Data recovered from Gene Roddenberry's floppies—but what's on them?


[Read the post]


Did his custom computers accept voice commands spoken through the mouse?


And it could speak back! Confusingly, it sounded just like his wife…


I’m guessing a copy of this. :sunglasses:


Three months? Someone was being paid by the hour methinks.


In the Before Time, each computer manufacturer rolled their own disk controller logic and their own disk file system. All the disks were the same physically, but a disk formatted for one machine could not be read on another machine without special software (and even then only some disks could be read, as other hardware families used non-standard disk motor timings or some such). With no working machine and no data on what the hardware or OS that created the files actually was, three months seems perfectly reasonable. Be glad that the data was word processing files (90% ascii), if they’d been trying to retrieve binary data like spreadsheets, then they would have needed to reverse engineer the software file format as well.

I have “fond” early 90’s memories of stumbling across info on a DOS program that could read any CPM disk (in a computer magazine, no internet obviously), and then, after downloading the software from a BBS, spending hours looking for a computer on my all P/S-2 university campus that actually had a 5" floppy drive so I could retrieve and archive five years of files from my old, defunct Kaypro.


My guess: 115 Kilobytes of pixel pr0n!


I find it hard to believe any floppy disk actually retained usable data for more than a few years. Hell back in the day it seemed like every time I wrote something to one I ended up with a bad sector…


This feat of reading custom disks without documentation may not have been as difficult as the article makes it out to be.

I spent a couple years c.1980 working at a company that built STD-bus computers, which had little 4x6" boards and an 8-bit bus. We made Z80s. The DOS of the day was CP/M, which required a custom BIOS code library that lived in the EPROM on the computer, to allow the standard DOS to interface to the particular hardware, much as today.

I personally wrote a version of BIOS that could read all the flavors of floppy disks, both 5" and 8", with varying sector counts and sizes, data density and track density. And it fit in 4K of EPROM. And I was a kid, still in college.

But for the love of god, please publish Roddenberry’s writings!


I am dubious of the claim he “made his own computers”. Looking at the auction house photo, it seems clear the machine’s case is homemade, yes, but designing hardware and OS from scratch, no. Roddenberry was a TV producer, not a Woz-level hacker.

More likely he assembled it from a kit (which was a thing you could do back in the day), and under the hood it’s an unremarkable early 80’s system, but there were so. many. early 80’s hardware families, and at this late date it would be near-impossible to find out who made the kit or what kind of OS it ran, and all the heirs remember is that Roddenberry “built it himself.”


Which sector was that?



They found a script, titled


I thought it might be a murder mystery: “A:\ SETUP”


It’s easy when you have the documentation. Working from scratch, without documentation, without working hardware, determining which of the dozens of different standards the disks were written in, is non-trivial. Once the data recovery tech realized that the disks weren’t written by any of the usual hardware systems, without documentation he really didn’t have a choice except to brute force it, working out the track/sector arrangement and filesystem from first principles.


Yeah, you’d think so, but if they used the standard of the day, the Western Digital WD1791 family, then there were only so many permutations available. It’s just a matter of hooking a 'scope up to the read data line to see the sector size and data rate, then programming the chip to do a raw read of all tracks and sectors, and descrambling the sector interleave by seeing which sectors follow which. A day’s work. And it was way more likely than not to be CP/M.

On the other hand, if he used the Woz approach of building his own sector sequencer with a state machine, then yeah, it could be a mess. But Roddenberry was not the Woz.


So these were probably written on WordStar under CP/M.



Also, this is the company’s business…you’d like to think that they have built a toolset that would try all of the various permutations in 20 minutes or so. Or maybe they are just goldfish.


I’m no expert, but I’ve heard OmniFlop is fairly decent.


No, “home made” in this case is just a way to talk to the non-technical.

Yes, in the very early days of home computers, people did build from scratch. But except right at the beginning, those were generally limited systems. So lots built “Cosmac Elfs”, but they had limited memory, so they weren’t for practical purposes.

But it was a different time. You could buy boards (assembled or a kit) and put them together in your own case, and have a “standard” computer that was customized. Or have someone do it for you.

There were “Big Boards”, buy a board and parts and put it together. Then the Apple II clones, and when the “IBM PC” came long, clones of that. Solder boards together or buy them preassembled, and then put in your own case.

Even assembled, there were plenty of paths to the end, both in the CP/M days and the “IBM PC” days, the operating system putting some level of standardization on the computer.

More than anything else, this is a computer without documentation. It likely is a combination of standard hardware and software, though perhaps the word processor was custom. There was a period when writers hoped to have their “dream typewriter” and did have custom software written. And a lot if those writers stuck with what came first, familiarity being more important than new features. Every so often we do hear about some writer still using an Apple II, or someone else hoping to find a replacement for their familiar old computer that thy need to keep writing.

But with no documentation, it’s easier to call it “custom”.

The non-working computer might just need some cleaning. But in itself it’s a map. What’s the CPU? What’s the vintage? The ICs will have a date code. Are they really home made boards, or boards with a company name, so information can be found? There’s probably lots of information in the computer to make the next steps easier, but the article wants to dazzle the reader, rather than inform them of a tedious task.

That computer provides hints that would limit the iterations needed to recover the data.