May/June 2003  Volume 29, Issue 4
May/June 2003 Issue
   Format for Printing        Send Feedback
Technology / Innovating with Erik J. Heels
Zen and the Art of Data Restoration
Computers fail, people make mistakes, data gets lost. But what if it's a treasured legacy file?

In Robert M. Pirsig's Zen and the Art of Motorcycle Maintenance, the storyteller Faedrus says, "Motorcycle maintenance gets frustrating. Angering. Infuriating. That's what makes it interesting." So, too, with computers.

A big difficulty with computers is that standardization is elusive, even though there are so many standards to choose from… or perhaps because there are so many. Nevertheless, for many years I have been pursuing the elusive goal of digitizing, standardizing and archiving all my data. In this particular chapter of the journey, I ended up purchasing a 15-year-old computer to restore one file.

When Well-Laid Plans Don't Work
Through the years, I have modified my data backup procedures to account for the ever-increasing amount of data I create each year. There have, however, been notable failures along the way.

  • In May 1988, a Macintosh floppy disk containing my thesis project failed the day before my thesis was due, resulting in the loss of three key hours of work.

  • In May 1999, a denial-of-service attack on my Web site led to a hard disk crash and partial loss of three months of recent data, as well as most of my data from MIT and law school. My thesis was spared, thanks in part to excellent data recovery work by Ontrack (

  • In September 2000, while restoring a full backup to CD-ROM, my then-online backup provider failed to include the preceding month of my Windows data—and then the vendor went out of business. So I switched to another online provider for my Windows backups, (, with whom I'm still satisfied. Because, unlike my first provider, doesn't support both Windows and Macintosh, I found a second online provider for my Macintosh backups as well.

  • In January 2002, I changed the name of my Macintosh's top-level directory, which contained all the data that I back up. My second Macintosh backup provider ended up making two copies of all my data because the vendor's algorithm (unlike Connected's) wasn't sophisticated enough to figure out that the underlying files had not changed.

  • In March 2002, while doing a test restore from a Windows 98 backup from my first online backup provider, I discovered that my Windows 2000 computer could not read the Windows 98 backup CDs. Neither could any other computer. It made me want to sue that first provider, but, of course, it was out of business. So I started writing about the experience instead.

  • In April 2002, I began standardizing my Windows and Macintosh file names. I adjusted the MIME settings in Windows, Mac OS, Eudora and Netscape to correctly handle downloads and attachments. I eliminated nonstandard characters from file names and converted old files to more portable formats (Jfax files to PDFs, MacPaint files to GIFs and the like). For Macintosh files, I added three- or four-letter extensions to each file name, using two programs at once to rename batches of files: UtilityDog ( www.prob for Macintosh files and Magic File Renamer ( http://mfr.query for Windows files.

    There were about five times as many file types on the Macintosh as on the Windows machine, but the majority of files on both platforms were in a handful of formats.

  • In May 2002, I decided to move all my archived Eudora e-mail from my Macintosh to my Windows computer. Then I decided to move the rest of my files so that I could have all my data on one computer and backed up by one reliable online provider. I then canceled my account with my second Macintosh online backup provider.

What does all this have to do with the 15-year-old computer mentioned in the introduction? In October 2002, as I was preparing to purchase two new computers (discussed in my March 2003 column), I discovered that one Macintosh file—out of 12,500—from my May 2002 Macintosh-to-Windows data migration had been lost: the Macintosh program that I wrote for my thesis. Essential? No. Sentimental value? Absolutely yes.

Problem One: Finding an Old Macintosh
After checking with the MIT libraries, my thesis advisor and various backup providers—including my last Macintosh provider, which had already deleted all my backup files—I concluded that the only way to recover my thesis project was to recompile the application from the source code. I still had the source code, but I would need to purchase a compatible computer and operating system, plus the compiler.

I decided to try to purchase exactly the same environment that I had in 1988: a Macintosh SE with 4MB of RAM, a 20MB hard disk running Mac OS 5.1 and the Think Lightspeed C 2.13 compiler. Thanks to eBay, I "settled" for a Macintosh SE with 4MB of RAM and a 40MB hard disk ($32 including postage) and Think C 3.0 ($8 including postage). On Apple's Web site, I found a newer version of the operating system—Macintosh System 7.1—that would run on my Mac SE.

Problem Two: Reading and Writing Macintosh Disks in Windows
Although the Mac SE included the built-in "SuperDrive" to read/write both Macintosh and DOS diskettes, it didn't include the software (PC Exchange, Access PC or DOSMounter) for enabling this feature. I had no analog modem, and the Ethernet card had no driver, so a floppy disk was the only way to copy files to the Mac SE.
Fortunately, I found a piece of software for Windows, TransMac 5.4 (, that was able to read/write Macintosh HFS disks. Equally fortunate, I had purchased a Dell laptop with a floppy disk drive, so I ended up with the software and hardware necessary to get data onto my Mac SE-that is, from the Internet, to the Dell, to TransMac, to a floppy disk, to the Mac SE.

Problem Three: Connecting to the Internet
The Mac Driver Museum ( was an excellent source of drivers for Macintosh network interface cards (NICs)—but I didn't know which brand of Ethernet card was installed in my Mac SE. I used Google to figure out the brand by searching for the random text that was visible on the back of the Ethernet card. Discovering that it was an Asante card, I then was able to find and install the correct driver.

The Mac SE doesn't use the newer OpenTransport networking but instead uses the older "classic networking," which doesn't support dynamic IP addressing. However, my network uses a DHCP router (Asante FR3004LC) to dynamically assign IP addresses. I also have a router at home, and eventually plan to connect the two networks with a Virtual Private Network. As such, I've configured the routers so that my work router assigns IP addresses ending in 100-149 and my home router assigns IP addresses ending in 150-200. So I assigned my Mac SE a static IP address out of the range of IP addresses dynamically assigned by the routers.

Problem Four: Opening Stuffit Archives
I wanted to cut out the PC middleman and download software directly from the Internet to my Mac SE. But to do that, I needed to install various Internet programs (Web browsers and the like). The problem was that I was able to open some Stuffit archives on my new iMac but not on the Mac SE. For those who don't know, Stuffit files (.sit) are to the Mac what WinZIP files (.zip) are to Windows.

I did my research and learned that if I wanted to open Stuffit archives on the Mac SE, they would have to be encoded with the older Stuffit Lite 3.5 (which the SE can run), as opposed to the now-current Stuffit Deluxe 7.0.1 (which the SE can't run). That meant I had to use my new iMac to read Stuffit archives and then re-archive them with Stuffit Lite 3.5 so that they could properly expand on the Mac SE. To make sure that all of my old Stuffit archives could be unstuffed on the Mac SE, I restuffed those and used ".sit35" for their file name extension. Lastly, I learned that, because Stuffit archives (and ZIP archives, for that matter) are binary (8-bit) files, they must be sent using FTP in binary mode.

Problem Five: Finding Old Macintosh Software
Some of the original FTP shareware archives were operational but had only a smattering of shareware for older Macintoshes. (What would Orwell say?) Fortunately, much of this software has also been archived (albeit sporadically) on the Web at four chief sites:

Voilà, My Thesis Recompiled: Lessons Learned and Next Steps
Once I managed to get my Mac SE configured and on the Internet, it was actually rather easy to recompile my thesis. I had to change a couple of things owing to a few syntax changes between Think Lightspeed C 2.13 (my original compiler) and Think C 3.0. But whenever I got an error message, I simply searched Google and found the solution to the problem. So, I now have version 1.1 of my thesis. And it is very well backed up.

I keep the Mac SE in my office as a reminder of why good data backup and restore procedures are important, why they should be periodically tested and, also, why old Macs are cool.

It took me a while to restore my thesis, and a long time to write this article but, again quoting Zen and the Art of Motorcycle Maintenance, "When you want to hurry something, that means you no longer care about it and want to get on to other things." I do care about the quality of my data. And, based on this chapter and others in my computing journey, I do feel a need to move away from proprietary file formats and operating systems to standards-based ones.

In fact, earlier this year, I moved to a UNIX-like directory structure. Does that signal a move from Windows to UNIX? Stay tuned.

Erik J. Heels ( is a patent attorney in Maynard, MA.

These sites are good sources for current, and occasionally some older, Macintosh shareware: