double spend ATTEMPTS have happened. but show me the actual tx's of a successful double spend (injection of fresh coins that were not mined)

That's not what double-spending is.

Successful double spends happen, usually against people who accept 0- or 1-confirmation transactions in situations where they have no recourse. The solution is not to do that.

Last time we polled the community to determine the accepted definition of "successful double spend," the results were as follows:

Given that BTC is still so new, I think the network has done a very good job preventing double spends.   No new system is ever going to be perfect, but the one we have going so far has worked far better than anyone probably could have initially imagined...

You're still ignoring what the community defines as double spending just to twist this around and give your argument more credibility. When your comment isn't what I'm talking about and you know it.

Imagine that an observer creates two transactions, A and B, that both spend the same outputs. He structures A so that it sends a few bits to an on-chain miner because he knows Eligius won't accept it into mempool.  He broadcasts A to the network at t=0 s according to his clock and then broadcasts B at t=10 s.  Eligius finds the next block and B gets mined into the blockchain instead of A.  

For the observer to deny that A was actually sent first is to deny objective reality.  Transaction A was sent first!  The fact that the rest of the world has no way to prove this fact does not make it false.  

There is no such thing as objective reality in a distributed system (unless one is God).  There are two problems here, first physical reality itself is not totally ordered, and second, each person has access to only his individual sense perceptions and measurements unless he trusts others.  In short, your "objective" reality is not mine, especially if I don't happen to trust you.  (From recent forum posts, it would seem prudent not to trust a large number of posters, so please don't take my remark personally.)

It is logically impossible to achieve reliable consensus in a distributed asynchronous system.  This was proven decades ago by Nancy Lynch et. al., and the proof didn't even require malicious nodes, just nodes that could fail or run slowly. Bitcoin doesn't circumvent this proof.  That's because Bitcoin does not ever give 100% confirmation that a transaction was committed.  Bitcoin does not and can not do what is logically impossible. One can circumvent this situation by positing the existence of a reliable central authority who controls the network and serves as "the decider", but then everyone will have to trust this "decider". (And when this "decider" goes "out to lunch" one will still have to wait to see if one's transaction was confirmed.)   This is not what bitcoin is about.

One can make a case, possibly a good case, that the bitcoin protocol can be improved to minimize the time constants involved, which are presently nowhere near the underlying physical limits because each node does extensive computation prior to forwarding transactions and blocks.  But this is not going to change the underlying situation, which is that there simply is no such thing as "objective reality" just a ceaseless flow of events that are relative to time, place and person.

There are double spends from time to time.

You can find some here. https://blockchain.info/double-spends

They usually come from orphaned blocks though. What I understand as a doublespend is:

Transaction A spends coins from A₁ to B₁ and gets confirmed at least once. So A is in one block and there have been 0 or more blocks after this block.
Now if this block or the chain of blocks after it get orphaned and the identical transaction A is in another block, thats no double spend, because the coins have been send to the very samd B¹ it just took them longer.
A successfull doublespend would be a new transaction E which sends coins from A₁ to E₁.

Why is this a double spend?

Because whoever made A could have gotten some good from whoever controls B₁. The owner of B₁ has no coins and no goods, thus the coins can be spend again (twice, thus the word doublespend). With the transaction E the attacker makes sure that the old transaction A can not be confirmed because the inputs have been used otherwise, thus A is no longer a valid TX.

no one can show a true double spend where a single input has 2 outputs which both have confirmations.

the double spend as people call it is not a bug on the protocol. the protocol has no double spend bugs. the bug is human error.. lazyness to not do checks and not wait for confirmations.

bitcoin is a thing of beauty that truly works, yet humans choose to ignore the rules of bitcoin, humans choose to aimlessly accept unconfirmed payments, humans choose to send out payments twice because someone complains they never received the funds.

you cant fix a human no matter what code you put in its way.

even if new code was added to somehow prevent the relaying of a tx input twice.. humans would find a way to be stupid and lose funds by ignoring certain aspects of bitcoins function

P2P pools are not as easy as joining ghash.

More pools like ghash should be created, until that happens ghash will continue to be the facebook of mining pools.