In essence this is really a question about how winning tickets are generated during the manufacturing process.
I assume that the process is random in the broadest sense. But since purely random assignment would mean that you could get significant clusters of winning tickets, I wonder if there isn’t some logic built into the computers that do the actual assignment to help smooth out any clumpiness.
For example, with a purely random assignment, you might get several tickets in a row that win significant prizes. For this example I’ll assume any prize that is more than double the ticket price is significant. Since one of the ways they encourage people to play the instant games is by having lots of small prizes such that if you play regularly you will get some portion of your money back. If you have very long runs of tickets with no prizes, then you end up discouraging players.
So ideally you want to have a distribution of winning tickets of all amounts that doesn’t demonstrate a high level of clustering. While the people who buy into such a cluster are very happy, it necessarily means that other players are seeing long runs of losing tickets.
So it would seem that there is a strong incentive to make the distribution of winning tickets at any given point in time be random in addition to having a random distribution overall. What I’m trying to say is that if there are 10 top level prizes for a particular game, ideally, you would want to see 1 prize claimed after 1/10th of the tickets were sold, 2 after 2/10th’s, etc. Except that your goal would really be to see this kind of time dependent distribution for all prize levels to some extent so as to encourage as many players as possible for the longest amount of time.
I do understand from the point of view of the math that there is no point in trying to massage the distribution as it relates to time. It is irrelevant. But the question is not one of math. I’m sure the companies that make the tickets know about random number generators. My question is do they do any filtering to either eliminate or at least smooth out any clumping of winning or losing tickets that you would get from a purely random process.
Obviously this would have to be encoded in the manufacturing system since you couldn’t trust humans with that sort of knowledge. Further, knowledge of the code shouldn’t give you an advantage since all it really tells you is that you won’t get many consecutive winners - or at least not as many as you would under normal probabilistic circumstances.
My guess is that it is probably illegal to do anything like this, but since the lotteries are state operated and the laws are state enforced, it’s not beyond the realm of possibility that some effort is made to create ideal distributions for the state’s benefit.
If done properly, it shouldn’t be at all obvious since rolls of tickets get sold at varying rates by a variety of vendors. So even if one were to build in this time dependent randomness, wouldn’t it be difficult to detect if the only data you could use to test the hypothesis were the purchase dates of winning tickets? That I’m not sure about. Not to go off on a tangent, but it seems to me that there might be ways you could “back into” what the time-based distribution would have been.