TOP: Game theory and bluffing

I've finished reading the chapter for the 18th time now and I get excited (wow, you can win everytime by picking random cards and bluffing if they come, sign me up) everytime I read it but then come to my senses. When and how is it applicable since:
a) You should let good judgement prevail (if you have a read on someone bluffing profusely, you need not to fold 1 out of 5 times if you get 1:5 odds on seeing the last street. ie. opponent will c-bet every flop)
b) You should use it against better players (players that have an edge or unknowns)
c) You figure your opponent's hand can only beat a bluff (we need a great read on someone with don't know or who is better)

These situations do not come very often and we figure that this play will not help us very much, especially if we take notes on players or play with regular players at the tables (they are not unknown anymore). Additionally, we would be a fool to play with tougher competition unless forced (during a tournament for example) and then, we have our M and blinds, and prize equity that influence our decision. We cannot rely on Game theory alone to insure success since our money will be won not if we repeat the play X times but if we place ITM. If we bust 6 times to win 1 time 200 chips more, we have actually lost a lot of money while gaining a small edge.

Plus, since everybody sees your last card in Hold'em, your bluff isn't al well hidden (or hidden at all) from your opponents, so I very much doubt that you can gain an edge by picking random cards since:
a)It'll become painfully obvious what you are doing to anyone with half a brain, especially if it doesn't comput with the way you played the hand so far
b) That community can improve your opponents hand too (unless in Stud type games where the card is hidden and you can represent strength/weakness by the way your board shapes up during the hand.

Another serious consideration online, because the pace of play is faster than live, you don't have as much time to calculate and pick the right random cards. The examples in TOP are pretty straightforward, but let's say you're facing an unknown and have 6 outs to improve your hand (opponent has only one small pair with marginal kicker on the turn against your AK). Any Ace and any King and you're golden. Pot is 55$ and the bet is 20$. You thus bet giving your opponent 7.5:2 odds to call your bet. Since you have 6 outs, you should be bluffing for 1,6 cards to replicate the ratio. Right there, we have an approximation of huge proportion due to the odds we give and the number of outs, since we either give 6:1 or 3:2 odds, not nearly what the Optimum ratio should be. And these calculations aren't easy to make on the fly, especially since bluffing situations are usually stressfull.

Plus, the problem lies in trying to figure out wich card you'd like to be "live" bluffs: you cannot pick anything less than overcards and very often and on a board such as 72T5 (drawless or almost), these overcards could very well be the kicker for our small pair (somebody with Q2 or J7 might stay in). For our bluff to work in a community card game such as Hold'em, the river shoudl appear to have made us a hand (since it is not hidden from view): picking the 3 of spade might be good since it has a small % of chance to increase our opponents' holding, but he will probably think the same of us on such a board and call our 20$ on the river everytime if it hits.

Any thoughts on this or am I completly misunderstanding the concept (wich I very well could be since it is the meatiest, but less explained thing in the book IMHO) ?