CT: Hope you're not offended at being made a "Partner in Crime!"
First of all, if we're talking about a mechanical ball machine draw, cracking "codes" is not relevant, IMO. In the case of computerized draws, saying that, "the lotteries just might have people employed to make sure the code can't be cracked..." is implying that there is a conspiracy at the Lottery which is producing a series of draws that could conceivably be deciphered and predicted by unraveling, or "cracking" some aspect of the computer program producing the draws. Just think about that for a minute...
Let's pretend you and I are the principal IT people in charge of the programs and computers that generate the winning sets in the Missouri Show Me Five. Realistically, there are very likely more than 2 software savvy people at our level, our IT manager, several levels of admin above him/her, and possibly even a security firm that monitors the software and hardware on a daily basis, but we'll pretend it's simpler. One night after work, over a beer, just for fun, we get to talking about how we might change the draw program in such a way that we would know what combos to buy. Now, given all the uncertainties caused by the testing by the admin people with their "dry runs," etc., and the passwords required at every turn, we get a little frustrated looking for ideas.
Given this scene, do you think we would end up settling on a very complex and convoluted algorithm, forcing us to reverse engineer a "system" that a normal player (like we want to appear to be) could design? I ask this because the system designers here at LP are assuming that this is what we would do. Never mind that there is a virtual "army" of data collectors out there analyzing lottery results, looking for deviations from what statistical chance predicts. One of the largest "divisions" in this army are the statisticians employed by the lotteries themselves, paid to uncover fraud. The kind of fraud that requires the installation of complex code that would continue to produce predictable combos for extended periods of time would be giving the security people extended periods of time to discover our buggy code!
OR, would we decide to work on a simple scheme that could be imbedded in the draw algorithm with a few bytes of code, like a Trojan? It could be triggered by a ruse as simple as us looking for an invalid password, a "special" "wrong" password that we slyly act on! The "action" we perform upon discovering this password could be camouflaged code that produces a combo based on, say, a static value in a specific memory location, that we are aware of. Hopefully, you get the idea and I needn't go further.
The point I hope I am making is that if there is fraud at the lottery at [computerized] draw time, it is much more likely to be a "one shot" event, rather than something requiring complex programs looking back over previous draws. And unless you are REALLY tight with me or Coin Toss, you ain't gonna' cash in!
Since most mathematicians are aware that on any given day, the draw could be anything, the fraudulent, non-persistent, inserted draw will be virtually undetectable! Where this sort of fraud is most likely to be detected is when the tickets are cashed. If we are too greedy and buy 5 identical copies of the combo we know is going to win, eyebrows will be raised!
IMHO, the fact that analysis of historical data comparing ball drop versus RNG draws has produced no statistically significant evidence of differences is support for this conjecture.