The reason why they tell you to just draw what you see and not what you know is simply because they are trying to demystify drawing for people who feel like it’s some kind of supernatural skill. People who haven’t reached a certain level of competence in drawing will put that skill up on a pedestal and think you need to have some kind of special intuition in order to draw well, and they don’t understand that when dealing with depicting likeness in a realistic manner from a defined source (still life, life model, landscape, photo…etc), there’s actually a lot of visual calculation and measuring going on, and often it’s not about “artistic thinking” at all, but very mathematical and calculated. When you draw what you see, you will think along the lines of:
(These are just made-up examples, not rules.)
“The corners of the mouth ends at the center of the pupil if you draw a vertical line to measure.”
“The forehead on that profile is slanting at roughly 25 degrees”
“That spot of highlight is completely white, although the local color is actually purple”
“The distance between the inner elbow and the wrist on that girl is exactly the length of her hand.”
“The value of that cast shadow towards the edge is exactly the same as the value at the terminating spot of the object.”
When you break what you see down to “exactly” what you see, then you start to think with mathematical precision, which will lead to technically accurate likeness. It is possible to draw exactly what’s in front of you even if you never learned your foundations but have honed your ability to capture just what’s in front of you, but then you become just a human copy machine. No artistic interpretation, no creativity, just technical skill.
So I agree that the saying of “Draw what you see and not what you know” mostly apply to situations where you are doing an exact likeness of a source in front of you. While it’s great that you hone your ability to do that, you should also apply the foundation knowledge you have learned and make creative decisions, such as composition (there is no viewfinder in nature, you have to frame the imagery yourself), what details to focus on and what details to leave out, emphasis on specific colors for artistic effect, simplifying of details (you never draw every single strand of hair–you do an approximation of clumps, strands, layers…etc), deployment of creative brushwork, and so on. But before you get all Picasso on the world, you really should first become proficient at simply drawing what you see. Once you ace that, you can go and get crazy. Not that there’s anything wrong with artists who can’t draw likeness or any kind of decent representational work, and only do abstract or highly stylized work, but knowing the general interests of the cgtalk members, it’s safe to assume most have interests beyond abstract and very stylized works.
And if you want to be able to draw any kind of representational imagery out of your head at all, then you absolutely must be able to draw what you know. Learning the foundations is the most effective way to get you there.