That's a good point. On the other hand, the SST measurements are a lot lower impact (in terms of the number of times they're used in graphs that get into your newspaper) than air temperatures.

Which may be not be logical. But humans aren't logical.

For that matter, increases in air temperature in the 20th century by no means constitute definitive evidence of anthropogenic warming, since the planet has been warming for about four centuries. Finding the anthropogenic "signal" is a matter of number-massaging statistical wizardry, not at all unlike general circulation modeling.

If we had a sample of about 20 independent planets we could experiment on, it'd be no sweat. But science is hard. There's no law of nature that says the definitive experiment actually has to be practical. The definition of science is "we know," not "this is the best we can do."