Put a bunch of Venetian blinds at a spot permanently between the earth and the sun (know to orbital engineers as the first earth-sun Lagrange point, a.k.a. "L1"). These blinds would be refractive rather than reflective to minimize the stationkeeping needed to prevent being "blown" away by the pressure of sunlight. At L1 they only need to refract the light by less than one degree of angle for the light to miss the earth. Less than 2% of sunlight needs to be deflected away from earth to offset the expected global warming from "greenhouse gases," primarily carbon dioxide.
Venetian blinds, unlike this proposal, and this one, can be opened or closed to allow continued control over the amount of sunlight hitting earth. This is essential since we can't really predict with great accuracy the degrees of global warming that we will need to combat. The blinds I propose, rather than being a number of joined slats as in a normal Venetian blind, are simply a large number of separate satellites each with a refractive slat that can be manuevered to any angle with the sun (i.e. from fully "closed" vertical to fully "open" horizontal).
I propose that the blinds be placed precisely at the region where they will deflect primarily light headed for equatorial regions on earth. That will even out temperatures a bit between northern and equatorial latitudes and reduce the energy (caused by temperature differences) available for storm formation, and thus probably the intensity of storms (although the effect, like the effect of global warming itself, on storms may be negligible). But the blinds should be maneuverable enough, by using solar electric propulsion or "sailing" on sunlight pressure, to redeploy to deflect sunlight headed to extreme latitudes if polar ice sheet melting becomes a larger problem than equatorial heat.
Instead of launching these vast (or vast number of) panels from the very deep gravity well of earth or manufacturing them on the moon (which probably lacks the proper organic ingredients and even sufficient water, a crucial industrial input), we can extract the raw materials from any near-earth asteroids containing water and methane ice or, if there are none such, from Jupiter-family comets. Besides the proper raw materials being available in sufficient volume in such ice, but probably not on the moon, there are a number of other advantages to microgravity ice mining explained in that linked article. If extracting material from comets and manufacturing the blinds can be automated the entire vast project can be conducted by launching just a few dozen of today's rockets using the recently-proven solar electric upper stages; otherwise it will require manned missions similar to those recently proposed to Mars, but again using solar-electric upper stages. (Solar electric rockets are only required for the first trip; much lower cost solar-thermal ice rockets can be used in subsequent trips, again as explained in that linked article).
Of course shades don't combat other effects of excess carbon dioxide such as ocean acidification and faster and differential plant growth, but these effects alone do not justify the vast costs of reducing carbon dioxide emissions. Indeed faster plant growth is probably a major economic benefit.
My off-the-cuff estimate of cost, which is probably as good as any recent NASA estimate for any of their gigaprojects, is that the project would cost $10 billion per year over 50 years. That's for the manned version; if it could be automated it would cost far less. This is far cheaper than other space-based sunshade proposals because once initial capital costs are made and the first two or three bootstrapping cycles have been undertaken, the ongoing transport costs of material from comet to the earth-sun L1 point are extremely small.
As a side benefit it would (unlike a mission to the moon or Mars or a "base" or any similar project) develop infrastructure needed for large-scale space industry and colonization. I propose funding the project, not with a tax, but by treating the blinds like a carbon dioxide sink and auctioning extra carbon dioxide credits for a global carbon dioxide market, since the shade allows us to be far less drastic in reducing growth of carbon dioxide emissions. If instead of a market we go with carbon taxes the project could be funded from those. $10 billion per year is much lower than even the most optimistic estimates (at 1% of world GDP about $500 billion per year) of the costs of reducing carbon dioxide emissions. Furthermore, there are a variety of other markets for large-scale industry that can be exploited once the basic ice mining infrastructure is developed: space tourism and making manned missions and other transportation around the solar system vastly less expensive, for starters, as well as providing propellant, tankage, and industrial raw materials in earth orbits. As I stated in 1994:
If the output of the icemaking equipment is high, even 10% of the original mass [returned to earth orbit by rockets using water propellant extracted from the comet ice] can be orders of magnitude cheaper than launching stuff from Earth. This allows bootstrapping: the cheap ice can be used to propel more equipment out to the comets, which can return more ice to Earth orbit, etc. Today the cost of propellant in Clarke [geosynchronous] orbit, the most important commercial orbit, is fifty thousand dollars per kilogram. The first native ice mission might reduce this to a hundred dollars, and to a few cents after two or three bootstrapping cycles.The savings for transporting materials to the earth-sun L1 point, instead of launching them from earth, are even larger.