It has been recommended to fire injector earlier rather then later.

Firing the injectors too late can cause the fuel to wash down the cylinder wall causing excessive wear and oil contamination.

While firing the injector early will puddle fuel on the top of the valve, this does allow the fuel to absorb heat from the intake track and valve. This heat will help keep the fuel close to vaporization.

The longer it sits on top of the valve the more heat it can absorb. But this could also be very hazardous under boost, were you want the coolest charge possible.
It has been stated that fuel on the valve could cause carbon buildup.
I don’t believe this to be the case. Carbon buildup is most likely caused at the point where the exhaust valve is closing, the intake is opening and there is residual pressure in the chamber is higher then intake pressure (nearing TDC, just as the intake stroke starts).

Much like any other spring-mass system being driven by an electric current, opening of the injector is not an exact step change.

It takes a small amount of time to build up enough energy in the coil to begin to move the pintle of the injector off the seat and allow fuel to flow into the manifold.
The initial delay is known as “dead-time,” but is also followed by a period of exponential movement of the pintle until it hits the fully open position.
Likewise, closing the valve takes time as well. Once the current to the coil is removed, the pintle is pushed back to its seat by internal spring pressure as well as fuel pressure behind it. The difference between the opening delay and closing delays is called “injector offset.”

Remembering that fuel injectors are actuated by electromagnets, it is important to further understand how their performance can change. The strength of the electromagnet in the injector varies relative to voltage.

Having more voltage across the field of the coil increases the strength and allows the injector to open quicker. This in turn means that fuel begins to flow into the manifold slightly sooner if voltage is higher.

Knowing that cars almost never have constant voltage, the PCM needs to be able to adjust. A failed alternator, dead battery, or even normal cranking can send voltage to 11.5 or lower. Normal charging usually keeps voltage around 14 volts, and a failed voltage regulator can send output above 17 volts.

The bottom line here is that the same injector under these varying conditions can change its actual output by 40% or more. All modern PCMs have tables built into their software code to model this change, even if they aren’t overtly visible to the calibrator.
Various injectors exhibit different voltage compensation curves. While all injectors change relative to voltage in a similar manner, the exact offsets at a given voltage are slightly different as internal construction of the injector changes.
To best model the actual fuel delivery to the engine, it is ideal to accurately input the voltage compensation for the injector used. A quick Internet search can often yield the exact voltage offsets for most injectors.

To add more complexity, the actual flow rate changes based on pulsewidth. As the injector first opens, more fuel flows for the split-second that pressure differences are the highest.

Additionally, when the PCM commands an injector-opening event for a short duration, there is a tendency for the injector’s spring-mass system to overshoot the desired duration.
The net result is that at small pulsewidths, the injector tends to deliver fuel at a rate slightly higher than the static flow rate of the injector.

This often leads to modeling the injector with two different flow rates, one for the normal pulsewidths of cruising and power delivery and another for the shorter pulsewidths of idle and starting.

This in turn leads to the need to determine where this change, known as the “break point,” occurs. This break point is usually relatively small, on the order of 1 to 3 ms, so the effect is often only seen at idle and very low loads.
Again, entering this break point into the PCM routine allows for more accurate modeling of exactly how much fuel can be expected to enter the engine for a given commanded pulsewidth.