IME most people don't know that using Excel to open and save a csv will silently mangle data. In our application leading zeros are significant, so we constantly get screwed by people trying to do quick manual edits and breaking the data. If we're lucky it breaks so badly the import fails. It's worse when the mangling results in structurally valid but wrong data.
I think what you’re saying is accurate, but it’s also important to be practical about stuff.
These are pretty well know excel limitations by now.
And really, anyone using excel who is somehow not aware of that limitation is probably not someone yet experienced enough to be working on a larger and/or mission critical dataset to begin with.
Are there exceptions? Sure. You might be tempted to cite the example of the incident where this happened to some biologists not too long ago, but mistakes happen. I’ve seen people make mistakes building android or iPhone using the right (TM) tools.
What is the exact number of mistakes where you make the decision to jump to a new format?
I’m not sure. This does happen eventually, but the author didn’t make a strong case here imo.
