Because percentages don't tell you the whole picture, imagine you have a group of 100 people, with 2 of them being of a certain group, e.g. Linux users, also 20 of them are of a different group, e.g. Chinese speaking. In percentages that means 2% for one and 20% for the other group. If next month the 20% group increases to 33.3% and the other drops to 1.6% there are a couple of alternatives, but the simplest explanation is that 20 new people from the second group were added to the total, meaning that while the percentage decreased for the first group the total amount of people in it did not.

So, when you only have a percentage it's hard to know if the total number of people increased, decreased, or remained the same because you don't know if the total is the same. Since people don't just decide to switch to Chinese it's expected than when the amount of a language changes significantly that most likely means the total amount of people changed, and you can guess by how much, doing that calculation you can see that every time the Linux users decreased its likely the total number of users increased and the number of Linux users remained the same or even grew, just not by the same margin as the total amount of users did.