I had the experience of working in healthcare doing data visualization for doctors. The first version of my software I took a very "engineering" approach, a simple, no-frills just-the-facts visualization of the data, defaulting to showing as much data as possible so the well-qualified doctors could decide for themselves what was important. I found very quickly that did not work well at all. Giving a "dumb" scale of graphed data obscured the fact that your blood pressure can vary 20% in a normal conversation, while a change of 10% (F) in your temperature would be generally fatal, that streams of information from redundant sensors make it hard to discern variations in unrelated ones, that a sudden spike in heart rate is likely medically interesting, while a sudden spike in blood pressure is often an artifact of the way that information is collected. In order to effectively present information even to a highly informed audience I needed to contextualize, prune, edit and carefully present that data because by not doing the system was actually creating more confusion and misleading through data overload.

The challenge I'm sure that these climate scientists have is, they see this data not as dry, boring information, but more like health information for the planet. They would like the rate of increase of CO2 to slow and give the planet (which they live on) more time to adapt to the rapid warming which is potentially disastrous. They are trying to properly frame the fact that these are all quite bad signs so we take appropriate actions. A doctor with a patient with a fever of 107º should not present the patient with a report saying "The 2 week moving-average of your temperature is looking normal", or "your temperature is up 9% today", or "while your brain temperature is trending higher, your extremities are cooling so your average temperature is staying even". Those all might be true, but would be medical malpractice.