Plenty of race cars use a locked differential. Not only is it consistant, it doesnt wear out like a clutched diff will. A lot of formula d cars use a spool in the diff.
I think that (wear and reliability) is an important factor to consider. I see your point. I imagine that a spool differential is the most dramatic to drift with, and contributes greatly to the eruptions of smoke and consistency of traction (or lack thereof) in a D1 car. While I don't think it serves the purposes here, it can be useful to achieve the effects seen in D1 easier, especially when they drift in low roll scenarios where this technique would not work. (I explain more on that later on in the post)
I think the merit in an Open diff lies in the proportion, instantaneously, of torque being spent on the lower-traction (inside) tire VS the outside. In an open differential car, upon corner entry applied throttle causes the outside to push the entire car into oversteer much easier as the inside wheel does not create understeer from also having significant power. During this phase, the difference is significant because of the roll of the car forward during braking. A locked differential would not gain that benefit. The rear inside wheel, as it loses traction from the rotation of the car during oversteer, loses traction as well during low grip scenarios, so that when the outside wheel is overwhelmed with power, the inside wheel is as well, though to a lesser extent than a locked differential. This, coupled with the fact that an open differential does not mean that the inside wheel gets
no power (in fact, it gets plenty) but rather gets power based on the roll of a car and how much weight is on that particular tire, makes it especially useful for low grip drifting, such as using Comfort Hard tires like many cars do.
As a result, to reap the speedy entry benefits of an open differential, it requires:
1. Low weight transfer (from a low center of mass, low eight, low tire G rating and narrow wheelbase)
2. Consistent and even torque curve.
This prefers lower power cars, as a curve that peaks at 276 ft/lbs but has a minimum of only 20-30 ft/lbs less over a 3000 rpm band is going to be a greater advantage than a car which has a torque peak that is higher, but with proportionally lower minimums. To narrow the percentage of minimum/maximum torque, the band has to be lowered, which requires an even higher amount of torque to arrive at the rear wheels. Eventually, the added power is either a detriment to throttle control, or cannot be achieved without the use of parts that ruin the shape of the curve.
To put it simply, in an open differential you are not giving an even distribution of power to the rear wheels, obviously. But to drift optimally, you do not
want an even distribution of power to the rear wheels. I find that cornering speeds are an average of 2-3 mph higher, grip is improved, and throttle modulation response is widened (making the car easier to drive, as a wider input on the throttle mid-corner has a smoother impact on angle/speed) at the same angle. This is easy to understand since some of the power is being bled off,
you can determine the instantaneous right/left split of torque in the rear wheels by how you use the throttle and roll of the car. So long as car's weight transfer does not cause the difference to become so great as to lift a wheel, it is an all-around improvement in speed and handling.
To summarize, the intention is to use the weight transfer of the car, along with the throttle response, to determine a more beneficial left-right ratio, different for each corner, for a variety of drifts, rather than rely on a constant split as determined by an LSD to be applied to all corners.
I do, however, see where this technique would fail. Using an open differential in a car that produces significant inside lift forces (My 240ZG does not, the unsprung weight is considerably more than the small lift forces on the inside wheels during cornering) might show how this technique is probably less useful in those scenarios.
On an ending note, I would not make this thread if using a locked differential was so universally better than an open differential that the reasons to explain why took only a sentence. I only made this thread after a few weeks of my own testing between the two and wanted to expand what I found to other drifters to experiment with.