Some data IS manufactured, of course, but a major component of the scientific method is replication, and we have data points over time from institutions that hate each other yet agreeing that CO2 is up and so are temperatures. Unevenly distributed, but of course...

Problem = 90% real
Solutions = 90% fake