If we aren't being LS1 specific then the argument boils down to, would you rather try to measure something directly or estimate it? There are lots of analogies to real world situations.

eg. would you work out your house electricity usage by looking at the rating of all your appliances, measure how long you had them all on for each day and then add them for your final result? Or would you just read the meter?

the MAF is a means of directly measuring the airflow into the engine. The main errors involved here are:
1. the MAFs accuracy which depends a lot on the flow characteristics (ie. smooth vs. turbulent)
2. an assumption based on airflow distribution to each cylinder (most code i have seen assumes equal distribution). The more symmetric your intake design is the better.

SD is a way of estimating the airflow into the engine using RPM, MAP and IAT. The errors involved in SD are:
1. the VE table assumes your engines efficiency never changes. Obviously any engine modifications change this. Also, the VE is often calibrated by trail and error using other measurement techniques such as WBO2 sensors etc. which also introduce extra error.
2. estimation of the inlet charge temperature, this is not just the measured IAT value (which is usually far from the cylinders) and can have a complex model of manifold and cylinder head heating effects. This can be a significant error depending on where and how the IAT sensor is located and the techniques used to estimate the cylinder air charge temperature.
3. Manifold pressure, assumed uniform and accurate.

By it's nature SD introduces more error into the calculation and hence why it is often used as a backup to the MAF. But, SD also allows a lot of freedom in tuning especially on wild intake setups where having all the air run thru a single point is not desired. eg. 8 throttle body setups, some twin turbo's etc.

SD is also often used when the airflow requirements of the engine are beyond the MAF's measuring capability, typically on a 5.7L LS1 this is around 8psi boost.

You can see that the MAF offers the most "set and forget" value but has some limitations in serious performance applications. SD offers the most tunability and flexibility but you need to keep an eye on it more and understand it is no panacea of exactness.

For most OEM applications that use a torque-based control model, the MAF is a key component to indicating the torque estimation calculation is giving accurate results and often if the MAF fails many other systems resort to a "default" behavior. This is very evident in the later vehicles that have TCM's fitted where basically the whole trans operation depends on a valid torque signal being transmitted from the ECM to the TCM. Unpicking the puzzle of what depends on what to make SD work the way you would expect is sometimes not trivial. In many cases the complexity of SD far outweighs the perceived gain by removing the MAF. IMO, unless the gains are clear and obvious leave the MAF on.

Hope that helps,

Chris...
__________________
HP Tuners - Performance at your fingertips...

HumpinSS did a test with SD and a MAF. He is runing a 90/90 setup and used a 85mm MAF. There was no noticable difference on the dyno between the two.