If you look at a spreadsheet, you do not see all the underlying dependencies. This is problematic, as users often have the need to see them, to fully understand the spreadsheet model. Sohon researches existing approaches for dependency tracing, but still work is still not used in practice. In practice people still use Excel, whose dependency visualization is… not really the best way:
So, why do the tools do not make it to industry? For this work, Sohon has selected the following 5 different papers proposing a visualization technique for spreadsheet dependencies.
All of them had, according to Sohon, innovative ideas and they all built a prototype of their work. So far so good. But, two of the prototypes no longer work, one has compatibility issues, and two seemed to work still (in java and C#).
Only two of the prototypes were tested, one in unrealistic setting (<400 cells) and two were not tested at all. Finally, only one of the tool was exposed to people from industry, and Sohon explains that this is probably the reason that tools aren’t making it through to industry. Also, it may be difficult to scale up the performance (spreadsheets can be huge as we saw before).