Somewhere in the British Library’s Boston Spa facility in West Yorkshire, a researcher is working with gear in a temperature-controlled storage room that most people would pass by in a charity store. A tiny, rectangular device that resembles a deck of cards and is heavier than you might anticipate. It has a screen that illuminates when you press the correct button and a tiny pen that slides into its side. Windows Mobile 6 is installed on it. The calendar program launches. There are no contacts in the list. The GPS chip is searching for satellites that are not present in this structure. The researcher is interested in the firmware that powers all of this—the code that launches the device, operates its user interface, controls its radio stack, and synchronizes GPS, Wi-Fi, and Bluetooth. It’s not the hardware. the code.
For archivists, the preservation of old software is not a novel issue. Since the 1990s, when it became clear that files created on one generation of computing platforms were becoming unreadable on the next—not because the data had been destroyed, but rather because the software and hardware that could interpret it had been discarded—libraries have been considering digital obsolescence. For the preservation of born-digital content, such as text files, databases, early web pages, and video game ROMs, the British Library, the Internet Archive, and a network of research universities have established procedures and guidelines. The realization that mobile firmware should be treated equally and that there is far less time to gather it before it vanishes than first thought is more recent.
Two convergent issues are the source of the particular urgency surrounding Glofiish firmware and comparable Windows Mobile devices. First, the original firmware’s storage medium is malfunctioning. The Glofiish devices that were sold in 2006 and 2007 are getting close to or have surpassed the projected ten to twenty-year functional lifespan of flash memory under appropriate storage conditions. Bit rot, a slow, undetectable deterioration that eventually results in read errors and subsequently unreadable sectors, is affecting the internal flash that contains the ROM, the bootloader, the radio firmware, and the operating system files. A device that turned on last year might not turn on the following year, and when it does, the firmware inside of it also fails.
The informal archival architecture that the tech enthusiast community developed around Windows Mobile during its active years is the second issue. A vast collection of firmware files, custom ROMs, patch sets, and documentation was gathered between 2004 and 2010 via a network of forums, with XDA Developers being the most well-known and dozens of smaller sites coexisting. Users uploaded everything to hosting services that appeared reliable at the time, shared original firmware images, reverse-engineered parts, and created modified versions. These hosting services have undergone numerous migrations, shutdowns, or changes. 404 errors are now returned by the URLs that previously connected to a Glofiish X600 ROM image. The files are no longer there. There is still a discussion topic on them, but it doesn’t lead anywhere.
Even though it takes some time to feel intuitive when the subject is a Taiwanese smartphone from the mid-2000s that sold in small quantities, libraries are treating this content as cultural heritage rather than technological artifact when they prioritize it. A particular era in the history of mobile computing is represented by the Glofiish and similar devices: the time when PDAs had reached their pinnacle of sophistication but before the iPhone had solidified the touchscreen paradigm as the sole paradigm worth pursuing.
With a new set of presumptions—physical keyboards, stylus interfaces, multitasking operating systems, and enterprise synchronization as the main use case—these devices attempted to address the smartphone issue. Running the applications is necessary to comprehend how they operated, what their security models looked like, and how their interfaces handled multi-application administration. The firmware is necessary to run the software. The firmware is vanishing.
Instead than using passive storage, the emulation research approach links the preservation task to active scholarship. The ability to execute original software on precise hardware emulators is essential for researchers examining the development of mobile computing user experience. This calls for binary ROM images, which are the actual binary code that can be loaded into an emulator and run rather than descriptions of how the program operated. The same is true for security researchers looking into the authentication and cryptographic models used in early Windows Mobile devices, some of which are relevant to comprehending vulnerabilities in systems that have inherited their designs.
