Also note that the error bars depend on the model, and no model to date has passed the simplest verification of inputting a decade or three of historic data, running the model till last year and seeing if the result actually corresponds, without massive amounts of "correction coefficients" ( or fudge to fit the curve you want irrespective of the input info) either during or in post processing.
While you had in the 1990's "global cooling OMG we are going to be popsicles" the same data now is "Global Warming OMG we are going to cook". While the climate is forever changing ( you know, entropy, a slightly variable star providing most heat input, varying amounts of cosmic dust compressing the star's magnetic field, the cooling of the core being expressed by varying volcanic activity) there is no "perfect climate" ever, you will always have some place too hot, some place too cold, some place where it is drought, some place where there are floods, and it will never be the same from even one year to the next, let alone for decades.
I am for reducing waste, reducing energy use and recycling what we have, along with having it last for a decent period and be relatively easy to maintain and repair. Buying a new one because the colour has changed, or the new one has a feature that you never use, or because it is "new shiny flavour of this month", is not a sustainable model.