For fans of the spreadsheet (@Duncan Cragg et al): > In essence, a spreadsheet is immensely powerful because it has three properties: > > 1. All of its data is contained in cells, each of them with an address. > 2. A cell can reference another cell. > 3. When a cell changes, all of the cells depending on it change as well. I wrote the above just now. Do you think I'm on the money here, or would you summarize the spreadsheet's power with different points?

You might be interested in Jonathan’s keynote about this, which focuses more on how sprradsheets augment data with computation (your first point) and less about the programming model (2 & 3) https://m.youtube.com/watch?v=4GOeYylCMJI&pp=ygUUQWNtIGxpdmUgc3BsYXNoIDIwMjQ%3D

Thanks for sharing @Christopher Shank!

I think of spreadsheets in terms of applications as follows: 1. A spreadsheet application is defined as cells and their attributes (type, value, alignment, etc.). 2. Cell values can be defined using formulas that refer to other cell values. 3. A spreadsheet application "runs" when cell values change and the changes propagate to cells whose values depend on the changed cell values.

Dan, I like your summary. I think each of my points maps to yours. However, your point 2 nails it better, because the value of a cell can have a reference to another cell, but also have things that are not references (constants and function calls).

Observable notebooks arguably fulfill properties 1-3. If you believe they meet the “immense power” of a spreadsheet, then you might have a nice formula, if not, then you’re missing something.

Good point, Joshua! And indeed notebooks are very dynamic. In practice, I felt them to work too "pipeline-y", in that the granularity was too big (a 10-20 line script to generate the next cell) and basically worked as a multi-pass data pipeline (with great feedback, though). And the code itself is not expressed in a self-similar way; that is, Javascript (or Python) is itself not reactive and not decomposable into cells.