I think to non-scientists, or at least people who don't work with log scales, the use of linear scales makes the graph easier to understand. While I wouldn't expect to see this in Nature, this article reads like it is intended for a general audience.

A linear scale for that graph is completely appropriate: it’s showing how one data series has flattened out and the other continues to grow at an increasing rate. I would also argue that since one of those data series is clearly not exponential (it’s more of a sideways S-shape) it would be harder to interpret what that series actually looked like on a log scale.

If you choose to plot things on a logarithmic scale, you open a whole can of worms with expecting your target audience to be educated numerically and understand the implications of a line on a log-scale (which looks far less dire than an exponential on a linear scale).

Because you need to get the attention of (1) the man in the street, and (2) politicians. And if the politicians aren't taking notice, they will get a quick wake-up call from their electorate.

If you understand exponentials log plots are the way to go. But most people need a lesson in what they mean before nature gives them a catastrophic one.

It’s reasonable if there is a logistical curve on the same diagram.

because log charts are for quitters https://xkcd.com/1162/