Better to create an open ecosystem that people adopt for that very reason - that it is open. In the early days, Excel would actually obfuscate its output, making it extremely difficult to parse. Now things are a bit better, but you're still trying to hit a moving target.
If we have an open format, and a corresponding high-quality, open application ecosystem, I can see many corporates making the switch. Not easy, but I can see a future without total dependence on Excel.
Excel never obfuscated their save code. I've seen it in person. They did just memcpy their data structures, which was the most efficient way to save on tiny machines of the day. Lotus did that too.
I've also seen and helped implement the Open Document spec. This is not a simple problem and you are doing yourself a disservice by blinding yourself to the scale of the task.
If you want to make an impact and don't have Google level budgets then leveraging OpenOffice is the only sane approach.
We don't have Google level budgets, but I think we can do something worthwhile with what we have. Might take a while, but I'm confident we will manage something good in the budget we are hoping to allocate to the task.
However the code is literally a memcpy of a struct that is used internally. Not being designed for others to easily read isn't obfuscation. FWIW the XML formats designed for that purpose still fail horribly on that front.
Anyways I've worked on spreadsheets for 10 years. File compatibility is the hard part and the rest is easy by comparison (but still pretty hard).
A grid of cells is the easy part. Compatibility is hard.