If somebody could explain what that means using a database analogy, that would be nice. I don’t quite understand why adding another column to a table should lead to such drastic changes all over the place.
In any event, so are the issues here mentioned really a big headache? Are there tools (presumably not mentioned in the article) intended to help the user do this sort of changes? Or is this problem so inherently complex that new columns are never added after spreadsheet has been created?
First, most spreadsheets I have seen don’t even have any downstream tables. So this doesn’t apply at all.
Secondly, if your referencing is done properly, you can insert rows or columns into the first table, and the downstream ones will adjust automatically. You do have to insert the rows or columns one at a time, which can be tedious if you are changing the dimensions greatly.
Personally, I don’t like systems built out of many downstream, cascading spreadsheet tables – such complicated systems ought to be databases rather than spreadsheets, IMHO. These are usually designed by someone without system design skills, and in my experience, are often neither cleanly designed nor fully tested.
What they are talking about is how much effort the spreadsheet app (e.g. Excel) has to go through to update its internal model when the user does some simple UI action like add a column.
There are some spreadsheet-based models used in the finance industry that are so large they take hours to recalc on modern workstation quality PCs. It’s common that within the model some sheets are mostly data, while others are summary calcs of the data. And others are summaries or crosstabs of the summaries, etc. The dependency graph is truly ginormous. This can go on across multiple workbooks each with dozens or even hundreds of interlocking worksheets.
In situations like that, adding a row or column to one of the upstream sheets= tables requires not only recalcing almost all the values in all the downstream sheets, it also requires recalcing many of the references in many / most of the formulas in the first-order downstream sheets.
How the spreadsheet app manages all this and does so as efficiently as possible is not simple.
Does any of this matter in the slightest to somebody’s departmental vacation tracking sheet with 100 rows & 17 columns? Not even remotely.
