Copyright Office: Jacqueline Charlesworth
Stacy Cheney (NTIA)
Proposed Class 26: Software – 3D printers
This proposed class would allow circumvention of TPMs on firmware or software in 3D printers to allow use of non-manufacturer-approved feedstock in the printer.
Proponents: Sherwin Siy, Public Knowledge: be specific about the copyrighted work at issue. The software embedded in the printer, not anything produced by the printer, and not even necessarily the software embedded on the chip. Since 2003, when 2D printers came up, Lexmark has moved that a bit more out of the arena of discussion. Adverse effect: Stratasys is quite clear about how they view their ability to lock TPMs to particular printers can benefit them, and their incentives are clear.
Charlesworth: tell me more about the software as you understand it at issue here.
A: we’re concerned about whether accessing the software in the printer would constitute access under 1201.
Charlesworth: reading through submissions, seemed that there are two components: (1) the chip on the cartridge that (2) then locks into the software on the machine. How does your exemption deal with both, or does it?
A: Ultimately what we want is to be able to use a chip that was not created by the original mfgr or use feedstock not created by the original mfgr. Both programs are involved, but the © work at issue would be the one in the printer, as in the Lexmarkcase.
Charlesworth: what’s the nature of the circumvention?
A: number of ways. Could copy software on the chip. That would interface w/the computer program on the printer. Q is whether access by using the printer and creating potential RAM copies triggers 1201. Substitute chip. You could also reuse/modify an existing chip.
Charlesworth: and in copying software is there circumvention involved in the chip?
A: I don’t think there’s circumvention. It’s likely to access a noncopyrighted program. But to the extent that’s necessary … It would vary based on the nature of the system.
Charlesworth: what TPM is on a chip?
A: it protects the interface b/t the chip and the printer itself. In many cases there is no overt TPM, it’s just a question of reading off the chip itself. It depends on other variables.
C: there might be a © issue but not a TPM issue.
C: and the machine?
A: the circumvention would be circumventing the measures that require a mfgr-created chip to operate. Authentication. The TPM controls the authentication.
Q: are you doing anything to the software itself?
A: it depends on what the feedstock is. You might want to change some of the variables in the program itself. It would vary depending on the model of the machine and other factors.
Michael Weinberg, former VP at Public Knowledge, now Shapeways, here in personal capacity: Part of the issue is that once you get past an abstract level: feedstock container and chip verifying that the container came from mfgr—there are a number of technical ways to implement that verification. You could make the chip very dumb, with most info/action on the software side. You could structure the system w/ more info in the feedstock container chip so the two pieces talk to each other more intensively. Depends on how you implement it. The key thing is that at a high level of abstraction, there is at a minimum software on the printer waiting to be used until it can verify that it has approved input and that software is likely protected by © so the only way you can access the interface is to convince it you have approved feedstock. Implementation varies from machine to machine and mfgr to mfgr.
C: the TPM is the need for the chip to make the printer run? Explain what the TPM is.
Siy: the system in its entirety that requires the presence of the chip. Whether the code is on the chip or more likely on the printer, that software code is the TPM, the part that requests authentication.
Weinberg: it’s easier to describe a dumb chip on a cartridge and a smarter printer, but I want to be clear that we aren’t restricting our request to that specific type of system.
C: what do you mean by a dumb chip?
Weinberg: may be as simple as serial number or RFID. Information presenting to be read.
C: like a key.
Weinberg: like a swipe card to get into a building. It’s not doing any processing, but w/out it the locks on the building won’t work. You could have a smartphone that would communicate back and forth with the building. The level of technical sophistication in keycard v. phone is different.
Q: is this intended primarily for consumer use or at the manufacturer level? Larger 3D printers?
Weinberg: Motivated by consumer use, but no reason to exclude manufacturing/more sophisticated commercial players. Worth noting there may be and likely are noncopyright related restrictions on what owners or possessors are doing—contractual, warranty. The focus is the ©/DRM part.
Q: This is just about feedstock bypass, not software that reads 3D modeling.
Weinberg: exactly correct. Nothing to do w/models going into machine or coming out. Not even related to software running the machine except for the feedstock part.
Q: does breaking TPM for feedstock affect 3D modeling software?
A: the process that reads and verifies the input doesn’t have to be linked to the process linked to slice the model. One thing the machine will do is slice model horizontally into thin layers and establish a traveling path through each layer to create it. While the path is determined by the type of material, it’s not determined by the origin. It could be implemented in a tied way on a machine.
Opponents: Ed Kerry (sp?), Stratasys Ltd. (or designated alternate Stratasys witness): The output is very dependent on the material and authenticity, though the input is not.
Siy: though that only means product quality, not © status.
C: might different TPMs protect the basic software v. the modeling software? Or once you break it do you have access to everything?
Weinberg: there’s no technical reason. Mfgr can decide how to implement it differently. You can choose one that simply governs feedstock source.
C: what’s in the market?
Weinberg: I don’t know b/c no one is particularly interested in making unauthorized copies of the software that runs the machine. The reason people worry about these is that they want to use unapproved feedstock; not discussion about accessing the software to copy it.
C: do you need to copy the software to modify the feedstock?
A: depends on the mfgr.
C: What is the legal basis for your exemption?
Weinberg: two core harms. Both flow from a cloud of ambiguity over whether this behavior even triggers 1201.
C: what is the noninfringing use? [um, making stuff, which doesn’t itself implicate ©?]
Siy: We want to figure out §106 uses—RAM copies made in the utilization in the printer itself, or modifications necessary to use the feedstock. Both fall within §117.
C: if you alter the software you might be creating a derivative work.
Siy: allowed by §117: adaptationsor copies created as essential step. They aren’t being made to distribute or even leave the machine.
C: who owns the computer program in the machine?
Siy: the person who owns the machine.
C: what’s the evidence of that?
Siy: you could recognize that to the extent owned by the hardware owner, §117 requires; MDY ruled that any contracts were covenants when the Q is whether the use of the software is licensed or not.
C: what are the typical mfgr practices?
Siy: regardless of language of license, that will allow you to use the printer.
C: but are they even claiming it’s under license?
Siy: it will vary from mfgr to mfgr.
Weinberg: especially in the consumer market, while I don’t have a study of current licenses, there are probably 70 desktop 3D printer companies of highly variable legal sophistication. It would be highly surprising if you didn’t see almost every version of © license theory applied to software in this space, including silence.
C: have you seen any purported licenses?
W: not in context of these proceedings.
C: any you’re sure are sold without a claimed license?
W: there are printers that are open licensed.
Q: have you looked at the printers using TPMs for their licenses?
Kerry: VP for Stratasys. Deal w/ top 120 manufacturers in the world. Prototypes/mfgr of parts. Those same customers buy low end Makerbots as well. New 3D companies this year: 150. We have revolutionary technologies, including using inkjet heads. Our software calculates droplets and can use 3 different materials at a time. My customers also use our service business, where we make parts for customers. We are helping them go to direct digital manufacturing of tools and end use parts.
A large airplane manufacturer is putting 1000 plastic parts on a plane; we spent years certifying this plastic for airplane parts. The highly integrated machine that prints it as FAA certified part is very important.
C: Are the airline companies making those parts?
A: they’re airline companies and downstream manufacturers.
C: why do they care?
A: customers don’t want anybody to be able to get into that integrated system. They want a reliable part that is exactly what they declared it to be. They don’t want people using cheaper feedstock. Traceable and reliable to be kept for years. [And that is related to © how?] Important that no derivative works are created and no feedstock introduced.
C: are there specific regs on making airplane parts?
A: The FAA. The FDA regulates medical devices. I’m not an expert. [Why is the © Office making decisions about this?] We help our customers certify parts and materials to the FAA and FDA.
C: these are manufacturing type customers, not Makerbot customers.
Q: can you test the part itself to make sure it’s used the right material?
A: The customer does test the part, though I don’t know if they test the material. Test it as they’d test from any manufacturer.
C: do they test every part? [© Office lawyer as airplane mechanic.]
A: depends on what the part is.
Q: Is the fear that someone not in the normal manufacturing stream gets ahold of inferior materials and then tries to put that part into the stream via passing off?
A: that’s my customers’ fear. They have multiple suppliers who supply counterfeit parts, which would affect our brand and that of the customers. [Wouldn’t that already be illegal and violate the suppliers’ contracts and probably trademarks?]
Our printers are fully integrated systems. They’re servers, not printers. We’re talking about the operating system and derivative works. Have to meet tight tolerances. Controlling the slicing is important. We track lot numbers, enabling verification of parts. Inkjets sold 600 million printers last year; we’ve only sold 120,000 systems and we’re the biggest—nascent industry.
C: answer Makerbot v. more sophisticated: how do you handle the software?
A: by license.
C: do you provide ongoing updates?
A: yes, we have an ongoing relationship. Thingiverse. We accept files made by different design software.
Weinberg: But Thingiverse doesn’t update the software.
C: do you send out patches to the Makerbot?
A: don’t know, but on the other lines we do.
Weinberg: I own a no longer supported Makerbot. There are machines that are no longer upgraded.
C: how do you get upgrades if they were there?
Weinberg: SD cards were required. As machines get more sophisticated they might be able to do direct downloads; it depends.
C: in many cases, you might be receiving upgrades for a consumer device.
A: it’s possible.
Q: is that software encrypted?
Kerry: it’s compiled to run on the printer. The feedstock chip is encrypted.
C: what is your warranty policy if someone modifies your printer?
Kerry: it’s no longer warranted. 1 year warranty otherwise.
Q: are you claiming the feedstock chips are copyrightable?
Kerry: not today, but we anticipate it might be.
3D printing is more complicated than 2D printing. IP is critical to the industry to justify investments for future development. I’m not a lawyer or DMCA expert, but I understand it was enacted to prevent circumvention of TPMs designed to prevent copying of IP. Exemptions were to be exceptional—a fail safe where there are substantial adverse effects on noninfringing uses.
Petitioners’ proposed exemption would deprive industry of useful tool during critical period. They aren’t seeking lawful access, but misusing the exemption process to encourage users to bypass controls where DMCA doesn’t apply and encourage users to infringe. Showing is insufficient.
C: Cheaper feedstock is a benefit, no?
Siy: and different.
Kerry: today that chip is not copyright protected, though we do plan to have one.
C: are you saying you don’t object to doing something w/ the chip?
Kerry: we highly object to counterfeiting the chip and to the exemption. This industry is just starting. They will not do medical devices, airplane parts, or car parts if they can be hacked.
C: on the consumer end, if I buy a makerbot and want to use different feedstock and break my warranty, so what?
Kerry: if the part comes out not right it affects our brand. Bigger concern on high end b/c derivative works of our server system is a huge issue b/c of our competitors in other companies. Opening that up for competition and un-integrating the system will stunt the industry.
C: are you familiar with the Lexmark case saying that using a different cartridge didn’t violate ©?
Kerry: yes, but that wasn’t a server. Printer toner is basically all the same. We print 100s of different kinds of plastics in different ways. Not a fair comparison. People rely on the objects we make.
C: respond to concern about the integrity of mfg chain?
Siy: in both consumer and commercial context, the person making use of the exemption is the person using the printer. They’ll be fully aware they’re the ones using 3d party feedstock. That’s their responsibility. [And has nothing to do with ©!] As for the commercial context, the FAA regulations will dictate what’s relevant. If you need a certain plastic, then the certification will say so.
C: but the concern is that bad guys will sneak bad stuff into the chain, b/c it will seem legitimate to use cheaper stuff.
Siy: Q of integrity of product as functional object depends on manufacturer and supplier. If they’re counterfeit, they’re counterfeit. The TPMs don’t solve the problem of unreliable suppliers.
C: makes it more likely that suppliers would circumvent and use inferior feedstock.
Siy: people would be violating their contracts.
Kerry: there are plenty of open printers w/ no encryption.
Siy: people who want to produce parts on the cheap can do that already then.
C: do you have specific examples in the record of people who want to circumvent commercial/mfg type printers as opposed to consumer printers?
Weinberg: there are companies actively developing alternative feedstocks for industrial printers. They see themselves engaging in an activity unrelated to © and don’t understand why they should come to this proceeding. One thing that can easily get lost: you have the idea of the primary benefit being lower costs of existing materials, but there also new materials for existing printers that are already owned. They’re looking at specific printers b/c different printers compete on different tech/functional capabilities—they aren’t a commodity.
C: why not just target it to open printers?
Weinberg: there are in desktop space, less so commercial. But the reason is some printers are the only printers that can achieve particular tech goals. If some are locked down you can’t necessarily achieve the same goal.
Weinberg: Stratasys says there are things only they can do.
Weinberg: the people who’ve developed open bio-3 printing started w/ a Stratasys machine b/c that was the machine w/ the technical ability they required for initial stages of their process.
Kerry: we regularly offer licenses for research/development for this exact reason. There are a lot of developing materials, and we’re the #1 material developer. It’s our greatest area of investment.
Weinberg: some of the techniques are patented, and by definition they’re only found in one printer. 20 years of patents tied to specific manufacturer.
Q: how useful 1201(f) could be?
Siy: That was pre Lexmark—we don’t know about the printer engine. About reverse engineering, not about interoperability. Copying chip directly makes 1201(f) uncertain alternative. As for reverse engineering the software itself: the Q is what the allegedly infringing use. If the act is reverse engineering, 1201(f) works, but the use itself doesn’t get certainty from (f). It does not obviate the need for an exemption across the uses necessary depending on the circumvention at issue.
Siy: no b/c of wide number of ways in which TPMs can be implemented. Kerry just said they wanted to include more sophisticated/copyrightable software on chips. Where the circumvention is necessary might change.
§117 is one of the ways in which the use by a consumer of a 3D printer of third party is feedstock is noninfringing. There are other reasons for noninfringing uses—use with permission. Even the most restrictive license will provide for the use of software. The only way in which use of a third party feedstock could infringe would be if you believed that a functional condition of the grant was not using third party feedstock.
C: but you might be altering the software.
Siy: but you have the right to use. [I don’t know why alterations in parameters would create a derivative work—if it’s swapping one set of physical facts about a material for another, that’s not enough creativity to create a separate work.] Also fair use.
Q: how much change in the software would be necessary?
Kerry: we test and tune the machine for new feedstock, and sometimes up to a year before the machine produces a high number of reliable parts. Motion control, heating, distribution, and layering b/c customers demand precision parts.
Q: but how much of the code is changed?
A: don’t know.
Q: any other distinguishing characteristics b/t high end and low end consumers? We say prosumer, tends to be professional. Home market is lower end. Pro engineer will use one on his desk.