Except that's not how this works. What yella is engaging in is trying to fit the evidence to his conclusion rather than the other way around, thus coming up with a convoluted theory about shard coding with no basis
Expected value =/ actual results, that is what this boils down to, combined with a woefully small sample size, ignoring the existence of outliers (see above post for outlier in other direction) and a laundry list of biases
will pass along the request for a pity system but I think that would see the drop rates of prisms get changed quite a bit
If I include free shards, those from events, and those from my ftp accounts, it is 118 pulls for 3 legos, or 2.5%. There can be fluctuations but with a larger sample size they should average out to the nominal rate, and that is not happening.
Not only that, but previous experiences with pool events typically require considerably more pulls than the summon chance would suggest, so the total is actually a lot larger than 118. In those earlier events I did fewer pulls, so I put that down to bad luck, but now I am not so sure. Having bad luck every time seems suspicious to me, but the observed numbers are consistent with a 3% chance.
In any event, they need to double check that they do not have a bug in the event description.
I am providing a significant data set, it is not just a handfull of numbers. I don't think it is appropriate to kneejerk reject my observations just because.