Yes, it's very easy to fool the system when using direct
flash, one has to be very careful with composition indeed, with any near objects in the frame the the camera is stiffed with a choice of loose loose whatever it does, either the far object is exposed ok with the near one suffering the neutron bomb effect, or the near one exposed ok and the far one lost in the murk, there isn't normally any escape from the inverse square law whether using ADI or TTL.
Bounce TTL of course would be the best if there are near and far objects as usually most things are at a similar distance from the ceiling and thus stand more of a chance of being evenly lit, but that can bring additional difficulties if you need too use an f/stop that is deep enough to bring near and far objects/subjects into focus...then you start to need a fair bit of power from the flash gun.
I did notice one time though when shooting along the length of long table with guests on both sides (and backing up further away not being an option, nor was ceiling bounce) that the system can fooled by the relatively wide-ish angle of lens zoom into setting the zoom on the flash to match the lens angle of view, (the focus was about 2/3 down the length of the table), that flash zoom setting was not going to work as the nearby people were going too be over exposed and the far ones way under exposed, (a photo confirmed this), it was possible though to overcome that too a degree I later discovered, by overriding the flash zoom manually and zooming it to 85mm-ish even though the lens was at something like 20mm, this allowed the exposure system much more latitude to expose the far end with the punch of the zoom flash and the nearby subjects getting lit by the spray coming off the edges, can work ok with a little bit of experimenting, but far above the capabilities of the system to do by itself.
Also on the weekend (against my better judgement), I decided to do a little investigating of the mysterious realm of taking direct flash photos of shiny/reflective things that occupy most of the frame. My first subject was a china cabinet that had a glass front, a mirror back internally, glass shelves and crystal whatnots on the shelves. I had the edge of the cabinet door (one of them) near the edge of the VF frame so that almost the entire frame was glass and crystal and mirror, and too make things as tough as possible I used ISO200 at f8. The A100 + F56 actually performed well above my expectations either with ADI or TTL including the AEL button on some for ambi balance, (standard flash sync.) most exposures were reasonable with a couple slightly overexposed, the only exposure that was quite overexposed (maybe two stops), was when I tried HSS flash.
I then moved the tripod over in front of the old grandfather clock and did a re-run, which included adjusting the tripod height until the face could reflect squarely back into the camera, as soon as I got the camera square on I instantly got a near black frame, just the rim of the face slightly visible and the brilliant reflected spot (on TTL), nothing else visible. Thinking that ADI would fix it was wrong...it was just about the same again, maybe because by this stage with so many photos of reflective things the system had finally caved.
I did manage to get it back though just by shifting the tripod a foot to the side and taking a couple of photos zoomed back out a tad...and presto back too normal again.
I'm not sure what conclusions one can actually draw from any of this, except too say, try too avoid strong reflections with direct flash as much as possible...because the only way I overcame the clockface reflection was to revert to M mode both on the camera and flash and adjusting settings accordingly to get a decent exposure.
I was also left wondering if the flash didn't spot the "other" flash and send it a signal to stop shining in "my" face which it then picked up from the reflection too shut itself off.