while we do have maf's, the ve table is referenced during open loop. It's also referenced in closed loop at lower rpm and on throttle transients b/c the maf is slow to react to the sudden change in airflow. At 4000 rpm and up the maf is the source of airflow calculation. Below this, it's a mix of the maf and ve table. It's a good idea to tune the ve so that you don't have lean spots when you hit the throttle and it will also give you better throttle response.

the VE table is used for throttle transient airflow estimation. The concept is simple.

1. you are running along at part throttle, with good vacuum in the inlet manifold, MAF is reading the amount of air that is entering the engine.
2. you suddenly go WOT and you get a big rush of air into the manifold to fill the vacuum. ie. this air enters the manifold and is measured by the MAF but it doesn't actually go into the cylinders.
3. At this point the MAF is reading far more air than the cylinders actually have entering and the result would be a rich condition, more measured air = more fuel. If you look at your MAF logs you can see this "spiking" during positive throttle transients. The opposite is true negative transients where you are snapping the throttle shut (eg. MAF stops reading airflow while the manifold builds vacuum, would result in a lean condition).
4. Now this is where the VE table steps in and says "the maf is almost right guys, but here's my take on the situation"
5. the two results are filtered together, also taking into account TPS values into quite a long predictive airflow filter routine, with varying coefficients depending what "zone" of operation the engine is in.
6. The net result is an airmass value that the PCM actually uses for the fueling and spark calculations.

If you disconnect the MAF you basically bypass all this predicitive filtering and essentially the PCM uses the airmass value that the VE table calculates at all times. You can think of this as being in transient mode all the time.

I like to think in analogies so here's something that may make it a little clearer. You can think of this whole thing like a sponge when washing your car.

Hold the dry sponge in your hand over your car and turn the hose on it. For a while there you see a constant rate of water going onto the sponge from the hose but nothing is coming off the sponge onto your car (the sponge is absorbing it). After a while as the sponge saturates you will see the same amount of water running off the sponge as you are applying with the hose.

Now think of the MAF as a flowmeter on the end of the hose, and the engine cylinders as receiving the water coming off the sponge. In this dry sponge example the flowmeter (MAF) doesn't accurately reflect the amount of water running off the sponge until it saturates (reaches steady state), and depending on the size of the sponge it may read high for quite some time.

Now imagine you turned off the hose, for a while there water still flows off the sponge as it slowly dries out. But now the flowmeter (MAF) is reading zero (cos you turned the hose off completely) but your eyes can plainly see that water is still running off the sponge.

This is exctly what happens to the MAF when you have a positive or negative throttle transient, only the sponge is now your intake manifold.

The, VE table on the other hand is a "calculation" thats tries to guess how much water is coming off the sponge at any given time - note the big difference here is "guess" rather than "measure" as is the case for the flowmeter. In our sponge case it might weigh the sponge to seehow much water is on it at the current time and look up a table that is outflow vs. mass. The point here is it measures a bunch of other things and does a calculation to guess the answer. In the VE table case it looks at MAP and RPM and looks up a human defined table that says for this MAP and RPM this is how much air is in the cylinder.

So you can see that while things are at steady state (or close to it) the flowmeter gives a very accurate indication of the water running off the sponge onto the car. But as soon as you get into a transient condition it can be wildly inaccurate.

So why bother with the flowmeter at all? Good question. The thing here is that the flowmeter is a universal device that "measures" flow very accurately in the case of water, and reasonably accurately in the case of air. You put one on an engine and you don't need to spend a whole bunch of time calibrating and refining your "guessing" calculation as most flowmeters are temperature independant devices (ie. they measure accurately at all temps). The carmaker can be assured that for the majority of cases across wide operating conditions and applications the engine is performing to the standards required. And if optionally backed up by a decent transient model the whole thing can perform even better giving accurate transient contions (again with very little calibration work). ie. transients are short and generally so long as the transient "guess" is close to the mark once steady state returns (usually very quickly) the flowmeter takes over again (with a smoothed blending in and out).

For those electrically minded replace the water with a variable current source, the sponge with a capacitor and the engine with a light bulb. ANyone who send me a VE table in volts vs. load impedance wins a prize...
and there is a new show on Nickelodeon called Spongebob that discusses this in great detail