John Shoch on Clean Rooms
Thursday, December 19, 1996 by Dave Winer.
John Shoch, email@example.com, a former Xerox exec, ran the division that produced the Xerox Star, the earliest graphic desktop system. Many of the ideas behind the software we use today came from innovations at Xerox.
Shoch is a computer scientist, executive and most recently a venture captialist at Asset Management in Palo Alto. He has a lot of experience with intellectual property in software, and has strong opinions on whether syntax can or should be copyrightable.
This essay came in response to my Does Anyone Own Syntax? 12/6/96.
Tangible property is easy to protect because it is, well, tangible. You can lock it up, and you can tell when it has been stolen. Intangible propery is hard to protect. You can steal an idea or copy a song, but the original owner does not know that it is gone.
Computer software is not the first form of valuable intellectual property -- our IP laws go back centuries.
Patents provide protection for an underlying idea, independent of any particular representation, with the limit that the idea be new, non-obvious, and useful.
Copyrights provide protection for a particular representation, but not an underlying abstraction or idea.
You can get a patent on the idea behind a transistor design. You can get a copyright on the artwork for a particular silicon implementation. You may need a license to the patent on the idea, in order to implement your copyrighted design.
If you have either a patent or a copyright, you can decide if you want to share these with others freely, license them, or keep them proprietary. Your choice may determine your personal satisfaction, or your economic success, but it is your choice.
AT&T decided to license the patent on the transistor, not seeing it as part of their core business. We could argue about the ultimate economic effect, but it was their choice. They would have been well within their rights to keep the patent, and do the investment to exploit it on their own, thus monopolizing this invention. That's the purpose of the patent law.
Polaroid patented many aspects of instant film, chose not to license it, and invested substantial amounts to take it to market. Kodak later tried to enter this field, and lost a massive patent suit. That was good for the Polaroid shareholders and bad for the Kodak shareholders. As much as we all liked instant film, however, Polaroid was under no obligation to license it just because we might have wanted cheaper film.
Unfortunately, the IP laws have had some serious problems when applied to software. In the 1970s and 80s the Patent and Trademark Office (PTO) would not grant patents on ideas to be embodied in software, thinking of them as being only abstract ideas like mathematical formulas, and not eligible for patent protection.
At Xerox, for example, we could get a patent on innovative hardware, such as the ball mouse, but not on the software. Thus, at that time we could not even apply for a patent on the new, innovative ideas which were the result of years of labor: the desktop metaphor, symbolic icons, multiple hierarchical windows, WYSIWYG editing.
We should have been able to patent these broader ideas, independent of their specific implementations, be it in Smalltalk, Bravo, Alto OS, Star, Mac, Windows. So we were only left with the alternative of seeking copyright protection on the specific representation -- the code and the visual displays. That stopped someone from copying the code, but did not stop anyone from copying the ideas and implementing them slightly differently.
I view it as one of the major tragedies of the computer era that the PTO took such a narrow view during that time, depriving many imaginative individuals, and their employers and shareholders, of the protection their ideas deserved. More recently, the PTO started accepting patent applications for ideas embodied in software, but it was too late to protect the earlier work.
Because of their late start in allowing patents on software, the PTO has let some unfortunate ones slip through. When a patent examiner looks at a patent application, his major source for evaluation is the prior collection of patents. Thus, some examinations have allowed patents on things which were actually subject to extensive prior art, but were not in the patent files. These are mistakes, which can sometimes be corrected, but it is an expensive nuisance.
Some people, however, have seen this happen and then claim that we should not allow patents on ideas relating to software. I think that is wrong -- we just need to stamp out the spurious patents which are not genuinely new. The community needs to rally to help strike some of these down, but also come to the aid of those who have patented something truly new and different.
There has emerged a gray area between patented ideas and copyrighted representations. Consider a research team that spends a year exploring alternative user interfaces and command sequences. They run detailed experiments, watch subjects, and identify a set of command sequences which clearly reduce confusion and produce a new, better UI. But this may not meet the test for an "idea" that can be patented, nor a representation which can be copyrighted.
Your question about syntax falls into this category -- how do you protect a grammar for a new language? Is it a protectable idea? Is it copyrightable? In this situation, many people have tried to stretch the notion of copyright protection to cover these forms of behavior or interfaces.
The most notable example in this category, I think, is PostScript. Developed by Adobe, it's a description language for documents, defined by a grammar. I don't believe that it qualified for patent protection, but their specific implementation of a PostScript interpreter is covered by copyright.
Adobe invested heavily to make the best PostScript interpreters, and got most of the printer manufacturers to buy their proprietary code. Some manufacturers tried to develop their own independent document formats (e.g., H-P), while others have implemented and sold independent PostScript interpreters.
As one who is generally supportive of providing protection for intellectual creativity, I am sympathetic with efforts to protect this kind of work. But I know it's a grey area. If you can get broad patent or copyright protection, more power to you. If not, you will face more intense competition.
If you build a better widget, and customers are willing to pay for it, that's to your credit. If you get a patent on your widget, others can't copy it, and customers can only buy it from you. That's what Polaroid did with instant film, and Kodak got clobbered when they infringed on those patents.
If you can lock your customers in, that's a testimonial to your creativity. But if you price it too high then they will go somewhere else, or abandon the product (look what 1-hour developing has done to Polaroid). You can make an enlighted judgement that, in the long run, you will grow the market much larger if you license your rights (as Xerox did with Ethernet), but that's a decision to be made by the innovator.
If someone is inflexible with their approach, go work on something different, which will be better. If you can't dislodge them, then their lock-in strategy has worked, to their benefit and the benefit of their shareholders. Buy their stock, or buy something else, or else just don't use it.
Some people try to step around copyrights with the notion of a "clean-room copy". This seems to have survived legal examination, but does not seem at all right.
Consider this scenario: I decide I want to create a new cartoon character, a mouse with big ears. So I sit in a room with pictures of Mickey Mouse, and I write a textual "spec" for my new character. Draw a mouse with round eyes about so big, and a nose like this, etc.
Then I toss this written spec into a clean room, where there's an artist who has never seen a Mickey Mouse cartoon. He takes my spec, draws a mouse, and passes it back to me. I tell him to put the eyes closer together, add a tail, etc.
After some number of iterations, we now have our new mouse -- one who looks just like Mickey! But the artist never saw the copyrighted image of Mickey, he "independently" produced it.
I don't think Walt Disney would stand for it.
I don't think so either!