I agree that a clarification and hopefully a more consistent experience can make this easier for a novice.
One tip is that while writing a formula, the datatype is clearly displayed as you are typing. The coloured variables have symbols next to them that display this information (see this post). Another tip is that after you have written a formula (or just selected a column type), and you are experimenting with your result, you can tell the difference by hovering over the element. The references will show a pop-up with the actual data on a small card. Simple “values” will not.
Finally, I just want to make something a bit more clear you might be confusing: Your use of the words “pointer / reference / link” are all correct conceptualisations of what is happening, but “cascade” is not. In databases, you can have “triggers” that cause actions to “cascade” across datasets to update values; however, in Coda, when you change a value in one place that has references elsewhere, there is no cascading, the values simply are different everywhere all at once. When you reference things, the data is only stored in one place, and it is just viewed in other places. For this reason, there is no cascading or changes propagating, rather, simple the same data with different views.
I think we have got their attention with this post (as well as many others across the forums) that this is something that needs to be addressed. I’m looking forward to the solution they propose and implement. Thanks for the great discussion