So it would still be a bar graph. I thought your argument was that it wasn’t the right type of graph for this dataset, not that you disagreed with the dataset used in the first place. I agree that the one you proposed is easier to understood, but I don’t really agree that using it like they did is improper. It is certainly useful if you want to save space.
Let’s say 2 people used TempleOS, Symbian, MS-DOS or some other extremely weird thing to load the website in 2023 and then 42 in 2024 because it gets referenced in a porn video or something. The relative increase bar of 2000% would dwarf the changes in major OSs’ usage even though the absolute number is miniscule in comparison. Yes, they get lumped into “other” but this kind of discrepancy is already happening with the somewhat exaggerated Linux increase: Windows actually gained 3x more users than Linux did in absolute terms but the Linux bar is 3x bigger.
So it would still be a bar graph. I thought your argument was that it wasn’t the right type of graph for this dataset, not that you disagreed with the dataset used in the first place. I agree that the one you proposed is easier to understood, but I don’t really agree that using it like they did is improper. It is certainly useful if you want to save space.
Let’s say 2 people used TempleOS, Symbian, MS-DOS or some other extremely weird thing to load the website in 2023 and then 42 in 2024 because it gets referenced in a porn video or something. The relative increase bar of 2000% would dwarf the changes in major OSs’ usage even though the absolute number is miniscule in comparison. Yes, they get lumped into “other” but this kind of discrepancy is already happening with the somewhat exaggerated Linux increase: Windows actually gained 3x more users than Linux did in absolute terms but the Linux bar is 3x bigger.
Yeah, that’s a good point.