Is there anywhere transcripts I can look at? In the matter of taint I think it's an either or issue. Either taint is something to be handled or nothing is to be done at all. If it is going to be addressed it will likely fracture bitcoins into some sort of two tiered legal and illegal bitcoins. It should come to the forefront fairly quickly though in Mr Trendon Shavers case.
Developer Maxwell is a strong proponent of mixing to preserve privacy, and of the necessity of fungibility to preserve the health of bitcoin. He also has a thread titled "I taint rich" that gives some deeper insight into transactions and the current ideas of taint.
I find his arguments persuasive, but I also find myself deeply unhappy about the theft issue. Before I studied his arguments, I developed the following scenario. This is a broad brush concept, the devil & his details have not yet been subjugated.
I have tried to find a scenario where miner's make clean coins, theft is reported to the block chain, the original victim gets restitution, bad guys can't benefit from false theft reports, and most coins are clean.
A claim of stolen coins could be made by sending a payment from the stolen address to a well known address 1Stolenwhatever, another payment to the theft receipt address, and change to an address <new good address> with a public comment giving the TXid of the theft. The payment to the theft receipt address is important so that they have (legal) notice of the theft claim. A simultaneous transaction would deposit coins into the stolen address.
In the case of coins that are stolen, I would assign (a user defined) value to the stolen inputs to the theft transaction, e.g., "0"
I would then trace all the inputs to the transaction being considered starting from freshly generated bitcoins. That transaction would have a face value input, and a "tainted value" input. All outputs from that transaction would carry the weighted value of output.
Please notice that inputs from the compromised address before the theft transaction carry their full value.
Any transactions in the block with the notice are still "clean."
By choosing coins that have a large number of confirms (a possible wallet source change), it is very unlikely that stolen coins will be propagated quickly, so exposure of innocent bystanders is minimized.
Now, there is a policy decision.
I would force a miner's fee to a tainted transaction.
Inputs that are 1 transaction after the notice forfeit 10% of the taint value to the miner's fee. Thus miners have incentive to adopt this policy.
The new address receives half of this miners fee as new coins, so the miner and the victim split this insurance payment.
The taint of the outputs is re-calculated to account for the amount of restitution (not including the miner's fee).
So, in time, the victim gets restitution, miner's are paid for protecting bitcoins, the theft victim gets to report coins as stolen, and old coins are presumably clean and can be spent with confidence.
It might be necessary to reduce the amount of restitution if the victim does not make a prompt report. For example, we don't want coins reported as stolen 5 years after the theft.
This is not a polished solution, and the numerical amounts need to be adjusted. I do think it is a good starting point.