Technological and Human Failures

Let me start with a mea cupla: had I properly archived the old drawings discussed below in 1996 none of the effort described would be necessary today. While there’s enough blame to go around – and in some ways, the problem is the result of blind circumstance and maybe no one’s fault – I can’t escape the fact that I made a serious mistake thirty years ago.

Our old drawings are rotting. I don’t mean literally, as we went down the paperless, digital path starting in 2017, but metaphorically. Actually, all of our old documents are rotting, but the good news is that things for text documents have already gotten as bad as they can get, so it bothers me less. The issue is documents created between 1992 and 1998. The first year is when I started Friedman Structural Engineering, the second is when that business merged with what was then the LZA Group, now Thornton-Tomasetti. Records since 2002, when I started up on my own again, are in somewhat better shape. The specifics:

• Our old text documents were created in a number of different, all now defunct, word processors. The files are readable as plain text files: after you scroll through several hundred lines of gibberish (internal program commands rendered as text) you find the actual meat of the document at the end. All formatting and images are gone, but plain .txt files are not going away, so these files will remain readable. As I said, it’s as bad as it can get and so won’t get worse.
• From 1992 to 1996, our drawings were created in a DOS-based program called Generic Cadd. The name, I believe, referred to the idea that it was CAD for any purpose rather than being optimized for architecture, or machine design, and so on. Terrible name aside, it was a good 2D drafting program: fast, with a reasonably comprehensive tool set, and all commands could be entered via two-letter codes (out of the box – I believe some other drafting programs allowed you to map codes to commands, which sounds like a lot of work). It was a small player competing with the big names, which was more likely to succeed in the 1990s world of software than it is today. AutoDesk bought it, attempted to combine it with (Windows-based) AutoCad LT, and eventually spun it off again. There were a few successors that maintained file compatibility with the GCD files of Generic, of which two, Visual Cad and General Cad Pro, survive.
• When we switched from Generic to AutoCad, we exported (from within Generic) files for active projects to AutoCad’s DWG format. (Of course, DWG has changed a great deal since the 90s, but current programs can read the old files with only minor weirdness happening on import.) As I stated at the beginning, I made the decision back then to not go back through the 300 or so projects that were not current and export their GCD files.
• Old Structures has, since 2007, run on Macs. There are many good aspects to doing so, which is why we do, but backward compatibility with DOS is not high on that list.
• In the past week, I’ve been contacted about new work at two buildings we worked on 1995, for which we have GCD files. We can get scanned images of our old drawings for both, but we’d have to redraft.
• First adventure: I still have the install floppy disks for Generic, saved as disk images. Using DosBox-X, I was able to get the program running and open the drawings. But the export to DWG function is broken.

• Second adventure: using UTM, I was able to get Generic installed in a virtual machine of Win98. Again, export was broken.
• Third adventure: using Crossover, I was able to get the demo version of General Cad Pro running and open the drawings. But the demo version has saving disabled.

The outcome: I’ve bought a license to General, which I don’t need for drafting, so I can open files we’ve had archived for thirty years, with the program running in emulation. This is, as the saying goes, suboptimal, but it’s better than losing the data embedded in those files entirely.

We now draft using Vectorworks, which I like quite a bit. It seems other people in the office do as well. The differences between CAD programs are smaller than they used to be: if you know AutoCad, Vectorworks is quite easy to pick up. Pretty much every year both AutoCad and Vectorworks update their file formats, so old files slowly rot, although as I said, today’s programs can read older versions of their own formats with only small-scale problems. It’s the nature of software development: if new features are added, the file format is likely to change to support them, so the past drifts away. It’s one reason I prefer tangible technology: we can put new bricks in an old brick wall, and new bolts in old rivet holes. There may be compatibility issues, but they are – at least so far in the 38 years I’ve been working – solvable.