I have made a pretty big deal about knowing what you are measuring before leaping to conclusions. The first is the Telescope Jockey total solar energy available, (1-albedo)TSI/4, approximation. That is an approximation for a homogeneous spherical surface. That gets you in the ballpark but sooner or later you have to bite the bullet and deal with the in-homogenous and asymmetry of the actual surface.
Since solar is absorbed by various surfaces at various depths of the atmosphere/oceans, each absorption layer has to be considered independently. The oceans have sub-surface absorption and liquid oceans on average only exist fro 70S to 70N. That means the maximum average energy available to the ocean sub-surface is 1082 Wm-2 and the minimum available since cloud albedo average is 50% would be 379 Wm-2. The ocean minimum is greater as the global (1-albedo)TSI/4 half day average at the true surface. as a result the "average" SST from 65S to 65N is about 18.5 C degrees versus initial estimates of 16 to 17 C because those were based on sub-surface temperatures collected by buckets and marine engine cooling water intakes. The actual SST measurements are much more complex due to the surface/atmosphere interface the "skin" layer, which is what radiant physics should require. The temperature an actual infrared photon might "see".
Visible light, above the red spectrum penetrates the oceans to some depth. 10% to 22% are ranges of the percentage of visible solar short wave energy that makes it 10 meters or greater. The euphotic zone ranges to nearly 200 meters and requires approximately 0.05% of the total solar energy available. Assuming 14% of the available TSI is absorbed sub-surface, there would be 14% to 28% impact due to interaction on this energy with the true surface. That energy is already underestimated with the (1-albedo)TSI/4 approximation, so the impact would be greater. So much for the faint young sun paradox. There can't be a paradox caused by error, that is called a mistake or parlor trick.
This now greater than 14% error is naturally smoothed by absorption at various depths producing a lagged impact at the true surface.
To add insult to injury, now the temperature of the deep ocean to 2000 meters has been added to the data mix. If there is an ~94 month lag in the 0-300 meter depths, there is likely a longer lag in the 300 to 2000 meter depths. Give that there are semi-regular ocean oscillations of 21 to 62 years there is a likelihood that the current ocean heat content increase is due to conditions 21 to 62 year ago. With recovery from the little ice age, there could be 300 plus year impacts of a few hundredths to tenths of degrees.
This is way Chaos Math is often mentioned, but one thing, the source of the greatest chaos is really mentioned, the human factor. If anything in the universe can screw up a simple math problem, it is a human.