granted credit 0.00 ???

YOU seem to be having a different problem...you seem to be returning the units but then the Project is re-issuing those same units, apparently thinking you did not crunch them properly, and giving the credits to someone else. For how long you are putting into those units, and getting no credits for them, I would stop crunching until they figure out what is wrong.

Showing a Mod where he is mistaken can get me banned but here goes anyway:
You gave me a link to click on for Tracy's workunits saying they have been return with errors. Okay I clicked and see this:
117825887 48475297 15 Feb 2009 17:02:50 UTC 1 Mar 2009 17:02:50 UTC In progress --- New --- --- ---
117531685 48337491 13 Feb 2009 18:29:35 UTC 19 Feb 2009 22:59:15 UTC Over Success Done 245,171.10 53.50 pending
117239986 48201583 11 Feb 2009 2:16:42 UTC 16 Feb 2009 4:49:15 UTC Over Success Done 216,807.60 53.70 0.00
117066736 48120693 8 Feb 2009 19:39:26 UTC 14 Feb 2009 6:19:37 UTC Over Success Done 235,145.00 54.21 0.00
116257890 47756031 1 Feb 2009 5:15:51 UTC 10 Feb 2009 18:19:46 UTC Over Success Done 245,172.50 56.24 0.00

She has 5 units on the page, 4 returned successfully and one still in progress. Where does it say there are errors in her crunching? At the bottom of the Task ID page it does say "invalid", okay but why would one look there when Einstein clearly says "success" above? Maybe it is a Boinc problem, maybe it is an Einstein problem, either way Tracy is getting no credits for crunching workunits. She is spending alot of time crunching and receiving no credits for it, hence my suggestion for her to find another project.

If I'm wrong about something, I would expect to be challenged about it. If you think you are correct, you have every right to present your reasons in a calm and dispassionate manner. This would never get you banned at this project unless the challenge contained some sort of deliberately hostile and insulting terminology that went beyond the bounds of reasonable behaviour. In your case there is nothing of that nature in your response so there is no risk of banishment.

Dagorath has correctly highlighted the key point that you seem to be missing. An outcome of "success" for a task does not mean that the answers returned are correct. If you look at the page I linked to, you will see the "explain" link immediately under the "Outcome" heading on the column where "success" was reported. If you follow that link you get to the definition of "success", which is simply,

A computer completed and reported the task successfully. This makes no claim about the validity or otherwise of the answers that were succesfully returned. That's the job of the validator and the only way to see what the validator thinks is to drill down into each TaskID and have a look at the "Validate state" heading.

If you take a look at the Server status page (link is on the home page) you can actually see how many invalid results currently exist in the database. At the moment that number is 110 which is a tiny fraction of the number of valid results also listed there. Each of those 110 invalid results will have caused the creation and sending of an extra task in order to finalise the quorum. Until that extra task is returned, the validate state will be recorded as CBNC (checked but no consensus) as mentioned in my previous response. There is one CBNC task in the list in question - the one where the "Granted Credit" column is still "pending".

It's unlikely be a BOINC problem because all the things that BOINC is responsible for (communications with the project, downloading of tasks and data, uploading of results, etc) are occurring normally.

It's possible that the science app contains a bug that is allowing incorrect results to be calculated in certain circumstances but if that were so, you would expect to see many more invalid results on the server status page.

It's possible that the validator contains a bug so that results that are really "close enough" are being reported as invalid. This is particularly possible where results are calculated on different platforms because of the different ways that different platforms handle floating point calculations. This has been a problem in the past and has led to the results of some minor platforms sometimes being adjudged as "invalid" when compared against two "Windows" results (which would tend to support each other) for example. A lot of work has gone into the validator to ensure this happens as infrequently as possible. There is no evidence that this issue has come back to haunt us at the moment.

The third possibility, probably the most likely possibility, is that there really are incorrect answers within the results returned from that computer, probably caused by some sort of hardware issue.

Wouldn't it be better to actually try to identify the real cause of the problem? Why "bad mouth" the project so vehemently as if there were no other possible explanations for the problem? You said

as if there were no doubt about the "guilt" of the project. If you're not really sure about something and haven't considered all the alternatives, you should resist the urge to make such statements.

The project Admins do look at invalid results in order to see if the invalidity is from the project itself or from something that is happening on the client. In the meantime, the best action is to see if there is anything happening at the client end to account for the problem. That's why I asked some questions in the hope that Tracy might be able to give more details about the previous history of the machine.