The typical BP system only samples approximately 10% of the total flow rate of the FF. That 10% value is the average of flow. Some will see less; some more. That is an average.
The "average" particle will go around the system 9 times before being caught (if it's large enough) in the BP. Some particles will experience a bit of misfortune and immediately go to the BP on the first or second trip, but some may actually make 13 or 20 trips into the FF before ever being sent to the BP. The "sample rate" of 10% is an average; hence statistically there are 9 trips to the FF before one to the BP.
It's not unlike the topic of conversation about fluid exchanges, such as in trannies. If you have a total of 8 qrts in the system, and change 4 qrts the first time, you get 50% out. The next OCI gets out 50% of the fluid, but half of that second OCI is "new" oil and half is "old" oil, so the net result is 25% net old fluid left. Again, and again, and again. It takes a LONG time before you get a thorough exchange. But at some practical point, you reach a law of diminishing return. At some point, we even have to engage the concept of the law of entropy; things that tend towards one side of a spectrum have the ever increasing opportunity to reverse, given a truly abstract selection methodology. (Note - there is chemical entropy, mathematical entropy, thermal entropy ... and they all have a slightly different meaning but convey a similar concept).
The filtration concept is similar. The BP system only samples the total FF flow rate, and because there is no control over the exact path of each particle, it's just a law of average math application.
The upside to BP is that the pump will produces about a bazillion cycles of the total system capacity over an OCI (OK - I do exaggerate here a bit, but you get the point). So eventually it's highly likely that any particle, sized large enough to be caught, eventually will be. The larger the particle, the more like it is to be caught.
This is a conditional statement where TWO criteria must be met simultaneously to be effective.
1) the particle must sent into the BP path
2) the particle must be large enough to be caught in the BP element
If the particle is smaller than the effective BP sizing, the particle may see literally thousands of trips through the pump, and therefore 10% of those thousands (represented as hundreds) of trips to the BP. But if it ain't large enough to be caught, it's moot! If the particle is large enough to be caught by the BP, it has (on average) a 90% chance of going to the engine first, based upon sampling flow rate.
We almost have to sub-divide this topic into three discussions:
a) what happens to a particle that is too small for any filter? A near infinite ticket to ride around the engine. The upside is that such a small particle is unlikely to do much damage
b) what happens to a particle that is large enough to be caught by the BP, but not be caught by the FF? 9 trips to the engine before 1 trip to the BP, on average (some more; some less) These can do damage
c) what happens to particles large enough to be caught by the FF? A 100% chance of being stopped because 90% of the time they are stopped by the FF and 10% of the time they are stopped by the BP (anything large enough to be stopped by the FF is also large enough to be stopped by the BP ...) These would assuredly do damage, but are also assuredly going to be caught before they do so.
But the main point to understand here is that soot particles simply don't get that big that fast. Putting a BP system on a piece of equipment and then doing "normal" OCIs, isn't going to produce any significant shift in wear-rates, or assure wear reduction. The soot particles are too small to do much damage and also too small to be caught even by the BP element, because they haven't had time to overwhelm the anti-agglomerates yet. Regardless how many times they may or may not enter the BP filter, they aren't big enough to be selected with any decent efficiency rate.
As you state, I sometimes like to drive it down almost too far for easy consumption, but I'll try to summarize:
the average sample rate of 10% means just that; it's an average particle expectation. Speaking only to the concept of particles in class "b" as defined above, some will take more trips before being caught in the BP; some less. The average is 10% based upon the selected flow rate. Most BP systems portion off about 10% of the flow that the FF element sees, therefore they only have a 10% chance to catch something relative to the total system flow. For every particle that only went 5x before being sent to the BP, there was one that went well more than 10x, on average. This is the effect of random sampling and particle sizing.
BP systems are a fantastic tool when you leave the oil in the sump for a long time, especially on a large capacity system. They can reward you with a very useful fluid suitable for fiscal savings via reduced quantity of OCIs. But used in small sump capacities, and with short OCIs, BP filters are a complete and total waste of time/effort/cash. Hence my stalwart objection to the aforementioned recommendation by GDubFlue and others who think BP systems are the right answer for all things ...