Tianchu Gao is an IPilogue Writer and a 1L JD Candidate at Osgoode Hall Law School.

Rising to the challenges and opportunities posed by the disruptive technology of artificial intelligence, 91��ɫ’s Artificial Intelligence and Society Task Force leveraged collegial expertise from a wide range of disciplines to explore the possibility of research, development, and innovation of artificial intelligence and its impacts on humanity and society. The Task Force, Co-Chaired by Prof. Pina D’Agostino and Prof. James Elder, hosted a virtual symposium “” on Nov. 16. The summarizes current AI research, teaching, and learning activities at 91��ɫ and offers advice on further expansion of AI-related initiatives in the near future. After welcoming remarks from 91��ɫ’s President & Vice-Chancellor Rhonda Lenton, VPRI Amir Asif, and Mayor Frank Scarpitti of Markham, the symposium featured a keynote speech by , Head of JP Morgan Chase AI Research and Herbert A. Simon University Professor in the School of Computer Science at Carnegie Mellon University. It was followed by a panel discussion led by Prof. D’Agostino and Prof. Elder. The panel featured Dr. Veloso, Neetika Sathe (Vice President, Alectra’s GRE&T Centre), James Goel (Director of Engineering Technical Standards, Qualcomm) and Androu Waheeb (JD Candidate 2023, Osgoode Hall Law School). The livestream of the complete symposium is now available on .

Artificial intelligence is “.” Professor Veloso began her speech by tracing the origin of AI to a ten-man workshop that lasted two months at Dartmouth College in 1956. The workshop proceeded based on the conjecture that The workshop’s goal to simulate human perception and cognition on machines continues to challenge scientists today. Despite the revolutionary breakthroughs in the past decades, artificial intelligence is still a young science that has a long journey to advance ahead.

Professor Veloso shared in the speech a few projects she and her team have developed for JP Morgan Chase when she worked as the head of its AI Research department since 2018. One example is the use of image-based decision making and prediction in stock market analysis. Machines have learned to recognize and classify objects; we see the implementation of this technology in self-driving cars, as the car needs to “see” the environment—identify roads, cars, and pedestrians—to navigate in the traffic safely. Similarly, this technology is used to analyze the stock market as pure images. The researchers labeled the images with “buy” and “no buy” decisions made by humans from historical data, and they can use visual signals to predict transactions in the stock market with 95% accuracy.

Other important tasks that AI can tackle for the financial market include the generation of synthetic data for development and exploration. Synthetic data is easier to access and process than real data. It allows banks to explore the impacts of the decisions they made and develop new approaches to problems like fraud and money laundering. Professor Veloso’s team is also working on the automated standardization of financial data represented in different formats. It is an AI-driven task because the computer program needs to recognize visual and semantic cues in order to understand the meaning of a document and convert it into the standard form. These are a few examples of the various tasks that AI can accomplish and continuously refine for the financial market and beyond.

The challenges AI poses for us do not concern technology alone. As panelist James Goel pointed out, the tremendous investments (nearly a trillion dollars) that AI-related markets will attract in the next five to ten years will generate transformative impacts on society at large. The question remains: what could academia and government policy do in response to this social challenge?

91��ɫ is taking the initiative to foster an interdisciplinary approach to explore AI as a key area of accelerated research and its social implications. The report by the Task Force is a first step of the journey; it studies the existing AI terrain in 91��ɫ in terms of research, teaching, and learning activities and offers guidance on how to grow AI initiatives in the university. 91��ɫ’s new Markham campus will further provide greater resources for AI-related research and innovation. Substantial activities suggest that 91��ɫ has a great potential to be a great center of teaching and research in AI.

The post Recap of “Fostering the Future of Artificial Intelligence: Report from the 91��ɫ Task Force on AI & Society” appeared first on IPOsgoode.

NVIDIA is a graphics chip manufacturer based in Santa Clara, California.  Since 1993 they have been responsible for many innovations in Graphical Processing Unit (GPU) design that have made sweeping changes in the way graphics can be processed on PCs. These improvements have been revolutionary in many industries such as smartphone design, video gaming, medical imaging, and computer generated imagery and special effects.

When building its smart phones, Samsung incorporates GPU chips that it has either designed in house or obtained from Qualcomm, Inc. NVIDIA alleges that the two companies are violating its patents by using the technology without a licence. The alleged patent infringement in the case refers to the Adreno GPUs used in Qualcomm's Snapdragon processors, and the PowerVR and Mali GPUs used in Samsung's Exynos processors.

NVIDIA's ITC complaint refers to seven of its patents:  US Patent Nos. ("the '488 Patent") -- "Transform, lighting and rasterization system embodied on a single semiconductor platform", ("the '667 Patent") -- "Single semiconductor graphics platform system and method with skinning, swizzling and masking capabilities", ("the '685 Patent") -- "Programmable graphics processor for multithreaded execution of programs", ("the '913 Patent") -- "Method and apparatus for multithreaded processing of data in a programmable graphics processor", ("the '063 Patent") -- "Rendering pipeline", ("the '140 Patent") -- "System, method and article of manufacture for a programmable vertex processing model with instruction set", and ("the '372 Patent") -- "System, method and article of manufacture for shadow mapping".

What is interesting and distinguishing from the  case, is the fact that NVIDIA is not filing a civil suit but is instead asking the ITC to commence an investigation. The is a favourable venue because of its speed:  the ITC usually responds within 30 calendar days of a filing, can provisionally accept motions for temporary relief within 35 days, and typically wraps up proceedings for a case in about a year.  The complaint is also under section 337, which enables the ITC to conduct investigations into unfair practices in imports. Under section 337, the primary remedy at the agency's disposal is an exclusion order that prohibits the importation of the infringing goods into the Unites States.  Furthermore, and more importantly for Qualcomm, the ITC can issue exclusion orders that affect downstream products, meaning that potentially any smartphone using Qualcomm's Snapdragon processors (which include the Adreno GPU) could potentially be blocked from importation.  This appears to be the most damaging part of the filing, as the Snapdragon is used not only in Samsung smartphones, but also those made by LG, Asus, Sony, Sharp, and Nokia.

The ITC first claimed these downstream rights in .  In that investigation, the ITC identified several factors determining the balance of downstream remedies:

"In performing this balancing, the Commission may consider such matters as the value of the infringing articles compared to the value of the downstream products in which they are incorporated, the identity of the manufacturer of the downstream products ( hare, the downstream products manufactured by the party found to have committed the unfair act, or by third parties), the incremental value to complainant of the exclusion of downstream products, the incremental detriment to respondents of such exclusion, the burdens imposed on third parties resulting from exclusion of downstream products, the availability of alternative downstream products which do not contain the infringing articles, the likelihood that imported downstream products actually contain the infringing articles and are thereby subject to exclusion, the opportunity for evasion of an exclusion order which does not include downstream products, the enforceability of an order by Customs" -- at page 207

For Samsung, this remedy would be crippling; its handsets represent a of the smartphone market in 2013. Google and the android ecosystem would also be affected,  as the Snapdragon is a very popular processor used by many of the various android manufacturers.

Will the showdown turn to the courts? Based on the amount of damages awarded to Apple in their suit against Samsung, one might think to answer that question in the affirmative. Apple v. Samsung was a personal fight however; Steve Jobs was famously (and acrimoniously) "I'm going to destroy Android, because it's a stolen product. I'm willing to go thermonuclear war on this." and "I will spend my last dying breath if I need to, and I will spend every penny of Apple's $40 billion [£25bn] in the bank, to right this wrong". Due to the venue the plaintiff has chosen, it's far more likely that NVIDIA will pursue a negotiated settlement. Alternatively, this investigation may serve to position NVIDIA for a future iPad and iPhone licensing deal with Apple; this is not a far reach considering it currently provides GPUs for several of the company's Mac desktop and laptop products.

Paul Blizzard is an IPilogue Editor and a JD candidate at Osgoode Hall Law School. 

Twitter:

The post Up the creek without a paddle: downstream exclusion threatens Qualcomm. appeared first on IPOsgoode.

This is the second part of Professor O'Connor's feature blog post "Does Qualcomm's Value Chain Licensing System Survive Its Settlement With Broadcom?" Part one can be read .

Spring-boarding off the Quanta decision, Broadcom's declaratory judgment action argued that Qualcomm was misusing its patents to force upstream and downstream parties to license those patents in a "double recovery" or "double compensation" scheme.  Qualcomm's response was that the upstream and downstream parties were instead simply "sharing" the market value royalty in a beneficial diversification of the risk for either party, rather than either of them having to pay the full royalty and "collect" or "recoup" portions of it from others. 

As a preliminary point of analyzing the parties' claims, we should acknowledge that �ϳܲ��Գٲ�'����"substantial embodiment" and method exhaustion holding brought into starker relief the fact that Qualcomm will now be seen to be licensing at least some patent claims twice.  In other words, it is harder now for Qualcomm to argue that it is licensing only the chip claims to the upstream chipmaker (under the APLA), and only the system claims to the downstream handset maker (under the SULA). This means that Qualcomm arguably now can only assert that its licensing system is proper on the grounds that it can license the same thing to two parties.  While neither party raised the following in their filings, there must be an implicit argument that vertical multiple licenses are more questionable somehow than horizontal multiple licenses.  Clearly, a patent owner can license the same patent to multiple unaffiliated parties in horizontal relation -- i.e., multiple chipmakers. So the second issue is why there would be anything wrong with licensing the same patent to multiple vertical parties?

Broadcom argued that because there is/will be a sale of the licensed chips from the chipmaker to the handset maker, the patent is exhausted as to the sold chips and thus Qualcomm has no legitimate property/rights to use as consideration in contracting with the handset maker to enter a SULA.  To follow this out, if the handset makers are only, or primarily, entering into the SULA under pain of a patent infringement suit, then Qualcomm is misusing its patents because it is extracting value beyond the legitimate patent monopoly in the realm of the patent-exhausted, sold (or to be sold) chips.  Of course, this presumes that the handset maker could buy the chips from the chipmaker absent the SULA, which of course it cannot because of Qualcomm's restrictions on the chipmaker through the APLA.  But Broadcom was trying to argue that Qualcomm is still misusing its patents because it has taken both upstream and downstream parties hostage, in a manner of speaking, by setting out a double license situation that is take-it-or-leave-it for both upstream and downstream parties. This is somewhat persuasive if one thinks of the licenses as being offered/granted at different times, such that the first offered/granted should preclude the ability to demand the second on pain of patent misuse.  But, when presented as a kind of simultaneous grant/offer to both upstream and downstream parties that merely partitions the full market value royalty and license fees between then, the system does not seem so problematic.

While I don't see anything invalid or illegal with Qualcomm's upstream and downstream licensing system from a pure patent licensing perspective (I am not commenting on whether such a system could be viewed as anti-competitive in some circumstances under antitrust/competition law), I do think that Broadcom raised an interesting additional issue.  Because the authorized chip sales from the APLA licensed chipmaker to the SULA licensed handset maker are indeed likely subject to patent exhaustion, if the handset maker does anything with the purchased chips in contravention of the SULA, then Qualcomm's only legal action is likely limited to contract remedies, and not patent infringement remedies.  This is because Qualcomm does not seem to have required APLA licensed chip makers to engage in only conditioned sales of chips to SULA licensed assemblers (which would raise the question about whether Mallinckrodt is still good law).  Many of us had been focused on the fact that post-Quanta there was still nothing improper about limiting a manufacturing license sell/vend right to a certain category of purchasers (such as Qualcomm's APLA limitation on chipmakers to sell only to SULA licensed handset makers).  But that is a completely separate issue from the fact that once the authorized sale of chips to a SULA licensed handset maker occurs, then exhaustion has likely applied to any further patent claims for those sales/chips.  It is possible that a licensing system in which the manufacturer's sell/vend license is limited not just to a category of purchasers, but also to appropriately conditioned sales to those purchasers, would preserve the right of the patent owner to pursue patent infringement claims against a wayward downstream licensee.  This will depend on the continued validity of Mallinckrodt as good law. But as to authorized unconditioned sales, Qualcomm may only assert breach of contract claims for the wayward SULA licensee who turns around and resells the chips or otherwise operates outside the scope of the SULA.  This is not to treat contract remedies as inconsequential, but unless extraordinary relief is provided for in the contract, then remedies on par with potent patent remedies such as injunctions will be unavailable.

When the district court dismissed Broadcom's declaratory judgment action -- for want of an active case or controversy in part because Broadcom is not a party to an APLA or SULA, and was not actively being threatened with a specific patent infringement suit from Broadcom at the time -- that denied us the chance to get a direct ruling on Qualcomm's licensing system. But the hope for an appeal that might have kept the case alive has been entirely snuffed out with the Settlement Agreement that includes the out of court settlement of all active disputes between the parties, including any appeal of the declaratory judgment dismissal.  Further, the Settlement Agreement is quite clear that by its own terms it does not question or disrupt any other licenses and agreements that the parties have with third parties, with the exceptions of some limited third party beneficiaries that are largely either Broadcom's or Qualcomm's customers.  Thus, Broadcom has effectively dropped, and released forever, its claims against Qualcomm's licensing system.  Further, while I am uncertain because of the redactions in the publicly released version of the Agreement, it does seem that its terms preserve the contours of Qualcomm's licensing system even as to Broadcom and its customers (with the provisions for "exhaustive" and "nonexhaustive" licenses, with the former perhaps being similar to SULAs and the latter similar to APLAs).  At the same time, the major division -- especially as Broadcom sums up the terms of the Agreement in the text of the 8-K itself -- seems to be that Qualcomm gets the combined patent portfolio under the cross license only for cellular products, while Broadcom gets the portfolio only for non-cellular products.  Further, Qualcomm will buy a substantial number of Broadcom's patents outright through an assignment provision in the deal.

Presumably the Settlement was signed off on by the various courts involved. And ending this patent war is likely good for the telecommunications and chip manufacturer community overall.  But one does have to worry about the anticompetitive effects of both Qualcomm's licensing system and the division of the cellular and non-cellular market between the parties.  On the other hand, I am generally in favor of at least the value chain licensing part as I think it promotes commercialization over the increasingly long value chains presented by modern high tech commercialization pathways.  So on balance, I'm more concerned that the Settlement "stick" and not fall apart too soon.  I have been watching the Showtime series "The Tudors" on DVD recently and I have visions of the Broadcom-Qualcomm Settlement as yet another of the "grand alliances" variously among England, France, and Spain of the historical period that would be honored while convenient, and cast away when not.

The post Does Qualcomm's Value Chain Licensing System Survive Its Settlement With Broadcom? (Part Two) appeared first on IPOsgoode.

Qualcomm and Broadcom finally seem to have ended their long running, epic patent litigation war -- or at least executed a truce for the time being.  With battle fronts in the U.S., Korea, Japan, and the EC, this conflict has been a poster child for patent reform advocates.  But as a multi-jurisdictional affair, any reform lessons to be learned from it need be multi-jurisdictional as well.  My interest, however, is the effect (if any) the settlement will have on Qualcomm's hotly contested value chain licensing system that secures royalties from both upstream and downstream parties in the chip and handset manufacturing and distribution chain.  While many might have believed the U.S. Supreme Court decision in Quanta Computer, Inc. v. LG Electronics, Inc., 128 S. Ct. 2109 (2008), overturned the validity of such licensing systems under the doctrine of patent exhaustion, a careful analysis of both Quanta and Qualcomm's actual system refutes this easy assumption.

Because Broadcom deemed the Settlement and Patent License and Non-Assert Agreement ("Settlement Agreement"), executed on April 26, to be a material agreement under U.S. securities laws, it released a on April 29 as an exhibit to a SEC material event report on Form 8-K.  Qualcomm appears to have taken the position that the Settlement was not a material agreement to it, as it has not released even a redacted version even in SEC filings that mention the Settlement, such as its , the same day as Broadcom's 8-K filing.  Admittedly, Qualcomm is a much bigger entity than Broadcom, and thus under U.S. securities cases what is material to Broadcom in terms of value and scope of an agreement is not necessarily material to Qualcomm.  Nonetheless, because of the extent of the litigation across multiple jurisdictions, the distraction it provided to management, and the final settlement payments totaling US$891M, I might have expected Qualcomm to file the Settlement Agreement itself with its 10-Q that disclosed the existence and outline of the Agreement.

The Settlement Agreement is quite long and detailed, and the SEC-permitted redactions in the publicly released version obscure what may be some key terms and provisions.  Nevertheless, what seems fairly clear is that Qualcomm's licensing system remains intact as to everyone other than Broadcom and its customers.  And even as to Broadcom's customers, the Agreement makes distinctions as to "exhaustive licenses" vs. "non-exhaustive licenses", "cellular products" vs. "non-cellular products", and the existence or absence of "pass through rights" in different circumstances.  At the same time, Broadcom had earlier brought a declaratory judgment action against Qualcomm arguing that the latter's licensing system was illegal, especially in light of the Quanta decision.  The case was filed in the U.S. District Court for the Southern District of California as Broadcom Corporation v. Qualcomm Incorporated, Case No. 08cv1829-WQH-LSP (October 7, 2008). The case was dismissed without prejudice on March 12, 2009 based on the court's decision that Broadcom presented insufficient claims showing an immediate and actual case or controversy.  Thus, it did not rule on the merits of Broadcom's allegations that Qualcomm's licensing system was illegal.

So what is Qualcomm's licensing system?  Qualcomm described it in an amicus brief it filed in the Quanta case:

"Qualcomm has provided chipmakers nontransferable, worldwide, nonexclusive, restricted licenses to its portfolio of technically necessary patents through licensing agreements called ASIC Patent License Agreements ("APLAs").  Chipmaker-licensees typically pay Qualcomm an up-front license fee and a running royalty (paid quarterly) that is an agreed upon percentage of the defined Net Selling Price of the chips produced by the licensee.  As is common in the industry and has been found by the Court plainly to be within the rights of patent holders, see, e.g., Gen. Talking Pictures Corp. v. W. Elec. Co., 305 U.S. 124, 127 (1938) ("That a restricted license is legal seems clear.") (citing Mitchell v. Hawley, 83 U.S. (16 Wall.) 544 (1872)), the license that Qualcomm has granted in APLAs is limited in both scope and conditioned upon the licensee acting within the bounds of its limited license. An APLA provides the chipmaker-licensees with a license to make (or have made) its own ASICs [ed. "Application Specific Integrated Circuits"]. An APLA also provides the chipmaker-licensee with a restricted license to sell ASICs, but only to handset makers that that APLA defines to be an "Authorized Purchaser" for incorporation into fully assembled handsets.  Authorized Purchasers are those handset makers that themselves have a license from Qualcomm through their own Subscriber Unit License Agreement ("SULA") to make, use and sell fully assembled handsets that, in the absence of a SULA, would infringe Qualcomm's patents.  Importantly, by their express terms, APLAs do not grant a license to the chipmaker to use the ASICs--i.e., licensed chipmakers may not themselves use or pass on to others the right to use the chipmaker's ASICs to make, operate or sell handsets or any other product. APLAs explicitly state that the rights to use the ASICs to make, operate or sell handsets are only conferred by licensing agreements between Qualcomm and Authorized Purchasers (i.e., by SULAs). APLAs also expressly state that the license granted is only for the limited scope laid out, that no other license is granted or implied and that if the chipmaker-licensee sells ASICs to entities that are not Authorized Purchasers, the licensee has materially breached the APLA, which gives Qualcomm the right to terminate the agreement, including the license granted."

"As previously mentioned, producers of chips that are licensed through APLAs are granted, inter alia, a license to sell such chips only to handset makers that have entered into a SULA with Qualcomm. The standard terms of the SULAs have granted handset makers a nontransferable, worldwide, nonexclusive, unrestricted license to Qualcomm's patents to make (and have made), import and use handsets, and to sell (and offer to sell) completed handsets. SULAs typically provide for an up-front licensing fee to be paid to Qualcomm, along with a running royalty (paid quarterly) that is set as a percentage of the Net Selling Price of the handsets sold. The broad license typically provided in Qualcomm's SULAs exhausts Qualcomm's patent rights when handset makers sell finished handsets to their customers--typically wireless carriers such as Verizon Wireless--and pay Qualcomm the royalties due under the SULA. Qualcomm has not sought to license its patents to or receive royalties from wireless carriers or end users."

"Qualcomm is also in the business of developing and selling its own chips and software for wireless handsets. Qualcomm typically sells chips only to those handset manufacturers that are licensees to Qualcomm's patents under a SULA. Such chip sales are pursuant to Components Supply Agreements, in which handset makers agree to pay Qualcomm an agreed upon price for the chips sold by Qualcomm. Components Supply Agreements provide that the buyer-handset makers may only incorporate the chips purchased from Qualcomm into fully assembled handsets that are the subject of the SULA between Qualcomm and the handset maker.  The sale of Qualcomm chips pursuant to Components Supply Agreements does not provide the buyer-handset makers any license to Qualcomm's patents, and Components Supply Agreements expressly state that no patent license, express or implied, is granted by the sale of the chips or the Components Supply Agreement.  Rather, all patent licensing is covered by the SULA.  Components Supply Agreements contain a representation and warranty by the buyer-handset maker that the chips purchased pursuant to the Components Supply Agreement will be used solely to develop and manufacture handsets for sale subject to and in accordance with the SULA between Qualcomm and the buyer-handset maker."

Because this system allows Qualcomm to collect royalties on the same patents from both chip makers (under the APLA) and handset makers (under the SULA), it has been criticized by others and alleged to be invalid or illegal after the Quanta ruling, such as by Broadcom in its declaratory judgment action. But I think that the Supreme Court left exactly enough space in its Quanta holding for Qualcomm's system to remain valid.  Despite what I sometimes hear as an interpretation of Quanta that the Supreme Court invalidated all "double royalty", multiple licensing, or value chain licensing systems, it does no such thing.  The case instead turned on whether there was an authorized and unconditional sale of products substantially embodying the patents at issue. If so, then exhaustion applies and the patent owner cannot bring an infringement action against the purchaser of the product, or demand that it take a license.  In the LG-Intel license at the heart of the Quanta case, the parties had an unlimited license to make and sell chips embodying each other's patents. A separate agreement required Intel to notify its customers that it (Intel) had no authority to grant a license to them to use the chips.  Intel did so and thus complied with this separate agreement.  The problem was that the customers, including Quanta, realized that because Intel was fully licensed to make and unconditionally sell chips to them, the patents were exhausted upon such sale and thus no license was needed from LG.  The Supreme Court agreed with this position, but was quite careful to explain that this holding rested on the exact nature of the unlimited make and sell license.  The Court seemed to endorse some of its earlier holdings, such as General Talking Pictures (see above), that allowed for restricted or limited licenses to cut off exhaustion defenses by certain classes of end users or other downstream purchasers.  This rests on the need for the sale to be "authorized".  Thus, in General Talking Pictures, the patent owner licensed a manufacturer to make and sell amplifiers but only to amateur users and not to commercial users.  The Supreme Court in that case held that when the manufacturer knowingly made and sold amplifiers to commercial users as well (who also knew the license limitation on the manufacturer), these actions were the same as if the manufacturer were operating with no license at all (as to these sales). And because the sales were not therefore authorized, then the purchasers were treated as having knowingly purchased what were essentially counterfeit goods, and thus no exhaustion had occurred and the patent owner could sue them for infringement as well.

Accordingly, following Quanta, my advice to patent owners has been to make sure that they limit the scope of their licenses to only the activities and sales they truly wanted to authorize, with the understanding that whatever was authorized would then be subject to the exhaustion doctrine.  Essentially, this was just a reminder that the "make" ("have made"), "use" ("have used"), and "sell" ("have sold") patent rights are indeed severable and can be individually limited.  An open question is whether "conditioned sales" under the U.S. Court of Appeals for the Federal Circuit's decision in Mallinckrodt Inc. v. Medipart, Inc., 976 F.2d 700 (Fed. Cir. 1992) can still hold off the application of exhaustion to goods sold under those terms after Quanta.  But that could easily be the topic for another blog entry.  When asked whether Quanta "changed" the law as to any of this, I have generally said no.  Intel and LG's licensing counsel should have known that an unlimited make and sell right was an invitation for exhaustion claims by downstream purchasers when they established the cross license in the first place.  Yet, before we rush to judge them too hastily, keep in mind that any negotiated deal can have certain sticking points that the parties cannot resolve and the resultant memorialized agreement may intentionally contain vague or suboptimal clauses (from a legal perspective). And indeed at a conference in which I mentioned this very point in the context of the LG-Intel license, an audience member stated that he knew one of the attorneys involved and that there were indeed business and negotiation issues that precluded the use of what we as licensing lawyers would have seen as the optimal make and sell right grant language, given the goal to limit some downstream uses.

In the second part of this feature blog post, Professor O'Connor will examine Broadcom's declaratory judgment action and further discuss Qualcomm's value chain licensing system.

The post Does Qualcomm's Value Chain Licensing System Survive Its Settlement With Broadcom? (Part One) appeared first on IPOsgoode.