Trademarking Infocom, again, part two

I posted yesterday about a company called SmartMonsters, who are running a port of (MIT) Zork in a MUD framework. They are trying to register the "Infocom" trademark.

But it turns out that someone else is trying to do the same thing. If you look at the URL infocom.xyz, you'll see a bare-bones site which claims "Infocom and the Infocom logo are trademarks of Infocom LLC." According to business records, Infocom LLC is a company formed in Colorado Springs in 2015.

Now, a US trademark search turns up no mention of this crew. And it looks like they've been claiming the "Infocom" trademark for years with no registration. But I am told that they are objecting to SmartMonsters' use of it. I don't really know how the trademark-tussling process works, so let's just say it's "in contention".

(I need hardly say that registering the Infocom trademark gives you no rights at all to the Infocom games. Those are covered by copyright, an entirely separate matter.)

So what are they doing with the trademark? The answer is a job posting that appeared last week:

In this position you will be provided with the source code for a proprietary assembler that consists of slightly under 4,000 lines of code. The source code you will study was written in assembly language to run on the TOPS-20 operating system on the PDP-10 mainframe computer. [...]

In this position you will play an important role by writing a functional specification document that describes the functions, program flow, error handling, and other information of the assembler that the person operating inside the clean room will need to know to develop a compatible replacement program. The replacement program is expected to be able to process the same input files and to generate bit-identical output files. [...]

The final specification will be made available under GPL-3.0-or-later. The software developed inside the clean room will be released under AGPL-3.0-or-later.

"Freelance Specification Writer" posted by Infocom LLC

This is an exact description of Infocom's ZAP assembler, which was part of their ZIL toolchain. (ZILCH turned ZIL code into Z-code assembly; ZAP turned the assembly into a Z-code game file.)

The source code in question turned up in the Infocom source-dump which appeared last year. Nobody noticed it right off. But a few months ago, a sharp-eyed user spotted ZAP buried in the MiniZork source directory.

The file "zap.mid" is MIDAS assembly code, an MIT variant of PDP-10 machine language. And it is indeed about 3800 lines long.

The job post describes a classical clean-room setup. You do this if you want to make a work-alike copy of someone else's program that isn't derived from their source code. The result does the same job -- "identical output", like the post says -- but you own it. This is legal because algorithms aren't copyrightable. (It may be ethically sketchy -- that's another whole question. But it's legal.) (Unless the algorithm is patented, but that's not the case here.)

So that's what this company is trying to do. The next question is, why? This is where my brain falls flat on the floor. And not just mine. I asked Jason Scott. He passed the word along to the old Infocom folks. Nobody, I mean nobody, can figure out what the point is.

The infocom.xyz site has two games up, which look like Lode Runner clones -- Linux only. This doesn't give much clue what line of business they're in, other than "not in it for the money".

(Yes, I sent them email, in case they were willing to tell me. They said "no further comment beyond what has already been made public.")

Let's be clear about what already exists. There are several open-source compilers that handle Z-code assembly. zasm does it; Inform 6 and ZILF both include the capability. We also have throrough descriptions of the Z-machine architecture, both Infocom's original document and the modern reconstruction. And of course there are dozens of open-source interpreters which play Z-code games.

All of these tools derive from the reverse-engineering work that went on in the late 1980s. The InfoTaskForce's seminal Z-code interpreter is archived here.

That was no kind of a "clean-room" project! The InfoTaskForce group dug into Infocom's proprietary games and interpreters, figured out how it all worked, and reimplemented it. (The Infocom spec document I linked didn't turn up until years later.) If Infocom, the original company, had wanted to make a legal issue of it in 1989, they probably could have. But they didn't.

After that, everything discovered by the ITF was public knowledge. The modern Z-machine spec (originally written by Graham Nelson) was a collation of that knowledge; Graham did not have to decompile Infocom interpreters. That spec has a Creative Common license (BY-SA-4, noted here). It's freely usable in every practical sense.

You can say that all modern Z-code/ZIL tools are "tainted evidence", due to the original ITF reverse-engineering. But it's a tenuous argument. And it still leaves the question of what you'd use a "less-tainted" ZAP assembler for.

Academic purposes? Studying Infocom's tools and processes is a worthwhile (and fascinating) goal. But it makes no sense to use a clean-room tool for that. You want to study every scrap of information available!

Compiling Infocom's ZIL source code for fun? There are plenty of people doing that already, using existing open-source tools. Some folks are even tackling bug fixes and modernization. (Yes, Activision's copyrights are a question here. The concensus is that volunteer updates to the source code are fan activity and basically okay. Don't go selling them, is all.)

Compiling Infocom's code for profit? A clean-room compiler or assembler doesn't give you any leverage there. You're still building a game file derived from proprietary source code. Again, selling it without Activision's permission would be right out.

Writing new, original ZIL games for fun? As I said, this is already a popular hobby. The forum is buzzing with ZIL programming chatter.

Writing new, original games for profit? I gotta tell you, ZIL is not the right tool for that job. Even if you think you're going to get rich off parser IF (tricky at best), you'll want a modern tool which can handle dense, highly-detailed games. Of all the Infocom alumni who revisited parser IF (Marc Blank, Mike Berlyn, Bob Bates), none of them chose ZIL.

For the sheer challenge of the hack? Maybe? But people usually don't put down money for that sort of fun. This setup involves hiring a documentation writer and a copyright lawyer, for a start.

So I'm left with nothing. My best guess is that they want to write unauthorized sequels to the Infocom canon, but they don't understand either the legal realities (a clean-room assembler gains you nothing) or the IF community. My second guess is that they want to contribute a legally unencumbered open-source tool, but they don't understand that this has no practical value. I dunno. They're certainly doing nothing to dispel the impression that they might have sketchy intentions.

On top of which, they consistently present themselves as "Infocom". Not "Infocom enthusiasts" or even "a new generation of Infocom". This tweet (from last year) is an eye-rolling example. I also see a discussion thread in which someone said they were (briefly) selling the old Infocom games without permission? I can't verify that, but jeez.

I guess we'll find out, or else they'll sink without a trace and we never will.