Fundamentally, the difference is in the timing. If one is linear and another parallel, how many linear units (bitwidth) can be processed at a time? Depends on the base being used. As of stardate sun-2010AD, I, along with all humans I know of, communicate with our computers using binary time conversion. That means our computing languages and compilers are calculated with only 1/0 (binary) to time increments (linear-infinite). Comparatively, a human mind is the plane between two values: two hemispheres.
Back to areas of attention: text attracts one hemisphere, images attract the other. The brain needs to quickly identify how to process the "splat", then make a decision on what to pay attention to: the image of the pretty lady or the
More thinking: how big is the font? can I recognize any body parts relevant to my interests? FACES is what people look at first, subconsciously. There's studies with babies as the test subjects. Go read it. Anyway, more thinking, more fatigue. The less interrupted the flow is, the more efficient, the more rewarding. Interrupt is entropy, losses are negative. Got it?
So here is Will, the conscious, constantly waking up the subconscious. Let's call her Akasha. Note projection of opposite sexes and the allusion of memory -- a modern narcissus and echo.
An efficient algorithm will take into consideration this balance: good font size, balanced image to text ratio. What the right balance is, is the subject for a subsequent post. Hint: it has to do with numbers.