I wonder if there's an opportunity for a product that introduces a more powerful programming language and steps away from doing all the calculations on a grid, but that keeps the intuitiveness of Excel.
So the reason excel won as the common mans programming language is that it doesn't force you to learn everything all at once. It starts you off as just a table of sorts (useful on its own), then you can graph stuff (useful), then you learn about sum(a1:a30) (useful), and you just keep going down the rabbit hole until you've mastered everything.
The only thing that I really hate about Excel is that Procs cannot be run in parallel, otherwise so many other cool things would be possible.
last time I checked, at least.
I think Excel 2007 and further, Excel 2010 improved on that. Pivot tables, formula calculations and many other addins (not sure about Solver) use multi-threading.
In the programmability space there's also SIAG (Scheme in a Grid).
For analysis you might also look at Quantrix and SuperTable for inspiration.
The problem is that the type of apps you can build are usually pretty narrow- and if your needs fall outside of that then you are screwed.
To riff on the car mechanic comparison - dental work is expensive and not everyone has insurance that covers it - doesn't mean I should perform my own fillings with a home-kit.
I don't buy in to this answer. People obviously connect with the idea of creating content on their computer whether it's word processing, creating music, or editing photos. I think people like the idea of creating their own programs too.
The problem with general programming is that the output is too ambiguous, so it's impossible to build simple tools to express them. Lego for instance has a drag and drop tool for programming mindstorm legos but the input and output are a known quantity. With PCs it is impossible to anticipate everything that can and will be built.