MC.Net As A Platform For Machine Learning?

One big question I have about MC.net is whether it will allow developers to maintain or pass data between different tests of an optimization.

Let me explain what I mean:

When you currently run an optimization in MultiCharts, the results of the particular test you're running will be calculated based on whatever inputs you set, including the ones being optimized, and how they inform the trading rules that buy and sell the market history (chart) being analyzed. However, there is no way for one test to *know* the results of a previous test. It's like each test occurs in a vacuum. From what I understand, this is part and parcel of how EasyLanguage/PowerLanguage is executed.

What I'm wondering is this: given the expanded programming capabilities of MC.net, is it possible for subsequent backtests to adjust their input settings based on the results of a previous backtest?

This would be helpful if you're looking to develop a methodology using an approach that incorporates aspects of machine learning, such as genetic algorithms or swarm technology. Right now, I'm not aware of any way to pass this information between different backtests. As I understand it, the fact that MultiCharts runs these tests in parallel makes it even more difficult to synchronize the passing of data between different tests.

Will MC.Net help to overcome this hurdle?

(I apologize if this question is not particularly clear - I will clarify if necessary)