This project is too big for me to do a "pioneering" aspect. I was just hoping that Captivate would be a good work around in case the other authoring tool can't work out it's serious kinks.
Since the experts haven't ventured this far, it's not a good idea for me to do it on this one.
If anyone else has done something like this I would love to hear it.
I'm not suggesting you build the whole project just to test out whether or not Captivate can support a dozen question pools with 200-300 questions. All you need to do is take a blank project file, insert 10 questions from each question type (takes a few seconds), create 10-15 quiz pools, (another few minutes), select a group of questions and duplicate them to quickly add more questions until you have hundreds (another few minutes gone), then select and drag 10 questions at a time into each question pool until all of them are assigned. Seriously, you can do all this in less than an hour.
Then try and run your project. If your project is as important as you say, surely you can invest an hour or two into this type of testing in order to prove the concept.
I've just finished a project where I used question pools and random questions. I have 35 pools with each pool holding between 5 and 36 questions, for a total of ~380 questions.
I have the preload set to 50%, and on our corporate intranet it gets to the first slide in about 9.5 seconds.
I haven't noticed any negative effects on the learners' side, the only problem I really had was when trying to publish. If I had too many other applications open publish would fail near the end with some kind of "failed to generate project backup" type message.
hope that gives you more information to move forward.