mind — and Chris Pearson reiterated this to me many times — it is an application for a patent, not a published patent.

A patent on Thesis 2, or all web templates?

The main patent is titled, Systems, Servers, and Methods for Managing Websites. Chris Pearson is listed as the inventor and DIYthemes the assignee.

The patent never mentions WordPress or WordPress themes, however both in the abstract and in the text, it does have many similarities to what one may expect as a description of a general template mechanism for a website, versus a specific description of the Thesis 2 technology.

Here is the abstract in full (and here is the full patent application PDF, including art):

Systems, servers, and methods for managing websites. Some embodiments provide methods which include, according to a user selection of a website skin, activating the selected skin. The skin comprises at least one structural box further comprising structural code for the website. The method also includes receiving a request (for instance a call to a hook) to serve the website. Further, the method includes, responsive to the request, outputting (by echoing PHP code if desired) the structural code with style related code applied thereto according to a reference between the box and a style package (which comprises the stylistic code). The outputting can occur in an order associated with the boxes. In some situations, another skin can be activated. Moreover, a change (or, perhaps, an edit) can be made to another skin and displayed without affecting the website. Moreover, another skin can be selected and associated with the website.

I discussed the patent at length with Chris Pearson, and while much of that conversation is off the record, I can share what I believe his motivations are with the patent application, and what I think the potential implications for this new chapter of Pearson versus Mullenweg are.

Discovery and publicity of the patent

This patent and another that’s since been rescinded were discovered by Automattic’s lawyers during the UDRP proceedings. There is debate as to how members of the WordPress community discovered the patents’ existence — Jeff Chandler of WP Tavern and Carl Hancock of Gravity Forms were two of the first to discuss it publicly — but there is speculation from Chris Pearson and others that the community discovery of these patents was leaked by Mullenweg himself in order to deflect the attention at the time away from the domain issue and onto the patents and their potential implications.

I honestly don’t care how they came up, though Pearson’s questioning of Jeff Chandler’s journalistic integrity were uncalled for. It is anyone’s right and ability to tip someone off to legitimate news — and these patent applications are legitimate news — and I don’t believe for a second that Chandler has played puppet to his boss’s wishes. He has dutifully and to the best of his ability written about whatever news matters to the community, and I respect him for it.

Nevertheless, this patent does potentially have significantly more newsworthiness, depending on if it is approved and how it is defended by Pearson if it is approved.

Patent law and litigation

Patents often get a bad reputation, and their role in the software world is quite murky. I apologize in advance for this long sidebar on the wild world of patents.

Patent trolls

Most folks have heard of “patent trolls” that prey on companies using vague or overly generic patents, demanding big payouts.

Patent law is weird, and the lawsuits that result are infamously unpredictable and cause a scenario ripe for abuse. For example, filing patent lawsuits in one district over another can have enormous impact, like the case of Marshall, Texas, which is a hotbed for patent trolls:

Patent litigation is a growing business across the country; Marshall is just the most visible example. Among the weightier issues behind the mushrooming of its patent docket is whether the elements that have made it expand — hungry plaintiffs’ lawyers, speedy judges and plaintiff-friendly juries — are encouraging an excess of expensive litigation that is actually stifling innovation.

Some say yes. “A lot of the cases being filed in Marshall are by patent holding companies, or patent trolls, as they’re called, whose primary and only assets are patents,” Mr. Tyler said.

Companies that deal in patents but do not utilize the patented technology are called non-practicing entities (NPEs).

One of the concerns with Pearson’s patent would be if it were approved and he sold it to an NPE. It’s not uncommon for NPEs to acquire patents with the express purpose to enforce them:

Finally, of course, some entities buy patents with the express purpose of licensing them aggressively. For instance, about 25% of “parent” NPEs tracked by PatentFreedom are enforcing only patents that they had acquired. Another 60% are asserting patents originally assigned to them, and the remaining 15% are asserting a blend of originally assigned and acquired patents. However, if we add in the more than 2,100 subsidiaries and affiliates of these entities and treat them all as standalone entities, we find that 19% of them are originally assignees, and 69% are acquirers, and 12% are blends.

Regardless of the important variations in their origin and behavior, NPEs present a fundamentally different challenge than operating company patent assertions.

Software patents unpredictability and “obviousness”

The concept of software patents at all has been in dispute for a long time. Thousands have been awarded, but there are a handful of past court rulings that seem to govern the US Patent and Trademark office’s interpretations for making decisions when reviewing software patents.

Obviousness is a key term in the patent world. Patent applicants aim to create “meaningful” patents, but “at a minimum you must have claims that embody patent eligible subject matter, demonstrate a useful invention, cover a novel invention and which are non-obvious in light of the prior art. Obviousness is typically the real hurdle to patentability, and unfortunately the law of obviousness can be quite subjective and difficult to understand. At times obviousness determinations almost seems arbitrary.”

“Art” is the outlay of the invention by the applicant, and the invention’s ability to be patented depends on “prior art” not deeming the invention as obvious. Establishing non-obviousness for software has a contentious history. If it can be shown that, “any combination of prior art references that when put together would be the invention in question,” then the applicant is in trouble.

But there is a great deal of potential for subjectivity from thousands of patent examiners:

Still, ever since the Supreme Court’s decision in KSR [reference] there has been a great deal of subjectivity in the application of the law of obviousness, which is apparent if you look at the patents that issue, patents that are finally rejected and ultimately abandoned, and the patents the Federal Circuit ultimately finds to include obvious patent claims. There is little to no predictability at the edges.

The Alice case

In my research, the Alice case came up many times as a pivotal case for helping to define the legitamacy of software patents. Martin Goetz is the holder of the first ever software patent, and has an excellent write-up on the importance of the Alice case.

I have been asked for my opinion based my long history in the software industry and from my perspective as someone that has followed that controversial question “Is Software patentable”? That question first began to be publicly debated when I received the first software patent in 1968 for an invention on a new way of sorting data on a computer. Shortly thereafter a publication printed a page one headline “First Patent is Issued for Software, Full Implications Are Not Yet Known.”

Forty five years later a variation of that question was again before the Supreme Court when it agreed to hear the appeal of the Alice v. CLS Bank case.

Goetz argues that the Alice case is a victory for software patents on both sides: that it helps true inventions gain patents (he and others assign a high standard to define “invention”), and it also helps prevent abuse of overly vague or non-inventive “obvious” patents to be denied:

The Alice v CLS Bank Supreme Court decision in June 2014 was a great victory for those that believe that inventors that use a digital computer to innovate can get a patent on their invention. It is also a victory for those people and organizations that recognize how the patent System has been abused for many years by trolls and others where there was no invention. Since the Supreme Court decision in June, thousands of patents that should never have been issued are now being deemed invalid by the US Courts and by the Patent Office.

Obviousness and invention for Pearson’s patent

This long and boring description of patents is necessary because Pearson’s patent application is still just an application, and it can be challenged, both by the patent reviewer, but also by third parties.

As patent obviousness is “so unevenly applied,” there is some risk in not challenging Pearson’s patent, if a third party like Automattic (or a myriad of other web template providers) is worried about the potential implications of the patent. Although, the Alice case does seem — in my very amateur opinion — to offer better guidance to reject the patent based on a lack of true invention.

Pearson’s reasoning for a patent, and its likelihood for success

Patents are not cheap to apply for. The patent application Pearson submitted is 34 pages of art diagrams and text describing the inventiveness of Thesis 2, though Thesis 2 is not specifically named.

I asked Pearson why he applied for the patent, which he did not want to share the specifics of his position due to the open nature of the application. I’ve racked my brain to try and determine the potential causes as well.

There are only a couple of decent outcomes for Pearson with this patent application. The most likely, and not good outcome for him, is that he is denied the patent; and in this case he would have spent a great deal of money for nothing.

In the event he does get the patent — or perhaps even before it is fully reviewed for approval — he could face a challenge from Automattic or other parties that may be concerned his invention’s description could apply more broadly than Thesis 2.

If he gets the patent, and he survives a challenge, I see three ways he could theoretically use it:

  • He can do nothing.
  • He can sell it to the highest bidding NPE, which would be a dreadful outcome for any web entities that sell templates.
  • He can keep it in his back pocket, in case someone threatens his business or his software license, wherein he can initiate a lawsuit.

Honestly, the whole patent route seems odd. I don’t love the idea of this patent being approved, because the application does seem overly broad toward all web templates to me, from the title to the meat of the application’s art. However, there are loads of patents in the world for incredibly silly things that have never really impacted a lot of folks’ life; it’s just that it doesn’t mean a silly patent couldn’t become a problem. The Electronic Frontier Foundation has mountains of evidence of patent holders causing havoc.

I’m not much of a gambler, but if I had to gamble on this I’d put my money on this patent never being approved, and definitely never truly impacting the web or WordPress industry at scale.

I don’t think Pearson is a bad guy for wanting to patent his work. When I requested comment about the patents, he told me, “If I were ever to consider selling my business, things like trademarks and patents show up on the balance sheet and add to the bottom line,” but that he views them as, “one of many expensive, ridiculous options for bolstering one’s business,” versus a way to celebrate and protect his achievements as I previously characterized his intentions.

The GPL

Most agree that the GPL has not been well tested in court, though a software license is a “legal instrument.” There is often confusion over whether a license is a contract or not.

License versus contract

One of my favorite things I’ve read on this is by former Adobe Associate General Counsel Robert Pierce:

A license is not a contract. This much I know.

Rather, a license is a permission granted by one party to another allowing use of a property without fear of lawsuit brought by the granting party. A license does not include a return promise (i.e., consideration) from the licensee. So, as we all learned in law school, a license cannot be a contract under law. This is not to say that a license cannot be an element of a contract under which two parties trade promises, one of such promises being a license. This is commonly known as a “license agreement.” But a bald license, a one-way promise, is enforceable outside of contract law. It is something apart. It exists and is enforceable under property law doctrine.

What makes things difficult is that the scope of a license’s grant, and the conditions and restrictions on the license (or all of them together) can make what is intended to be a one-way license look a lot like a contract. The precise wording used becomes critical.

The distinction can be significant because, “contracts are enforceable by contract law, whereas licenses are enforced under copyright law,” though even this rule depends on the jurisdiction where the matter is being discussed. His larger point is that a license is a one-way street, whereas a contract is agreed upon by both sides.

Spirit of the GPL

There is little debate that a “Split GPL” or “PHP GPL” license is perfectly GPL compatible, though Mullenweg doesn’t consider that the “spirit of the GPL,” and companies like Envato’s ThemeForest and others have felt the consequences of not adopting 100% GPL licenses.

From Milestones:

While not everyone liked the fact that the WordPress project would only support 100% GPL products, most people had accepted it. Many, however, were taken by surprise, by a sudden flare-up around not just the legalities of the GPL, but the “spirit” of the license. In a 2008 interview, Jeff Chandler asks Matt about the spirit of the GPL. Matt says that the spirit of the GPL is about user empowerment, about the four freedoms: to use, distribute, modify, and distribute modifications of the software. Software distributed with these four freedoms is in the spirit of the GPL. WordPress was created and distributed in this spirit, giving users full freedom with regard to the software.

The Software Freedom Law Center’s opinion — with regards to WordPress themes, however — gives developers a loophole, one that helps them achieve GPL compliance, but denies the same freedoms as WordPress. PHP in themes must be GPL, but the CSS, images, and JavaScript do not have to be GPL. This is how Thesis released with a split license — the PHP was GPL, while the rest of the code and files were proprietary. This split license ensures that the theme is GPL compliant yet it isn’t released under the same spirit as the GPL’s driving user-freedom ethos.

The loophole may have kept theme sellers in line with the GPL, but WordPress.org didn’t support that approach. In a 2010 interview, Matt says “in the philosophy there are no loopholes: you’re either following the principles of it or you’re not, regardless of what the specific license of the language is.” Theme sellers that sell their themes with a 100% GPL license are supported by WordPress. Those that aren’t don’t get any support or promotion on WordPress.org or on official resources. This is also one of the WordCamp guidelines, introduced in 2010; that WordCamps should promote WordPress’ philosophies. If a speaker, volunteer, or organizer is distributing a WordPress product it needs to be 100% GPL, i.e., the CSS, JavaScript, and other assets need to be GPL, just like the PHP.

Mullenweg believes that Thesis 2 is not only not in the spirit of the GPL, but flagrantly operates in total violation of it, as Thesis 2 carries a 100% proprietary license. Considering the implications for folks that make stuff compatible with the GPL, it’s little surprise that Mullenweg has taken the stance and actions he has toward Pearson.

Thesis 2 carries a proprietary license

Chris Pearson’s comment on his blog post describes that the theme has always been a proprietary license, and he describes why it is okay to be such:

In October 2012, I released an all-new version of Thesis that carried the same name as the original (which had a split-GPL license), but that’s where the similarities stopped.

The new Thesis is not a Theme—it is an operating system for templates and design. This system runs Skins and Boxes, which are similar to Themes and Plugins, but with a boatload of built-in efficiencies that Themes and Plugins cannot provide.

It is worth noting the final paragraph of the SFLC’s opinion that Mullenweg has cited numerous times as justification against proprietary licenses does have a provision for avoiding WordPress’s copyright:

Finally, we note that it might be possible to design a valid WordPress theme that avoids the factors that subject it to WordPress’s copyright, but such a theme would have to forgo almost all the WordPress functionality that makes the software useful.

That’s exactly what Pearson believes Thesis 2 is. But the GPL has rarely been tested in a proper court, and never from a WordPress perspective. The SFLC’s opinion is just that, for now, whether Mullenweg likes it or not.

The GPL in court

The Free Software Foundation maintains the copyright on the text of the GPL itself, and between the FSF and the SFLC, a small number of lawsuits have occurred, and a German court upheld the GPL as a “valid, legally binding” license, but most of these tests have occurred outside of the United States. From what I can tell, cases involving the GPL have largely settled outside of court when based in the United States, or were argued on whether the GPL was legally applied, like in the case of SCO vs IBM, rather than whether the GPL was legally binding itself.

In another case, Wallace vs FSF, Daniel Wallace compared the GPL to price fixing, as it required software to be free. The FSF won the case, as the judge cited, “The GPL encourages, rather than discourages, free competition and the distribution of computer operating systems, the benefits of which directly pass to consumers. These benefits include lower prices, better access and more innovation.”

A much newer case involving VMware again tests the GPL. The Software Freedom Conservancy, “claims VMware is using the Linux kernel without respecting the terms of its copyright license, the GPL.” This case may offer a better precedent for WordPress and its derivative works, as it is centered on “module loading” in VMware, “with an insulating layer to allow its kernel to use unmodified Linux drivers.” The case gets murkier than that, as it may not have been as isolated as it was attempted, but the result could be decent precedent for similarly loading WordPress themes and plugins, in my opinion.

Limited guidance

Few lawyers want to be the first to test something in court. It’s easier to make a case when there are many cases before you to provide guidance. With the GPL, there is what’s called “limited guidance,” meaning that it’s untested, and therefore the outcome of a GPL case in the US could be very difficult to predict.

If a lawsuit does occur, it could prove costly to all parties involved, and I think it’s clear that Pearson and Mullenweg both wish that litigation was not happening, though both of them maintain a bit of a “you started it” attitude.

Without a formal court proceeding, which could last years, it’s going to be near impossible for Matt Mullenweg to fully prevent non-GPL compatible licenses to exist for WordPress themes and plugins. His best method to prevent it is to do what he’s done so far: make an example of bad offenders and cause anyone thinking of using a non-GPL compatible license to reconsider.

Are all WordPress themes derivative works?

A key question to whether themes and plugins must be GPL compatible licensed is whether the theme or plugin is a derivative of WordPress itself. If it is derivative, then it is under the umbrella of the GPL’s Copyleft nature.

Folks disagree a good bit on how themes and plugins should be considered as derivative works, though most either agree with Mullenweg’s strict “spirit of the GPL” view, or at least have molded to avoid being an outcast.

The strongest argument I’ve seen in favor of all themes being derivative of WordPress — no matter how much or little they rely on WordPress functionality — is from WordPress lead developer Mark Jaquith:

There is a tendency to think that there are two things: WordPress, and the active theme. But they do not run separately. They run as one cohesive unit. They don’t even run in a sequential order. WordPress starts up, WordPress tells the theme to run its functions and register its hooks and filters, then WordPress runs some queries, then WordPress calls the appropriate theme PHP file, and then the theme hooks into the queried WordPress data and uses WordPress functions to display it, and then WordPress shuts down and finishes the request. On that simple view, it looks like a multi-layered sandwich. But the integration is even more amalgamated than the sandwich analogy suggests.

Here is one important takeaway: themes interact with WordPress (and WordPress with themes) the exact same way that WordPress interacts with itself. Give that a second read, and then we’ll digest.

The same core WordPress functions that themes use are used by WordPress itself. The same action/filter hook system that themes use is used by WordPress itself. Themes can thus disable core WordPress functionality, or modify WordPress core data. Not just take WordPress’ ultimate output and change it, but actually reach into the internals of WordPress and change those values before WordPress is finished working with them. If you were thinking that theme code is a separate work because it is contained in a separate file, also consider that many core WordPress files work the same way. They define functions, they use the WordPress hook system to insert themselves at various places in the code, they perform various functions on their own but also interact with the rest of WordPress, etc. No one would argue that these core files don’t have to be licensed under the GPL — but they operate in the same way that themes do!

It isn’t correct to think of WordPress and a theme as separate entities. As far as the code is concerned, they form one functional unit. The theme code doesn’t sit “on top of” WordPress. It is within it, in multiple different places, with multiple interdependencies. This forms a web of shared data structures and code all contained within a shared memory space. If you followed the code execution for Thesis as it jumped between WordPress core code and Thesis-specific code, you’d get a headache, because you’d be jumping back and forth literally hundreds of times. But that is an artificial distinction that you’d only be aware of based on which file contained a particular function. To the PHP parser, it is all one and the same. There isn’t WordPress core code and theme code. There is merely the resulting product, which parses as one code entity.

Jaquith’s argument that the theme and WordPress execute together to form a joint “modified work” is the key phrase, I believe. As he states, and I tend to agree, it does not matter that the files are separate or that they can be distributed independently; together, when executed, they are so intertwined that they become a single work.

That said, the theme is clearly dependent on WordPress itself, which is another common justification that themes are derivative. Explaining this concept is simple: WordPress can be distributed without any theme but those that ship with it by default. But a distributed theme, like Thesis, must be installed and activated using WordPress’s own schema for loading a template, and cannot operate independently of WordPress.

What about the WordPress REST API?

Thus far, we’ve discussed the derivative nature of WordPress themes and plugins, which require they operate within the WordPress install. It is a different matter if we consider applications that consume data or interact with WordPress as an outside application.

The WordPress REST API enables one to interact with or consume data from WordPress, while being wholly independent of the WordPress install. Jaquith makes a clear exception for a scenario like this (and also applies it to technologies like RSS and XML-RPC):

Something that interacts with these APIs sits entirely outside of WordPress. Google Reader doesn’t become part of WordPress by accessing your feed, and MarsEdit doesn’t become part of WordPress when you use it to publish a post on your WordPress blog. These are separate applications, running separately, on separate codebases. All they are doing is communicating. Applications that interact with WordPress this way are separate works, and the author can license them in any way they have authority to do so.

The GNU’s take

The GNU agrees with Jaquith’s take. They provide an FAQ to answer, “”What is the difference between an ‘aggregate’ and other kinds of ‘modified versions’?” The emphasis provided is my own:

An “aggregate” consists of a number of separate programs, distributed together on the same CD-ROM or other media. The GPL permits you to create and distribute an aggregate, even when the licenses of the other software are non-free or GPL-incompatible. The only condition is that you cannot release the aggregate under a license that prohibits users from exercising rights that each program’s individual license would grant them.

Where’s the line between two separate programs, and one program with two parts? This is a legal question, which ultimately judges will decide. We believe that a proper criterion depends both on the mechanism of communication (exec, pipes, rpc, function calls within a shared address space, etc.) and the semantics of the communication (what kinds of information are interchanged).

If the modules are included in the same executable file, they are definitely combined in one program. If modules are designed to run linked together in a shared address space, that almost surely means combining them into one program.

By contrast, pipes, sockets and command-line arguments are communication mechanisms normally used between two separate programs. So when they are used for communication, the modules normally are separate programs. But if the semantics of the communication are intimate enough, exchanging complex internal data structures, that too could be a basis to consider the two parts as combined into a larger program.

The GNU argument falls very much in line with Jaquith’s, though admits itself that judges must decide whether it’s the case, in the end.

The case against The GNU position on derivative works

The University of Washington School of Law has a section of their website devoted to the, “business, legal and technical consequences of choosing Open Source Software (OSS) or proprietary software.” They cover many of the topics I’ve outlined in this post so far, and in the case of the GPL and derivative works, they believe the GNU is overstepping with an “expansive definition” of derivative works with consequences, “counter to the goals of the proponents of Free Software.”

The most compelling of multiple derivative works examples they provide is that of subclasses. For example, imagine a class, Some_Theme_Class that extends Some_Core_WordPress_Class. The GPL FAQ is very hardline on the topic (and for what it’s worth, Thesis 2 does extend some WordPress core classes). Washington believes the GNU stance on inheritance is too over-reaching:

Example 5: Programmer X wishes to write a class D, that is a subclass of existing class B. Class B is subject to the terms of the GPL. If X distributes D, does it have to be licensed under the terms of the GPL?

The answer given in the GPL FAQ is short and to the point: “Subclassing is creating a derivative work.” In our example, this makes D a work derived from B, and thereby makes D subject to the terms of the GPL upon distribution. This approach attempts to further broaden the reach of the GPL, but it again leads to counter-intuitive results.

Typical object oriented programming languages include a standard class hierarchy. This hierarchy provides a framework within which application developers can build their programs. The standard classes typically provide useful classes that represent user interface elements (e.g. windows, buttons, etc.), collection classes (for handling collections of data), and input-output abstractions (e.g. files and networking connections). In many object oriented languages, each class must be a subclass of exactly one superclass. And for this reason, the class hierarchies are rooted by a highly generic, standard class called Object. (The question of the superclass of Object is beyond the scope of this article.) The class Object describes only the most general properties and behaviors. For instance, in Java, the class Object only performs a handful of functions. In Java, every class is a subclass (directly or indirectly) of the Object class. Under the GPL approach, then, every program written in Java is a derived work of Object, because every program written in Java by definition consists of classes that inherit from the Object class.

Whether this argument or any of the others Washington outlines would apply to WordPress themes and/or plugins would, again, need to be settled in court. But Washington does give a compelling argument.

They conclude with the following:

In some ways, the apparent weaknesses in the GPL should come as no surprise, as the GPL was born of an era in which the central artifact of software development and distribution was the monolithic executable. In such a universe, software development proceeded principally by modifying the existing source text of programs, compiling source modules, linking the corresponding object files, and distributing the resulting executable. This model of software development and distribution has become increasingly fractured in an era characterized by highly dynamic, late binding, object- and network-based systems. The GPL, consequently, strains to cover these newly arising scenarios.

To effectuate the goals of the free software movement, the drafters of the GPL urge a generally expansive definition of derivative work. The great irony is, of course, that such an expansive definition would have second order consequences that are exactly counter to the goals of the proponents of Free Software. A broad definition of derivative would give code authors less freedom to create software that they can truly call their own and do with as they please. And if naive analytic approaches such as “subclassing equals derivation” reign, then proprietary vendors such as Microsoft could arguably stake claim to every program ever written in C#, because they authored the original class hierarchy. And since it seems unlikely that courts would employ different standards depending on the goals or ideological motivations of licensors, proponents of free software might want to be careful what they wish for: what’s good for the GNU might not be good for the gander.

Aggressive license agreements

Both the GPL and DIYthemes’ proprietary license could be appropriately identified as aggressive. The Copyleft nature of the GPL annoys many open source advocates, who would prefer a less restrictive license for developers, like the MIT or BSD licenses. The GPL is absolutely an opinionated license.

Pearson’s proprietary license is also aggressive, in the other direction. I’ve never purchased a WordPress-centric product that so strongly forced me to accept a license. Usually, you have to look in the source code or a page on the website for a license; DIYthemes forces you to accept the terms of the proprietary license before you can download the product at all.

Derivative works are not a bright line

The GNU attempts to offer a “bright line” distinction for derivative works. A bright line, in much of the legal analysis I’ve read, is where thing x is so because of thing y, and can be applied across the board. You can clearly see the bright line, and when it has been crossed.

Washington proves the point quite well that the GNU’s bright line approach to derivation is quite challengeable. But I don’t think their arguments prove that WordPress themes in particular are not derivative. I believe, from a legal perspective, it’s fuzzier than a bright line approach, and if I were Mullenweg or anyone defending GPL software, I would not be excited to take the issue to court.

The “spirit of the GPL” is to offer users liberal freedoms, even while restricting developers building on a GPL licensed application. And I believe there is merit in the fact that WordPress, its co-founders, its lead developers, and the vast majority of its copyright holders (contributors) wish to defend the spirit of the license, even if it’s not been tested in court.

Pearson is not in the majority opinion by using a proprietary license, but he is also not definitively in a position of legal wrongdoing. His desire to protect his works from user freedoms with a proprietary license may well be tested all the way to the courts, and he must be prepared to deal with that, but I don’t believe there is clear wrongdoing, legally, with his license.

A tale of idealism

Matt Mullenweg and Chris Pearson are two of the most idealistic people I’ve ever observed. They are near polar opposites, from their business belief systems and even their general world views.

One of the most depressing components of my research was something Pearson told me. I asked him why he doesn’t just get out of it all and do something else. He’s not married to the culture of WordPress bestowed on it by its leaders. He called it, the “zeitgeist of western culture,” with its openness, zen attitude, and more that he feels no need to embrace if he doesn’t want to.

But he admits that WordPress, “is the most used piece of software to build a website in the world. WordPress was the beneficiary of impeccable timing.” And it’s a tool for him to make his living; it’s his job, and he doesn’t see a need to be in love with every aspect of his job. This is what made me sad, because for most of us that make our living within this space, we were able to escape the “it’s just a job” mentality and be emotionally enriched by what is possible on the open web.

Matt could probably drop his various issues with Pearson and life would go on. The vast majority of WordPress businesses could understand the status quo and live by it, and those that don’t can keep living their lives outside of the approval of Mullenweg, and for that matter, the official WordPress project and website. But he too insists to stand up for his ideals and the web he believes in. He sees himself as a defender of the user, and his defense of the GPL is an extension of his core beliefs on software.

They will never agree on licensing, that much is clear. The question of what’s next is multi-layered.

Will litigation continue?

Undoubtedly, yes, litigation will continue. But the litigation should be viewed as three distinct parts:

  • In regard to the thesis.com domain, it’s really a sideline issue that resulted from more deep-rooted differences in ideals that turned into a personal spat. Pearson may continue in court to try and get the domain back, but I doubt it. I don’t know if Automattic will relent on the trademark cancellation requests, but I wouldn’t be surprised if they dropped it in some form of settlement.
  • The patent issue is not over. I believe Automattic and potentially other organizations will challenge Pearson’s patent application using a variety of legal options available. The chances the patent gets approved or holds up long term are unlikely (but yes, it is possible), and I don’t believe there is a significant chance it will have longstanding implications on the WordPress project.
  • I believe the GPL will continue to be tested in court, and eventually we may have a proper precedent set to put current questions aside. I won’t pretend a guess how it will turn out, because I think it truly depends on many, many factors. I do think that if Matt Mullenweg pursues a case himself, he will be joined by a number of interested organizations, including the Free Software Foundation; or the GPL debate could be settled in court in a completely different dispute, unrelated to WordPress — but have a longstanding impact on products made for WordPress.

Is this debate bad for WordPress?

The way this debate has occurred is bad for WordPress. Neither Matt Mullenweg nor Chris Pearson looks like a saint right now. And parts of the whole thing don’t do a whole lot to further the conversation.

At the root of the debate is licensing, and that debate is worth having.

It is important that we separate the intent and the legal interpretation of the GPL. It is also important that we separate one’s legal ability to not license distributed WordPress products as GPL compatible, versus the business and community consequences that may result from such a decision.

Endmatter

This post would not be possible without the Post Status Partners and Members that fund the website, and my ability to write about WordPress full time. If you enjoyed this post, please consider becoming a Post Status member to fund more free content, plus loads of great members-only benefits, including a daily-ish newsletter that keeps you covered on the happenings of the WordPress world.

I would also like to thank Matt Mullenweg and Chris Pearson for the interviews they provided me in preparation of this post.

And I’d like to thank my lawyer, Keith Lee (a WordPress fan and blogger himself!) for helping me think through some of the legal matters discussed — though the opinions themselves are my own.

Finally, I’d like to thank the folks that helped me review the post, consider my positions, and organize my thoughts. You know who you are.