The problem is that the "patterns" do not seem to affect the long term payout, so you will still see 95% in the long term. What they do is enhance the streakiness of the games. If they played uniformly, they would be boring, streakiness adds to the excitement by providing a great winning session that encourages further play. This is a psycological trick, and has been used here in the UK since the 1970's on our "Fruit Machines" when the internal chips became good enough to support slot "cycles". Many UK fruit machines also use the "heartstopper" to make players believe they are getting many near misses. They were also found by the regulators to be "cheating", by deliberately creating a "enhanced" situation on the player's last credit to encourage them to insert another coin to play it out - this was against the rules, and they were told off - but they still did it thinking they could get away with it.
Microgaming have been found out "cheating" on both the Video Poker double game AND the older 5 reel video slots. This "cheating" does not affect the long term payout, but it DOES deceive the player as they are not playing the game they believe they are from what is displayed in front of them.
The VP Double game is predetermined, and the player has NO influence on the outcome, despite having a choice of 4 cards.
The old video slots have weighted reels, hence there will be more "near misses" than by chance on the highest wins. They behave more like three reel slots. Unlike the VP example, there IS a problem, as if players bet on less than the maximum number of lines on these weighted video slots, the long term payout IS affected, and it has been worked out that it can be depressed to 90% from 95% or so, which is NOT the case with random video slots, where playing fewer lines has no effect on the overall payout.
I suspect this goes beyond the old slots. I have seen a pattern on Cabin Fever. You can get the "sun" symbol almost every spin somewhere on that reel at times, yet at other times it hardly shows itself in hundreds of spins. I can't see how this can be the behaviour if each reel stop were equally likely, and completely independent of the last, for each new spin. I suspect the same is true of the Thunderstuck clones, but is harder to spot clearly as any three scatters triggers the bonus, not just a pair on two specific reels as with Cabin Fever & clones.
