Uncovering Wang Xiaobo: The Writer Who Pioneered Chinese Software in the Early 1990s

Background

In the late 1980s and early 1990s, Wang Xiaobo, a Chinese economist and statistician, taught himself FORTRAN, assembly, and C in order to write his own statistical and text‑processing tools. The scarcity of Chinese software and the limited availability of IBM‑compatible PCs in China forced him to develop a Chinese editor and a four‑tone pinyin input method from scratch.

Key Technical Contributions

1988 December – Requested copies of IBM‑compatible statistical packages (SAS, SPSS, Statistx) and an APL implementation, highlighting the lack of Chinese‑language software on domestic machines.

1990 January – While performing statistical work for Peking University, noted that only SPSS was usable and inquired about the S language for matrix and distribution‑function calculations.

1990 May – Discussed the rapid growth of the IBM‑PC compatible market, the upcoming Intel CPU generations, and a theoretical statistical model based on multiple dummy (binary) variables forming a joint distribution on a hypersphere.

1991 February – Reported incompatibility of IBM Chinese software on certain graphics‑monitor machines and the limited presence of Macintosh computers in China.

1991 March – Compared his own B‑tree‑based phrase‑dictionary input program with the “Yan‑shi 2.0A” software, praising its four‑tone pinyin scheme.

1991 May (first entry) – Discovered that his software required a floating‑point coprocessor (8087/80287), which many domestic PCs lacked.

1991 May (second entry) – Invented a glyph‑generator‑based method for rendering Chinese characters on Western software windows, enabling Chinese text in applications such as SPSS.

1991 September – Rebuilt a C‑based Chinese input program after a floppy disk was damaged; credited a friend for teaching him C.

1992 January – Compilation Details tcc -mc -ewka:wk*.c a:wk5.obj graphics.lib The program consisted of five C source files, two header files, and one assembly module (compiled to wk5.obj ). It required three support files: yindian , cclib , and egavga.bgi . The resulting executable was about 55 KB, compact but with limited graphics update speed compared to the Yan‑shi version.

1992 July – Upgraded an aging PC/XT, added recursive features to his editor, and planned to write C code that could run in virtual‑address protected mode on a new 286 machine.

1992 September – Shifted focus entirely to custom C tools for novel writing; abandoned reliance on third‑party IBM software.

1993 March – Purchased a 286 and completed a protected‑mode editor capable of loading and editing files up to 400 KB. The system ran with 1 MB RAM but lacked a hard disk, limiting further expansion. The editor used low‑level port I/O for performance, which reduced compatibility on newer hardware.

Technical Observations

The correspondence illustrates several recurring challenges of early Chinese computing:

Hardware scarcity: many machines lacked floating‑point units, hard drives, or sufficient RAM.

Software incompatibility: Chinese language support was often missing or broken on IBM PCs.

Low‑level programming: Wang frequently mixed C with assembly, dealt with memory models (compact vs. small), and managed graphics via BGI libraries.

Data‑structure focus: He continuously refined his B‑tree phrase dictionary and explored recursive algorithms to improve editor functionality.