When mixing two products of different quality levels, the resulting QL is truncated. I understand that this is to prevent abuse of the system since rounding can cause a QL increase.
However, there are some instances where this truncation is too severe.
Say I have 10 steel factories producing 10 units of steel each at QL10.
Say I also have 100 units of QL0 steel in the warehouse.
When a WU comes along, I will produce 100 units of QL10 steel to mix with 100 units of QL0 steel. I would expect to have 200 units of QL5 steel or something close to QL5. I don't.
When the WU is performed, each steel factory produces 10 units of QL10 steel and mixes the batch of 10 to the existing warehouse supply. The first batch of 10, QL10 mixes with 100, QL0 to produce 110, QL0 steel. The process is repeated and you end up with 200 units of QL0 steel. This is quite a bit different than the expected 200 units of QL5 or QL4 steel.
Here are two suggestions that would reduce the truncation of QL making the actual results closer to the expected results.
1) Record the QL in 10ths of a unit. Thus an old QL5 product would actually be stored as the integer 50. When calculating QL for display or sale, always perform an integer division by 10 to get the display QL level. If you did this with the above example the QL of the final product would be about 46, which would be displayed as "4". This is much closer to what I would expect. This could cause some transparency problems. By this I mean that there are hidden QL fractions that the user cannot see and cause results that are not expected. Say a user mixes one QL1 product with one QL0 product. With this new scheme you could get two QL1 or two QL0 products depending on the actual values stored. eg. QL1.9+QL0.9=QL2.8/2=QL1.4. This is displayed as QL1. eg QL1.0+QL0=QL1.0/2=QL0.5. This is displayed as QL0. This may be confusing.
2) Perform the WU calculations with greater precision (say floating point). Then truncate the results at the end of the WU. This would result in QL truncation of at most 1 unit compared to the example of QL5 units of truncation in the steel production example. All user initiated mixing would behave as before. Each mix can cause a QL truncation of at most 1 unit.
Either method would eliminated the multi-unit QL truncation that I currently find as counter intuitive.
Walter